Streams/streams
Streams/streams
1
0
Fork 0
mirror of https://codeberg.org/streams/streams.git synced 2024-09-20 03:55:18 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
streams/include/items.php

4744 lines
146 KiB
PHP
Raw Normal View History

start formatting for Doxygen
2013-02-26 01:09:40 +00:00
<?php /** @file */
feed generator
2010-07-18 13:02:55 +00:00
move encryption functions to crypto file
2011-08-20 11:53:11 +00:00
require_once('include/bbcode.php');
require_once('include/oembed.php');
require_once('include/crypto.php');
photo driver abstraction
2013-04-26 03:01:24 +00:00
require_once('include/photo/photo_driver.php');
this should make tag delivery (e.g. forums) work - but not yet fully tested
2013-01-14 02:35:12 +00:00
require_once('include/permissions.php');
feed generator
2010-07-18 13:02:55 +00:00
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
a bit more cleanup and a bug fix or two for the new private post security model
2014-08-08 00:14:11 +00:00
function collect_recipients($item,&$private_envelope) {
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
a bit of e_all cleanup
2012-12-08 22:18:02 +00:00
require_once('include/group.php');
a bit more cleanup and a bug fix or two for the new private post security model
2014-08-08 00:14:11 +00:00
$private_envelope = ((intval($item['item_private'])) ? true : false);
try to reduce the number of simulaneous deliveries of the same post when dealing with owner relays that have more than one channel instance. If things melt down in the next few hours and I'm not available please revert this. I've reviewed a few times and think it's OK, but this part of the delivery code is traditionally touchy.
2014-01-23 06:07:47 +00:00
$recipients = array();
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
if($item['allow_cid'] || $item['allow_gid'] || $item['deny_cid'] || $item['deny_gid']) {
try to reduce the number of simulaneous deliveries of the same post when dealing with owner relays that have more than one channel instance. If things melt down in the next few hours and I'm not available please revert this. I've reviewed a few times and think it's OK, but this part of the delivery code is traditionally touchy.
2014-01-23 06:07:47 +00:00
// it is private
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
$allow_people = expand_acl($item['allow_cid']);
insecure network filter
2014-08-31 00:03:26 +00:00
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
$allow_groups = expand_groups(expand_acl($item['allow_gid']));
insecure network filter
2014-08-31 00:03:26 +00:00
$allow_groups = filter_insecure($item['uid'],$allow_groups);
bug #150, if deny access is present but nobody has been allowed, create an allow list containing everybody in your address book. Then subtract those who are denied.
2013-09-23 03:55:55 +00:00
$recipients = array_unique(array_merge($allow_people,$allow_groups));
// if you specifically deny somebody but haven't allowed anybody, we'll allow everybody in your
// address book minus the denied connections. The post is still private and can't be seen publicly
// as that would allow the denied person to see the post by logging out.
if((! $item['allow_cid']) && (! $item['allow_gid'])) {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("select * from abook where abook_channel = %d and not (abook_flags & %d)>0 ",
bug #150, if deny access is present but nobody has been allowed, create an allow list containing everybody in your address book. Then subtract those who are denied.
2013-09-23 03:55:55 +00:00
intval($item['uid']),
minor optimisation
2014-07-02 04:24:43 +00:00
intval(ABOOK_FLAG_SELF|ABOOK_FLAG_PENDING|ABOOK_FLAG_ARCHIVED)
bug #150, if deny access is present but nobody has been allowed, create an allow list containing everybody in your address book. Then subtract those who are denied.
2013-09-23 03:55:55 +00:00
);
if($r) {
foreach($r as $rr) {
$recipients[] = $rr['abook_xchan'];
}
}
}
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
$deny_people = expand_acl($item['deny_cid']);
$deny_groups = expand_groups(expand_acl($item['deny_gid']));
$deny = array_unique(array_merge($deny_people,$deny_groups));
insecure network filter
2014-08-31 00:03:26 +00:00
// Don't deny anybody if nobody was allowed (e.g. they were all filtered out)
// That would lead to array_diff doing the wrong thing.
// This will result in a private post that won't be delivered to anybody.
if($recipients && $deny)
$recipients = array_diff($recipients,$deny);
a bit more cleanup and a bug fix or two for the new private post security model
2014-08-08 00:14:11 +00:00
$private_envelope = true;
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
}
else {
a bit more cleanup and a bug fix or two for the new private post security model
2014-08-08 00:14:11 +00:00
// if the post is marked private but there are no recipients and public_policy/scope = self,
// only add the author and owner as recipients. The ACL for the post may live on the hub of
// a different clone. We need to get the post to that hub.
// The post may be private by virtue of not being visible to anybody on the internet,
// but there are no envelope recipients, so set this to false. Delivery is controlled
// by the directives in $item['public_policy'].
$private_envelope = false;
place holder
2014-11-18 02:29:57 +00:00
require_once('include/identity.php');
$sys = get_sys_channel();
a bit more cleanup and a bug fix or two for the new private post security model
2014-08-08 00:14:11 +00:00
if(array_key_exists('public_policy',$item) && $item['public_policy'] !== 'self') {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("select abook_xchan, xchan_network from abook left join xchan on abook_xchan = xchan_hash where abook_channel = %d and not (abook_flags & %d)>0 ",
try to reduce the number of simulaneous deliveries of the same post when dealing with owner relays that have more than one channel instance. If things melt down in the next few hours and I'm not available please revert this. I've reviewed a few times and think it's OK, but this part of the delivery code is traditionally touchy.
2014-01-23 06:07:47 +00:00
intval($item['uid']),
minor optimisation
2014-07-02 04:24:43 +00:00
intval(ABOOK_FLAG_SELF|ABOOK_FLAG_PENDING|ABOOK_FLAG_ARCHIVED)
try to reduce the number of simulaneous deliveries of the same post when dealing with owner relays that have more than one channel instance. If things melt down in the next few hours and I'm not available please revert this. I've reviewed a few times and think it's OK, but this part of the delivery code is traditionally touchy.
2014-01-23 06:07:47 +00:00
);
if($r) {
privacy issue - restrictive stream permission setting with a non-targetted post cannot be enforced on remote networks. Restrict these posts to zot network.
2014-09-15 00:07:39 +00:00
// filter out restrictive public_policy settings from remote networks
// which don't have this concept and will treat them as public.
$policy = substr($item['public_policy'],0,3);
try to reduce the number of simulaneous deliveries of the same post when dealing with owner relays that have more than one channel instance. If things melt down in the next few hours and I'm not available please revert this. I've reviewed a few times and think it's OK, but this part of the delivery code is traditionally touchy.
2014-01-23 06:07:47 +00:00
foreach($r as $rr) {
privacy issue - restrictive stream permission setting with a non-targetted post cannot be enforced on remote networks. Restrict these posts to zot network.
2014-09-15 00:07:39 +00:00
switch($policy) {
case 'net':
case 'aut':
case 'sit':
case 'any':
case 'con':
if($rr['xchan_network'] != 'zot')
break;
case 'pub':
case '':
default:
$recipients[] = $rr['abook_xchan'];
break;
}
try to reduce the number of simulaneous deliveries of the same post when dealing with owner relays that have more than one channel instance. If things melt down in the next few hours and I'm not available please revert this. I've reviewed a few times and think it's OK, but this part of the delivery code is traditionally touchy.
2014-01-23 06:07:47 +00:00
}
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
}
place holder
2014-11-18 02:29:57 +00:00
// we probably want to check that discovery channel delivery is allowed before uncommenting this.
// if($policy === 'pub')
// $recipients[] = $sys['xchan_hash'];
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
}
}
this should make tag delivery (e.g. forums) work - but not yet fully tested
2013-01-14 02:35:12 +00:00
private forum issues
2014-10-21 23:33:35 +00:00
this should make tag delivery (e.g. forums) work - but not yet fully tested
2013-01-14 02:35:12 +00:00
// This is a somewhat expensive operation but important.
// Don't send this item to anybody who isn't allowed to see it
$recipients = check_list_permissions($item['uid'],$recipients,'view_stream');
doco
2014-03-31 23:29:46 +00:00
// remove any upstream recipients from our list.
// If it is ourself we'll add it back in a second.
// This should prevent complex delivery chains from getting overly complex by not
// sending to anybody who is on our list of those who sent it to us.
finish up the source route storage so we can prevent messages from flowing upstream in complicated delivery chains
2014-03-31 23:25:05 +00:00
if($item['route']) {
$route = explode(',',$item['route']);
source routing stuff
2014-03-30 04:53:50 +00:00
if(count($route)) {
$route = array_unique($route);
$recipients = array_diff($recipients,$route);
}
}
Add self to recipient lists. Sure we already have a copy but it's possible our nomadic channel clones don't.
2013-07-14 08:12:05 +00:00
// add ourself just in case we have nomadic clones that need to get a copy.
$recipients[] = $item['author_xchan'];
if($item['owner_xchan'] != $item['author_xchan'])
$recipients[] = $item['owner_xchan'];
more diaspora stuff - most of the basic bits are there except queueing and physical delivery (I'm keeping delivery turned off until some of the bugs are fixed so we don't cause inifinite loops or network meltdowns) - now it's just a matter of going through and methodically finding all the bugs
2014-08-26 05:43:44 +00:00
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
return $recipients;
this should make tag delivery (e.g. forums) work - but not yet fully tested
2013-01-14 02:35:12 +00:00
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
}
insecure network filter
2014-08-31 00:03:26 +00:00
/**
* If channel is configured to filter insecure members of privacy groups
* (those whose networks leak privacy via email notifications or other criteria)
* remove them from any privacy groups (collections) that were included in a post.
* They can still be addressed individually.
* Networks may need to be added or removed from this list as circumstances change.
doc change - read it.
2014-08-31 05:53:21 +00:00
*
* Update: this may need to be the default, which will force people to opt-in to sending stuff
* privately to insecure platforms.
insecure network filter
2014-08-31 00:03:26 +00:00
*/
function filter_insecure($channel_id,$arr) {
$insecure_nets = " and not xchan_network in ('diaspora', 'friendica-over-diaspora') ";
$ret = array();
if((! intval(get_config($channel_id,'system','filter_insecure_collections'))) || (! $arr))
return $arr;
$str = '';
foreach($arr as $rr) {
if(strlen($str))
$str .= ',';
$str .= "'" . dbesc($rr) . "'";
}
$r = q("select xchan_hash from xchan where xchan_hash in ($str) $insecure_nets ");
if($r) {
foreach($r as $rr) {
$ret[] = $rr['xchan_hash'];
}
}
return $ret;
}
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
function comments_are_now_closed($item) {
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
if($item['comments_closed'] !== NULL_DATE) {
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$d = datetime_convert();
if($d > $item['comments_closed'])
return true;
}
return false;
}
fix can_comment_on_post when viewing wall-to-wall
2013-09-28 11:41:12 +00:00
/**
* @function can_comment_on_post($observer_xchan,$item);
*
* This function examines the comment_policy attached to an item and decides if the current observer has
* sufficient privileges to comment. This will normally be called on a remote site where perm_is_allowed()
* will not be suitable because the post owner does not have a local channel_id.
* Generally we should look at the item - in particular the author['book_flags'] and see if ABOOK_FLAG_SELF is set.
* If it is, you should be able to use perm_is_allowed( ... 'post_comments'), and if it isn't you need to call
* can_comment_on_post()
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
* We also check the comments_closed date/time on the item if this is set.
close comment
2013-09-28 12:03:58 +00:00
*/
ability to unset a layout and subthread is probably close to working again.
2014-04-24 02:05:48 +00:00
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
function can_comment_on_post($observer_xchan,$item) {
fix can_comment_on_post when viewing wall-to-wall
2013-09-28 11:41:12 +00:00
// logger('can_comment_on_post: comment_policy: ' . $item['comment_policy'], LOGGER_DEBUG);
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
if(! $observer_xchan)
return false;
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
if($item['comment_policy'] === 'none')
return false;
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
if(comments_are_now_closed($item))
return false;
hopefully this will go over a bit better - check post owner permissions as well as local owner permissions for comments during submission, since the comment was being blindly accepted on the local system due to only checking the profile owner. Also change collections query to include only top-level posts by a member of the collection.
2013-09-21 00:58:11 +00:00
if($observer_xchan === $item['author_xchan'] || $observer_xchan === $item['owner_xchan'])
return true;
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
switch($item['comment_policy']) {
case 'self':
if($observer_xchan === $item['author_xchan'] || $observer_xchan === $item['owner_xchan'])
return true;
break;
case 'public':
fix can_comment_on_post when viewing wall-to-wall
2013-09-28 11:41:12 +00:00
// We don't allow public comments yet, until a policy
// for dealing with anonymous comments is in place with
// a means to moderate comments. Until that time, return
// false.
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
return false;
break;
add "any connections" to can_comment_on_post() - was missing
2014-08-23 22:22:26 +00:00
case 'any connections':
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
case 'contacts':
Provide comment box for PERMS_AUTHED items. Presently only commentable from owners wall.
2014-10-23 20:50:53 +00:00
case 'authenticated':
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
case '':
fix contact comment permissions
2013-09-21 02:14:57 +00:00
if(array_key_exists('owner',$item)) {
if(($item['owner']['abook_xchan']) && ($item['owner']['abook_their_perms'] & PERMS_W_COMMENT))
return true;
}
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
break;
default:
break;
}
typos
2013-06-17 06:07:04 +00:00
if(strstr($item['comment_policy'],'network:') && strstr($item['comment_policy'],'red'))
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
return true;
typos
2013-06-17 06:07:04 +00:00
if(strstr($item['comment_policy'],'site:') && strstr($item['comment_policy'],get_app()->get_hostname()))
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
return true;
return false;
}
ability to unset a layout and subthread is probably close to working again.
2014-04-24 02:05:48 +00:00
/**
* @function add_source_route($iid,$hash)
* Adds $hash to the item source route specified by $iid
* @param integer $iid
* item['id'] of target item
* @param string $hash
* xchan_hash of the channel that sent the item
* Modifies item pointed to by $iid
*
* $item['route'] contains a comma-separated list of xchans that sent the current message,
* somewhat analogous to the * Received: header line in email. We can use this to perform
* loop detection and to avoid sending a particular item to any "upstream" sender (they
* already have a copy because they sent it to us).
*
*/
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
source routing stuff
2014-03-30 04:53:50 +00:00
function add_source_route($iid,$hash) {
wrong var name in add_source_route
2014-04-01 00:21:16 +00:00
// logger('add_source_route ' . $iid . ' ' . $hash, LOGGER_DEBUG);
crank up logging
2014-04-01 00:14:56 +00:00
wrong var name in add_source_route
2014-04-01 00:21:16 +00:00
if((! $iid) || (! $hash))
source routing stuff
2014-03-30 04:53:50 +00:00
return;
finish up the source route storage so we can prevent messages from flowing upstream in complicated delivery chains
2014-03-31 23:25:05 +00:00
$r = q("select route from item where id = %d limit 1",
source routing stuff
2014-03-30 04:53:50 +00:00
intval($iid)
);
finish up the source route storage so we can prevent messages from flowing upstream in complicated delivery chains
2014-03-31 23:25:05 +00:00
if($r) {
$new_route = (($r[0]['route']) ? $r[0]['route'] . ',' : '') . $hash;
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
q("update item set route = '%s' where id = %d",
finish up the source route storage so we can prevent messages from flowing upstream in complicated delivery chains
2014-03-31 23:25:05 +00:00
(dbesc($new_route)),
source routing stuff
2014-03-30 04:53:50 +00:00
intval($iid)
);
}
}
turn naked links in posts into zrl's if we've got a hubloc for the site
2013-04-19 00:55:35 +00:00
/**
* @function red_zrl_callback
* preg_match function when fixing 'naked' links in mod item.php
* Check if we've got a hubloc for the site and use a zrl if we do, a url if we don't.
strip hard-wired zids from posted links as they will have the wrong identity when somebody tries to view the link
2014-02-16 22:13:26 +00:00
* Remove any existing zid= param which may have been pasted by mistake - and will have
* the author's credentials. zid's are dynamic and can't really be passed around like
* that.
turn naked links in posts into zrl's if we've got a hubloc for the site
2013-04-19 00:55:35 +00:00
*/
function red_zrl_callback($matches) {
remove some code duplication
2014-07-14 08:38:16 +00:00
require_once('include/hubloc.php');
$zrl = is_matrix_url($matches[2]);
strip hard-wired zids from posted links as they will have the wrong identity when somebody tries to view the link
2014-02-16 22:13:26 +00:00
$t = strip_zids($matches[2]);
if($t !== $matches[2]) {
$zrl = true;
$matches[2] = $t;
}
that's better
2014-02-08 20:08:07 +00:00
if($matches[1] === '#^')
$matches[1] = '';
turn naked links in posts into zrl's if we've got a hubloc for the site
2013-04-19 00:55:35 +00:00
if($zrl)
add bookmark tag to naked links if they're red sites. We can do this because red links are linkified when the post is submitted. Can't do this for other naked links because they only get linkified during display and won't have a taxonomy object attached to the message.
2014-02-06 21:54:39 +00:00
return $matches[1] . '#^[zrl=' . $matches[2] . ']' . $matches[2] . '[/zrl]';
make all naked links bookmark-able
2014-02-07 00:22:40 +00:00
return $matches[1] . '#^[url=' . $matches[2] . ']' . $matches[2] . '[/url]';
turn naked links in posts into zrl's if we've got a hubloc for the site
2013-04-19 00:55:35 +00:00
}
fix wall photos
2014-02-09 08:30:43 +00:00
// If we've got a url or zrl tag with a naked url somewhere in the link text,
// escape it with quotes unless the naked url is a linked photo.
function red_escape_zrl_callback($matches) {
fix the "every other link gets messed up" problem
2014-02-09 21:46:40 +00:00
// Uncertain why the url/zrl forms weren't picked up by the non-greedy regex.
if((strpos($matches[3],'zmg') !== false) || (strpos($matches[3],'img') !== false) || (strpos($matches[3],'zrl') !== false) || (strpos($matches[3],'url') !== false))
fix wall photos
2014-02-09 08:30:43 +00:00
return $matches[0];
return '[' . $matches[1] . 'rl' . $matches[2] . ']' . $matches[3] . '"' . $matches[4] . '"' . $matches[5] . '[/' . $matches[6] . 'rl]';
}
turn naked links in posts into zrl's if we've got a hubloc for the site
2013-04-19 00:55:35 +00:00
don't add bookmark tags to naked links inside code blocks
2014-02-10 09:11:59 +00:00
function red_escape_codeblock($m) {
better bookmark escaping
2014-02-10 09:44:21 +00:00
return '[$b64' . $m[2] . base64_encode($m[1]) . '[/' . $m[2] . ']';
don't add bookmark tags to naked links inside code blocks
2014-02-10 09:11:59 +00:00
}
function red_unescape_codeblock($m) {
better bookmark escaping
2014-02-10 09:44:21 +00:00
return '[' . $m[2] . base64_decode($m[1]) . '[/' . $m[2] . ']';
don't add bookmark tags to naked links inside code blocks
2014-02-10 09:11:59 +00:00
}
zrlify (turn into zmg) any relevant img tags in a post.
2014-05-14 03:17:52 +00:00
function red_zrlify_img_callback($matches) {
remove some code duplication
2014-07-14 08:38:16 +00:00
require_once('include/hubloc.php');
$zrl = is_matrix_url($matches[2]);
zrlify (turn into zmg) any relevant img tags in a post.
2014-05-14 03:17:52 +00:00
$t = strip_zids($matches[2]);
if($t !== $matches[2]) {
$zrl = true;
$matches[2] = $t;
}
if($zrl)
return '[zmg' . $matches[1] . ']' . $matches[2] . '[/zmg]';
return $matches[0];
}
wrapper function to create arbitrary activities (e.g. for plugins and apps) with a minimal amount of fuss, though the amount of fuss could increase significantly depending on the nature of the activity. For a simple wall post, just set the $arr['body'] to the desired text and we'll do the rest. If the activity requires special verbs, objects or targets, these will all need to be constructed.
2013-04-05 01:54:24 +00:00
/**
* @function post_activity_item($arr)
*
* post an activity
*
* @param array $arr
*
* In its simplest form one needs only to set $arr['body'] to post a note to the logged in channel's wall.
* Much more complex activities can be created. Permissions are checked. No filtering, tag expansion
* or other processing is performed.
*
* @returns array
* 'success' => true or false
* 'activity' => the resulting activity if successful
*/
function post_activity_item($arr) {
$ret = array('success' => false);
tagging posts now sends out a presumably legal activity message - all that's missing is catching it on the post owner's side, checking if people can tag this stream, and then add the tag to the parent message.
2013-06-05 03:07:13 +00:00
$is_comment = false;
if((($arr['parent']) && $arr['parent'] != $arr['id']) || (($arr['parent_mid']) && $arr['parent_mid'] != $arr['mid']))
$is_comment = true;
wrapper function to create arbitrary activities (e.g. for plugins and apps) with a minimal amount of fuss, though the amount of fuss could increase significantly depending on the nature of the activity. For a simple wall post, just set the $arr['body'] to the desired text and we'll do the rest. If the activity requires special verbs, objects or targets, these will all need to be constructed.
2013-04-05 01:54:24 +00:00
if(! x($arr,'item_flags')) {
tagging posts now sends out a presumably legal activity message - all that's missing is catching it on the post owner's side, checking if people can tag this stream, and then add the tag to the parent message.
2013-06-05 03:07:13 +00:00
if($is_comment)
$arr['item_flags'] = ITEM_ORIGIN;
else
$arr['item_flags'] = ITEM_ORIGIN | ITEM_WALL | ITEM_THREAD_TOP;
wrapper function to create arbitrary activities (e.g. for plugins and apps) with a minimal amount of fuss, though the amount of fuss could increase significantly depending on the nature of the activity. For a simple wall post, just set the $arr['body'] to the desired text and we'll do the rest. If the activity requires special verbs, objects or targets, these will all need to be constructed.
2013-04-05 01:54:24 +00:00
}
tagging posts now sends out a presumably legal activity message - all that's missing is catching it on the post owner's side, checking if people can tag this stream, and then add the tag to the parent message.
2013-06-05 03:07:13 +00:00
wrapper function to create arbitrary activities (e.g. for plugins and apps) with a minimal amount of fuss, though the amount of fuss could increase significantly depending on the nature of the activity. For a simple wall post, just set the $arr['body'] to the desired text and we'll do the rest. If the activity requires special verbs, objects or targets, these will all need to be constructed.
2013-04-05 01:54:24 +00:00
$channel = get_app()->get_channel();
$observer = get_app()->get_observer();
$arr['aid'] = ((x($arr,'aid')) ? $arr['aid'] : $channel['channel_account_id']);
$arr['uid'] = ((x($arr,'uid')) ? $arr['uid'] : $channel['channel_id']);
tagging posts now sends out a presumably legal activity message - all that's missing is catching it on the post owner's side, checking if people can tag this stream, and then add the tag to the parent message.
2013-06-05 03:07:13 +00:00
if(! perm_is_allowed($arr['uid'],$observer['xchan_hash'],(($is_comment) ? 'post_comments' : 'post_wall'))) {
wrapper function to create arbitrary activities (e.g. for plugins and apps) with a minimal amount of fuss, though the amount of fuss could increase significantly depending on the nature of the activity. For a simple wall post, just set the $arr['body'] to the desired text and we'll do the rest. If the activity requires special verbs, objects or targets, these will all need to be constructed.
2013-04-05 01:54:24 +00:00
$ret['message'] = t('Permission denied');
return $ret;
}
issue #573 - this may not fix everything in that issue but should resolve some of it.
2014-08-23 01:01:21 +00:00
$arr['public_policy'] = ((x($_REQUEST,'public_policy')) ? escape_tags($_REQUEST['public_policy']) : map_scope($channel['channel_r_stream'],true));
if($arr['public_policy'])
$arr['item_private'] = 1;
add basic input filtering to the simple activity posting
2013-05-23 01:22:41 +00:00
post_activity_item issues
2013-12-08 07:29:26 +00:00
if(! array_key_exists('mimetype',$arr))
$arr['mimetype'] = 'text/bbcode';
if(array_key_exists('item_private',$arr) && $arr['item_private']) {
$arr['body'] = z_input_filter($arr['uid'],$arr['body'],$arr['mimetype']);
if($channel) {
if($channel['channel_hash'] === $arr['author_xchan']) {
$arr['sig'] = base64url_encode(rsa_sign($arr['body'],$channel['channel_prvkey']));
$arr['item_flags'] = $arr['item_flags'] | ITEM_VERIFIED;
}
}
logger('Encrypting local storage');
$key = get_config('system','pubkey');
$arr['item_flags'] = $arr['item_flags'] | ITEM_OBSCURED;
if($arr['title'])
switch everything over to crypto_encapsulate()
2014-08-23 09:45:21 +00:00
$arr['title'] = json_encode(crypto_encapsulate($arr['title'],$key));
post_activity_item issues
2013-12-08 07:29:26 +00:00
if($arr['body'])
switch everything over to crypto_encapsulate()
2014-08-23 09:45:21 +00:00
$arr['body'] = json_encode(crypto_encapsulate($arr['body'],$key));
post_activity_item issues
2013-12-08 07:29:26 +00:00
}
add basic input filtering to the simple activity posting
2013-05-23 01:22:41 +00:00
wrapper function to create arbitrary activities (e.g. for plugins and apps) with a minimal amount of fuss, though the amount of fuss could increase significantly depending on the nature of the activity. For a simple wall post, just set the $arr['body'] to the desired text and we'll do the rest. If the activity requires special verbs, objects or targets, these will all need to be constructed.
2013-04-05 01:54:24 +00:00
$arr['mid'] = ((x($arr,'mid')) ? $arr['mid'] : item_message_id());
$arr['parent_mid'] = ((x($arr,'parent_mid')) ? $arr['parent_mid'] : $arr['mid']);
$arr['thr_parent'] = ((x($arr,'thr_parent')) ? $arr['thr_parent'] : $arr['mid']);
basic poke working
2013-04-18 04:40:40 +00:00
$arr['owner_xchan'] = ((x($arr,'owner_xchan')) ? $arr['owner_xchan'] : $channel['channel_hash']);
wrapper function to create arbitrary activities (e.g. for plugins and apps) with a minimal amount of fuss, though the amount of fuss could increase significantly depending on the nature of the activity. For a simple wall post, just set the $arr['body'] to the desired text and we'll do the rest. If the activity requires special verbs, objects or targets, these will all need to be constructed.
2013-04-05 01:54:24 +00:00
$arr['author_xchan'] = ((x($arr,'author_xchan')) ? $arr['author_xchan'] : $observer['xchan_hash']);
$arr['verb'] = ((x($arr,'verb')) ? $arr['verb'] : ACTIVITY_POST);
$arr['obj_type'] = ((x($arr,'obj_type')) ? $arr['obj_type'] : ACTIVITY_OBJ_NOTE);
$arr['allow_cid'] = ((x($arr,'allow_cid')) ? $arr['allow_cid'] : $channel['channel_allow_cid']);
$arr['allow_gid'] = ((x($arr,'allow_gid')) ? $arr['allow_gid'] : $channel['channel_allow_gid']);
$arr['deny_cid'] = ((x($arr,'deny_cid')) ? $arr['deny_cid'] : $channel['channel_deny_cid']);
$arr['deny_gid'] = ((x($arr,'deny_gid')) ? $arr['deny_gid'] : $channel['channel_deny_gid']);
undefined function
2013-06-17 06:08:52 +00:00
$arr['comment_policy'] = map_scope($channel['channel_w_comment']);
improvements to post_activity_item() so that it does not behave differently with delivery plugins than post_local via the API.
2013-05-22 04:51:02 +00:00
some minor cleanup on plinks for some very subtle permissions issues
2014-01-09 23:45:17 +00:00
if ((! $arr['plink']) && ($arr['item_flags'] & ITEM_THREAD_TOP)) {
$arr['plink'] = z_root() . '/channel/' . $channel['channel_address'] . '/?f=&mid=' . $arr['mid'];
}
improvements to post_activity_item() so that it does not behave differently with delivery plugins than post_local via the API.
2013-05-22 04:51:02 +00:00
// for the benefit of plugins, we will behave as if this is an API call rather than a normal online post
$_REQUEST['api_source'] = 1;
missing comma
2013-05-22 04:54:21 +00:00
call_hooks('post_local',$arr);
improvements to post_activity_item() so that it does not behave differently with delivery plugins than post_local via the API.
2013-05-22 04:51:02 +00:00
if(x($arr,'cancel')) {
logger('post_activity_item: post cancelled by plugin.');
return $ret;
}
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$post = item_store($arr);
post_activity_item issues
2013-12-08 07:29:26 +00:00
if($post['success'])
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$post_id = $post['item_id'];
wrapper function to create arbitrary activities (e.g. for plugins and apps) with a minimal amount of fuss, though the amount of fuss could increase significantly depending on the nature of the activity. For a simple wall post, just set the $arr['body'] to the desired text and we'll do the rest. If the activity requires special verbs, objects or targets, these will all need to be constructed.
2013-04-05 01:54:24 +00:00
if($post_id) {
$arr['id'] = $post_id;
call_hooks('post_local_end', $arr);
proc_run('php','include/notifier.php','activity',$post_id);
$ret['success'] = true;
fix app display
2013-05-22 03:52:18 +00:00
$r = q("select * from item where id = %d limit 1",
intval($post_id)
);
if($r)
$ret['activity'] = $r[0];
wrapper function to create arbitrary activities (e.g. for plugins and apps) with a minimal amount of fuss, though the amount of fuss could increase significantly depending on the nature of the activity. For a simple wall post, just set the $arr['body'] to the desired text and we'll do the rest. If the activity requires special verbs, objects or targets, these will all need to be constructed.
2013-04-05 01:54:24 +00:00
}
return $ret;
}
ability to unset a layout and subthread is probably close to working again.
2014-04-24 02:05:48 +00:00
/**
* @function get_public_feed($channel,$params)
* generate an Atom feed
*/
wrapper function to create arbitrary activities (e.g. for plugins and apps) with a minimal amount of fuss, though the amount of fuss could increase significantly depending on the nature of the activity. For a simple wall post, just set the $arr['body'] to the desired text and we'll do the rest. If the activity requires special verbs, objects or targets, these will all need to be constructed.
2013-04-05 01:54:24 +00:00
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
function get_public_feed($channel,$params) {
$type = 'xml';
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
$begin = NULL_DATE;
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
$end = '';
$start = 0;
$records = 40;
$direction = 'desc';
pages feed
2013-09-06 02:31:26 +00:00
$pages = 0;
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
if(! $params)
$params = array();
pages feed
2013-09-06 02:31:26 +00:00
$params['type'] = ((x($params,'type')) ? $params['type'] : 'xml');
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
$params['begin'] = ((x($params,'begin')) ? $params['begin'] : NULL_DATE);
pages feed
2013-09-06 02:31:26 +00:00
$params['end'] = ((x($params,'end')) ? $params['end'] : datetime_convert('UTC','UTC','now'));
$params['start'] = ((x($params,'start')) ? $params['start'] : 0);
$params['records'] = ((x($params,'records')) ? $params['records'] : 40);
$params['direction'] = ((x($params,'direction')) ? $params['direction'] : 'desc');
$params['pages'] = ((x($params,'pages')) ? intval($params['pages']) : 0);
issue #571 - use feed/nickname?top=1
2014-08-22 22:51:48 +00:00
$params['top'] = ((x($params,'top')) ? intval($params['top']) : 0);
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
switch($params['type']) {
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
case 'json':
header("Content-type: application/atom+json");
break;
case 'xml':
default:
header("Content-type: application/atom+xml");
break;
}
bug #372 - category feeds
2012-04-14 11:19:41 +00:00
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
return get_feed_for($channel,get_observer_hash(),$params);
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
}
feed generator
2010-07-18 13:02:55 +00:00
ability to unset a layout and subthread is probably close to working again.
2014-04-24 02:05:48 +00:00
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
function get_feed_for($channel, $observer_hash, $params) {
reverse sort order for public feeds
2010-09-10 08:45:58 +00:00
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
if(! channel)
http_status_exit(401);
more robust feed error handling, at the expense of performance. Pass profile owner through HTML (yuk) to the ajax updater - as browser pre-fetch totally buggers passing it via the server session.
2010-09-21 02:34:44 +00:00
bug #372 - category feeds
2012-04-14 11:19:41 +00:00
pages feed
2013-09-06 02:31:26 +00:00
if($params['pages']) {
if(! perm_is_allowed($channel['channel_id'],$observer_hash,'view_pages'))
http_status_exit(403);
}
else {
if(! perm_is_allowed($channel['channel_id'],$observer_hash,'view_stream'))
http_status_exit(403);
}
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
$items = items_fetch(array(
'wall' => '1',
'datequery' => $params['begin'],
'datequery2' => $params['end'],
'start' => $params['start'], // FIXME
'records' => $params['records'], // FIXME
'direction' => $params['direction'], // FIXME
pages feed
2013-09-06 02:31:26 +00:00
'pages' => $params['pages'],
issue #571 - use feed/nickname?top=1
2014-08-22 22:51:48 +00:00
'order' => 'post',
'top' => $params['top']
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
), $channel, $observer_hash, CLIENT_MODE_NORMAL, get_app()->module);
more robust feed error handling, at the expense of performance. Pass profile owner through HTML (yuk) to the ajax updater - as browser pre-fetch totally buggers passing it via the server session.
2010-09-21 02:34:44 +00:00
feed generator
2010-07-18 13:02:55 +00:00
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
$feed_template = get_markup_template('atom_feed.tpl');
feed generator
2010-07-18 13:02:55 +00:00
$atom = '';
$atom .= replace_macros($feed_template, array(
begin to cut the umbilical
2013-04-15 05:24:47 +00:00
'$version' => xmlify(RED_VERSION),
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
'$red' => xmlify(RED_PLATFORM),
outbound feed issue
2014-09-03 08:09:28 +00:00
'$feed_id' => xmlify($channel['xchan_url']),
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
'$feed_title' => xmlify($channel['channel_name']),
more lint
2010-10-31 23:38:22 +00:00
'$feed_updated' => xmlify(datetime_convert('UTC', 'UTC', 'now' , ATOM_TIME)) ,
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
'$hub' => '', // feed_hublinks(),
'$salmon' => '', // feed_salmonlinks($channel['channel_address']),
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
'$name' => xmlify($channel['channel_name']),
outbound feed issue
2014-09-03 08:09:28 +00:00
'$profile_page' => xmlify($channel['xchan_url']),
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
'$mimephoto' => xmlify($channel['xchan_photo_mimetype']),
'$photo' => xmlify($channel['xchan_photo_l']),
'$thumb' => xmlify($channel['xchan_photo_m']),
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
'$picdate' => '',
'$uridate' => '',
'$namdate' => '',
'$birthday' => '',
'$community' => '',
feed generator
2010-07-18 13:02:55 +00:00
));
more plugin hooks
2010-12-25 02:32:23 +00:00
call_hooks('atom_feed', $atom);
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
if($items) {
$type = 'html';
foreach($items as $item) {
if($item['item_private'])
keep private emails out of any offsite feeds
2011-08-11 04:46:19 +00:00
continue;
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
$atom .= atom_entry($item,$type,null,$owner,true);
}
feed generator
2010-07-18 13:02:55 +00:00
}
more plugin hooks
2010-12-25 02:32:23 +00:00
call_hooks('atom_feed_end', $atom);
mongo notification checkin, wall, comments, settings
2010-08-14 14:55:18 +00:00
$atom .= '</feed>' . "\r\n";
undo one of the hooks from last checkin - passing objects led to error
2010-12-25 03:04:40 +00:00
feed generator
2010-07-18 13:02:55 +00:00
return $atom;
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
}
function construct_verb($item) {
if($item['verb'])
return $item['verb'];
return ACTIVITY_POST;
}
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
function construct_activity_object($item) {
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
if($item['object']) {
$o = '<as:object>' . "\r\n";
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
$r = json_decode($item['object'],false);
fixed a few feed problems affecting activity objects/targets
2011-06-10 08:51:14 +00:00
if(! $r)
return '';
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
if($r->type)
item table rename parent-uri, target-type, object-type (more to come later)
2012-08-16 05:33:37 +00:00
$o .= '<as:obj_type>' . xmlify($r->type) . '</as:obj_type>' . "\r\n";
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
if($r->id)
activity objects weren't escaped properly
2010-10-11 03:02:29 +00:00
$o .= '<id>' . xmlify($r->id) . '</id>' . "\r\n";
hub anomolies & swat0 compliance
2010-11-05 03:47:44 +00:00
if($r->title)
$o .= '<title>' . xmlify($r->title) . '</title>' . "\r\n";
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
if($r->links) {
// FIXME!!
fixed a few feed problems affecting activity objects/targets
2011-06-10 08:51:14 +00:00
if(substr($r->link,0,1) === '<') {
$r->link = preg_replace('/\<link(.*?)\"\>/','<link$1"/>',$r->link);
hub anomolies & swat0 compliance
2010-11-05 03:47:44 +00:00
$o .= $r->link;
fixed a few feed problems affecting activity objects/targets
2011-06-10 08:51:14 +00:00
}
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
else
$o .= '<link rel="alternate" type="text/html" href="' . xmlify($r->link) . '" />' . "\r\n";
}
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
if($r->content)
activity objects weren't escaped properly
2010-10-11 03:02:29 +00:00
$o .= '<content type="html" >' . xmlify(bbcode($r->content)) . '</content>' . "\r\n";
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
$o .= '</as:object>' . "\r\n";
return $o;
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
}
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
return '';
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
}
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
function construct_activity_target($item) {
if($item['target']) {
$o = '<as:target>' . "\r\n";
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
$r = json_decode($item['target'],false);
fixed a few feed problems affecting activity objects/targets
2011-06-10 08:51:14 +00:00
if(! $r)
return '';
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
if($r->type)
item table rename parent-uri, target-type, object-type (more to come later)
2012-08-16 05:33:37 +00:00
$o .= '<as:obj_type>' . xmlify($r->type) . '</as:obj_type>' . "\r\n";
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
if($r->id)
$o .= '<id>' . xmlify($r->id) . '</id>' . "\r\n";
hub anomolies & swat0 compliance
2010-11-05 03:47:44 +00:00
if($r->title)
$o .= '<title>' . xmlify($r->title) . '</title>' . "\r\n";
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
if($r->links) {
// FIXME !!!
fixed a few feed problems affecting activity objects/targets
2011-06-10 08:51:14 +00:00
if(substr($r->link,0,1) === '<') {
if(strstr($r->link,'&') && (! strstr($r->link,'&amp;')))
$r->link = str_replace('&','&amp;', $r->link);
$r->link = preg_replace('/\<link(.*?)\"\>/','<link$1"/>',$r->link);
hub anomolies & swat0 compliance
2010-11-05 03:47:44 +00:00
$o .= $r->link;
fixed a few feed problems affecting activity objects/targets
2011-06-10 08:51:14 +00:00
}
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
else
$o .= '<link rel="alternate" type="text/html" href="' . xmlify($r->link) . '" />' . "\r\n";
}
if($r->content)
$o .= '<content type="html" >' . xmlify(bbcode($r->content)) . '</content>' . "\r\n";
$o .= '</as:target>' . "\r\n";
return $o;
}
return '';
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
}
/* limit_body_size()
*
* The purpose of this function is to apply system message length limits to
* imported messages without including any embedded photos in the length
*/
liking comments were reverted to liking the parent after sending item to another channel
2013-04-04 01:04:48 +00:00
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
function limit_body_size($body) {
$maxlen = get_max_import_size();
// If the length of the body, including the embedded images, is smaller
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
// than the maximum, then don't waste time looking for the images
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
if($maxlen && (strlen($body) > $maxlen)) {
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
$orig_body = $body;
$new_body = '';
$textlen = 0;
$max_found = false;
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
$img_start = strpos($orig_body, '[img');
$img_st_close = ($img_start !== false ? strpos(substr($orig_body, $img_start), ']') : false);
$img_end = ($img_start !== false ? strpos(substr($orig_body, $img_start), '[/img]') : false);
while(($img_st_close !== false) && ($img_end !== false)) {
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
$img_st_close++; // make it point to AFTER the closing bracket
$img_end += $img_start;
$img_end += strlen('[/img]');
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
if(! strcmp(substr($orig_body, $img_start + $img_st_close, 5), 'data:')) {
// This is an embedded image
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
if( ($textlen + $img_start) > $maxlen ) {
if($textlen < $maxlen) {
logger('limit_body_size: the limit happens before an embedded image', LOGGER_DEBUG);
$new_body = $new_body . substr($orig_body, 0, $maxlen - $textlen);
$textlen = $maxlen;
}
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
}
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
else {
$new_body = $new_body . substr($orig_body, 0, $img_start);
fix bugs
2012-07-08 00:47:13 +00:00
$textlen += $img_start;
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
}
$new_body = $new_body . substr($orig_body, $img_start, $img_end - $img_start);
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
}
else {
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
if( ($textlen + $img_end) > $maxlen ) {
if($textlen < $maxlen) {
$new_body = $new_body . substr($orig_body, 0, $maxlen - $textlen);
$textlen = $maxlen;
}
}
else {
$new_body = $new_body . substr($orig_body, 0, $img_end);
$textlen += $img_end;
}
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
}
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
$orig_body = substr($orig_body, $img_end);
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
if($orig_body === false) // in case the body ends on a closing image tag
$orig_body = '';
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
$img_start = strpos($orig_body, '[img');
$img_st_close = ($img_start !== false ? strpos(substr($orig_body, $img_start), ']') : false);
$img_end = ($img_start !== false ? strpos(substr($orig_body, $img_start), '[/img]') : false);
}
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
if( ($textlen + strlen($orig_body)) > $maxlen) {
if($textlen < $maxlen) {
$new_body = $new_body . substr($orig_body, 0, $maxlen - $textlen);
$textlen = $maxlen;
}
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
}
else {
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
$new_body = $new_body . $orig_body;
$textlen += strlen($orig_body);
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
}
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
return $new_body;
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
}
else
return $body;
liking comments were reverted to liking the parent after sending item to another channel
2013-04-04 01:04:48 +00:00
}
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
eliminate the obsolete "last-child" indicator
2012-07-19 00:08:03 +00:00
function title_is_body($title, $body) {
$title = strip_tags($title);
$title = trim($title);
$title = str_replace(array("\n", "\r", "\t", " "), array("","","",""), $title);
$body = strip_tags($body);
$body = trim($body);
$body = str_replace(array("\n", "\r", "\t", " "), array("","","",""), $body);
if (strlen($title) < strlen($body))
$body = substr($body, 0, strlen($title));
if (($title != $body) and (substr($title, -3) == "...")) {
$pos = strrpos($title, "...");
if ($pos > 0) {
$title = substr($title, 0, $pos);
$body = substr($body, 0, $pos);
}
}
return($title == $body);
}
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
mostly working item_decode, but without full input santising
2012-11-17 10:29:02 +00:00
function get_item_elements($x) {
get_item_elements (json version)
2012-07-30 05:43:51 +00:00
$arr = array();
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['body'] = (($x['body']) ? htmlspecialchars($x['body'],ENT_COMPAT,'UTF-8',false) : '');
get_item_elements (json version)
2012-07-30 05:43:51 +00:00
issues with diaspora comment signatures on relayed comments that are relayed through a redmatrix site (parent post is redmatrix) and involve a private post.
2014-10-02 09:43:07 +00:00
$key = get_config('system','pubkey');
implement max_import_size
2014-08-10 06:38:33 +00:00
$maxlen = get_max_import_size();
if($maxlen && mb_strlen($arr['body']) > $maxlen) {
$arr['body'] = mb_substr($arr['body'],0,$maxlen,'UTF-8');
logger('get_item_elements: message length exceeds max_import_size: truncated');
}
mostly working item_decode, but without full input santising
2012-11-17 10:29:02 +00:00
$arr['created'] = datetime_convert('UTC','UTC',$x['created']);
$arr['edited'] = datetime_convert('UTC','UTC',$x['edited']);
get_item_elements (json version)
2012-07-30 05:43:51 +00:00
if($arr['created'] > datetime_convert())
mostly working item_decode, but without full input santising
2012-11-17 10:29:02 +00:00
$arr['created'] = datetime_convert();
get_item_elements (json version)
2012-07-30 05:43:51 +00:00
if($arr['edited'] > datetime_convert())
mostly working item_decode, but without full input santising
2012-11-17 10:29:02 +00:00
$arr['edited'] = datetime_convert();
pull in some posts when we first connect with a new channel (if allowed to) - if not allowed to, do it if that condition changes
2013-11-18 03:22:24 +00:00
$arr['expires'] = ((x($x,'expires') && $x['expires'])
? datetime_convert('UTC','UTC',$x['expires'])
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
: NULL_DATE);
pull in some posts when we first connect with a new channel (if allowed to) - if not allowed to, do it if that condition changes
2013-11-18 03:22:24 +00:00
$arr['commented'] = ((x($x,'commented') && $x['commented'])
? datetime_convert('UTC','UTC',$x['commented'])
: $arr['created']);
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$arr['comments_closed'] = ((x($x,'comments_closed') && $x['comments_closed'])
? datetime_convert('UTC','UTC',$x['comments_closed'])
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
: NULL_DATE);
pull in some posts when we first connect with a new channel (if allowed to) - if not allowed to, do it if that condition changes
2013-11-18 03:22:24 +00:00
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['title'] = (($x['title']) ? htmlspecialchars($x['title'], ENT_COMPAT,'UTF-8',false) : '');
fix bug #84, make imported messages from your clones show up on your wall
2013-07-30 00:30:46 +00:00
if(mb_strlen($arr['title']) > 255)
$arr['title'] = mb_substr($arr['title'],0,255);
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['app'] = (($x['app']) ? htmlspecialchars($x['app'], ENT_COMPAT,'UTF-8',false) : '');
finish up the source route storage so we can prevent messages from flowing upstream in complicated delivery chains
2014-03-31 23:25:05 +00:00
$arr['route'] = (($x['route']) ? htmlspecialchars($x['route'], ENT_COMPAT,'UTF-8',false) : '');
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['mid'] = (($x['message_id']) ? htmlspecialchars($x['message_id'], ENT_COMPAT,'UTF-8',false) : '');
$arr['parent_mid'] = (($x['message_top']) ? htmlspecialchars($x['message_top'], ENT_COMPAT,'UTF-8',false) : '');
$arr['thr_parent'] = (($x['message_parent']) ? htmlspecialchars($x['message_parent'], ENT_COMPAT,'UTF-8',false) : '');
mostly working item_decode, but without full input santising
2012-11-17 10:29:02 +00:00
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['plink'] = (($x['permalink']) ? htmlspecialchars($x['permalink'], ENT_COMPAT,'UTF-8',false) : '');
$arr['location'] = (($x['location']) ? htmlspecialchars($x['location'], ENT_COMPAT,'UTF-8',false) : '');
$arr['coord'] = (($x['longlat']) ? htmlspecialchars($x['longlat'], ENT_COMPAT,'UTF-8',false) : '');
$arr['verb'] = (($x['verb']) ? htmlspecialchars($x['verb'], ENT_COMPAT,'UTF-8',false) : '');
$arr['mimetype'] = (($x['mimetype']) ? htmlspecialchars($x['mimetype'], ENT_COMPAT,'UTF-8',false) : '');
$arr['obj_type'] = (($x['object_type']) ? htmlspecialchars($x['object_type'], ENT_COMPAT,'UTF-8',false) : '');
$arr['tgt_type'] = (($x['target_type']) ? htmlspecialchars($x['target_type'], ENT_COMPAT,'UTF-8',false) : '');
first step first - just store the public_policy data, don't do anything with it yet.
2014-08-07 00:36:07 +00:00
$arr['public_policy'] = (($x['public_scope']) ? htmlspecialchars($x['public_scope'], ENT_COMPAT,'UTF-8',false) : '');
if($arr['public_policy'] === 'public')
$arr['public_policy'] = '';
finish up the source route storage so we can prevent messages from flowing upstream in complicated delivery chains
2014-03-31 23:25:05 +00:00
$arr['comment_policy'] = (($x['comment_scope']) ? htmlspecialchars($x['comment_scope'], ENT_COMPAT,'UTF-8',false) : 'contacts');
post signatures
2013-10-03 04:04:48 +00:00
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['sig'] = (($x['signature']) ? htmlspecialchars($x['signature'], ENT_COMPAT,'UTF-8',false) : '');
post signatures
2013-10-03 04:04:48 +00:00
issues with diaspora comment signatures on relayed comments that are relayed through a redmatrix site (parent post is redmatrix) and involve a private post.
2014-10-02 09:43:07 +00:00
$arr['diaspora_meta'] = (($x['diaspora_signature']) ? json_encode(crypto_encapsulate($x['diaspora_signature'],$key)) : '');
and there's the import side of things
2012-11-17 20:55:59 +00:00
$arr['object'] = activity_sanitise($x['object']);
$arr['target'] = activity_sanitise($x['target']);
get_item_elements (json version)
2012-07-30 05:43:51 +00:00
sort out the rest of the nasty stuff in the attachment/file api - time to move on to something else
2013-02-01 08:49:07 +00:00
$arr['attach'] = activity_sanitise($x['attach']);
and there's the import side of things
2012-11-17 20:55:59 +00:00
$arr['term'] = decode_tags($x['tags']);
get_item_elements (json version)
2012-07-30 05:43:51 +00:00
fix item_private flag
2012-11-27 01:07:23 +00:00
$arr['item_private'] = ((array_key_exists('flags',$x) && is_array($x['flags']) && in_array('private',$x['flags'])) ? 1 : 0);
fix some privacy leakage in logs
2013-08-01 01:57:14 +00:00
$arr['item_flags'] = 0;
You've got mail! (actually a post, mail is for another day)
2012-11-27 05:20:16 +00:00
if(array_key_exists('flags',$x) && in_array('deleted',$x['flags']))
basic changes for file activity
2015-01-09 14:18:45 +00:00
$arr['item_restrict'] = ITEM_DELETED;
if(array_key_exists('flags',$x) && in_array('hidden',$x['flags']))
$arr['item_restrict'] = ITEM_HIDDEN;
fix item_private flag
2012-11-27 01:07:23 +00:00
and there's the import side of things
2012-11-17 20:55:59 +00:00
// Here's the deal - the site might be down or whatever but if there's a new person you've never
// seen before sending stuff to your stream, we MUST be able to look them up and import their data from their
// hub and verify that they are legit - or else we're going to toss the post. We only need to do this
// once, and after that your hub knows them. Sure some info is in the post, but it's only a transit identifier
// and not enough info to be able to look you up from your hash - which is the only thing stored with the post.
get_item_elements (json version)
2012-07-30 05:43:51 +00:00
include re-organisation and more doco, post_to_red fix ampersands in categories
2013-12-04 08:19:29 +00:00
if(($xchan_hash = import_author_xchan($x['author'])) !== false)
$arr['author_xchan'] = $xchan_hash;
and there's the import side of things
2012-11-17 20:55:59 +00:00
else
return array();
get_item_elements (json version)
2012-07-30 05:43:51 +00:00
It's really not supposed to be quite this chatty on the network - ah that's why. Oops.
2013-01-29 04:51:37 +00:00
// save a potentially expensive lookup if author == owner
clean up more code duplication
2014-07-15 04:21:24 +00:00
if($arr['author_xchan'] === make_xchan_hash($x['owner']['guid'],$x['owner']['guid_sig']))
It's really not supposed to be quite this chatty on the network - ah that's why. Oops.
2013-01-29 04:51:37 +00:00
$arr['owner_xchan'] = $arr['author_xchan'];
else {
include re-organisation and more doco, post_to_red fix ampersands in categories
2013-12-04 08:19:29 +00:00
if(($xchan_hash = import_author_xchan($x['owner'])) !== false)
$arr['owner_xchan'] = $xchan_hash;
It's really not supposed to be quite this chatty on the network - ah that's why. Oops.
2013-01-29 04:51:37 +00:00
else
return array();
}
get_item_elements (json version)
2012-07-30 05:43:51 +00:00
post signatures
2013-10-03 04:04:48 +00:00
if($arr['sig']) {
$r = q("select xchan_pubkey from xchan where xchan_hash = '%s' limit 1",
dbesc($arr['author_xchan'])
);
if($r && rsa_verify($x['body'],base64url_decode($arr['sig']),$r[0]['xchan_pubkey']))
$arr['item_flags'] |= ITEM_VERIFIED;
else
logger('get_item_elements: message verification failed.');
}
// if it's a private post, encrypt it in the DB.
// We have to do that here because we need to cleanse the input and prevent bad stuff from getting in,
// and we need plaintext to do that.
optionally include wall items (posts) in import/export
2014-09-26 03:21:23 +00:00
post signatures
2013-10-03 04:04:48 +00:00
if(intval($arr['item_private'])) {
$arr['item_flags'] = $arr['item_flags'] | ITEM_OBSCURED;
if($arr['title'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$arr['title'] = json_encode(crypto_encapsulate($arr['title'],$key));
post signatures
2013-10-03 04:04:48 +00:00
if($arr['body'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$arr['body'] = json_encode(crypto_encapsulate($arr['body'],$key));
post signatures
2013-10-03 04:04:48 +00:00
}
optionally include wall items (posts) in import/export
2014-09-26 03:21:23 +00:00
if(array_key_exists('revision',$x)) {
// extended export encoding
$arr['revision'] = $x['revision'];
$arr['allow_cid'] = $x['allow_cid'];
$arr['allow_gid'] = $x['allow_gid'];
$arr['deny_cid'] = $x['deny_cid'];
$arr['deny_gid'] = $x['deny_gid'];
$arr['layout_mid'] = $x['layout_mid'];
$arr['postopts'] = $x['postopts'];
$arr['resource_id'] = $x['resource_id'];
$arr['resource_type'] = $x['resource_type'];
$arr['item_restrict'] = $x['item_restrict'];
$arr['item_flags'] = $x['item_flags'];
$arr['attach'] = $x['attach'];
}
get_item_elements (json version)
2012-07-30 05:43:51 +00:00
return $arr;
}
y'all got mail
2012-12-06 00:44:07 +00:00
and there's the import side of things
2012-11-17 20:55:59 +00:00
function import_author_xchan($x) {
It's really not supposed to be quite this chatty on the network - ah that's why. Oops.
2013-01-29 04:51:37 +00:00
include re-organisation and more doco, post_to_red fix ampersands in categories
2013-12-04 08:19:29 +00:00
$arr = array('xchan' => $x, 'xchan_hash' => '');
call_hooks('import_author_xchan',$arr);
if($arr['xchan_hash'])
return $arr['xchan_hash'];
It's really not supposed to be quite this chatty on the network - ah that's why. Oops.
2013-01-29 04:51:37 +00:00
include re-organisation and more doco, post_to_red fix ampersands in categories
2013-12-04 08:19:29 +00:00
if((! array_key_exists('network', $x)) || ($x['network'] === 'zot')) {
project "snakebite"
2014-02-17 22:30:02 +00:00
$y = import_author_zot($x);
It's really not supposed to be quite this chatty on the network - ah that's why. Oops.
2013-01-29 04:51:37 +00:00
}
allow stuff with Diaspora authors to federate in zot conversations even if the local site has diaspora disabled. This way it won't look like folks are talking to themselves.
2014-09-04 08:13:01 +00:00
if(! $y)
$y = import_author_diaspora($x);
It's really not supposed to be quite this chatty on the network - ah that's why. Oops.
2013-01-29 04:51:37 +00:00
project "snakebite"
2014-02-17 22:30:02 +00:00
if($x['network'] === 'rss') {
$y = import_author_rss($x);
}
correct item author for feeds with multiple or different authors
2014-09-12 00:27:49 +00:00
if($x['network'] === 'unknown') {
$y = import_author_unknown($x);
}
project "snakebite"
2014-02-17 22:30:02 +00:00
return(($y) ? $y : false);
}
allow stuff with Diaspora authors to federate in zot conversations even if the local site has diaspora disabled. This way it won't look like folks are talking to themselves.
2014-09-04 08:13:01 +00:00
function import_author_diaspora($x) {
if(! $x['address'])
return false;
if(discover_by_webbie($x['address'])) {
$r = q("select xchan_hash from xchan where xchan_addr = '%s' limit 1",
dbesc($x['address'])
);
if($r)
return $r[0]['xchan_hash'];
}
return false;
}
project "snakebite"
2014-02-17 22:30:02 +00:00
function import_author_rss($x) {
if(! $x['url'])
return false;
$r = q("select xchan_hash from xchan where xchan_network = 'rss' and xchan_url = '%s' limit 1",
dbesc($x['url'])
);
if($r) {
logger('import_author_rss: in cache' , LOGGER_DEBUG);
return $r[0]['xchan_hash'];
}
$name = trim($x['name']);
import_author_rss not succeeding, missing url
2014-09-05 02:15:01 +00:00
$r = q("insert into xchan ( xchan_hash, xchan_guid, xchan_url, xchan_name, xchan_network )
values ( '%s', '%s', '%s', '%s', '%s' )",
dbesc($x['guid']),
dbesc($x['guid']),
project "snakebite"
2014-02-17 22:30:02 +00:00
dbesc($x['url']),
minor stuff, some doco, auth cleanup, and make "unknown" more translateable by context.
2014-02-22 22:58:12 +00:00
dbesc(($name) ? $name : t('(Unknown)')),
project "snakebite"
2014-02-17 22:30:02 +00:00
dbesc('rss')
);
import_profile_photo array warnings
2014-09-08 23:06:15 +00:00
if($r && $x['photo']) {
project "snakebite"
2014-02-17 22:30:02 +00:00
import_profile_photo array warnings
2014-09-08 23:06:15 +00:00
$photos = import_profile_photo($x['photo']['src'],$x['url']);
project "snakebite"
2014-02-17 22:30:02 +00:00
if($photos) {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("update xchan set xchan_photo_date = '%s', xchan_photo_l = '%s', xchan_photo_m = '%s', xchan_photo_s = '%s', xchan_photo_mimetype = '%s' where xchan_url = '%s' and xchan_network = 'rss'",
project "snakebite"
2014-02-17 22:30:02 +00:00
dbesc(datetime_convert('UTC','UTC',$arr['photo_updated'])),
dbesc($photos[0]),
dbesc($photos[1]),
dbesc($photos[2]),
dbesc($photos[3]),
dbesc($x['url'])
);
if($r)
return $x['url'];
}
}
It's really not supposed to be quite this chatty on the network - ah that's why. Oops.
2013-01-29 04:51:37 +00:00
include re-organisation and more doco, post_to_red fix ampersands in categories
2013-12-04 08:19:29 +00:00
return false;
project "snakebite"
2014-02-17 22:30:02 +00:00
and there's the import side of things
2012-11-17 20:55:59 +00:00
}
get_item_elements (json version)
2012-07-30 05:43:51 +00:00
correct item author for feeds with multiple or different authors
2014-09-12 00:27:49 +00:00
function import_author_unknown($x) {
if(! $x['url'])
return false;
$r = q("select xchan_hash from xchan where xchan_network = 'unknown' and xchan_url = '%s' limit 1",
dbesc($x['url'])
);
if($r) {
logger('import_author_unknown: in cache' , LOGGER_DEBUG);
return $r[0]['xchan_hash'];
}
$name = trim($x['name']);
$r = q("insert into xchan ( xchan_hash, xchan_guid, xchan_url, xchan_name, xchan_network )
values ( '%s', '%s', '%s', '%s', '%s' )",
dbesc($x['url']),
dbesc($x['url']),
dbesc($x['url']),
dbesc(($name) ? $name : t('(Unknown)')),
dbesc('unknown')
);
if($r && $x['photo']) {
$photos = import_profile_photo($x['photo']['src'],$x['url']);
if($photos) {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("update xchan set xchan_photo_date = '%s', xchan_photo_l = '%s', xchan_photo_m = '%s', xchan_photo_s = '%s', xchan_photo_mimetype = '%s' where xchan_url = '%s' and xchan_network = 'unknown'",
correct item author for feeds with multiple or different authors
2014-09-12 00:27:49 +00:00
dbesc(datetime_convert('UTC','UTC',$arr['photo_updated'])),
dbesc($photos[0]),
dbesc($photos[1]),
dbesc($photos[2]),
dbesc($photos[3]),
dbesc($x['url'])
);
if($r)
return $x['url'];
}
}
return false;
}
use the more portable encoded_item format for exported items - but with added attributes so we can use it as a reasonably complete item backup. The encoded_item format gives us extended author and owner information in case we need to probe them to bring the entry back. It also contains taxonomy entries. Importing and/or recovering will best be accomplished in chunks. It could take some time and some memory to chew through this.
2014-09-16 10:33:48 +00:00
function encode_item($item,$mirror = false) {
start on item feed structures
2012-11-15 07:09:25 +00:00
$x = array();
"heavy lifting" - zot messages flowing, now just need to be parsed and stored at the other end.
2012-11-20 04:45:12 +00:00
$x['type'] = 'activity';
add an encoding type to formatted zot communication structures so that we can later add alternate encodings.
2014-07-24 22:55:01 +00:00
$x['encoding'] = 'zot';
start on item feed structures
2012-11-15 07:09:25 +00:00
post signatures
2013-10-03 04:04:48 +00:00
// logger('encode_item: ' . print_r($item,true));
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
found it - but I'll do some more testing before undoing the bypass
2013-06-17 06:20:36 +00:00
$r = q("select channel_r_stream, channel_w_comment from channel where channel_id = %d limit 1",
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
intval($item['uid'])
);
a couple of "not quite public" fixes
2014-08-08 22:27:17 +00:00
if($r)
We will need a per item comment policy to be able to determine in advance if we have permission to comment on something , and we'll need to send it out with all communications. The current check is not only flawed but also a huge performance hit. Also provide the ability for an item to disable commenting completely - such as for a webpage or wherever you want to prevent comments on one item, without requiring a change to your entire permission scheme. All of this is only partially implemented at the moment but we need the structures in place on several sites in order to finish it without breaking everything.
2013-06-17 02:14:01 +00:00
$comment_scope = $r[0]['channel_w_comment'];
a couple of "not quite public" fixes
2014-08-08 22:27:17 +00:00
else
We will need a per item comment policy to be able to determine in advance if we have permission to comment on something , and we'll need to send it out with all communications. The current check is not only flawed but also a huge performance hit. Also provide the ability for an item to disable commenting completely - such as for a webpage or wherever you want to prevent comments on one item, without requiring a change to your entire permission scheme. All of this is only partially implemented at the moment but we need the structures in place on several sites in order to finish it without breaking everything.
2013-06-17 02:14:01 +00:00
$comment_scope = 0;
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
booboo
2014-08-09 22:21:15 +00:00
$scope = $item['public_policy'];
a couple of "not quite public" fixes
2014-08-08 22:27:17 +00:00
if(! $scope)
$scope = 'public';
We will need a per item comment policy to be able to determine in advance if we have permission to comment on something , and we'll need to send it out with all communications. The current check is not only flawed but also a huge performance hit. Also provide the ability for an item to disable commenting completely - such as for a webpage or wherever you want to prevent comments on one item, without requiring a change to your entire permission scheme. All of this is only partially implemented at the moment but we need the structures in place on several sites in order to finish it without breaking everything.
2013-06-17 02:14:01 +00:00
$c_scope = map_scope($comment_scope);
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
use the more portable encoded_item format for exported items - but with added attributes so we can use it as a reasonably complete item backup. The encoded_item format gives us extended author and owner information in case we need to probe them to bring the entry back. It also contains taxonomy entries. Importing and/or recovering will best be accomplished in chunks. It could take some time and some memory to chew through this.
2014-09-16 10:33:48 +00:00
$key = get_config('system','prvkey');
encrypt private messages on disk - there are still a couple of places where the text is leaked in the logs during processing.
2013-07-29 04:04:03 +00:00
if(array_key_exists('item_flags',$item) && ($item['item_flags'] & ITEM_OBSCURED)) {
if($item['title'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$item['title'] = crypto_unencapsulate(json_decode_plus($item['title']),$key);
encrypt private messages on disk - there are still a couple of places where the text is leaked in the logs during processing.
2013-07-29 04:04:03 +00:00
if($item['body'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$item['body'] = crypto_unencapsulate(json_decode_plus($item['body']),$key);
encrypt private messages on disk - there are still a couple of places where the text is leaked in the logs during processing.
2013-07-29 04:04:03 +00:00
}
use the more portable encoded_item format for exported items - but with added attributes so we can use it as a reasonably complete item backup. The encoded_item format gives us extended author and owner information in case we need to probe them to bring the entry back. It also contains taxonomy entries. Importing and/or recovering will best be accomplished in chunks. It could take some time and some memory to chew through this.
2014-09-16 10:33:48 +00:00
// If we're trying to backup an item so that it's recoverable or for export/imprt,
// add all the attributes we need to recover it
if($mirror) {
$x['id'] = $item['id'];
$x['parent'] = $item['parent'];
$x['uid'] = $item['uid'];
$x['allow_cid'] = $item['allow_cid'];
$x['allow_gid'] = $item['allow_gid'];
$x['deny_cid'] = $item['deny_cid'];
$x['deny_gid'] = $item['deny_gid'];
$x['revision'] = $item['revision'];
$x['layout_mid'] = $item['layout_mid'];
$x['postopts'] = $item['postopts'];
$x['resource_id'] = $item['resource_id'];
$x['resource_type'] = $item['resource_type'];
$x['item_restrict'] = $item['item_restrict'];
$x['item_flags'] = $item['item_flags'];
$x['attach'] = $item['attach'];
}
finish basic item encoding, add plugin compatibility tag, notes on improving $global_perms for edge cases
2012-11-16 21:57:55 +00:00
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$x['message_id'] = $item['mid'];
$x['message_top'] = $item['parent_mid'];
$x['message_parent'] = $item['thr_parent'];
$x['created'] = $item['created'];
$x['edited'] = $item['edited'];
prevent database hackery from leaking over the wire
2014-11-22 19:54:40 +00:00
// always send 0's over the wire
$x['expires'] = (($item['expires'] == '0001-01-01 00:00:00') ? '0000-00-00 00:00:00' : $item['expires']);
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$x['commented'] = $item['commented'];
$x['mimetype'] = $item['mimetype'];
$x['title'] = $item['title'];
$x['body'] = $item['body'];
$x['app'] = $item['app'];
$x['verb'] = $item['verb'];
$x['object_type'] = $item['obj_type'];
$x['target_type'] = $item['tgt_type'];
$x['permalink'] = $item['plink'];
$x['location'] = $item['location'];
$x['longlat'] = $item['coord'];
$x['signature'] = $item['sig'];
$x['route'] = $item['route'];
$x['owner'] = encode_item_xchan($item['owner']);
$x['author'] = encode_item_xchan($item['author']);
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
if($item['object'])
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$x['object'] = json_decode_plus($item['object']);
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
if($item['target'])
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$x['target'] = json_decode_plus($item['target']);
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
if($item['attach'])
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$x['attach'] = json_decode_plus($item['attach']);
finish basic item encoding, add plugin compatibility tag, notes on improving $global_perms for edge cases
2012-11-16 21:57:55 +00:00
if($y = encode_item_flags($item))
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$x['flags'] = $y;
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
if($item['comments_closed'] !== NULL_DATE)
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$x['comments_closed'] = $item['comments_closed'];
$x['public_scope'] = $scope;
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
We will need a per item comment policy to be able to determine in advance if we have permission to comment on something , and we'll need to send it out with all communications. The current check is not only flawed but also a huge performance hit. Also provide the ability for an item to disable commenting completely - such as for a webpage or wherever you want to prevent comments on one item, without requiring a change to your entire permission scheme. All of this is only partially implemented at the moment but we need the structures in place on several sites in order to finish it without breaking everything.
2013-06-17 02:14:01 +00:00
if($item['item_flags'] & ITEM_NOCOMMENT)
$x['comment_scope'] = 'none';
else
$x['comment_scope'] = $c_scope;
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
if($item['term'])
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$x['tags'] = encode_item_terms($item['term']);
start on item feed structures
2012-11-15 07:09:25 +00:00
issues with diaspora comment signatures on relayed comments that are relayed through a redmatrix site (parent post is redmatrix) and involve a private post.
2014-10-02 09:43:07 +00:00
if($item['diaspora_meta'])
$x['diaspora_signature'] = crypto_unencapsulate(json_decode($item['diaspora_meta'],true),$key);
cheange encode_item() logging to logger_data because it's way too noisy and had no loglevel set
2014-03-24 01:26:48 +00:00
logger('encode_item: ' . print_r($x,true), LOGGER_DATA);
post signatures
2013-10-03 04:04:48 +00:00
start on item feed structures
2012-11-15 07:09:25 +00:00
return $x;
registration dangling code fragment that should've been removed
2012-08-16 06:15:29 +00:00
}
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
basic 'notpublic' bits working
2014-08-07 02:24:46 +00:00
function map_scope($scope,$strip = false) {
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
switch($scope) {
public scope
2013-02-15 22:13:58 +00:00
case 0:
return 'self';
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
case PERMS_PUBLIC:
basic 'notpublic' bits working
2014-08-07 02:24:46 +00:00
if($strip)
return '';
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
return 'public';
case PERMS_NETWORK:
return 'network: red';
basic 'notpublic' bits working
2014-08-07 02:24:46 +00:00
case PERMS_AUTHED:
return 'authenticated';
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
case PERMS_SITE:
return 'site: ' . get_app()->get_hostname();
Implements a new permission level PERMS_PENDING With this a user can allow some action to any user which connects to them, even before they've connected back. Ref. https://mobiliza.org.br/display/478d9e71eaf55748dc646d3990651d6d34cfb7db5c38360538ec730ca3ccf908@zothub.com Also some code cleanup and an alternative logic for handling notifications of permission changes in zot.php. This assumes that private posts are still restricted to people in your addressbook. Regardless of your global permissions, a pending channel won't get private posts, even if the post only has a deny clause not matching the pending channel.
2014-06-17 07:26:18 +00:00
case PERMS_PENDING:
return 'any connections';
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
case PERMS_CONTACTS:
default:
return 'contacts';
}
}
basic 'notpublic' bits working
2014-08-07 02:24:46 +00:00
function translate_scope($scope) {
if(! $scope || $scope === 'public')
cleanup some text where we said "public" or "visible to anybody on the internet" when that isn't necessarily true.
2014-08-08 01:22:50 +00:00
return t('Visible to anybody on the internet.');
basic 'notpublic' bits working
2014-08-07 02:24:46 +00:00
if(strpos($scope,'self') === 0)
return t('Visible to you only.');
if(strpos($scope,'network:') === 0)
return t('Visible to anybody in this network.');
if(strpos($scope,'authenticated') === 0)
return t('Visible to anybody authenticated.');
if(strpos($scope,'site:') === 0)
return sprintf( t('Visible to anybody on %s.'), strip_tags(substr($scope,6)));
if(strpos($scope,'any connections') === 0)
return t('Visible to all connections.');
if(strpos($scope,'contacts') === 0)
return t('Visible to approved connections.');
let's make this "specific" since that's what it is.
2014-08-28 11:14:49 +00:00
if(strpos($scope,'specific') === 0)
return t('Visible to specific connections.');
basic 'notpublic' bits working
2014-08-07 02:24:46 +00:00
}
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
function encode_item_xchan($xchan) {
registration dangling code fragment that should've been removed
2012-08-16 06:15:29 +00:00
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
$ret = array();
$ret['name'] = $xchan['xchan_name'];
$ret['address'] = $xchan['xchan_addr'];
import_author_rss not succeeding, missing url
2014-09-05 02:15:01 +00:00
$ret['url'] = (($xchan['hubloc_url']) ? $xchan['hubloc_url'] : $xchan['xchan_url']);
include re-organisation and more doco, post_to_red fix ampersands in categories
2013-12-04 08:19:29 +00:00
$ret['network'] = $xchan['xchan_network'];
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
$ret['photo'] = array('mimetype' => $xchan['xchan_photo_mimetype'], 'src' => $xchan['xchan_photo_m']);
$ret['guid'] = $xchan['xchan_guid'];
$ret['guid_sig'] = $xchan['xchan_guid_sig'];
return $ret;
}
registration dangling code fragment that should've been removed
2012-08-16 06:15:29 +00:00
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
function encode_item_terms($terms) {
$ret = array();
transmit, receive, and parse bookmarks
2014-02-04 03:38:15 +00:00
$allowed_export_terms = array( TERM_UNKNOWN, TERM_HASHTAG, TERM_MENTION, TERM_CATEGORY, TERM_BOOKMARK );
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
if($terms) {
foreach($terms as $term) {
if(in_array($term['type'],$allowed_export_terms))
and there's the import side of things
2012-11-17 20:55:59 +00:00
$ret[] = array('tag' => $term['term'], 'url' => $term['url'], 'type' => termtype($term['type']));
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
}
}
return $ret;
}
function termtype($t) {
transmit, receive, and parse bookmarks
2014-02-04 03:38:15 +00:00
$types = array('unknown','hashtag','mention','category','private_category','file','search','thing','bookmark');
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
return(($types[$t]) ? $types[$t] : 'unknown');
}
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
and there's the import side of things
2012-11-17 20:55:59 +00:00
function decode_tags($t) {
if($t) {
$ret = array();
foreach($t as $x) {
$tag = array();
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$tag['term'] = htmlspecialchars($x['tag'], ENT_COMPAT,'UTF-8',false);
$tag['url'] = htmlspecialchars($x['url'], ENT_COMPAT,'UTF-8',false);
and there's the import side of things
2012-11-17 20:55:59 +00:00
switch($x['type']) {
case 'hashtag':
$tag['type'] = TERM_HASHTAG;
break;
case 'mention':
$tag['type'] = TERM_MENTION;
break;
case 'category':
$tag['type'] = TERM_CATEGORY;
break;
case 'private_category':
$tag['type'] = TERM_PCATEGORY;
break;
case 'file':
$tag['type'] = TERM_FILE;
break;
case 'search':
$tag['type'] = TERM_SEARCH;
break;
transmit, receive, and parse bookmarks
2014-02-04 03:38:15 +00:00
case 'thing':
$tag['type'] = TERM_THING;
break;
case 'bookmark':
$tag['type'] = TERM_BOOKMARK;
break;
and there's the import side of things
2012-11-17 20:55:59 +00:00
default:
case 'unknown':
$tag['type'] = TERM_UNKNOWN;
break;
}
$ret[] = $tag;
}
return $ret;
}
return '';
}
profile messages for directory synchronisation
2012-12-22 12:33:32 +00:00
// santise a potentially complex array
and there's the import side of things
2012-11-17 20:55:59 +00:00
function activity_sanitise($arr) {
if($arr) {
Another try at issue #61 and #62 - an earlier fix was partially working but the issue persisted - this extends it a bit.
2013-08-18 07:04:38 +00:00
if(is_array($arr)) {
$ret = array();
foreach($arr as $k => $x) {
if(is_array($x))
$ret[$k] = activity_sanitise($x);
else
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$ret[$k] = htmlspecialchars($x, ENT_COMPAT,'UTF-8',false);
Another try at issue #61 and #62 - an earlier fix was partially working but the issue persisted - this extends it a bit.
2013-08-18 07:04:38 +00:00
}
return $ret;
}
else {
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
return htmlspecialchars($arr, ENT_COMPAT,'UTF-8', false);
and there's the import side of things
2012-11-17 20:55:59 +00:00
}
}
return '';
}
finish basic item encoding, add plugin compatibility tag, notes on improving $global_perms for edge cases
2012-11-16 21:57:55 +00:00
profile messages for directory synchronisation
2012-12-22 12:33:32 +00:00
// sanitise a simple linear array
function array_sanitise($arr) {
if($arr) {
$ret = array();
foreach($arr as $x) {
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$ret[] = htmlspecialchars($x, ENT_COMPAT,'UTF-8',false);
profile messages for directory synchronisation
2012-12-22 12:33:32 +00:00
}
return $ret;
}
return '';
}
finish basic item encoding, add plugin compatibility tag, notes on improving $global_perms for edge cases
2012-11-16 21:57:55 +00:00
function encode_item_flags($item) {
// most of item_flags and item_restrict are local settings which don't apply when transmitted.
// We may need those for the case of syncing other hub locations which you are attached to.
$ret = array();
move the deleted flag setting to encode_item_flags() where it belongs now that we aren't giving deleted posts special treatment.
2014-04-10 23:26:27 +00:00
if($item['item_restrict'] & ITEM_DELETED)
$ret[] = 'deleted';
basic changes for file activity
2015-01-09 14:18:45 +00:00
if($item['item_restrict'] & ITEM_HIDDEN)
$ret[] = 'hidden';
more heavy lifting on delivery - now we can find all the recipients on the system who have any acceptable permissions for receiving a post. Next we just need to process edits and deletes, and otherwise store the message - yay. Still a lot of work, but we've come a long way baby.
2012-11-27 00:48:04 +00:00
if($item['item_flags'] & ITEM_THREAD_TOP)
finish basic item encoding, add plugin compatibility tag, notes on improving $global_perms for edge cases
2012-11-16 21:57:55 +00:00
$ret[] = 'thread_parent';
more heavy lifting on delivery - now we can find all the recipients on the system who have any acceptable permissions for receiving a post. Next we just need to process edits and deletes, and otherwise store the message - yay. Still a lot of work, but we've come a long way baby.
2012-11-27 00:48:04 +00:00
if($item['item_flags'] & ITEM_NSFW)
finish basic item encoding, add plugin compatibility tag, notes on improving $global_perms for edge cases
2012-11-16 21:57:55 +00:00
$ret[] = 'nsfw';
fix item_private flag
2012-11-27 01:07:23 +00:00
if($item['item_private'])
mostly working item_decode, but without full input santising
2012-11-17 10:29:02 +00:00
$ret[] = 'private';
finish basic item encoding, add plugin compatibility tag, notes on improving $global_perms for edge cases
2012-11-16 21:57:55 +00:00
return $ret;
}
y'all got mail
2012-12-06 00:44:07 +00:00
function encode_mail($item) {
$x = array();
$x['type'] = 'mail';
add an encoding type to formatted zot communication structures so that we can later add alternate encodings.
2014-07-24 22:55:01 +00:00
$x['encoding'] = 'zot';
y'all got mail
2012-12-06 00:44:07 +00:00
better mail obscuring
2013-07-31 09:32:41 +00:00
if(array_key_exists('mail_flags',$item) && ($item['mail_flags'] & MAIL_OBSCURED)) {
$key = get_config('system','prvkey');
if($item['title'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$item['title'] = crypto_unencapsulate(json_decode_plus($item['title']),$key);
better mail obscuring
2013-07-31 09:32:41 +00:00
if($item['body'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$item['body'] = crypto_unencapsulate(json_decode_plus($item['body']),$key);
better mail obscuring
2013-07-31 09:32:41 +00:00
}
y'all got mail
2012-12-06 00:44:07 +00:00
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$x['message_id'] = $item['mid'];
$x['message_parent'] = $item['parent_mid'];
y'all got mail
2012-12-06 00:44:07 +00:00
$x['created'] = $item['created'];
allow private mail sender to set an expiration on their messages. Once expired the message is destroyed at both ends (subject to the granularity of the polling interval) and is gone. Officially it takes some form of language independent string like 2013/11/22, but English speakers can use anything that strtotime() understands, like "+30 minutes" or "next Tuesday".
2013-11-07 02:28:36 +00:00
$x['expires'] = $item['expires'];
store diaspora meta info in the item table. It has to go there or it will kill us with complex joins. We can phase out the sign table once this all checks out.
2014-09-03 12:09:43 +00:00
$x['diaspora_meta'] = $item['diaspora_meta'];
y'all got mail
2012-12-06 00:44:07 +00:00
$x['title'] = $item['title'];
$x['body'] = $item['body'];
$x['from'] = encode_item_xchan($item['from']);
$x['to'] = encode_item_xchan($item['to']);
email attachments (and fix email photos)
2013-08-20 04:19:39 +00:00
if($item['attach'])
$x['attach'] = json_decode_plus($item['attach']);
private mail is just a little more private now. Not encrypted and the obfuscation is easily reversible, but not casually readable by browsing logfiles or mysql dumps. This isn't backward compatible - folks will have to upgrade if they can't read their mail.
2013-07-19 09:45:44 +00:00
$x['flags'] = array();
mail recall - just set mail_flags = mail_flags & MAIL_RECALLED and tell the notifier. We'll need a special button for this. Also should probably report whether it was seen or not. Will do that another day.
2013-02-28 04:38:33 +00:00
if($item['mail_flags'] & MAIL_RECALLED) {
private mail is just a little more private now. Not encrypted and the obfuscation is easily reversible, but not casually readable by browsing logfiles or mysql dumps. This isn't backward compatible - folks will have to upgrade if they can't read their mail.
2013-07-19 09:45:44 +00:00
$x['flags'][] = 'recalled';
mail recall - just set mail_flags = mail_flags & MAIL_RECALLED and tell the notifier. We'll need a special button for this. Also should probably report whether it was seen or not. Will do that another day.
2013-02-28 04:38:33 +00:00
$x['title'] = '';
$x['body'] = '';
}
y'all got mail
2012-12-06 00:44:07 +00:00
return $x;
}
function get_mail_elements($x) {
$arr = array();
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['body'] = (($x['body']) ? htmlspecialchars($x['body'], ENT_COMPAT,'UTF-8',false) : '');
$arr['title'] = (($x['title'])? htmlspecialchars($x['title'],ENT_COMPAT,'UTF-8',false) : '');
y'all got mail
2012-12-06 00:44:07 +00:00
$arr['created'] = datetime_convert('UTC','UTC',$x['created']);
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
if((! array_key_exists('expires',$x)) || ($x['expires'] === NULL_DATE))
$arr['expires'] = NULL_DATE;
allow private mail sender to set an expiration on their messages. Once expired the message is destroyed at both ends (subject to the granularity of the polling interval) and is gone. Officially it takes some form of language independent string like 2013/11/22, but English speakers can use anything that strtotime() understands, like "+30 minutes" or "next Tuesday".
2013-11-07 02:28:36 +00:00
else
$arr['expires'] = datetime_convert('UTC','UTC',$x['expires']);
y'all got mail
2012-12-06 00:44:07 +00:00
mail recall - just set mail_flags = mail_flags & MAIL_RECALLED and tell the notifier. We'll need a special button for this. Also should probably report whether it was seen or not. Will do that another day.
2013-02-28 04:38:33 +00:00
$arr['mail_flags'] = 0;
if($x['flags'] && is_array($x['flags'])) {
if(in_array('recalled',$x['flags'])) {
private mail is just a little more private now. Not encrypted and the obfuscation is easily reversible, but not casually readable by browsing logfiles or mysql dumps. This isn't backward compatible - folks will have to upgrade if they can't read their mail.
2013-07-19 09:45:44 +00:00
$arr['mail_flags'] |= MAIL_RECALLED;
}
mail recall - just set mail_flags = mail_flags & MAIL_RECALLED and tell the notifier. We'll need a special button for this. Also should probably report whether it was seen or not. Will do that another day.
2013-02-28 04:38:33 +00:00
}
better mail obscuring
2013-07-31 09:32:41 +00:00
$key = get_config('system','pubkey');
$arr['mail_flags'] |= MAIL_OBSCURED;
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['body'] = htmlspecialchars($arr['body'],ENT_COMPAT,'UTF-8',false);
better mail obscuring
2013-07-31 09:32:41 +00:00
if($arr['body'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$arr['body'] = json_encode(crypto_encapsulate($arr['body'],$key));
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['title'] = htmlspecialchars($arr['title'],ENT_COMPAT,'UTF-8',false);
better mail obscuring
2013-07-31 09:32:41 +00:00
if($arr['title'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$arr['title'] = json_encode(crypto_encapsulate($arr['title'],$key));
mail recall - just set mail_flags = mail_flags & MAIL_RECALLED and tell the notifier. We'll need a special button for this. Also should probably report whether it was seen or not. Will do that another day.
2013-02-28 04:38:33 +00:00
y'all got mail
2012-12-06 00:44:07 +00:00
if($arr['created'] > datetime_convert())
$arr['created'] = datetime_convert();
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['mid'] = (($x['message_id']) ? htmlspecialchars($x['message_id'], ENT_COMPAT,'UTF-8',false) : '');
$arr['parent_mid'] = (($x['message_parent']) ? htmlspecialchars($x['message_parent'], ENT_COMPAT,'UTF-8',false) : '');
y'all got mail
2012-12-06 00:44:07 +00:00
email attachments (and fix email photos)
2013-08-20 04:19:39 +00:00
if($x['attach'])
$arr['attach'] = activity_sanitise($x['attach']);
include re-organisation and more doco, post_to_red fix ampersands in categories
2013-12-04 08:19:29 +00:00
if(($xchan_hash = import_author_xchan($x['from'])) !== false)
$arr['from_xchan'] = $xchan_hash;
y'all got mail
2012-12-06 00:44:07 +00:00
else
return array();
include re-organisation and more doco, post_to_red fix ampersands in categories
2013-12-04 08:19:29 +00:00
if(($xchan_hash = import_author_xchan($x['to'])) !== false)
$arr['to_xchan'] = $xchan_hash;
y'all got mail
2012-12-06 00:44:07 +00:00
else
return array();
return $arr;
}
profile messages for directory synchronisation
2012-12-22 12:33:32 +00:00
function get_profile_elements($x) {
$arr = array();
include re-organisation and more doco, post_to_red fix ampersands in categories
2013-12-04 08:19:29 +00:00
if(($xchan_hash = import_author_xchan($x['from'])) !== false)
$arr['xprof_hash'] = $xchan_hash;
profile messages for directory synchronisation
2012-12-22 12:33:32 +00:00
else
return array();
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['desc'] = (($x['title']) ? htmlspecialchars($x['title'],ENT_COMPAT,'UTF-8',false) : '');
profile messages for directory synchronisation
2012-12-22 12:33:32 +00:00
$arr['dob'] = datetime_convert('UTC','UTC',$x['birthday'],'Y-m-d');
add age to directory profile - requires updating on each birthday and that part is still missing
2013-06-30 07:38:02 +00:00
$arr['age'] = (($x['age']) ? intval($x['age']) : 0);
profile messages for directory synchronisation
2012-12-22 12:33:32 +00:00
issue #240 - we were using htmlentities instead of htmlspecialchars in several places, and this was a bit greedy in the set of characters which were converted from utf-8 to HTML entities. Also brought mail attachments up to date so they are rendered identically to item attachments.
2013-12-12 07:13:36 +00:00
$arr['gender'] = (($x['gender']) ? htmlspecialchars($x['gender'], ENT_COMPAT,'UTF-8',false) : '');
$arr['marital'] = (($x['marital']) ? htmlspecialchars($x['marital'], ENT_COMPAT,'UTF-8',false) : '');
$arr['sexual'] = (($x['sexual']) ? htmlspecialchars($x['sexual'], ENT_COMPAT,'UTF-8',false) : '');
$arr['locale'] = (($x['locale']) ? htmlspecialchars($x['locale'], ENT_COMPAT,'UTF-8',false) : '');
$arr['region'] = (($x['region']) ? htmlspecialchars($x['region'], ENT_COMPAT,'UTF-8',false) : '');
$arr['postcode'] = (($x['postcode']) ? htmlspecialchars($x['postcode'], ENT_COMPAT,'UTF-8',false) : '');
$arr['country'] = (($x['country']) ? htmlspecialchars($x['country'], ENT_COMPAT,'UTF-8',false) : '');
profile messages for directory synchronisation
2012-12-22 12:33:32 +00:00
$arr['keywords'] = (($x['keywords'] && is_array($x['keywords'])) ? array_sanitise($x['keywords']) : array());
return $arr;
}
y'all got mail
2012-12-06 00:44:07 +00:00
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
function get_atom_elements($feed,$item,&$author) {
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
$best_photo = array();
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
$res = array();
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$found_author = $item->get_author();
if($found_author) {
$author['author_name'] = unxmlify($found_author->get_name());
$author['author_link'] = unxmlify($found_author->get_link());
if no rss/atom element author, use feed level contact
2014-09-24 21:27:18 +00:00
$author['author_is_feed'] = false;
better handling of troublesome feeds.
2011-04-05 02:36:18 +00:00
}
else {
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_name'] = unxmlify($feed->get_title());
$author['author_link'] = unxmlify($feed->get_permalink());
if no rss/atom element author, use feed level contact
2014-09-24 21:27:18 +00:00
$author['author_is_feed'] = true;
better handling of troublesome feeds.
2011-04-05 02:36:18 +00:00
}
snakebite continued
2014-08-20 04:31:33 +00:00
fix url mismatches of a single trailing slash for feed author/owner comparisons; update diaspora_compat
2014-09-24 03:55:22 +00:00
if(substr($author['author_link'],-1,1) == '/')
$author['author_link'] = substr($author['author_link'],0,-1);
encode rss message-id's - warning this will double store all existing rss feed items and probably make a mess of things
2014-09-09 23:28:33 +00:00
$res['mid'] = base64url_encode(unxmlify($item->get_id()));
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
$res['title'] = unxmlify($item->get_title());
$res['body'] = unxmlify($item->get_content());
provide permalink for posts, include permalinks in feeds and imported from feeds
2011-02-16 10:07:11 +00:00
$res['plink'] = unxmlify($item->get_link(0));
diaspora private mail seems to work now - but there's an obfuscation leak via the conversation structure that needs to be dealt with.
2014-09-24 03:36:10 +00:00
$res['item_flags'] = ITEM_RSS;
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
eliminate the obsolete "last-child" indicator
2012-07-19 00:08:03 +00:00
// removing the content of the title if its identically to the body
// This helps with auto generated titles e.g. from tumblr
wrong attribution on likes of comments in email notifications. Let's also call the thing they liked a comment instead of a status
2013-05-28 08:40:27 +00:00
eliminate the obsolete "last-child" indicator
2012-07-19 00:08:03 +00:00
if (title_is_body($res["title"], $res["body"]))
$res['title'] = "";
use reltoabs - primarily for github activity feeds
2012-01-18 00:30:32 +00:00
if($res['plink'])
$base_url = implode('/', array_slice(explode('/',$res['plink']),0,3));
else
$base_url = '';
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
// look for a photo. We should check media size and find the best one,
// but for now let's just find any author photo
$rawauthor = $item->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_10,'author');
if($rawauthor && $rawauthor[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['link']) {
$base = $rawauthor[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['link'];
foreach($base as $link) {
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
if(!x($author, 'author_photo') || ! $author['author_photo']) {
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
if($link['attribs']['']['rel'] === 'photo' || $link['attribs']['']['rel'] === 'avatar')
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_photo'] = unxmlify($link['attribs']['']['href']);
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
}
}
eliminate the obsolete "last-child" indicator
2012-07-19 00:08:03 +00:00
}
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
$rawactor = $item->get_item_tags(NAMESPACE_ACTIVITY, 'actor');
item table rename parent-uri, target-type, object-type (more to come later)
2012-08-16 05:33:37 +00:00
if($rawactor && activity_match($rawactor[0]['child'][NAMESPACE_ACTIVITY]['obj_type'][0]['data'],ACTIVITY_OBJ_PERSON)) {
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
$base = $rawactor[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['link'];
if($base && count($base)) {
foreach($base as $link) {
snakebite continued
2014-08-20 04:31:33 +00:00
if($link['attribs']['']['rel'] === 'alternate' && (! $res['author_link']))
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_link'] = unxmlify($link['attribs']['']['href']);
if(!x($author, 'author_photo') || ! $author['author_photo']) {
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
if($link['attribs']['']['rel'] === 'avatar' || $link['attribs']['']['rel'] === 'photo')
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_photo'] = unxmlify($link['attribs']['']['href']);
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
}
}
}
}
correct item author for feeds with multiple or different authors
2014-09-12 00:27:49 +00:00
// check for a yahoo media element (github etc.)
if(! $author['author_photo']) {
$rawmedia = $item->get_item_tags(NAMESPACE_YMEDIA,'thumbnail');
if($rawmedia && $rawmedia[0]['attribs']['']['url']) {
$author['author_photo'] = strip_tags(unxmlify($rawmedia[0]['attribs']['']['url']));
}
}
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
// No photo/profile-link on the item - look at the feed level
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
if((! (x($author,'author_link'))) || (! (x($author,'author_photo')))) {
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
$rawauthor = $feed->get_feed_tags(SIMPLEPIE_NAMESPACE_ATOM_10,'author');
if($rawauthor && $rawauthor[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['link']) {
$base = $rawauthor[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['link'];
foreach($base as $link) {
if no rss/atom element author, use feed level contact
2014-09-24 21:27:18 +00:00
if($link['attribs']['']['rel'] === 'alternate' && (! $author['author_link'])) {
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_link'] = unxmlify($link['attribs']['']['href']);
if no rss/atom element author, use feed level contact
2014-09-24 21:27:18 +00:00
$author['author_is_feed'] = true;
}
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
if(! $author['author_photo']) {
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
if($link['attribs']['']['rel'] === 'photo' || $link['attribs']['']['rel'] === 'avatar')
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_photo'] = unxmlify($link['attribs']['']['href']);
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
}
}
eliminate the obsolete "last-child" indicator
2012-07-19 00:08:03 +00:00
}
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
$rawactor = $feed->get_feed_tags(NAMESPACE_ACTIVITY, 'subject');
item table rename parent-uri, target-type, object-type (more to come later)
2012-08-16 05:33:37 +00:00
if($rawactor && activity_match($rawactor[0]['child'][NAMESPACE_ACTIVITY]['obj_type'][0]['data'],ACTIVITY_OBJ_PERSON)) {
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
$base = $rawactor[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['link'];
if($base && count($base)) {
foreach($base as $link) {
snakebite continued
2014-08-20 04:31:33 +00:00
if($link['attribs']['']['rel'] === 'alternate' && (! $res['author_link']))
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_link'] = unxmlify($link['attribs']['']['href']);
if(! (x($author,'author_photo'))) {
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
if($link['attribs']['']['rel'] === 'avatar' || $link['attribs']['']['rel'] === 'photo')
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_photo'] = unxmlify($link['attribs']['']['href']);
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
}
}
}
}
}
item source
2011-06-21 02:08:40 +00:00
$apps = $item->get_item_tags(NAMESPACE_STATUSNET,'notice_info');
if($apps && $apps[0]['attribs']['']['source']) {
don't allow tags in app source
2011-06-23 22:58:43 +00:00
$res['app'] = strip_tags(unxmlify($apps[0]['attribs']['']['source']));
item source
2011-06-21 02:08:40 +00:00
}
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
provide ability to survive transport mangling when travelling through html networks
2011-02-01 13:10:04 +00:00
/**
* If there's a copy of the body content which is guaranteed to have survived mangling in transit, use it.
*/
$have_real_body = false;
$rawenv = $item->get_item_tags(NAMESPACE_DFRN, 'env');
if($rawenv) {
$have_real_body = true;
$res['body'] = $rawenv[0]['data'];
$res['body'] = str_replace(array(' ',"\t","\r","\n"), array('','','',''),$res['body']);
make sure content is unaltered
2011-02-02 01:50:49 +00:00
// make sure nobody is trying to sneak some html tags by us
$res['body'] = notags(base64url_decode($res['body']));
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
// We could probably turn these old Friendica bbcode bookmarks into bookmark tags but we'd have to
// create a term table item for them. For now just make sure they stay as links.
regex missing terminator char
2014-09-29 22:41:47 +00:00
$res['body'] = preg_replace('/\[bookmark(.*?)\](.*?)\[\/bookmark\]/','[url$1]$2[/url]',$res['body']);
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
provide ability to survive transport mangling when travelling through html networks
2011-02-01 13:10:04 +00:00
}
add function to exclude embedded images when checking message length
2012-07-06 23:09:23 +00:00
$res['body'] = limit_body_size($res['body']);
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
// It isn't certain at this point whether our content is plaintext or html and we'd be foolish to trust
// the content type. Our own network only emits text normally, though it might have been converted to
fix some linebreak issues
2011-01-18 03:50:18 +00:00
// html if we used a pubsubhubbub transport. But if we see even one html tag in our text, we will
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
// have to assume it is all html and needs to be purified.
// It doesn't matter all that much security wise - because before this content is used anywhere, we are
// going to escape any tags we find regardless, but this lets us import a limited subset of html from
// the wild, by sanitising it and converting supported tags to bbcode before we rip out any remaining
// html.
try slightly better detection of html in feeds
2012-04-01 10:19:37 +00:00
if((strpos($res['body'],'<') !== false) && (strpos($res['body'],'>') !== false)) {
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
use reltoabs - primarily for github activity feeds
2012-01-18 00:30:32 +00:00
$res['body'] = reltoabs($res['body'],$base_url);
stuff - background the external network notifications (Facebook, Twitter, Statusnet) - add vimeo support - consolidate editor video options (insert video, regardless of source) - add "Connector Settings" page purely for configuring connectors - moved mailbox settings to Connectors - Move the "Connect/Follow" dialogue out of "find people" and move to top. - Add "Find People" dialogue - minor theme edits on duepuntozero and testbubble to make all this stuff work - A bit of theming on duepunto notifications to eliminate the gigantic Diaspora profile photos.
2011-10-12 01:24:37 +00:00
$res['body'] = html2bb_video($res['body']);
patches for youtube's change in link and embed formats, please test
2011-05-20 22:46:41 +00:00
trying to solve the double encoding issue
2011-02-02 02:20:25 +00:00
$res['body'] = oembed_html2bbcode($res['body']);
sanitise all incoming url's - also stop them from getting mangled by simplepie
2011-02-17 01:32:15 +00:00
wrong attribution on likes of comments in email notifications. Let's also call the thing they liked a comment instead of a status
2013-05-28 08:40:27 +00:00
$res['body'] = purify_html($res['body']);
undo debug statements
2010-09-09 09:00:54 +00:00
try slightly better detection of html in feeds
2012-04-01 10:19:37 +00:00
$res['body'] = @html2bbcode($res['body']);
iframes were getting stripped at the rss/atom parser level. We can preserve these because we will strip them in the purifier after checking for supported video sources
2012-06-20 12:04:51 +00:00
fix some linebreak issues
2011-01-18 03:50:18 +00:00
}
slight improvement in handling of plaintext feeds
2012-04-01 22:43:57 +00:00
elseif(! $have_real_body) {
// it's not one of our messages and it has no tags
// so it's probably just text. We'll escape it just to be safe.
$res['body'] = escape_tags($res['body']);
}
make it the default
2014-09-24 09:18:13 +00:00
if($res['plink'] && $res['title']) {
$res['body'] = '#^[url=' . $res['plink'] . ']' . $res['title'] . '[/url]' . "\n\n" . $res['body'];
$terms = array();
$terms[] = array(
'otype' => TERM_OBJ_POST,
'type' => TERM_BOOKMARK,
'url' => $res['plink'],
'term' => $res['title'],
);
}
elseif($res['plink']) {
$res['body'] = '#^[url]' . $res['plink'] . '[/url]' . "\n\n" . $res['body'];
$terms = array();
$terms[] = array(
'otype' => TERM_OBJ_POST,
'type' => TERM_BOOKMARK,
'url' => $res['plink'],
'term' => $res['plink'],
);
just an experiment - don't do anything with this until I'm finished with it.
2014-09-24 08:45:57 +00:00
}
make it the default
2014-09-24 09:18:13 +00:00
display the fact that a conversation is private without disclosing the details of who can/cannot see it.
2010-12-08 04:47:53 +00:00
$private = $item->get_item_tags(NAMESPACE_DFRN,'private');
create third privacy state - public post but not searchable or publicly visible
2012-06-29 00:43:29 +00:00
if($private && intval($private[0]['data']) > 0)
snakebite continued
2014-08-20 04:31:33 +00:00
$res['item_private'] = ((intval($private[0]['data'])) ? 1 : 0);
display the fact that a conversation is private without disclosing the details of who can/cannot see it.
2010-12-08 04:47:53 +00:00
else
snakebite continued
2014-08-20 04:31:33 +00:00
$res['item_private'] = 0;
display the fact that a conversation is private without disclosing the details of who can/cannot see it.
2010-12-08 04:47:53 +00:00
can't use ActivityStreams for location. Grrr.
2010-08-24 23:25:02 +00:00
$rawlocation = $item->get_item_tags(NAMESPACE_DFRN, 'location');
added location tagging
2010-08-20 21:33:15 +00:00
if($rawlocation)
$res['location'] = unxmlify($rawlocation[0]['data']);
allow RSS for feed contacts, but no comments
2011-01-25 03:36:20 +00:00
$rawcreated = $item->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_10,'published');
if($rawcreated)
$res['created'] = unxmlify($rawcreated[0]['data']);
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
$rawedited = $item->get_item_tags(SIMPLEPIE_NAMESPACE_ATOM_10,'updated');
if($rawedited)
get correct article timestamp for feeds with no 'published' and with non-utc stamps
2011-02-02 00:50:19 +00:00
$res['edited'] = unxmlify($rawedited[0]['data']);
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
get correct article timestamp for feeds with no 'published' and with non-utc stamps
2011-02-02 00:50:19 +00:00
if((x($res,'edited')) && (! (x($res,'created'))))
typo
2011-02-02 01:05:17 +00:00
$res['created'] = $res['edited'];
allow RSS for feed contacts, but no comments
2011-01-25 03:36:20 +00:00
if(! $res['created'])
get correct article timestamp for feeds with no 'published' and with non-utc stamps
2011-02-02 00:50:19 +00:00
$res['created'] = $item->get_date('c');
allow RSS for feed contacts, but no comments
2011-01-25 03:36:20 +00:00
if(! $res['edited'])
get correct article timestamp for feeds with no 'published' and with non-utc stamps
2011-02-02 00:50:19 +00:00
$res['edited'] = $item->get_date('c');
allow RSS for feed contacts, but no comments
2011-01-25 03:36:20 +00:00
force last-child on foster children, disallow time travel
2011-03-15 01:34:58 +00:00
// Disallow time travelling posts
$d1 = strtotime($res['created']);
$d2 = strtotime($res['edited']);
$d3 = strtotime('now');
if($d1 > $d3)
$res['created'] = datetime_convert();
if($d2 > $d3)
$res['edited'] = datetime_convert();
various rss fixes
2014-09-01 03:51:05 +00:00
$res['created'] = datetime_convert('UTC','UTC',$res['created']);
$res['edited'] = datetime_convert('UTC','UTC',$res['edited']);
fixup image on wall
2010-08-03 02:06:36 +00:00
$rawowner = $item->get_item_tags(NAMESPACE_DFRN, 'owner');
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
if(! $rawowner)
$rawowner = $item->get_item_tags(NAMESPACE_ZOT,'owner');
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
if($rawowner[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['name'][0]['data'])
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['owner_name'] = unxmlify($rawowner[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['name'][0]['data']);
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
elseif($rawowner[0]['child'][NAMESPACE_DFRN]['name'][0]['data'])
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['owner_name'] = unxmlify($rawowner[0]['child'][NAMESPACE_DFRN]['name'][0]['data']);
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
if($rawowner[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['uri'][0]['data'])
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['owner_link'] = unxmlify($rawowner[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['uri'][0]['data']);
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
elseif($rawowner[0]['child'][NAMESPACE_DFRN]['uri'][0]['data'])
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['owner_link'] = unxmlify($rawowner[0]['child'][NAMESPACE_DFRN]['uri'][0]['data']);
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
if($rawowner[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['link']) {
$base = $rawowner[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['link'];
foreach($base as $link) {
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
if(!x($author, 'owner_photo') || ! $author['owner_photo']) {
eliminate the obsolete "last-child" indicator
2012-07-19 00:08:03 +00:00
if($link['attribs']['']['rel'] === 'photo' || $link['attribs']['']['rel'] === 'avatar')
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['owner_photo'] = unxmlify($link['attribs']['']['href']);
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
}
}
}
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
use browser geolocation when desired
2010-10-20 03:52:05 +00:00
$rawgeo = $item->get_item_tags(NAMESPACE_GEORSS,'point');
if($rawgeo)
$res['coord'] = unxmlify($rawgeo[0]['data']);
pull out follow, unfollow and as:actor for handling
2010-10-22 04:48:22 +00:00
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
$rawverb = $item->get_item_tags(NAMESPACE_ACTIVITY, 'verb');
convert received ostatus unfollow verb to activity streams stop-following
2011-01-04 10:20:00 +00:00
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
// select between supported verbs
convert received ostatus unfollow verb to activity streams stop-following
2011-01-04 10:20:00 +00:00
if($rawverb) {
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
$res['verb'] = unxmlify($rawverb[0]['data']);
convert received ostatus unfollow verb to activity streams stop-following
2011-01-04 10:20:00 +00:00
}
// translate OStatus unfollow to activity streams if it happened to get selected
if((x($res,'verb')) && ($res['verb'] === 'http://ostatus.org/schema/1.0/unfollow'))
$res['verb'] = ACTIVITY_UNFOLLOW;
atom categories and tag import
2011-04-06 00:41:02 +00:00
$cats = $item->get_categories();
if($cats) {
more testing
2014-09-24 08:58:24 +00:00
if(is_null($terms))
$terms = array();
atom categories and tag import
2011-04-06 00:41:02 +00:00
foreach($cats as $cat) {
$term = $cat->get_term();
if(! $term)
$term = $cat->get_label();
$scheme = $cat->get_scheme();
catchup
2012-07-11 11:29:47 +00:00
$termurl = '';
if($scheme && $term && stristr($scheme,'X-DFRN:')) {
$termtype = ((substr($scheme,7,1) === '#') ? TERM_HASHTAG : TERM_MENTION);
$termurl = unxmlify(substr($scheme,9));
}
else {
get out the sunnies
2014-09-01 05:15:00 +00:00
$termtype = TERM_CATEGORY;
catchup
2012-07-11 11:29:47 +00:00
}
$termterm = notags(trim(unxmlify($term)));
if($termterm) {
various rss fixes
2014-09-01 03:51:05 +00:00
$terms[] = array(
catchup
2012-07-11 11:29:47 +00:00
'otype' => TERM_OBJ_POST,
'type' => $termtype,
'url' => $termurl,
'term' => $termterm,
);
}
atom categories and tag import
2011-04-06 00:41:02 +00:00
}
}
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
one step closer
2014-09-24 09:08:35 +00:00
if(! is_null($terms))
$res['term'] = $terms;
support feed attachments/enclosures, and fix for bug #30
2011-04-07 02:41:16 +00:00
$attach = $item->get_enclosures();
if($attach) {
snakebite continued
2014-08-20 04:31:33 +00:00
$res['attach'] = array();
support feed attachments/enclosures, and fix for bug #30
2011-04-07 02:41:16 +00:00
$att_arr = array();
foreach($attach as $att) {
convert feed attach to bbcode
2011-04-13 08:53:40 +00:00
$len = intval($att->get_length());
$link = str_replace(array(',','"'),array('%2D','%22'),notags(trim(unxmlify($att->get_link()))));
$title = str_replace(array(',','"'),array('%2D','%22'),notags(trim(unxmlify($att->get_title()))));
$type = str_replace(array(',','"'),array('%2D','%22'),notags(trim(unxmlify($att->get_type()))));
if(strpos($type,';'))
$type = substr($type,0,strpos($type,';'));
support feed attachments/enclosures, and fix for bug #30
2011-04-07 02:41:16 +00:00
if((! $link) || (strpos($link,'http') !== 0))
continue;
if(! $title)
$title = ' ';
if(! $type)
$type = 'application/octet-stream';
fix phantom attachments in feeds
2014-09-02 03:59:53 +00:00
$res['attach'][] = array('href' => $link, 'length' => $len, 'type' => $type, 'title' => $title );
support feed attachments/enclosures, and fix for bug #30
2011-04-07 02:41:16 +00:00
}
}
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
$rawobj = $item->get_item_tags(NAMESPACE_ACTIVITY, 'object');
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
if($rawobj) {
snakebite continued
2014-08-20 04:31:33 +00:00
$obj = array();
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
$child = $rawobj[0]['child'];
item table rename parent-uri, target-type, object-type (more to come later)
2012-08-16 05:33:37 +00:00
if($child[NAMESPACE_ACTIVITY]['obj_type'][0]['data']) {
$res['obj_type'] = $child[NAMESPACE_ACTIVITY]['obj_type'][0]['data'];
snakebite continued
2014-08-20 04:31:33 +00:00
$obj['type'] = $child[NAMESPACE_ACTIVITY]['obj_type'][0]['data'];
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
}
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
if(x($child[SIMPLEPIE_NAMESPACE_ATOM_10], 'id') && $child[SIMPLEPIE_NAMESPACE_ATOM_10]['id'][0]['data'])
snakebite continued
2014-08-20 04:31:33 +00:00
$obj['id'] = $child[SIMPLEPIE_NAMESPACE_ATOM_10]['id'][0]['data'];
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
if(x($child[SIMPLEPIE_NAMESPACE_ATOM_10], 'link') && $child[SIMPLEPIE_NAMESPACE_ATOM_10]['link'])
snakebite continued
2014-08-20 04:31:33 +00:00
$obj['link'] = encode_rel_links($child[SIMPLEPIE_NAMESPACE_ATOM_10]['link']);
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
if(x($child[SIMPLEPIE_NAMESPACE_ATOM_10], 'title') && $child[SIMPLEPIE_NAMESPACE_ATOM_10]['title'][0]['data'])
snakebite continued
2014-08-20 04:31:33 +00:00
$obj['title'] = $child[SIMPLEPIE_NAMESPACE_ATOM_10]['title'][0]['data'];
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
if(x($child[SIMPLEPIE_NAMESPACE_ATOM_10], 'content') && $child[SIMPLEPIE_NAMESPACE_ATOM_10]['content'][0]['data']) {
$body = $child[SIMPLEPIE_NAMESPACE_ATOM_10]['content'][0]['data'];
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
if(! $body)
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
$body = $child[SIMPLEPIE_NAMESPACE_ATOM_10]['summary'][0]['data'];
preserve body content in feed to extract microformat stuff
2011-01-12 21:21:37 +00:00
// preserve a copy of the original body content in case we later need to parse out any microformat information, e.g. events
snakebite continued
2014-08-20 04:31:33 +00:00
$obj['orig'] = xmlify($body);
trying to solve the double encoding issue
2011-02-02 02:20:25 +00:00
if((strpos($body,'<') !== false) || (strpos($body,'>') !== false)) {
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
$body = purify_html($body);
fix some linebreak issues
2011-01-18 03:50:18 +00:00
$body = html2bbcode($body);
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
}
snakebite continued
2014-08-20 04:31:33 +00:00
$obj['content'] = $body;
like, dislike, activity streams, etc.
2010-09-17 10:10:19 +00:00
}
snakebite continued
2014-08-20 04:31:33 +00:00
$res['object'] = $obj;
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
}
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
$rawobj = $item->get_item_tags(NAMESPACE_ACTIVITY, 'target');
if($rawobj) {
snakebite continued
2014-08-20 04:31:33 +00:00
$obj = array();
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
$child = $rawobj[0]['child'];
item table rename parent-uri, target-type, object-type (more to come later)
2012-08-16 05:33:37 +00:00
if($child[NAMESPACE_ACTIVITY]['obj_type'][0]['data']) {
snakebite continued
2014-08-20 04:31:33 +00:00
$res['tgt_type'] = $child[NAMESPACE_ACTIVITY]['obj_type'][0]['data'];
$obj['type'] = $child[NAMESPACE_ACTIVITY]['obj_type'][0]['data'];
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
}
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
if(x($child[SIMPLEPIE_NAMESPACE_ATOM_10], 'id') && $child[SIMPLEPIE_NAMESPACE_ATOM_10]['id'][0]['data'])
snakebite continued
2014-08-20 04:31:33 +00:00
$obj['id'] = $child[SIMPLEPIE_NAMESPACE_ATOM_10]['id'][0]['data'];
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
if(x($child[SIMPLEPIE_NAMESPACE_ATOM_10], 'link') && $child[SIMPLEPIE_NAMESPACE_ATOM_10]['link'])
snakebite continued
2014-08-20 04:31:33 +00:00
$obj['link'] = encode_rel_links($child[SIMPLEPIE_NAMESPACE_ATOM_10]['link']);
if(x($child[SIMPLEPIE_NAMESPACE_ATOM_10], 'title') && $child[SIMPLEPIE_NAMESPACE_ATOM_10]['title'][0]['data'])
$obj['title'] = $child[SIMPLEPIE_NAMESPACE_ATOM_10]['title'][0]['data'];
if(x($child[SIMPLEPIE_NAMESPACE_ATOM_10], 'content') && $child[SIMPLEPIE_NAMESPACE_ATOM_10]['content'][0]['data']) {
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
$body = $child[SIMPLEPIE_NAMESPACE_ATOM_10]['content'][0]['data'];
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
if(! $body)
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
$body = $child[SIMPLEPIE_NAMESPACE_ATOM_10]['summary'][0]['data'];
preserve body content in feed to extract microformat stuff
2011-01-12 21:21:37 +00:00
// preserve a copy of the original body content in case we later need to parse out any microformat information, e.g. events
snakebite continued
2014-08-20 04:31:33 +00:00
$obj['orig'] = xmlify($body);
trying to solve the double encoding issue
2011-02-02 02:20:25 +00:00
if((strpos($body,'<') !== false) || (strpos($body,'>') !== false)) {
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
$body = purify_html($body);
fix some linebreak issues
2011-01-18 03:50:18 +00:00
$body = html2bbcode($body);
snakebite continued
2014-08-20 04:31:33 +00:00
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
}
snakebite continued
2014-08-20 04:31:33 +00:00
$obj['content'] = $body;
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
}
snakebite continued
2014-08-20 04:31:33 +00:00
$res['target'] = $obj;
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
}
various rss fixes
2014-09-01 03:51:05 +00:00
$res['public_policy'] = 'specific';
$res['comment_policy'] = 'none';
more plugin hooks
2010-12-25 23:01:02 +00:00
$arr = array('feed' => $feed, 'item' => $item, 'result' => $res);
call_hooks('parse_atom', $arr);
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
logger('get_atom_elements: author: ' . print_r($author,true),LOGGER_DATA);
feeds with '@' in the URL were being handled as webfinger addresses and didn't fall through to RSS discovery. Issue #599
2014-09-21 21:46:12 +00:00
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
logger('get_atom_elements: ' . print_r($res,true),LOGGER_DATA);
more plugin hooks
2010-12-25 02:32:23 +00:00
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
return $res;
}
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
function encode_rel_links($links) {
$o = '';
if(! ((is_array($links)) && (count($links))))
return $o;
snakebite continued
2014-08-20 04:31:33 +00:00
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
foreach($links as $link) {
$o .= '<link ';
if($link['attribs']['']['rel'])
$o .= 'rel="' . $link['attribs']['']['rel'] . '" ';
if($link['attribs']['']['type'])
$o .= 'type="' . $link['attribs']['']['type'] . '" ';
if($link['attribs']['']['href'])
activity object link had incorrect attribute, highlight critical install steps
2011-01-18 00:06:22 +00:00
$o .= 'href="' . $link['attribs']['']['href'] . '" ';
minor stuff
2010-11-04 07:37:29 +00:00
if( (x($link['attribs'],NAMESPACE_MEDIA)) && $link['attribs'][NAMESPACE_MEDIA]['width'])
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
$o .= 'media:width="' . $link['attribs'][NAMESPACE_MEDIA]['width'] . '" ';
minor stuff
2010-11-04 07:37:29 +00:00
if( (x($link['attribs'],NAMESPACE_MEDIA)) && $link['attribs'][NAMESPACE_MEDIA]['height'])
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
$o .= 'media:height="' . $link['attribs'][NAMESPACE_MEDIA]['height'] . '" ';
$o .= ' />' . "\n" ;
}
return xmlify($o);
}
important bits we need to allow php executable content. These must be explicitly allowed - but only if the account has ACCOUNT_ROLE_ALLOWCODE and *only* for web pages and profile fields. This content cannot be transmitted to other sites.
2013-08-11 23:56:06 +00:00
function item_store($arr,$allow_exec = false) {
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
item store hooks
2013-11-29 02:10:04 +00:00
$d = array('item' => $arr, 'allow_exec' => $allow_exec);
call_hooks('item_store', $d );
$arr = $d['item'];
$allow_exec = $d['allow_exec'];
Red API: /api/red/item/new ; which is like /api/statuses/update except it uses Red semantics and returns success/failure indication.
2014-03-05 03:57:54 +00:00
$ret = array('success' => false, 'item_id' => 0);
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
call item_translate hook if source language is not acceptable, ignore message if it cannot be translated.
2012-07-16 10:32:42 +00:00
if(! $arr['uid']) {
logger('item_store: no uid');
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'No uid.';
various diaspora issues
2014-08-30 00:31:40 +00:00
return $ret;
call item_translate hook if source language is not acceptable, ignore message if it cannot be translated.
2012-07-16 10:32:42 +00:00
}
preserve the source owner when creating a delivery fork so that we can uplink back to them without any ambiguity.
2013-10-14 01:36:28 +00:00
$uplinked_comment = false;
Sheesh. Three lines of Javascript. All you have to do is find which of the three lines is doing something wrong. Or FILE A BUG. Or use a different browser (not *try* a different browser, *use* a different browser), or turn off freaking tinymce. But no - let's argue about browsers instead.
2013-09-02 02:15:08 +00:00
// If a page layout is provided, ensure it exists and belongs to us.
if(array_key_exists('layout_mid',$arr) && $arr['layout_mid']) {
$l = q("select item_restrict from item where mid = '%s' and uid = %d limit 1",
dbesc($arr['layout_mid']),
intval($arr['uid'])
);
if((! $l) || (! ($l[0]['item_restrict'] & ITEM_PDL)))
unset($arr['layout_mid']);
}
fix some issues that showed up in the logs.
2013-07-16 01:07:23 +00:00
// Don't let anybody set these, either intentionally or accidentally
if(array_key_exists('id',$arr))
unset($arr['id']);
if(array_key_exists('parent',$arr))
unset($arr['parent']);
pass item by reference so title can be altered
2013-08-02 02:18:05 +00:00
$arr['mimetype'] = ((x($arr,'mimetype')) ? notags(trim($arr['mimetype'])) : 'text/bbcode');
important bits we need to allow php executable content. These must be explicitly allowed - but only if the account has ACCOUNT_ROLE_ALLOWCODE and *only* for web pages and profile fields. This content cannot be transmitted to other sites.
2013-08-11 23:56:06 +00:00
if(($arr['mimetype'] == 'application/x-php') && (! $allow_exec)) {
logger('item_store: php mimetype but allow_exec is denied.');
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'exec denied.';
return $ret;
important bits we need to allow php executable content. These must be explicitly allowed - but only if the account has ACCOUNT_ROLE_ALLOWCODE and *only* for web pages and profile fields. This content cannot be transmitted to other sites.
2013-08-11 23:56:06 +00:00
}
profile edit - missing visibility and drop link on non-default profiles, re-arrange order of replacing red#matrix smilie so it works correctly, accept a post with body content of '0' which was interpreted by x() as nothing (was treated as integer).
2014-08-21 23:46:24 +00:00
$arr['title'] = ((array_key_exists('title',$arr) && strlen($arr['title'])) ? trim($arr['title']) : '');
$arr['body'] = ((array_key_exists('body',$arr) && strlen($arr['body'])) ? trim($arr['body']) : '');
pass item by reference so title can be altered
2013-08-02 02:18:05 +00:00
store diaspora meta info in the item table. It has to go there or it will kill us with complex joins. We can phase out the sign table once this all checks out.
2014-09-03 12:09:43 +00:00
$arr['diaspora_meta'] = ((x($arr,'diaspora_meta')) ? $arr['diaspora_meta'] : '');
pass item by reference so title can be altered
2013-08-02 02:18:05 +00:00
$arr['allow_cid'] = ((x($arr,'allow_cid')) ? trim($arr['allow_cid']) : '');
$arr['allow_gid'] = ((x($arr,'allow_gid')) ? trim($arr['allow_gid']) : '');
$arr['deny_cid'] = ((x($arr,'deny_cid')) ? trim($arr['deny_cid']) : '');
$arr['deny_gid'] = ((x($arr,'deny_gid')) ? trim($arr['deny_gid']) : '');
$arr['item_private'] = ((x($arr,'item_private')) ? intval($arr['item_private']) : 0 );
$arr['item_flags'] = ((x($arr,'item_flags')) ? intval($arr['item_flags']) : 0 );
webpage content-type -- needs cleaning up and a security check once all the important bits are in place.
2013-09-02 08:38:17 +00:00
post signatures
2013-10-03 04:04:48 +00:00
ensure that no unencrypted content leaks through item_store which is private - we typically do this in mod/item, but some functions bypass mod/item to create private posts
2013-08-02 01:50:36 +00:00
// only detect language if we have text content, and if the post is private but not yet
// obscured, make it so.
if(! ($arr['item_flags'] & ITEM_OBSCURED)) {
post signatures
2013-10-03 04:04:48 +00:00
ensure that no unencrypted content leaks through item_store which is private - we typically do this in mod/item, but some functions bypass mod/item to create private posts
2013-08-02 01:50:36 +00:00
$arr['lang'] = detect_language($arr['body']);
fix private message encryption since the input_filter mangled it
2013-09-04 05:48:08 +00:00
// apply the input filter here - if it is obscured it has been filtered already
$arr['body'] = z_input_filter($arr['uid'],$arr['body'],$arr['mimetype']);
ensure that no unencrypted content leaks through item_store which is private - we typically do this in mod/item, but some functions bypass mod/item to create private posts
2013-08-02 01:50:36 +00:00
post signatures
2013-10-03 04:04:48 +00:00
if(local_user() && (! $arr['sig'])) {
$channel = get_app()->get_channel();
if($channel['channel_hash'] === $arr['author_xchan']) {
$arr['sig'] = base64url_encode(rsa_sign($arr['body'],$channel['channel_prvkey']));
$arr['item_flags'] |= ITEM_VERIFIED;
}
}
ensure that no unencrypted content leaks through item_store which is private - we typically do this in mod/item, but some functions bypass mod/item to create private posts
2013-08-02 01:50:36 +00:00
$allowed_languages = get_pconfig($arr['uid'],'system','allowed_languages');
call item_translate hook if source language is not acceptable, ignore message if it cannot be translated.
2012-07-16 10:32:42 +00:00
ensure that no unencrypted content leaks through item_store which is private - we typically do this in mod/item, but some functions bypass mod/item to create private posts
2013-08-02 01:50:36 +00:00
if((is_array($allowed_languages)) && ($arr['lang']) && (! array_key_exists($arr['lang'],$allowed_languages))) {
$translate = array('item' => $arr, 'from' => $arr['lang'], 'to' => $allowed_languages, 'translated' => false);
call_hooks('item_translate', $translate);
if((! $translate['translated']) && (intval(get_pconfig($arr['uid'],'system','reject_disallowed_languages')))) {
logger('item_store: language ' . $arr['lang'] . ' not accepted for uid ' . $arr['uid']);
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'language not accepted';
return $ret;
ensure that no unencrypted content leaks through item_store which is private - we typically do this in mod/item, but some functions bypass mod/item to create private posts
2013-08-02 01:50:36 +00:00
}
$arr = $translate['item'];
call item_translate hook if source language is not acceptable, ignore message if it cannot be translated.
2012-07-16 10:32:42 +00:00
}
ensure that no unencrypted content leaks through item_store which is private - we typically do this in mod/item, but some functions bypass mod/item to create private posts
2013-08-02 01:50:36 +00:00
if($arr['item_private']) {
$key = get_config('system','pubkey');
$arr['item_flags'] = $arr['item_flags'] | ITEM_OBSCURED;
if($arr['title'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$arr['title'] = json_encode(crypto_encapsulate($arr['title'],$key));
ensure that no unencrypted content leaks through item_store which is private - we typically do this in mod/item, but some functions bypass mod/item to create private posts
2013-08-02 01:50:36 +00:00
if($arr['body'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$arr['body'] = json_encode(crypto_encapsulate($arr['body'],$key));
ensure that no unencrypted content leaks through item_store which is private - we typically do this in mod/item, but some functions bypass mod/item to create private posts
2013-08-02 01:50:36 +00:00
}
call item_translate hook if source language is not acceptable, ignore message if it cannot be translated.
2012-07-16 10:32:42 +00:00
}
trying to solve the double encoding issue
2011-02-02 02:20:25 +00:00
input cleansing
2013-01-07 21:58:25 +00:00
if((x($arr,'object')) && is_array($arr['object'])) {
activity_sanitise($arr['object']);
$arr['object'] = json_encode($arr['object']);
}
if((x($arr,'target')) && is_array($arr['target'])) {
activity_sanitise($arr['target']);
$arr['target'] = json_encode($arr['target']);
}
sort out the rest of the nasty stuff in the attachment/file api - time to move on to something else
2013-02-01 08:49:07 +00:00
if((x($arr,'attach')) && is_array($arr['attach'])) {
activity_sanitise($arr['attach']);
$arr['attach'] = json_encode($arr['attach']);
}
Can't see any posts currently - after the big shakeup, but we can now post top level activities and store them.
2012-10-04 05:28:19 +00:00
$arr['aid'] = ((x($arr,'aid')) ? intval($arr['aid']) : 0);
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$arr['mid'] = ((x($arr,'mid')) ? notags(trim($arr['mid'])) : random_string());
Can't see any posts currently - after the big shakeup, but we can now post top level activities and store them.
2012-10-04 05:28:19 +00:00
$arr['author_xchan'] = ((x($arr,'author_xchan')) ? notags(trim($arr['author_xchan'])) : '');
$arr['owner_xchan'] = ((x($arr,'owner_xchan')) ? notags(trim($arr['owner_xchan'])) : '');
lint
2010-11-10 04:38:24 +00:00
$arr['created'] = ((x($arr,'created') !== false) ? datetime_convert('UTC','UTC',$arr['created']) : datetime_convert());
$arr['edited'] = ((x($arr,'edited') !== false) ? datetime_convert('UTC','UTC',$arr['edited']) : datetime_convert());
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
$arr['expires'] = ((x($arr,'expires') !== false) ? datetime_convert('UTC','UTC',$arr['expires']) : NULL_DATE);
pull in some posts when we first connect with a new channel (if allowed to) - if not allowed to, do it if that condition changes
2013-11-18 03:22:24 +00:00
$arr['commented'] = ((x($arr,'commented') !== false) ? datetime_convert('UTC','UTC',$arr['commented']) : datetime_convert());
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
$arr['comments_closed'] = ((x($arr,'comments_closed') !== false) ? datetime_convert('UTC','UTC',$arr['comments_closed']) : NULL_DATE);
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
new item view, sort by received date - not posted date.
2011-06-24 00:56:59 +00:00
$arr['received'] = datetime_convert();
lint
2010-11-10 04:38:24 +00:00
$arr['changed'] = datetime_convert();
$arr['location'] = ((x($arr,'location')) ? notags(trim($arr['location'])) : '');
$arr['coord'] = ((x($arr,'coord')) ? notags(trim($arr['coord'])) : '');
well that really stuffed things up...
2013-09-11 05:45:04 +00:00
$arr['parent_mid'] = ((x($arr,'parent_mid')) ? notags(trim($arr['parent_mid'])) : '');
liking comments were reverted to liking the parent after sending item to another channel
2013-04-04 01:04:48 +00:00
$arr['thr_parent'] = ((x($arr,'thr_parent')) ? notags(trim($arr['thr_parent'])) : $arr['parent_mid']);
lint
2010-11-10 04:38:24 +00:00
$arr['verb'] = ((x($arr,'verb')) ? notags(trim($arr['verb'])) : '');
new item structure to item_store
2012-10-06 08:17:25 +00:00
$arr['obj_type'] = ((x($arr,'obj_type')) ? notags(trim($arr['obj_type'])) : '');
lint
2010-11-10 04:38:24 +00:00
$arr['object'] = ((x($arr,'object')) ? trim($arr['object']) : '');
new item structure to item_store
2012-10-06 08:17:25 +00:00
$arr['tgt_type'] = ((x($arr,'tgt_type')) ? notags(trim($arr['tgt_type'])) : '');
lint
2010-11-10 04:38:24 +00:00
$arr['target'] = ((x($arr,'target')) ? trim($arr['target']) : '');
provide permalink for posts, include permalinks in feeds and imported from feeds
2011-02-16 10:07:11 +00:00
$arr['plink'] = ((x($arr,'plink')) ? notags(trim($arr['plink'])) : '');
attachment storage
2011-04-07 04:59:07 +00:00
$arr['attach'] = ((x($arr,'attach')) ? notags(trim($arr['attach'])) : '');
item source
2011-06-21 02:08:40 +00:00
$arr['app'] = ((x($arr,'app')) ? notags(trim($arr['app'])) : '');
fix some issues that couldn't be tested yesterday - as was working offline
2012-10-07 23:39:29 +00:00
$arr['item_restrict'] = ((x($arr,'item_restrict')) ? intval($arr['item_restrict']) : 0 );
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
first step first - just store the public_policy data, don't do anything with it yet.
2014-08-07 00:36:07 +00:00
$arr['public_policy'] = ((x($arr,'public_policy')) ? notags(trim($arr['public_policy'])) : '' );
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
$arr['comment_policy'] = ((x($arr,'comment_policy')) ? notags(trim($arr['comment_policy'])) : 'contacts' );
new item structure to item_store
2012-10-06 08:17:25 +00:00
further progress on new network/home notifications
2012-11-30 04:46:45 +00:00
$arr['item_flags'] = $arr['item_flags'] | ITEM_UNSEEN;
there are now 2 permalinks - one is the remote original post (plink) and the other is a pointer to a local copy (llink). mod_display is not longer tied to a particular id or channel so if a message is publicly viewable it will appear without any ownership info.
2012-12-16 10:18:26 +00:00
implement 'can_comment_on_post()' which doesn't require a separate DB lookup per item.
2013-06-17 03:44:29 +00:00
if($arr['comment_policy'] == 'none')
$arr['item_flags'] = $arr['item_flags'] | ITEM_NOCOMMENT;
implement time travelling posts
2013-05-24 00:24:15 +00:00
// handle time travelers
// Allow a bit of fudge in case somebody just has a slightly slow/fast clock
force delayed_publish time comparisons to UTC.
2013-05-30 04:11:30 +00:00
$d1 = new DateTime('now +10 minutes', new DateTimeZone('UTC'));
$d2 = new DateTime($arr['created'] . '+00:00');
implement time travelling posts
2013-05-24 00:24:15 +00:00
if($d2 > $d1)
$arr['item_restrict'] = $arr['item_restrict'] | ITEM_DELAYED_PUBLISH;
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$arr['llink'] = z_root() . '/display/' . $arr['mid'];
there are now 2 permalinks - one is the remote original post (plink) and the other is a pointer to a local copy (llink). mod_display is not longer tied to a particular id or channel so if a message is publicly viewable it will appear without any ownership info.
2012-12-16 10:18:26 +00:00
some minor cleanup on plinks for some very subtle permissions issues
2014-01-09 23:45:17 +00:00
if(! $arr['plink'])
$arr['plink'] = $arr['llink'];
there are now 2 permalinks - one is the remote original post (plink) and the other is a pointer to a local copy (llink). mod_display is not longer tied to a particular id or channel so if a message is publicly viewable it will appear without any ownership info.
2012-12-16 10:18:26 +00:00
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
if($arr['parent_mid'] === $arr['mid']) {
give orphans a second chance, linkify naked links
2010-11-12 04:32:20 +00:00
$parent_id = 0;
Avoid a notice
2012-02-28 12:56:16 +00:00
$parent_deleted = 0;
give orphans a second chance, linkify naked links
2010-11-12 04:32:20 +00:00
$allow_cid = $arr['allow_cid'];
$allow_gid = $arr['allow_gid'];
$deny_cid = $arr['deny_cid'];
$deny_gid = $arr['deny_gid'];
booboo
2014-08-09 22:21:15 +00:00
$public_policy = $arr['public_policy'];
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$comments_closed = $arr['comments_closed'];
basic (very basic) communication showing up on network page
2012-11-27 07:52:16 +00:00
$arr['item_flags'] = $arr['item_flags'] | ITEM_THREAD_TOP;
give orphans a second chance, linkify naked links
2010-11-12 04:32:20 +00:00
}
else {
// find the parent and snarf the item id and ACL's
inherit 'wall' setting from parent post on remote item
2011-03-08 22:40:33 +00:00
// and anything else we need to inherit
give orphans a second chance, linkify naked links
2010-11-12 04:32:20 +00:00
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$r = q("SELECT * FROM `item` WHERE `mid` = '%s' AND `uid` = %d ORDER BY `id` ASC LIMIT 1",
dbesc($arr['parent_mid']),
give orphans a second chance, linkify naked links
2010-11-12 04:32:20 +00:00
intval($arr['uid'])
);
make zot_finger more bulletproof
2013-07-22 05:39:21 +00:00
if($r) {
flatten conversation hierarchy to one level on import but preserve all thread info in case somebody wants to write a plugin to display as multi-level thread
2010-12-05 07:16:16 +00:00
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
if(comments_are_now_closed($r[0])) {
logger('item_store: comments closed');
$ret['message'] = 'Comments closed.';
various diaspora issues
2014-08-30 00:31:40 +00:00
return $ret;
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
}
allow (our own) ajax urls to pass through bbcode intact.
2014-10-10 03:59:33 +00:00
flatten conversation hierarchy to one level on import but preserve all thread info in case somebody wants to write a plugin to display as multi-level thread
2010-12-05 07:16:16 +00:00
// is the new message multi-level threaded?
// even though we don't support it now, preserve the info
// and re-attach to the conversation parent.
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
if($r[0]['mid'] != $r[0]['parent_mid']) {
$arr['parent_mid'] = $r[0]['parent_mid'];
$z = q("SELECT * FROM `item` WHERE `mid` = '%s' AND `parent_mid` = '%s' AND `uid` = %d
storage race condition
2011-10-12 09:21:18 +00:00
ORDER BY `id` ASC LIMIT 1",
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
dbesc($r[0]['parent_mid']),
dbesc($r[0]['parent_mid']),
reparenting imported item was only partially reparenting.
2011-08-09 02:10:36 +00:00
intval($arr['uid'])
);
if($z && count($z))
$r = $z;
flatten conversation hierarchy to one level on import but preserve all thread info in case somebody wants to write a plugin to display as multi-level thread
2010-12-05 07:16:16 +00:00
}
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$parent_id = $r[0]['id'];
$parent_deleted = $r[0]['item_restrict'] & ITEM_DELETED;
$allow_cid = $r[0]['allow_cid'];
$allow_gid = $r[0]['allow_gid'];
$deny_cid = $r[0]['deny_cid'];
$deny_gid = $r[0]['deny_gid'];
$public_policy = $r[0]['public_policy'];
$comments_closed = $r[0]['comments_closed'];
new item structure to item_store
2012-10-06 08:17:25 +00:00
fix some issues that couldn't be tested yesterday - as was working offline
2012-10-07 23:39:29 +00:00
if($r[0]['item_flags'] & ITEM_WALL)
new item structure to item_store
2012-10-06 08:17:25 +00:00
$arr['item_flags'] = $arr['item_flags'] | ITEM_WALL;
bug #447 - make followed feeds private
2012-06-11 12:28:08 +00:00
preserve the source owner when creating a delivery fork so that we can uplink back to them without any ambiguity.
2013-10-14 01:36:28 +00:00
// An uplinked comment might arrive with a downstream owner.
// Fix it.
if($r[0]['owner_xchan'] !== $arr['owner_xchan']) {
$arr['owner_xchan'] = $r[0]['owner_xchan'];
$uplinked_comment = true;
}
bug #447 - make followed feeds private
2012-06-11 12:28:08 +00:00
// if the parent is private, force privacy for the entire conversation
start fixing all the item queries
2012-10-08 01:44:06 +00:00
if($r[0]['item_private'])
$arr['item_private'] = $r[0]['item_private'];
bug #447 - make followed feeds private
2012-06-11 12:28:08 +00:00
public forum author using private posting, correct the comment permissions downstream.
2012-06-19 00:19:56 +00:00
// Edge case. We host a public forum that was originally posted to privately.
// The original author commented, but as this is a comment, the permissions
// weren't fixed up so it will still show the comment as private unless we fix it here.
start fixing all the item queries
2012-10-08 01:44:06 +00:00
if((intval($r[0]['item_flags']) & ITEM_UPLINK) && (! $r[0]['item_private']))
$arr['item_private'] = 0;
give orphans a second chance, linkify naked links
2010-11-12 04:32:20 +00:00
}
else {
update manage table and some documentation
2013-03-28 02:02:01 +00:00
logger('item_store: item parent was not found - ignoring item');
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'parent not found.';
return $ret;
give orphans a second chance, linkify naked links
2010-11-12 04:32:20 +00:00
}
}
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
new item structure to item_store
2012-10-06 08:17:25 +00:00
if($parent_deleted)
$arr['item_restrict'] = $arr['item_restrict'] | ITEM_DELETED;
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$r = q("SELECT `id` FROM `item` WHERE `mid` = '%s' AND `uid` = %d LIMIT 1",
dbesc($arr['mid']),
storage race condition
2011-10-12 09:21:18 +00:00
intval($arr['uid'])
item_store: prevent storing a duplicate item with same uri+uid
2011-10-03 13:04:17 +00:00
);
got comments going back and forth - we really need local_deliver now since even local deliveries are going through zot and that isn't necessary or desirable in the long term
2012-11-28 01:09:01 +00:00
if($r) {
this should fix the foreign language like notifications (again)
2013-09-11 04:10:56 +00:00
logger('item_store: duplicate item ignored. ' . print_r($arr,true));
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'duplicate post.';
return $ret;
item_store: prevent storing a duplicate item with same uri+uid
2011-10-03 13:04:17 +00:00
}
item guids
2011-08-08 00:29:26 +00:00
A useful hook
2014-04-26 04:12:43 +00:00
call_hooks('item_store',$arr);
// This hook remains for backward compatibility.
plugin/addon API landing - still things left to do before it's useful
2010-12-22 22:16:22 +00:00
call_hooks('post_remote',$arr);
allow plugin to abort message posting, fix half-duplex D* relationship
2012-01-31 04:49:54 +00:00
if(x($arr,'cancel')) {
logger('item_store: post cancelled by plugin.');
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'cancelled.';
return $ret;
allow plugin to abort message posting, fix half-duplex D* relationship
2012-01-31 04:49:54 +00:00
}
catchup
2012-07-11 11:29:47 +00:00
// pull out all the taxonomy stuff for separate storage
$terms = null;
You've got mail! (actually a post, mail is for another day)
2012-11-27 05:20:16 +00:00
if(array_key_exists('term',$arr)) {
catchup
2012-07-11 11:29:47 +00:00
$terms = $arr['term'];
unset($arr['term']);
}
show like/dislike when no comments yet, bit of lint
2010-11-04 02:47:07 +00:00
logger('item_store: ' . print_r($arr,true), LOGGER_DATA);
Another try at issue #61 and #62 - an earlier fix was partially working but the issue persisted - this extends it a bit.
2013-08-18 07:04:38 +00:00
dbesc_array($arr);
use raw db queries wherever query items could contain '%'
2010-11-09 23:11:47 +00:00
$r = dbq("INSERT INTO `item` (`"
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
. implode("`, `", array_keys($arr))
. "`) VALUES ('"
. implode("', '", array_values($arr))
. "')" );
give orphans a second chance, linkify naked links
2010-11-12 04:32:20 +00:00
// find the item we just created
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
make comments work again
2013-09-11 06:25:56 +00:00
$r = q("SELECT * FROM `item` WHERE `mid` = '%s' AND `uid` = %d ORDER BY `id` ASC ",
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$arr['mid'], // already dbesc'd
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
intval($arr['uid'])
);
on rare occasions posts do not get a parent. The only thing which could cause it is if we can't find our new post. So we'll check for this and try one more time before failing.
2011-05-30 05:50:36 +00:00
make comments work again
2013-09-11 06:25:56 +00:00
more stuff we don't need.
2012-10-08 07:23:43 +00:00
if($r && count($r)) {
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
$current_post = $r[0]['id'];
this should fix the foreign language like notifications (again)
2013-09-11 04:10:56 +00:00
$arr = $r[0]; // This will gives us a fresh copy of what's now in the DB and undo the db escaping, which really messes up the notifications
more stuff we don't need.
2012-10-08 07:23:43 +00:00
logger('item_store: created item ' . $current_post, LOGGER_DEBUG);
show like/dislike when no comments yet, bit of lint
2010-11-04 02:47:07 +00:00
}
give orphans a second chance, linkify naked links
2010-11-12 04:32:20 +00:00
else {
change wording on item_store message so we can find failures fast with grep item_store logfile | grep -v created
2013-09-11 03:14:12 +00:00
logger('item_store: could not locate stored item');
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'unable to retrieve.';
return $ret;
mistpark 2.0 infrasturcture lands
2010-09-09 03:14:17 +00:00
}
storage race condition
2011-10-12 09:21:18 +00:00
if(count($r) > 1) {
logger('item_store: duplicated post occurred. Removing duplicates.');
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
q("DELETE FROM `item` WHERE `mid` = '%s' AND `uid` = %d AND `id` != %d ",
$arr['mid'],
storage race condition
2011-10-12 09:21:18 +00:00
intval($arr['uid']),
intval($current_post)
);
}
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
if((! $parent_id) || ($arr['parent_mid'] === $arr['mid']))
give orphans a second chance, linkify naked links
2010-11-12 04:32:20 +00:00
$parent_id = $current_post;
more error checking on posts
2011-02-04 21:37:04 +00:00
basic 'notpublic' bits working
2014-08-07 02:24:46 +00:00
if(strlen($allow_cid) || strlen($allow_gid) || strlen($deny_cid) || strlen($deny_gid) || strlen($public_policy))
display the fact that a conversation is private without disclosing the details of who can/cannot see it.
2010-12-08 04:47:53 +00:00
$private = 1;
else
encrypt private messages on disk - there are still a couple of places where the text is leaked in the logs during processing.
2013-07-29 04:04:03 +00:00
$private = $arr['item_private'];
display the fact that a conversation is private without disclosing the details of who can/cannot see it.
2010-12-08 04:47:53 +00:00
give orphans a second chance, linkify naked links
2010-11-12 04:32:20 +00:00
// Set parent id - and also make sure to inherit the parent's ACL's.
inherit parent ACL's on remote children
2010-09-29 05:32:48 +00:00
wrong attribution on likes of comments in email notifications. Let's also call the thing they liked a comment instead of a status
2013-05-28 08:40:27 +00:00
$r = q("UPDATE item SET parent = %d, allow_cid = '%s', allow_gid = '%s',
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
deny_cid = '%s', deny_gid = '%s', public_policy = '%s', item_private = %d, comments_closed = '%s'
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
WHERE id = %d",
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
intval($parent_id),
inherit parent ACL's on remote children
2010-09-29 05:32:48 +00:00
dbesc($allow_cid),
dbesc($allow_gid),
dbesc($deny_cid),
dbesc($deny_gid),
basic 'notpublic' bits working
2014-08-07 02:24:46 +00:00
dbesc($public_policy),
display the fact that a conversation is private without disclosing the details of who can/cannot see it.
2010-12-08 04:47:53 +00:00
intval($private),
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
dbesc($comments_closed),
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
intval($current_post)
);
this should fix the foreign language like notifications (again)
2013-09-11 04:10:56 +00:00
// These are probably redundant now that we've queried the just stored post
more stuff we don't need.
2012-10-08 07:23:43 +00:00
$arr['id'] = $current_post;
$arr['parent'] = $parent_id;
catchup
2012-07-11 11:29:47 +00:00
$arr['allow_cid'] = $allow_cid;
$arr['allow_gid'] = $allow_gid;
more stuff we don't need.
2012-10-08 07:23:43 +00:00
$arr['deny_cid'] = $deny_cid;
$arr['deny_gid'] = $deny_gid;
basic 'notpublic' bits working
2014-08-07 02:24:46 +00:00
$arr['public_policy'] = $public_policy;
encrypt private messages on disk - there are still a couple of places where the text is leaked in the logs during processing.
2013-07-29 04:04:03 +00:00
$arr['item_private'] = $private;
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$arr['comments_closed'] = $comments_closed;
more stuff we don't need.
2012-10-08 07:23:43 +00:00
// Store taxonomy
progress on tag delivery
2013-02-11 08:20:14 +00:00
catchup
2012-07-11 11:29:47 +00:00
if(($terms) && (is_array($terms))) {
foreach($terms as $t) {
q("insert into term (uid,oid,otype,type,term,url)
values(%d,%d,%d,%d,'%s','%s') ",
intval($arr['uid']),
intval($current_post),
progress on tag delivery
2013-02-11 08:20:14 +00:00
intval(TERM_OBJ_POST),
catchup
2012-07-11 11:29:47 +00:00
intval($t['type']),
dbesc($t['term']),
dbesc($t['url'])
);
}
$arr['term'] = $terms;
}
Add a hook for post-processing remote items after they've been stored in the database and have an ID
2012-06-12 05:09:50 +00:00
call_hooks('post_remote_end',$arr);
optimise the commented date view
2011-09-19 02:04:11 +00:00
// update the commented timestamp on the parent
update timestamp on activity change (comment/like)
2011-09-16 13:49:24 +00:00
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$z = q("select max(created) as commented from item where parent_mid = '%s' and uid = %d and not ( item_restrict & %d )>0 ",
pull in some posts when we first connect with a new channel (if allowed to) - if not allowed to, do it if that condition changes
2013-11-18 03:22:24 +00:00
dbesc($arr['parent_mid']),
when updating the commented timestamp on the parent post (item_store) don't include any time travelling posts. They can mess up the conversation sort order in a big way until they're actually published.
2014-08-29 00:25:00 +00:00
intval($arr['uid']),
intval(ITEM_DELAYED_PUBLISH)
pull in some posts when we first connect with a new channel (if allowed to) - if not allowed to, do it if that condition changes
2013-11-18 03:22:24 +00:00
);
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
q("UPDATE item set commented = '%s', changed = '%s' WHERE id = %d",
fix commented timestamp
2013-11-18 04:09:40 +00:00
dbesc(($z) ? $z[0]['commented'] : (datetime_convert())),
dbesc(datetime_convert()),
update timestamp on activity change (comment/like)
2011-09-16 13:49:24 +00:00
intval($parent_id)
);
work on notifications
2013-02-07 04:29:17 +00:00
send_status_notifications($current_post,$arr);
item tag notifications
2012-02-10 05:18:50 +00:00
tag_deliver($arr['uid'],$current_post);
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['success'] = true;
$ret['item_id'] = $current_post;
tgroup relays
2011-11-16 04:30:34 +00:00
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
return $ret;
more poll stuff ahead of item revamp
2010-07-19 03:49:10 +00:00
}
preliminary network abstraction, configurable debugging.
2010-09-26 23:30:21 +00:00
work on notifications
2013-02-07 04:29:17 +00:00
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
important bits we need to allow php executable content. These must be explicitly allowed - but only if the account has ACCOUNT_ROLE_ALLOWCODE and *only* for web pages and profile fields. This content cannot be transmitted to other sites.
2013-08-11 23:56:06 +00:00
function item_store_update($arr,$allow_exec = false) {
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
item store hooks
2013-11-29 02:10:04 +00:00
$d = array('item' => $arr, 'allow_exec' => $allow_exec);
call_hooks('item_store_update', $d );
$arr = $d['item'];
$allow_exec = $d['allow_exec'];
Red API: /api/red/item/new ; which is like /api/statuses/update except it uses Red semantics and returns success/failure indication.
2014-03-05 03:57:54 +00:00
$ret = array('success' => false, 'item_id' => 0);
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
if(! intval($arr['uid'])) {
logger('item_store_update: no uid');
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'no uid.';
return $ret;
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
}
if(! intval($arr['id'])) {
logger('item_store_update: no id');
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'no id.';
return $ret;
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
}
$orig_post_id = $arr['id'];
$uid = $arr['uid'];
fix a bunch of post editing bugs - though this consolidates some disparate code and may create some new bugs due to regression
2013-10-04 11:47:41 +00:00
$orig = q("select * from item where id = %d and uid = %d limit 1",
intval($orig_post_id),
intval($uid)
);
if(! $orig) {
logger('item_store_update: original post not found: ' . $orig_post_id);
$ret['message'] = 'no original';
return $ret;
}
prevent posts from floating to the top after minor edits
2013-10-20 06:38:07 +00:00
// override the unseen flag with the original
if($arr['item_flags'] & ITEM_UNSEEN)
$arr['item_flags'] = $arr['item_flags'] ^ ITEM_UNSEEN;
fix a bunch of post editing bugs - though this consolidates some disparate code and may create some new bugs due to regression
2013-10-04 11:47:41 +00:00
if($orig[0]['item_flags'] & ITEM_VERIFIED)
$orig[0]['item_flags'] = $orig[0]['item_flags'] ^ ITEM_VERIFIED;
issues with editing webpage permissions
2014-06-16 01:58:02 +00:00
if($orig[0]['item_flags'] & ITEM_OBSCURED)
$orig[0]['item_flags'] = $orig[0]['item_flags'] ^ ITEM_OBSCURED;
fix a bunch of post editing bugs - though this consolidates some disparate code and may create some new bugs due to regression
2013-10-04 11:47:41 +00:00
$arr['item_flags'] = intval($arr['item_flags']) | $orig[0]['item_flags'];
$arr['item_restrict'] = intval($arr['item_restrict']) | $orig[0]['item_restrict'];
turn off endless scroll if we have no more content to load - currently for mod_display, need to add to other content modules
2013-10-14 03:09:11 +00:00
fix a bunch of post editing bugs - though this consolidates some disparate code and may create some new bugs due to regression
2013-10-04 11:47:41 +00:00
if(array_key_exists('edit',$arr))
unset($arr['edit']);
important bits we need to allow php executable content. These must be explicitly allowed - but only if the account has ACCOUNT_ROLE_ALLOWCODE and *only* for web pages and profile fields. This content cannot be transmitted to other sites.
2013-08-11 23:56:06 +00:00
$arr['mimetype'] = ((x($arr,'mimetype')) ? notags(trim($arr['mimetype'])) : 'text/bbcode');
if(($arr['mimetype'] == 'application/x-php') && (! $allow_exec)) {
logger('item_store: php mimetype but allow_exec is denied.');
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'exec denied.';
return $ret;
important bits we need to allow php executable content. These must be explicitly allowed - but only if the account has ACCOUNT_ROLE_ALLOWCODE and *only* for web pages and profile fields. This content cannot be transmitted to other sites.
2013-08-11 23:56:06 +00:00
}
post signatures
2013-10-03 04:04:48 +00:00
if(! ($arr['item_flags'] & ITEM_OBSCURED)) {
important bits we need to allow php executable content. These must be explicitly allowed - but only if the account has ACCOUNT_ROLE_ALLOWCODE and *only* for web pages and profile fields. This content cannot be transmitted to other sites.
2013-08-11 23:56:06 +00:00
post signatures
2013-10-03 04:04:48 +00:00
$arr['lang'] = detect_language($arr['body']);
// apply the input filter here - if it is obscured it has been filtered already
$arr['body'] = z_input_filter($arr['uid'],$arr['body'],$arr['mimetype']);
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
post signatures
2013-10-03 04:04:48 +00:00
if(local_user() && (! $arr['sig'])) {
$channel = get_app()->get_channel();
if($channel['channel_hash'] === $arr['author_xchan']) {
$arr['sig'] = base64url_encode(rsa_sign($arr['body'],$channel['channel_prvkey']));
$arr['item_flags'] |= ITEM_VERIFIED;
}
}
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
post signatures
2013-10-03 04:04:48 +00:00
$allowed_languages = get_pconfig($arr['uid'],'system','allowed_languages');
if((is_array($allowed_languages)) && ($arr['lang']) && (! array_key_exists($arr['lang'],$allowed_languages))) {
$translate = array('item' => $arr, 'from' => $arr['lang'], 'to' => $allowed_languages, 'translated' => false);
call_hooks('item_translate', $translate);
if((! $translate['translated']) && (intval(get_pconfig($arr['uid'],'system','reject_disallowed_languages')))) {
logger('item_store: language ' . $arr['lang'] . ' not accepted for uid ' . $arr['uid']);
$ret['message'] = 'language not accepted';
return $ret;
}
$arr = $translate['item'];
important bits we need to allow php executable content. These must be explicitly allowed - but only if the account has ACCOUNT_ROLE_ALLOWCODE and *only* for web pages and profile fields. This content cannot be transmitted to other sites.
2013-08-11 23:56:06 +00:00
}
post signatures
2013-10-03 04:04:48 +00:00
if($arr['item_private']) {
$key = get_config('system','pubkey');
$arr['item_flags'] = $arr['item_flags'] | ITEM_OBSCURED;
if($arr['title'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$arr['title'] = json_encode(crypto_encapsulate($arr['title'],$key));
post signatures
2013-10-03 04:04:48 +00:00
if($arr['body'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$arr['body'] = json_encode(crypto_encapsulate($arr['body'],$key));
post signatures
2013-10-03 04:04:48 +00:00
}
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
post signatures
2013-10-03 04:04:48 +00:00
}
important bits we need to allow php executable content. These must be explicitly allowed - but only if the account has ACCOUNT_ROLE_ALLOWCODE and *only* for web pages and profile fields. This content cannot be transmitted to other sites.
2013-08-11 23:56:06 +00:00
post signatures
2013-10-03 04:04:48 +00:00
if((x($arr,'object')) && is_array($arr['object'])) {
activity_sanitise($arr['object']);
$arr['object'] = json_encode($arr['object']);
}
if((x($arr,'target')) && is_array($arr['target'])) {
activity_sanitise($arr['target']);
$arr['target'] = json_encode($arr['target']);
}
if((x($arr,'attach')) && is_array($arr['attach'])) {
activity_sanitise($arr['attach']);
$arr['attach'] = json_encode($arr['attach']);
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
}
post signatures
2013-10-03 04:04:48 +00:00
webpage edit - losing the ability to pass unfiltered html and perhaps some php due to unsetting uid too early in item_store_update()
2013-11-21 00:20:32 +00:00
unset($arr['id']);
unset($arr['uid']);
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
unset($arr['aid']);
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
unset($arr['mid']);
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
unset($arr['parent']);
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
unset($arr['parent_mid']);
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
unset($arr['created']);
unset($arr['author_xchan']);
unset($arr['owner_xchan']);
unset($arr['thr_parent']);
unset($arr['llink']);
$arr['edited'] = ((x($arr,'edited') !== false) ? datetime_convert('UTC','UTC',$arr['edited']) : datetime_convert());
webpage permissions weirdness
2014-06-14 23:35:38 +00:00
$arr['expires'] = ((x($arr,'expires') !== false) ? datetime_convert('UTC','UTC',$arr['expires']) : $orig[0]['expires']);
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
if(array_key_exists('comments_closed',$arr) && $arr['comments_closed'] != NULL_DATE)
Ability to close comments at a certain date/time - needed for loom.io emulation (and many other uses)
2014-08-28 23:56:13 +00:00
$arr['comments_closed'] = datetime_convert('UTC','UTC',$arr['comments_closed']);
else
$arr['comments_closed'] = $orig[0]['comments_closed'];
prevent posts from floating to the top after minor edits
2013-10-20 06:38:07 +00:00
$arr['commented'] = $orig[0]['commented'];
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
$arr['received'] = datetime_convert();
$arr['changed'] = datetime_convert();
missing source route after edit
2014-10-10 21:45:18 +00:00
$arr['route'] = ((array_key_exists('route',$arr)) ? trim($arr['route']) : $orig[0]['route']);
store diaspora meta info in the item table. It has to go there or it will kill us with complex joins. We can phase out the sign table once this all checks out.
2014-09-03 12:09:43 +00:00
$arr['diaspora_meta'] = ((x($arr,'diaspora_meta')) ? $arr['diaspora_meta'] : $orig[0]['diaspora_meta']);
fix a bunch of post editing bugs - though this consolidates some disparate code and may create some new bugs due to regression
2013-10-04 11:47:41 +00:00
$arr['location'] = ((x($arr,'location')) ? notags(trim($arr['location'])) : $orig[0]['location']);
$arr['coord'] = ((x($arr,'coord')) ? notags(trim($arr['coord'])) : $orig[0]['coord']);
$arr['verb'] = ((x($arr,'verb')) ? notags(trim($arr['verb'])) : $orig[0]['verb']);
$arr['obj_type'] = ((x($arr,'obj_type')) ? notags(trim($arr['obj_type'])) : $orig[0]['obj_type']);
$arr['object'] = ((x($arr,'object')) ? trim($arr['object']) : $orig[0]['object']);
$arr['tgt_type'] = ((x($arr,'tgt_type')) ? notags(trim($arr['tgt_type'])) : $orig[0]['tgt_type']);
$arr['target'] = ((x($arr,'target')) ? trim($arr['target']) : $orig[0]['target']);
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
$arr['plink'] = ((x($arr,'plink')) ? notags(trim($arr['plink'])) : $orig[0]['plink']);
webpage permissions weirdness
2014-06-14 23:35:38 +00:00
$arr['allow_cid'] = ((array_key_exists('allow_cid',$arr)) ? trim($arr['allow_cid']) : $orig[0]['allow_cid']);
$arr['allow_gid'] = ((array_key_exists('allow_gid',$arr)) ? trim($arr['allow_gid']) : $orig[0]['allow_gid']);
$arr['deny_cid'] = ((array_key_exists('deny_cid',$arr)) ? trim($arr['deny_cid']) : $orig[0]['deny_cid']);
$arr['deny_gid'] = ((array_key_exists('deny_gid',$arr)) ? trim($arr['deny_gid']) : $orig[0]['deny_gid']);
$arr['item_private'] = ((array_key_exists('item_private',$arr)) ? intval($arr['item_private']) : $orig[0]['item_private']);
profile edit - missing visibility and drop link on non-default profiles, re-arrange order of replacing red#matrix smilie so it works correctly, accept a post with body content of '0' which was interpreted by x() as nothing (was treated as integer).
2014-08-21 23:46:24 +00:00
$arr['title'] = ((array_key_exists('title',$arr) && strlen($arr['title'])) ? trim($arr['title']) : '');
$arr['body'] = ((array_key_exists('body',$arr) && strlen($arr['body'])) ? trim($arr['body']) : '');
fix a bunch of post editing bugs - though this consolidates some disparate code and may create some new bugs due to regression
2013-10-04 11:47:41 +00:00
$arr['attach'] = ((x($arr,'attach')) ? notags(trim($arr['attach'])) : $orig[0]['attach']);
$arr['app'] = ((x($arr,'app')) ? notags(trim($arr['app'])) : $orig[0]['app']);
// $arr['item_restrict'] = ((x($arr,'item_restrict')) ? intval($arr['item_restrict']) : $orig[0]['item_restrict'] );
// $arr['item_flags'] = ((x($arr,'item_flags')) ? intval($arr['item_flags']) : $orig[0]['item_flags'] );
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
fix a bunch of post editing bugs - though this consolidates some disparate code and may create some new bugs due to regression
2013-10-04 11:47:41 +00:00
$arr['sig'] = ((x($arr,'sig')) ? $arr['sig'] : '');
ability to unset a layout and subthread is probably close to working again.
2014-04-24 02:05:48 +00:00
$arr['layout_mid'] = ((array_key_exists('layout_mid',$arr)) ? dbesc($arr['layout_mid']) : $orig[0]['layout_mid'] );
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
first step first - just store the public_policy data, don't do anything with it yet.
2014-08-07 00:36:07 +00:00
$arr['public_policy'] = ((x($arr,'public_policy')) ? notags(trim($arr['public_policy'])) : $orig[0]['public_policy'] );
$arr['comment_policy'] = ((x($arr,'comment_policy')) ? notags(trim($arr['comment_policy'])) : $orig[0]['comment_policy'] );
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
call_hooks('post_remote_update',$arr);
if(x($arr,'cancel')) {
logger('item_store_update: post cancelled by plugin.');
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'cancelled.';
return $ret;
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
}
// pull out all the taxonomy stuff for separate storage
$terms = null;
if(array_key_exists('term',$arr)) {
$terms = $arr['term'];
unset($arr['term']);
}
dbesc_array($arr);
logger('item_store_update: ' . print_r($arr,true), LOGGER_DATA);
$str = '';
foreach($arr as $k => $v) {
if($str)
$str .= ",";
$str .= " `" . $k . "` = '" . $v . "' ";
}
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = dbq("update `item` set " . $str . " where id = " . $orig_post_id );
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
if($r)
logger('item_store_update: updated item ' . $orig_post_id, LOGGER_DEBUG);
else {
logger('item_store_update: could not update item');
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['message'] = 'DB update failed.';
return $ret;
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
}
$r = q("delete from term where oid = %d and otype = %d",
intval($orig_post_id),
intval(TERM_OBJ_POST)
);
if(($terms) && (is_array($terms))) {
foreach($terms as $t) {
q("insert into term (uid,oid,otype,type,term,url)
values(%d,%d,%d,%d,'%s','%s') ",
intval($uid),
intval($orig_post_id),
intval(TERM_OBJ_POST),
intval($t['type']),
dbesc($t['term']),
dbesc($t['url'])
);
}
$arr['term'] = $terms;
}
call_hooks('post_remote_update_end',$arr);
send_status_notifications($orig_post_id,$arr);
tag_deliver($uid,$orig_post_id);
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$ret['success'] = true;
$ret['item_id'] = $orig_post_id;
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
return $ret;
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
}
improved wall-to-wall detection for comments so we can handle Diaspora signing and wall-to-wall attribution correctly. Do it at the point of submission. This also fixes a potential bug in yesterday's wall-to-wall permission setting, if it was a local comment to a remote post.
2014-10-23 03:39:49 +00:00
function store_diaspora_comment_sig($datarray, $channel, $parent_item, $post_id, $walltowall = false) {
various diaspora issues
2014-08-30 00:31:40 +00:00
// We won't be able to sign Diaspora comments for authenticated visitors
// - we don't have their private key
// since Diaspora doesn't handle edits we can only do this for the original text and not update it.
require_once('include/bb2diaspora.php');
diaspora walltowall comment signature, force update.
2014-10-23 23:08:29 +00:00
$signed_body = bb2diaspora_itembody($datarray,$walltowall);
various diaspora issues
2014-08-30 00:31:40 +00:00
improved wall-to-wall detection for comments so we can handle Diaspora signing and wall-to-wall attribution correctly. Do it at the point of submission. This also fixes a potential bug in yesterday's wall-to-wall permission setting, if it was a local comment to a remote post.
2014-10-23 03:39:49 +00:00
if($walltowall) {
logger('wall to wall comment',LOGGER_DEBUG);
// post will come across with the owner's identity. Throw a preamble onto the post to indicate the true author.
$signed_body = "\n\n"
doh - this needs to be markdown, not bbcode
2014-10-24 21:18:02 +00:00
. '![' . $datarray['author']['xchan_name'] . '](' . $datarray['author']['xchan_photo_m'] . ')'
. '[' . $datarray['author']['xchan_name'] . '](' . $datarray['author']['xchan_url'] . ')' . "\n\n"
improved wall-to-wall detection for comments so we can handle Diaspora signing and wall-to-wall attribution correctly. Do it at the point of submission. This also fixes a potential bug in yesterday's wall-to-wall permission setting, if it was a local comment to a remote post.
2014-10-23 03:39:49 +00:00
. $signed_body;
}
logger('storing diaspora comment signature',LOGGER_DEBUG);
various diaspora issues
2014-08-30 00:31:40 +00:00
$diaspora_handle = $channel['channel_address'] . '@' . get_app()->get_hostname();
$signed_text = $datarray['mid'] . ';' . $parent_item['mid'] . ';' . $signed_body . ';' . $diaspora_handle;
if( $uprvkey !== false )
$authorsig = base64_encode(rsa_sign($signed_text,$channel['channel_prvkey'],'sha256'));
else
$authorsig = '';
store diaspora meta info in the item table. It has to go there or it will kill us with complex joins. We can phase out the sign table once this all checks out.
2014-09-03 12:09:43 +00:00
$x = array('signer' => $diaspora_handle, 'body' => $signed_body, 'signed_text' => $signed_text, 'signature' => base64_encode($authorsig));
encrypt diaspora_meta since it may contain private message contents
2014-09-11 02:02:00 +00:00
$key = get_config('system','pubkey');
$y = crypto_encapsulate(json_encode($x),$key);
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("update item set diaspora_meta = '%s' where id = %d",
encrypt diaspora_meta since it may contain private message contents
2014-09-11 02:02:00 +00:00
dbesc(json_encode($y)),
store diaspora meta info in the item table. It has to go there or it will kill us with complex joins. We can phase out the sign table once this all checks out.
2014-09-03 12:09:43 +00:00
intval($post_id)
);
various diaspora issues
2014-08-30 00:31:40 +00:00
if(! $r)
logger('store_diaspora_comment_sig: DB write failed');
return;
}
this might make edits update across the wire, or it might not
2013-03-01 02:20:42 +00:00
work on notifications
2013-02-07 04:29:17 +00:00
function send_status_notifications($post_id,$item) {
$notify = false;
$parent = 0;
$r = q("select channel_hash from channel where channel_id = %d limit 1",
intval($item['uid'])
);
if(! $r)
return;
// my own post - no notification needed
if($item['author_xchan'] === $r[0]['channel_hash'])
return;
// I'm the owner - notify me
if($item['owner_hash'] === $r[0]['channel_hash'])
$notify = true;
// Was I involved in this conversation?
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$x = q("select * from item where parent_mid = '%s' and uid = %d",
dbesc($item['parent_mid']),
work on notifications
2013-02-07 04:29:17 +00:00
intval($item['uid'])
);
if($x) {
foreach($x as $xx) {
if($xx['author_xchan'] === $r[0]['channel_hash']) {
$notify = true;
implement sender side of public scope setting (public post, network-wide, site-wide, etc....)
2013-02-15 04:17:30 +00:00
}
if($xx['id'] == $xx['parent']) {
$parent = $xx['parent'];
work on notifications
2013-02-07 04:29:17 +00:00
}
}
}
This would be about the 75th attempt to try and prevent duplicated email notifications for comments. Eventually we'll find something that works.
2014-02-05 23:01:53 +00:00
$link = get_app()->get_baseurl() . '/display/' . $item['mid'];
notification bug - don't use $r, we already set it to something else that we need further on.
2014-02-06 21:32:29 +00:00
$y = q("select id from notify where link = '%s' and uid = %d limit 1",
This would be about the 75th attempt to try and prevent duplicated email notifications for comments. Eventually we'll find something that works.
2014-02-05 23:01:53 +00:00
dbesc($link),
intval($item['uid'])
);
notification bug - don't use $r, we already set it to something else that we need further on.
2014-02-06 21:32:29 +00:00
if($y)
This would be about the 75th attempt to try and prevent duplicated email notifications for comments. Eventually we'll find something that works.
2014-02-05 23:01:53 +00:00
$notify = false;
work on notifications
2013-02-07 04:29:17 +00:00
if(! $notify)
return;
require_once('include/enotify.php');
notification(array(
'type' => NOTIFY_COMMENT,
'from_xchan' => $item['author_xchan'],
'to_xchan' => $r[0]['channel_hash'],
'item' => $item,
This would be about the 75th attempt to try and prevent duplicated email notifications for comments. Eventually we'll find something that works.
2014-02-05 23:01:53 +00:00
'link' => $link,
work on notifications
2013-02-07 04:29:17 +00:00
'verb' => ACTIVITY_POST,
'otype' => 'item',
'parent' => $parent,
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
'parent_mid' => $item['parent_mid']
work on notifications
2013-02-07 04:29:17 +00:00
));
return;
}
preliminary network abstraction, configurable debugging.
2010-09-26 23:30:21 +00:00
function get_item_contact($item,$contacts) {
if(! count($contacts) || (! is_array($item)))
return false;
foreach($contacts as $contact) {
if($contact['id'] == $item['contact-id']) {
return $contact;
break; // NOTREACHED
}
}
return false;
}
item tag notifications
2012-02-10 05:18:50 +00:00
function tag_deliver($uid,$item_id) {
tgroup relays
2011-11-16 04:30:34 +00:00
This should get community tagging pretty close to working - deleting a community tag is left as a FIXME
2013-06-05 05:52:17 +00:00
// Called when we deliver things that might be tagged in ways that require delivery processing.
// Handles community tagging of posts and also look for mention tags
// and sets up a second delivery chain if appropriate
new approach to secondary delivery fork
2011-11-19 06:20:08 +00:00
tgroup relays
2011-11-16 04:30:34 +00:00
$a = get_app();
item tag notifications
2012-02-10 05:18:50 +00:00
$mention = false;
tgroup relays
2011-11-16 04:30:34 +00:00
snakebite continued
2014-08-20 04:31:33 +00:00
/**
* Fetch stuff we need - a channel and an item
*/
more stuff we don't need.
2012-10-08 07:23:43 +00:00
$u = q("select * from channel where channel_id = %d limit 1",
item tag notifications
2012-02-10 05:18:50 +00:00
intval($uid)
tgroup relays
2011-11-16 04:30:34 +00:00
);
progress on tag delivery
2013-02-11 08:20:14 +00:00
if(! $u)
tgroup relays
2011-11-16 04:30:34 +00:00
return;
more stuff we don't need.
2012-10-08 07:23:43 +00:00
forum cleanup
2011-11-19 11:13:46 +00:00
$i = q("select * from item where id = %d and uid = %d limit 1",
tgroup relays
2011-11-16 04:30:34 +00:00
intval($item_id),
intval($uid)
);
This should get community tagging pretty close to working - deleting a community tag is left as a FIXME
2013-06-05 05:52:17 +00:00
if(! $i)
tgroup relays
2011-11-16 04:30:34 +00:00
return;
start on tag deliver rewrite
2013-01-13 02:06:34 +00:00
$i = fetch_post_tags($i);
tgroup relays
2011-11-16 04:30:34 +00:00
$item = $i[0];
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
if(($item['source_xchan']) && ($item['item_flags'] & ITEM_UPLINK)
&& ($item['item_flags'] & ITEM_THREAD_TOP) && ($item['edited'] != $item['created'])) {
// this is an update (edit) to a post which was already processed by us and has a second delivery chain
sourced items which are then edited at the source weren't setting up the second delivery chain.
2013-12-20 04:23:19 +00:00
// Just start the second delivery chain to deliver the updated post
proc_run('php','include/notifier.php','tgroup',$item['id']);
return;
}
This should get community tagging pretty close to working - deleting a community tag is left as a FIXME
2013-06-05 05:52:17 +00:00
snakebite continued
2014-08-20 04:31:33 +00:00
/**
* Seems like a good place to plug in a poke notification.
*/
poke notifications
2014-06-05 05:15:52 +00:00
if (stristr($item['verb'],ACTIVITY_POKE)) {
$poke_notify = true;
if(($item['obj_type'] == "") || ($item['obj_type'] !== ACTIVITY_OBJ_PERSON) || (! $item['object']))
$poke_notify = false;
$obj = json_decode_plus($item['object']);
if($obj) {
if($obj['id'] !== $u[0]['channel_hash'])
$poke_notify = false;
}
$verb = urldecode(substr($item['verb'],strpos($item['verb'],'#')+1));
if($poke_notify) {
require_once('include/enotify.php');
notification(array(
'to_xchan' => $u[0]['channel_hash'],
'from_xchan' => $item['author_xchan'],
'type' => NOTIFY_POKE,
'item' => $item,
'link' => $i[0]['llink'],
'verb' => ACTIVITY_POKE,
'activity' => $verb,
'otype' => 'item'
));
}
}
snakebite continued
2014-08-20 04:31:33 +00:00
/**
* Do community tagging
*/
This should get community tagging pretty close to working - deleting a community tag is left as a FIXME
2013-06-05 05:52:17 +00:00
if($item['obj_type'] === ACTIVITY_OBJ_TAGTERM) {
// We received a community tag activity for a post.
// See if we are the owner of the parent item and have given permission to tag our posts.
// If so tag the parent post.
community tag issues
2013-06-21 04:19:22 +00:00
logger('tag_deliver: community tag activity received');
This should get community tagging pretty close to working - deleting a community tag is left as a FIXME
2013-06-05 05:52:17 +00:00
if(($item['owner_xchan'] === $u[0]['channel_hash']) && (! get_pconfig($u[0]['channel_id'],'system','blocktags'))) {
issue #378, failure to correctly link tags where one tag contains a substring of the other (depends on the order they are discovered).
2014-03-31 03:53:59 +00:00
logger('tag_deliver: community tag recipient: ' . $u[0]['channel_name']);
convert all stored json calls to json_decode_plus()
2013-08-06 10:54:49 +00:00
$j_tgt = json_decode_plus($item['target']);
community tag issues
2013-06-21 04:19:22 +00:00
if($j_tgt && $j_tgt['id']) {
This should get community tagging pretty close to working - deleting a community tag is left as a FIXME
2013-06-05 05:52:17 +00:00
$p = q("select * from item where mid = '%s' and uid = %d limit 1",
community tag issues
2013-06-21 04:19:22 +00:00
dbesc($j_tgt['id']),
This should get community tagging pretty close to working - deleting a community tag is left as a FIXME
2013-06-05 05:52:17 +00:00
intval($u[0]['channel_id'])
);
if($p) {
convert all stored json calls to json_decode_plus()
2013-08-06 10:54:49 +00:00
$j_obj = json_decode_plus($item['object']);
community tag issues
2013-06-21 04:19:22 +00:00
logger('tag_deliver: tag object: ' . print_r($j_obj,true), LOGGER_DATA);
This should get community tagging pretty close to working - deleting a community tag is left as a FIXME
2013-06-05 05:52:17 +00:00
if($j_obj && $j_obj['id'] && $j_obj['title']) {
fix some community tagging bugs
2013-07-01 06:04:27 +00:00
if(is_array($j_obj['link']))
$taglink = get_rel_link($j_obj['link'],'alternate');
community tags don't show up in the tagged posts because we removed the displayed list of hashtags at the bottom (which are usually redundant since hashtags are usually present in the message text). This brings them back, but only for tags which aren't currently present in the message body (e.g. community tags)
2014-04-01 00:03:07 +00:00
fix some community tagging bugs
2013-07-01 06:04:27 +00:00
store_item_tag($u[0]['channel_id'],$p[0]['id'],TERM_OBJ_POST,TERM_HASHTAG,$j_obj['title'],$j_obj['id']);
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$x = q("update item set edited = '%s', received = '%s', changed = '%s' where mid = '%s' and uid = %d",
several things were not working correctly w/r/t community tagging. The preference vanished from settings at some point, and we also weren't updating the original post timestamp so that the changed taxonomy would propagate correctly as an edit.
2014-02-17 00:04:46 +00:00
dbesc(datetime_convert()),
dbesc(datetime_convert()),
dbesc(datetime_convert()),
dbesc($j_tgt['id']),
intval($u[0]['channel_id'])
);
This should get community tagging pretty close to working - deleting a community tag is left as a FIXME
2013-06-05 05:52:17 +00:00
proc_run('php','include/notifier.php','edit_post',$p[0]['id']);
}
}
}
}
community tag issues
2013-06-21 04:19:22 +00:00
else
logger('tag_deliver: tag permission denied for ' . $u[0]['channel_address']);
This should get community tagging pretty close to working - deleting a community tag is left as a FIXME
2013-06-05 05:52:17 +00:00
}
snakebite continued
2014-08-20 04:31:33 +00:00
/**
* A "union" is a message which our channel has sourced from another channel.
* This sets up a second delivery chain just like forum tags do.
* Find out if this is a source-able post.
*/
aggregate channels
2013-09-27 02:34:45 +00:00
$union = check_item_source($uid,$item);
if($union)
logger('check_item_source returns true');
fix tag delivery default permissions (use channel_allow_cid rather than allow_cid etc.)
2014-02-13 22:20:23 +00:00
// This might be a followup (e.g. comment) by the original post author to a tagged forum
doc update, plus possible fix to "complex delivery chains" issue. Basically when the top-level post creator replies to a thread he sent to a forum via tags, it isn't getting redelivered to the forum members.
2013-06-29 09:04:44 +00:00
// If so setup a second delivery chain
if( ! ($item['item_flags'] & ITEM_THREAD_TOP)) {
$x = q("select * from item where id = parent and parent = %d and uid = %d limit 1",
intval($item['parent']),
intval($uid)
);
issue #59 - this is a real bugger. Might seem fixed but read the code. This could potentially create a security/permissions issue
2013-09-05 06:15:53 +00:00
if(($x) && ($x[0]['item_flags'] & ITEM_UPLINK)) {
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
start_delivery_chain($u[0],$item,$item_id,$x[0]);
doc update, plus possible fix to "complex delivery chains" issue. Basically when the top-level post creator replies to a thread he sent to a forum via tags, it isn't getting redelivered to the forum members.
2013-06-29 09:04:44 +00:00
}
}
snakebite continued
2014-08-20 04:31:33 +00:00
/**
* Now we've got those out of the way. Let's see if this is a post that's tagged for re-delivery
*/
start on tag deliver rewrite
2013-01-13 02:06:34 +00:00
$terms = get_terms_oftype($item['term'],TERM_MENTION);
doc update, plus possible fix to "complex delivery chains" issue. Basically when the top-level post creator replies to a thread he sent to a forum via tags, it isn't getting redelivered to the forum members.
2013-06-29 09:04:44 +00:00
if($terms)
logger('tag_deliver: post mentions: ' . print_r($terms,true), LOGGER_DATA);
start on tag deliver rewrite
2013-01-13 02:06:34 +00:00
debug tag_deliver
2013-02-11 07:19:52 +00:00
$link = normalise_link($a->get_baseurl() . '/channel/' . $u[0]['channel_address']);
tgroup relays
2011-11-16 04:30:34 +00:00
start on tag deliver rewrite
2013-01-13 02:06:34 +00:00
if($terms) {
foreach($terms as $term) {
Make tag delivery work when channel name differs from profile name
2014-05-21 20:23:18 +00:00
if(link_compare($term['url'],$link)) {
item tag notifications
2012-02-10 05:18:50 +00:00
$mention = true;
start on tag deliver rewrite
2013-01-13 02:06:34 +00:00
break;
tgroup relays
2011-11-16 04:30:34 +00:00
}
}
start on tag deliver rewrite
2013-01-13 02:06:34 +00:00
}
tgroup relays
2011-11-16 04:30:34 +00:00
start on tag deliver rewrite
2013-01-13 02:06:34 +00:00
if($mention) {
debug tag_deliver
2013-02-11 07:19:52 +00:00
logger('tag_deliver: mention found for ' . $u[0]['channel_name']);
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("update item set item_flags = ( item_flags | %d ) where id = %d",
start on tag deliver rewrite
2013-01-13 02:06:34 +00:00
intval(ITEM_MENTIONSME),
intval($item_id)
);
tgroup relays
2011-11-16 04:30:34 +00:00
aggregate channels
2013-09-27 02:34:45 +00:00
// At this point we've determined that the person receiving this post was mentioned in it or it is a union.
// Now let's check if this mention was inside a reshare so we don't spam a forum
private tagged posts weren't creating second delivery chain - they were incorrectly being marked as "in a reshare" due to being obscured.
2013-11-05 00:44:23 +00:00
// If it's private we may have to unobscure it momentarily so that we can parse it.
$body = '';
if($item['item_flags'] & ITEM_OBSCURED) {
$key = get_config('system','prvkey');
if($item['body'])
Protocol: now set data['alg'] on all encapsulated encrypted packets, so that we can more easily retire 'aes256cbc' once it is no longer viable.
2013-11-20 23:20:12 +00:00
$body = crypto_unencapsulate(json_decode_plus($item['body']),$key);
private tagged posts weren't creating second delivery chain - they were incorrectly being marked as "in a reshare" due to being obscured.
2013-11-05 00:44:23 +00:00
}
else
$body = $item['body'];
ignore tagged items in reshares from notifications and tag_delivery
2013-01-13 02:35:27 +00:00
private tagged posts weren't creating second delivery chain - they were incorrectly being marked as "in a reshare" due to being obscured.
2013-11-05 00:44:23 +00:00
$body = preg_replace('/\[share(.*?)\[\/share\]/','',$body);
ignore tagged items in reshares from notifications and tag_delivery
2013-01-13 02:35:27 +00:00
this should fix mention notification that aren't plus tagged
2014-04-20 05:53:42 +00:00
$tagged = false;
$plustagged = false;
Make tag delivery work when channel name differs from profile name
2014-05-21 20:23:18 +00:00
$pattern = '/@\!?\[zrl\=' . preg_quote($term['url'],'/') . '\]' . preg_quote($term['term'],'/') . '\[\/zrl\]/';
this should fix mention notification that aren't plus tagged
2014-04-20 05:53:42 +00:00
if(preg_match($pattern,$body,$matches))
$tagged = true;
Make tag delivery work when channel name differs from profile name
2014-05-21 20:23:18 +00:00
$pattern = '/@\!?\[zrl\=' . preg_quote($term['url'],'/') . '\]' . preg_quote($term['term'] . '+','/') . '\[\/zrl\]/';
this should fix mention notification that aren't plus tagged
2014-04-20 05:53:42 +00:00
if(preg_match($pattern,$body,$matches))
$plustagged = true;
tag notifications now work.
2013-02-11 23:51:43 +00:00
this should fix mention notification that aren't plus tagged
2014-04-20 05:53:42 +00:00
if(! ($tagged || $plustagged)) {
aggregate channels
2013-09-27 02:34:45 +00:00
logger('tag_deliver: mention was in a reshare - ignoring');
return;
}
ignore tagged items in reshares from notifications and tag_delivery
2013-01-13 02:35:27 +00:00
A useful hook
2014-04-26 04:12:43 +00:00
$arr = array('channel_id' => $uid, 'item' => $item, 'body' => $body);
call_hooks('tagged',$arr);
this should fix mention notification that aren't plus tagged
2014-04-20 05:53:42 +00:00
snakebite continued
2014-08-20 04:31:33 +00:00
/**
* Kill two birds with one stone. As long as we're here, send a mention notification.
*/
more private forums, default privacy group for new contacts
2012-05-18 05:44:52 +00:00
aggregate channels
2013-09-27 02:34:45 +00:00
require_once('include/enotify.php');
notification(array(
'to_xchan' => $u[0]['channel_hash'],
'from_xchan' => $item['author_xchan'],
'type' => NOTIFY_TAGSELF,
'item' => $item,
'link' => $i[0]['llink'],
'verb' => ACTIVITY_TAG,
'otype' => 'item'
));
setup delivery chain for private groups (work in progress)
2012-05-18 02:35:24 +00:00
this should fix mention notification that aren't plus tagged
2014-04-20 05:53:42 +00:00
// Just a normal tag?
if(! $plustagged) {
logger('tag_deliver: not a plus tag', LOGGER_DEBUG);
return;
}
// plustagged - keep going, next check permissions
start on tag deliver rewrite
2013-01-13 02:06:34 +00:00
aggregate channels
2013-09-27 02:34:45 +00:00
if(! perm_is_allowed($uid,$item['author_xchan'],'tag_deliver')) {
logger('tag_delivery denied for uid ' . $uid . ' and xchan ' . $item['author_xchan']);
return;
}
add extra logging to tag_deliver
2013-08-06 04:42:55 +00:00
}
setup delivery chain for private groups (work in progress)
2012-05-18 02:35:24 +00:00
dav: throw exception if channel for requested DAV directory is deleted
2014-01-15 11:12:47 +00:00
if((! $mention) && (! $union)) {
logger('tag_deliver: no mention and no union.');
aggregate channels
2013-09-27 02:34:45 +00:00
return;
dav: throw exception if channel for requested DAV directory is deleted
2014-01-15 11:12:47 +00:00
}
aggregate channels
2013-09-27 02:34:45 +00:00
send notify emails to community pages as well as people
2012-02-10 08:08:23 +00:00
// tgroup delivery - setup a second delivery chain
// prevent delivery looping - only proceed
// if the message originated elsewhere and is a top-level post
new approach to secondary delivery fork
2011-11-19 06:20:08 +00:00
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
if(($item['item_flags'] & ITEM_WALL)
|| ($item['item_flags'] & ITEM_ORIGIN)
|| (!($item['item_flags'] & ITEM_THREAD_TOP))
|| ($item['id'] != $item['parent'])) {
add extra logging to tag_deliver
2013-08-06 04:42:55 +00:00
logger('tag_deliver: item was local or a comment. rejected.');
this should make tag delivery (e.g. forums) work - but not yet fully tested
2013-01-14 02:35:12 +00:00
return;
add extra logging to tag_deliver
2013-08-06 04:42:55 +00:00
}
tgroup relays
2011-11-16 04:30:34 +00:00
more progresss on tag delivery
2013-02-12 00:59:48 +00:00
logger('tag_deliver: creating second delivery chain.');
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
start_delivery_chain($u[0],$item,$item_id,null);
send notify emails to community pages as well as people
2012-02-10 08:08:23 +00:00
tgroup relays
2011-11-16 04:30:34 +00:00
}
snakebite continued
2014-08-20 04:31:33 +00:00
/**
* @function tgroup_check($uid,$item)
*
* This function is called pre-deliver to see if a post matches the criteria to be tag delivered.
* We don't actually do anything except check that it matches the criteria.
* This is so that the channel with tag_delivery enabled can receive the post even if they turn off
* permissions for the sender to send their stream. tag_deliver() can't be called until the post is actually stored.
* By then it would be too late to reject it.
*/
tgroup relays
2011-11-16 04:30:34 +00:00
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
function tgroup_check($uid,$item) {
$a = get_app();
$mention = false;
// check that the message originated elsewhere and is a top-level post
Tricky little bug. Allowed somebody to bypass comment permissions. Hopefully the fix will have no undesired side effects.
2013-12-17 02:07:41 +00:00
// or is a followup and we have already accepted the top level post as an uplink
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
potentially fixes the "federation" bug w/r/t/ tagged forums
2013-08-02 04:24:55 +00:00
if($item['mid'] != $item['parent_mid']) {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("select id from item where mid = '%s' and uid = %d and ( item_flags & %d )>0 limit 1",
potentially fixes the "federation" bug w/r/t/ tagged forums
2013-08-02 04:24:55 +00:00
dbesc($item['parent_mid']),
Tricky little bug. Allowed somebody to bypass comment permissions. Hopefully the fix will have no undesired side effects.
2013-12-17 02:07:41 +00:00
intval($uid),
intval(ITEM_UPLINK)
tgroup_check wasn't allowing comments through.
2013-07-29 23:51:44 +00:00
);
if($r)
return true;
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
return false;
tgroup_check wasn't allowing comments through.
2013-07-29 23:51:44 +00:00
}
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
if(! perm_is_allowed($uid,$item['author_xchan'],'tag_deliver'))
return false;
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
$u = q("select * from channel where channel_id = %d limit 1",
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
intval($uid)
);
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
if(! $u)
return false;
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
$terms = get_terms_oftype($item['term'],TERM_MENTION);
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
add channel_name to delivery logs because it is a bit more useful than an xchan_hash
2013-07-16 02:04:23 +00:00
if($terms)
logger('tgroup_check: post mentions: ' . print_r($terms,true), LOGGER_DATA);
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
$link = normalise_link($a->get_baseurl() . '/channel/' . $u[0]['channel_address']);
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
if($terms) {
foreach($terms as $term) {
Make tag delivery work when channel name differs from profile name
2014-05-21 20:23:18 +00:00
if(link_compare($term['url'],$link)) {
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
$mention = true;
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
break;
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
}
}
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
}
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
if($mention) {
logger('tgroup_check: mention found for ' . $u[0]['channel_name']);
}
else
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
return false;
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
// At this point we've determined that the person receiving this post was mentioned in it.
// Now let's check if this mention was inside a reshare so we don't spam a forum
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
tgroup_check - decrypt if obscured before checking body text for reshares
2014-11-07 21:06:56 +00:00
$body = $item['body'];
if(array_key_exists('item_flags',$item) && ($item['item_flags'] & ITEM_OBSCURED) && $body) {
$key = get_config('system','prvkey');
$body = crypto_unencapsulate(json_decode($body,true),$key);
}
$body = preg_replace('/\[share(.*?)\[\/share\]/','',$body);
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
Make tag delivery work when channel name differs from profile name
2014-05-21 20:23:18 +00:00
$pattern = '/@\!?\[zrl\=' . preg_quote($term['url'],'/') . '\]' . preg_quote($term['term'] . '+','/') . '\[\/zrl\]/';
hopefully this won't screw up everything - if it does, revert. Otherwise this should work at delivery time to check tag deliveries and bounce the message before it's stored if the channel doesn't allow you to post and you aren't allowed to tag deliver either. Previously this was handled after the post was already stored so you needed posting permission as well as tag deliver permission to get past the checks.
2013-06-16 07:57:39 +00:00
if(! preg_match($pattern,$body,$matches)) {
logger('tgroup_check: mention was in a reshare - ignoring');
return false;
}
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
return true;
}
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
/**
* Sourced and tag-delivered posts are re-targetted for delivery to the connections of the channel
* receiving the post. This starts the second delivery chain, by resetting permissions and ensuring
* that ITEM_UPLINK is set on the parent post, and storing the current owner_xchan as the source_xchan.
* We'll become the new owner. If called without $parent, this *is* the parent post.
*/
function start_delivery_chain($channel,$item,$item_id,$parent) {
// Change this copy of the post to a forum head message and deliver to all the tgroup members
// also reset all the privacy bits to the forum default permissions
$private = (($channel['channel_allow_cid'] || $channel['channel_allow_gid']
|| $channel['channel_deny_cid'] || $channel['channel_deny_gid']) ? 1 : 0);
$new_public_policy = map_scope($channel['channel_r_stream'],true);
if((! $private) && $new_public_policy)
$private = 1;
$flag_bits = $item['item_flags'] | ITEM_WALL|ITEM_ORIGIN;
unset item_nocomment when starting a new delivery chain
2014-09-05 01:52:37 +00:00
// unset the nocomment bit if it's there.
if($flag_bits & ITEM_NOCOMMENT)
$flag_bits = $flag_bits ^ ITEM_NOCOMMENT;
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
// maintain the original source, which will be the original item owner and was stored in source_xchan
// when we created the delivery fork
if($parent) {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("update item set source_xchan = '%s' where id = %d",
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
dbesc($parent['source_xchan']),
intval($item_id)
);
}
else {
$flag_bits = $flag_bits | ITEM_UPLINK;
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("update item set source_xchan = owner_xchan where id = %d",
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
intval($item_id)
);
}
$title = $item['title'];
$body = $item['body'];
if($private) {
if(!($flag_bits & ITEM_OBSCURED)) {
$key = get_config('system','pubkey');
$flag_bits = $flag_bits|ITEM_OBSCURED;
"false" issue
2014-08-23 00:34:18 +00:00
if($title)
switch everything over to crypto_encapsulate()
2014-08-23 09:45:21 +00:00
$title = json_encode(crypto_encapsulate($title,$key));
"false" issue
2014-08-23 00:34:18 +00:00
if($body)
switch everything over to crypto_encapsulate()
2014-08-23 09:45:21 +00:00
$body = json_encode(crypto_encapsulate($body,$key));
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
}
}
else {
if($flag_bits & ITEM_OBSCURED) {
$key = get_config('system','prvkey');
$flag_bits = $flag_bits ^ ITEM_OBSCURED;
"false" issue
2014-08-23 00:34:18 +00:00
if($title)
$title = crypto_unencapsulate(json_decode($title,true),$key);
if($body)
$body = crypto_unencapsulate(json_decode($body,true),$key);
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
}
}
$r = q("update item set item_flags = %d, owner_xchan = '%s', allow_cid = '%s', allow_gid = '%s',
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
deny_cid = '%s', deny_gid = '%s', item_private = %d, public_policy = '%s', comment_policy = '%s', title = '%s', body = '%s' where id = %d",
extensible profiles - add the input form elements. Still needs styling and we're not yet storing the results. And right now we're ignoring the type and only doing type=string. Oh yeah - and I split off starting the second delivery chain into its own function in tag_deliver since it was largely duplicate code.
2014-08-21 06:01:25 +00:00
intval($flag_bits),
dbesc($channel['channel_hash']),
dbesc($channel['channel_allow_cid']),
dbesc($channel['channel_allow_gid']),
dbesc($channel['channel_deny_cid']),
dbesc($channel['channel_deny_gid']),
intval($private),
dbesc($new_public_policy),
dbesc(map_scope($channel['channel_w_comment'])),
dbesc($title),
dbesc($body),
intval($item_id)
);
if($r)
proc_run('php','include/notifier.php','tgroup',$item_id);
else
logger('start_delivery_chain: failed to update item');
return;
}
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
aggregate channels
2013-09-27 02:34:45 +00:00
/**
* @function check_item_source($uid,$item)
* @param $uid
* @param $item
*
* @description
* Checks to see if this item owner is referenced as a source for this channel and if the post
* matches the rules for inclusion in this channel. Returns true if we should create a second delivery
* chain and false if none of the rules apply, or if the item is private.
*/
function check_item_source($uid,$item) {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("select * from source where src_channel_id = %d and ( src_xchan = '%s' or src_xchan = '*' ) limit 1",
aggregate channels
2013-09-27 02:34:45 +00:00
intval($uid),
sourced items which are then edited at the source weren't setting up the second delivery chain.
2013-12-20 04:23:19 +00:00
dbesc(($item['source_xchan']) ? $item['source_xchan'] : $item['owner_xchan'])
aggregate channels
2013-09-27 02:34:45 +00:00
);
unset item_nocomment when starting a new delivery chain
2014-09-05 01:52:37 +00:00
aggregate channels
2013-09-27 02:34:45 +00:00
if(! $r)
return false;
oops forgot about the privacy exclusion
2014-09-05 01:06:59 +00:00
$x = q("select abook_their_perms, abook_flags from abook where abook_channel = %d and abook_xchan = '%s' limit 1",
implement republish permission for use in sourced channels
2013-10-01 04:49:26 +00:00
intval($uid),
dbesc($item['owner_xchan'])
);
unset item_nocomment when starting a new delivery chain
2014-09-05 01:52:37 +00:00
implement republish permission for use in sourced channels
2013-10-01 04:49:26 +00:00
if(! $x)
return false;
unset item_nocomment when starting a new delivery chain
2014-09-05 01:52:37 +00:00
implement republish permission for use in sourced channels
2013-10-01 04:49:26 +00:00
if(! ($x[0]['abook_their_perms'] & PERMS_A_REPUBLISH))
return false;
oops forgot about the privacy exclusion
2014-09-05 01:06:59 +00:00
if($item['item_private'] && (! ($x[0]['abook_flags'] & ABOOK_FLAG_FEED)))
return false;
aggregate channels
2013-09-27 02:34:45 +00:00
if($r[0]['src_channel_xchan'] === $item['owner_xchan'])
return false;
if(! $r[0]['src_patt'])
return true;
convert to plaintext before matching words in body. Otherwise every post I make with a link in it is going to match a test for zot.
2013-09-27 02:58:24 +00:00
require_once('include/html2plain.php');
$text = prepare_text($item['body'],$item['mimetype']);
$text = html2plain($text);
aggregate channels
2013-09-27 02:34:45 +00:00
$tags = ((count($items['term'])) ? $items['term'] : false);
$words = explode("\n",$r[0]['src_patt']);
if($words) {
foreach($words as $word) {
if(substr($word,0,1) === '#' && $tags) {
foreach($tags as $t)
provide some interesting new options to channel sources
2014-02-01 04:02:59 +00:00
if(($t['type'] == TERM_HASHTAG) && ((substr($t,1) === substr($word,1)) || (substr($word,1) === '*')))
aggregate channels
2013-09-27 02:34:45 +00:00
return true;
}
convert to plaintext before matching words in body. Otherwise every post I make with a link in it is going to match a test for zot.
2013-09-27 02:58:24 +00:00
if(stristr($text,$word) !== false)
aggregate channels
2013-09-27 02:34:45 +00:00
return true;
}
}
return false;
}
y'all got mail
2012-12-06 00:44:07 +00:00
function mail_store($arr) {
if(! $arr['channel_id']) {
logger('mail_store: no uid');
return 0;
}
if((strpos($arr['body'],'<') !== false) || (strpos($arr['body'],'>') !== false))
$arr['body'] = escape_tags($arr['body']);
email attachments (and fix email photos)
2013-08-20 04:19:39 +00:00
if(array_key_exists('attach',$arr) && is_array($arr['attach']))
$arr['attach'] = json_encode($arr['attach']);
y'all got mail
2012-12-06 00:44:07 +00:00
$arr['account_id'] = ((x($arr,'account_id')) ? intval($arr['account_id']) : 0);
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$arr['mid'] = ((x($arr,'mid')) ? notags(trim($arr['mid'])) : random_string());
y'all got mail
2012-12-06 00:44:07 +00:00
$arr['from_xchan'] = ((x($arr,'from_xchan')) ? notags(trim($arr['from_xchan'])) : '');
include channel clones in PM deliveries (also some whitespace edits)
2015-01-09 00:02:15 +00:00
$arr['to_xchan'] = ((x($arr,'to_xchan')) ? notags(trim($arr['to_xchan'])) : '');
y'all got mail
2012-12-06 00:44:07 +00:00
$arr['created'] = ((x($arr,'created') !== false) ? datetime_convert('UTC','UTC',$arr['created']) : datetime_convert());
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
$arr['expires'] = ((x($arr,'expires') !== false) ? datetime_convert('UTC','UTC',$arr['expires']) : NULL_DATE);
y'all got mail
2012-12-06 00:44:07 +00:00
$arr['title'] = ((x($arr,'title')) ? notags(trim($arr['title'])) : '');
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$arr['parent_mid'] = ((x($arr,'parent_mid')) ? notags(trim($arr['parent_mid'])) : '');
y'all got mail
2012-12-06 00:44:07 +00:00
$arr['body'] = ((x($arr,'body')) ? trim($arr['body']) : '');
email attachments (and fix email photos)
2013-08-20 04:19:39 +00:00
y'all got mail
2012-12-06 00:44:07 +00:00
$arr['mail_flags'] = ((x($arr,'mail_flags')) ? intval($arr['mail_flags']) : 0 );
make mail replies work
2012-12-06 01:11:38 +00:00
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
if(! $arr['parent_mid']) {
make mail replies work
2012-12-06 01:11:38 +00:00
logger('mail_store: missing parent');
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$arr['parent_mid'] = $arr['mid'];
make mail replies work
2012-12-06 01:11:38 +00:00
}
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$r = q("SELECT `id` FROM mail WHERE `mid` = '%s' AND channel_id = %d LIMIT 1",
dbesc($arr['mid']),
y'all got mail
2012-12-06 00:44:07 +00:00
intval($arr['channel_id'])
);
if($r) {
logger('mail_store: duplicate item ignored. ' . print_r($arr,true));
return 0;
}
call_hooks('post_mail',$arr);
if(x($arr,'cancel')) {
logger('mail_store: post cancelled by plugin.');
return 0;
}
dbesc_array($arr);
logger('mail_store: ' . print_r($arr,true), LOGGER_DATA);
$r = dbq("INSERT INTO mail (`"
. implode("`, `", array_keys($arr))
. "`) VALUES ('"
. implode("', '", array_values($arr))
. "')" );
// find the item we just created
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$r = q("SELECT `id` FROM mail WHERE `mid` = '%s' AND `channel_id` = %d ORDER BY `id` ASC ",
$arr['mid'], // already dbesc'd
y'all got mail
2012-12-06 00:44:07 +00:00
intval($arr['channel_id'])
);
if($r) {
$current_post = $r[0]['id'];
logger('mail_store: created item ' . $current_post, LOGGER_DEBUG);
start rewrite of notification system delivery backend
2013-01-06 02:08:06 +00:00
$arr['id'] = $current_post; // for notification
y'all got mail
2012-12-06 00:44:07 +00:00
}
else {
logger('mail_store: could not locate created item');
return 0;
}
if(count($r) > 1) {
logger('mail_store: duplicated post occurred. Removing duplicates.');
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
q("DELETE FROM mail WHERE `mid` = '%s' AND `channel_id` = %d AND `id` != %d ",
$arr['mid'],
y'all got mail
2012-12-06 00:44:07 +00:00
intval($arr['channel_id']),
intval($current_post)
);
}
start rewrite of notification system delivery backend
2013-01-06 02:08:06 +00:00
else {
require_once('include/enotify.php');
$notif_params = array(
'from_xchan' => $arr['from_xchan'],
'to_xchan' => $arr['to_xchan'],
'type' => NOTIFY_MAIL,
'item' => $arr,
'verb' => ACTIVITY_POST,
'otype' => 'mail'
);
notification($notif_params);
}
y'all got mail
2012-12-06 00:44:07 +00:00
call_hooks('post_mail_end',$arr);
return $current_post;
}
don't change photo visibility if only album name was changed
2011-02-08 05:16:39 +00:00
/**
pull out follow, unfollow and as:actor for handling
2010-10-22 04:48:22 +00:00
*
* consume_feed - process atom feed and update anything/everything we might need to update
*
don't change photo visibility if only album name was changed
2011-02-08 05:16:39 +00:00
* $xml = the (atom) feed to consume - RSS isn't as fully supported but may work for simple feeds.
*
pull out follow, unfollow and as:actor for handling
2010-10-22 04:48:22 +00:00
* $importer = the contact_record (joined to user_record) of the local user who owns this relationship.
* It is this person's stuff that is going to be updated.
* $contact = the person who is sending us stuff. If not set, we MAY be processing a "follow" activity
* from an external network and MAY create an appropriate contact record. Otherwise, we MUST
* have a contact record.
rino recipient patch
2010-12-01 21:39:00 +00:00
* $hub = should we find a hub declation in the feed, pass it back to our calling process, who might (or
pull out follow, unfollow and as:actor for handling
2010-10-22 04:48:22 +00:00
* might not) try and subscribe to it.
document multi-pass mode on consume_feed
2011-10-04 11:38:58 +00:00
* $datedir sorts in reverse order
* $pass - by default ($pass = 0) we cannot guarantee that a parent item has been
* imported prior to its children being seen in the stream unless we are certain
* of how the feed is arranged/ordered.
* With $pass = 1, we only pull parent items out of the stream.
* With $pass = 2, we only pull children (comments/likes).
pull out follow, unfollow and as:actor for handling
2010-10-22 04:48:22 +00:00
*
document multi-pass mode on consume_feed
2011-10-04 11:38:58 +00:00
* So running this twice, first with pass 1 and then with pass 2 will do the right
* thing regardless of feed ordering. This won't be adequate in a fully-threaded
* model where comments can have sub-threads. That would require some massive sorting
* to get all the feed items into a mostly linear ordering, and might still require
* recursion.
pull out follow, unfollow and as:actor for handling
2010-10-22 04:48:22 +00:00
*/
fangs bared
2014-08-20 05:47:11 +00:00
function consume_feed($xml,$importer,&$contact,$pass = 0) {
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
bug #96 move libraries to library - better alignment of like rotator
2011-06-29 04:11:52 +00:00
require_once('library/simplepie/simplepie.inc');
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
allow custom avatar sizes - needed for Diaspora hcard/vcard
2011-08-04 04:05:39 +00:00
if(! strlen($xml)) {
logger('consume_feed: empty input');
return;
}
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
$feed = new SimplePie();
$feed->set_raw_data($xml);
$feed->init();
more ssl_policy cleanup, allow manual feed update per contact when hub is whacked, log feed parse errors
2011-01-31 03:38:03 +00:00
if($feed->error())
logger('consume_feed: Error parsing XML: ' . $feed->error());
better handling of troublesome feeds.
2011-04-05 02:36:18 +00:00
$permalink = $feed->get_permalink();
more ssl_policy cleanup, allow manual feed update per contact when hub is whacked, log feed parse errors
2011-01-31 03:38:03 +00:00
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
// Check at the feed level for updated contact name and/or photo
rework remote deletion
2011-02-08 12:25:27 +00:00
// process any deleted entries
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
rework remote deletion
2011-02-08 12:25:27 +00:00
$del_entries = $feed->get_feed_tags(NAMESPACE_TOMB, 'deleted-entry');
improve feed import performance
2011-10-04 10:06:34 +00:00
if(is_array($del_entries) && count($del_entries) && $pass != 2) {
rework remote deletion
2011-02-08 12:25:27 +00:00
foreach($del_entries as $dentry) {
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
$deleted = false;
rework remote deletion
2011-02-08 12:25:27 +00:00
if(isset($dentry['attribs']['']['ref'])) {
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$mid = $dentry['attribs']['']['ref'];
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
$deleted = true;
rework remote deletion
2011-02-08 12:25:27 +00:00
if(isset($dentry['attribs']['']['when'])) {
$when = $dentry['attribs']['']['when'];
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
$when = datetime_convert('UTC','UTC', $when, 'Y-m-d H:i:s');
}
else
$when = datetime_convert('UTC','UTC','now','Y-m-d H:i:s');
}
snakebite continued
2014-08-20 04:31:33 +00:00
pull out follow, unfollow and as:actor for handling
2010-10-22 04:48:22 +00:00
if($deleted && is_array($contact)) {
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
$r = q("SELECT * from item where mid = '%s' and author_xchan = '%s' and uid = %d limit 1",
encode rss message-id's - warning this will double store all existing rss feed items and probably make a mess of things
2014-09-09 23:28:33 +00:00
dbesc(base64url_encode($mid)),
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
dbesc($contact['xchan_hash']),
intval($importer['channel_id'])
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
);
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
if($r) {
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
$item = $r[0];
rework remote deletion
2011-02-08 12:25:27 +00:00
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
if(! ($item['item_restrict'] & ITEM_DELETED)) {
encode rss message-id's - warning this will double store all existing rss feed items and probably make a mess of things
2014-09-09 23:28:33 +00:00
logger('consume_feed: deleting item ' . $item['id'] . ' mid=' . base64url_decode($item['mid']), LOGGER_DEBUG);
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
drop_item($item['id'],false);
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
}
}
}
rework remote deletion
2011-02-08 12:25:27 +00:00
}
}
// Now process the feed
eliminate the obsolete "last-child" indicator
2012-07-19 00:08:03 +00:00
if($feed->get_item_quantity()) {
rework remote deletion
2011-02-08 12:25:27 +00:00
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
logger('consume_feed: feed item count = ' . $feed->get_item_quantity(), LOGGER_DEBUG);
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
$items = $feed->get_items();
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
rework remote deletion
2011-02-08 12:25:27 +00:00
foreach($items as $item) {
eliminate the obsolete "last-child" indicator
2012-07-19 00:08:03 +00:00
$is_reply = false;
encode rss message-id's - warning this will double store all existing rss feed items and probably make a mess of things
2014-09-09 23:28:33 +00:00
$item_id = base64url_encode($item->get_id());
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
logger('consume_feed: processing ' . $item_id, LOGGER_DEBUG);
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
$rawthread = $item->get_item_tags( NAMESPACE_THREAD,'in-reply-to');
if(isset($rawthread[0]['attribs']['']['ref'])) {
$is_reply = true;
encode rss message-id's - warning this will double store all existing rss feed items and probably make a mess of things
2014-09-09 23:28:33 +00:00
$parent_mid = base64url_encode($rawthread[0]['attribs']['']['ref']);
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
}
remove some un-needed code
2013-08-07 10:54:47 +00:00
if($is_reply) {
fix for errant top-level posts from feeds
2011-10-20 04:54:33 +00:00
if($pass == 1)
continue;
more graceful handling of some statusnet followups
2011-02-09 02:44:30 +00:00
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
// Have we seen it? If not, import it.
eliminate the obsolete "last-child" indicator
2012-07-19 00:08:03 +00:00
encode rss message-id's - warning this will double store all existing rss feed items and probably make a mess of things
2014-09-09 23:28:33 +00:00
$item_id = base64url_encode($item->get_id());
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author = array();
$datarray = get_atom_elements($feed,$item,$author);
more graceful handling of some statusnet followups
2011-02-09 02:44:30 +00:00
if no rss/atom element author, use feed level contact
2014-09-24 21:27:18 +00:00
if((! x($author,'author_name')) || ($author['author_is_feed']))
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_name'] = $contact['xchan_name'];
if no rss/atom element author, use feed level contact
2014-09-24 21:27:18 +00:00
if((! x($author,'author_link')) || ($author['author_is_feed']))
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_link'] = $contact['xchan_url'];
if no rss/atom element author, use feed level contact
2014-09-24 21:27:18 +00:00
if((! x($author,'author_photo'))|| ($author['author_is_feed']))
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_photo'] = $contact['xchan_photo_m'];
ability to link contacts that are "really me, but on another system/network".
2011-07-06 02:55:09 +00:00
correct item author for feeds with multiple or different authors
2014-09-12 00:27:49 +00:00
$datarray['author_xchan'] = '';
if($author['author_link'] != $contact['xchan_url']) {
Typo
2014-09-25 18:47:06 +00:00
$x = import_author_unknown(array('name' => $author['author_name'],'url' => $author['author_link'],'photo' => array('src' => $author['author_photo'])));
correct item author for feeds with multiple or different authors
2014-09-12 00:27:49 +00:00
if($x)
$datarray['author_xchan'] = $x;
}
if(! $datarray['author_xchan'])
$datarray['author_xchan'] = $contact['xchan_hash'];
$datarray['owner_xchan'] = $contact['xchan_hash'];
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
$r = q("SELECT edited FROM item WHERE mid = '%s' AND uid = %d LIMIT 1",
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
dbesc($item_id),
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
intval($importer['channel_id'])
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
);
more graceful handling of some statusnet followups
2011-02-09 02:44:30 +00:00
feeds: update content when updated/edited changes
2011-03-01 02:24:32 +00:00
// Update content if 'updated' changes
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
if($r) {
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
if((x($datarray,'edited') !== false)
&& (datetime_convert('UTC','UTC',$datarray['edited']) !== $r[0]['edited'])) {
reduce reverted edits due to edit arriving before original
2012-06-12 01:14:49 +00:00
// do not accept (ignore) an earlier edit than one we currently have.
move parens to proper place
2012-06-12 09:47:05 +00:00
if(datetime_convert('UTC','UTC',$datarray['edited']) < $r[0]['edited'])
reduce reverted edits due to edit arriving before original
2012-06-12 01:14:49 +00:00
continue;
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
update_feed_item($importer['channel_id'],$datarray);
feeds: update content when updated/edited changes
2011-03-01 02:24:32 +00:00
}
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
continue;
}
more graceful handling of some statusnet followups
2011-02-09 02:44:30 +00:00
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$datarray['parent_mid'] = $parent_mid;
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
$datarray['uid'] = $importer['channel_id'];
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
more tagging
2011-10-24 11:17:46 +00:00
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
logger('consume_feed: ' . print_r($datarray,true),LOGGER_DATA);
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
various rss fixes
2014-09-01 03:51:05 +00:00
$xx = item_store($datarray);
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$r = $xx['item_id'];
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
continue;
}
else {
feeds: update content when updated/edited changes
2011-03-01 02:24:32 +00:00
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
// Head post of a conversation. Have we seen it? If not, import it.
encode rss message-id's - warning this will double store all existing rss feed items and probably make a mess of things
2014-09-09 23:28:33 +00:00
$item_id = base64url_encode($item->get_id());
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author = array();
$datarray = get_atom_elements($feed,$item,$author);
feeds: update content when updated/edited changes
2011-03-01 02:24:32 +00:00
refactor "which link to show" logic
2011-05-06 13:30:33 +00:00
if(is_array($contact)) {
if no rss/atom element author, use feed level contact
2014-09-24 21:27:18 +00:00
if((! x($author,'author_name')) || ($author['author_is_feed']))
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_name'] = $contact['xchan_name'];
if no rss/atom element author, use feed level contact
2014-09-24 21:27:18 +00:00
if((! x($author,'author_link')) || ($author['author_is_feed']))
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_link'] = $contact['xchan_url'];
if no rss/atom element author, use feed level contact
2014-09-24 21:27:18 +00:00
if((! x($author,'author_photo'))|| ($author['author_is_feed']))
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
$author['author_photo'] = $contact['xchan_photo_m'];
refactor "which link to show" logic
2011-05-06 13:30:33 +00:00
}
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
if((! x($author,'author_name')) || (! x($author,'author_link'))) {
logger('consume_feed: no author information! ' . print_r($author,true));
attribution issue - don't take author name from contact if available from item.
2012-02-25 04:03:13 +00:00
continue;
}
correct item author for feeds with multiple or different authors
2014-09-12 00:27:49 +00:00
$datarray['author_xchan'] = '';
if($author['author_link'] != $contact['xchan_url']) {
doc updates and a couple of issues spotted in the logs (diaspora likes, feed author import)
2014-09-12 07:59:23 +00:00
$x = import_author_unknown(array('name' => $author['author_name'],'url' => $author['author_link'],'photo' => array('src' => $author['author_photo'])));
correct item author for feeds with multiple or different authors
2014-09-12 00:27:49 +00:00
if($x)
$datarray['author_xchan'] = $x;
}
if(! $datarray['author_xchan'])
$datarray['author_xchan'] = $contact['xchan_hash'];
$datarray['owner_xchan'] = $contact['xchan_hash'];
event sharing, cont.
2011-06-14 02:06:49 +00:00
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
$r = q("SELECT edited FROM item WHERE mid = '%s' AND uid = %d LIMIT 1",
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
dbesc($item_id),
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
intval($importer['channel_id'])
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
);
feeds: update content when updated/edited changes
2011-03-01 02:24:32 +00:00
// Update content if 'updated' changes
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
if($r) {
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
if((x($datarray,'edited') !== false)
&& (datetime_convert('UTC','UTC',$datarray['edited']) !== $r[0]['edited'])) {
reduce reverted edits due to edit arriving before original
2012-06-12 01:14:49 +00:00
// do not accept (ignore) an earlier edit than one we currently have.
move parens to proper place
2012-06-12 09:47:05 +00:00
if(datetime_convert('UTC','UTC',$datarray['edited']) < $r[0]['edited'])
reduce reverted edits due to edit arriving before original
2012-06-12 01:14:49 +00:00
continue;
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
update_feed_item($importer['channel_id'],$datarray);
feeds: update content when updated/edited changes
2011-03-01 02:24:32 +00:00
}
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
continue;
}
workflow for federated/non-dfrn followers
2010-10-25 03:39:24 +00:00
ability to link contacts that are "really me, but on another system/network".
2011-07-06 02:55:09 +00:00
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$datarray['parent_mid'] = $item_id;
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
$datarray['uid'] = $importer['channel_id'];
correct item author for feeds with multiple or different authors
2014-09-12 00:27:49 +00:00
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
if(! link_compare($author['owner_link'],$contact['xchan_url'])) {
logger('consume_feed: Correcting item owner.', LOGGER_DEBUG);
various rss fixes
2014-09-01 03:51:05 +00:00
$author['owner_name'] = $contact['name'];
$author['owner_link'] = $contact['url'];
$author['owner_avatar'] = $contact['thumb'];
restructure the author and owner info from feeds
2014-08-21 01:35:36 +00:00
}
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
correct item author for feeds with multiple or different authors
2014-09-12 00:27:49 +00:00
logger('consume_feed: author ' . print_r($author,true),LOGGER_DEBUG);
various rss fixes
2014-09-01 03:51:05 +00:00
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
logger('consume_feed: ' . print_r($datarray,true),LOGGER_DATA);
separate the public feed from everything else
2012-09-21 00:04:22 +00:00
various rss fixes
2014-09-01 03:51:05 +00:00
$xx = item_store($datarray);
provide detailed error to remote site for the myriad of things that can go wrong inside item_store(), !! this changes the return of item_store !!
2013-09-11 02:06:06 +00:00
$r = $xx['item_id'];
pubsubhubbub endpoint
2010-10-01 02:41:22 +00:00
continue;
}
}
}
RSS feeds as channel sources. It's feasible and within reach, however there's a lot of work left to do. How bad does anybody want it? Let's find out...
2013-09-30 03:34:05 +00:00
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
}
function update_feed_item($uid,$datarray) {
encode rss message-id's - warning this will double store all existing rss feed items and probably make a mess of things
2014-09-09 23:28:33 +00:00
logger('update_feed_item: not implemented! ' . $uid . ' ' . print_r($datarray,true), LOGGER_DATA);
greatly simplify consume_feed() - a lot of this was friendica specific and somebody may want to put some of it back in, but we'd be much better off building a zot compatible feed for friendica (hint,hint) - even if it didn't support the other protocol bits. Special purpose feed handlers for things like likes, dislikes, events, and community tags are better handled as zot arrays. We have all those handlers. We just need to pass them a JSON message instead of an XML/Atom message. Handlers for statusnet follow messages are another casualty. That's easy to bring back - but we need to keep driving nails into the XML coffin or the damn zombies will get out.
2014-08-20 11:38:17 +00:00
the rest of pubsubhubbub except for the UI that actually turns it on
2010-10-01 04:38:45 +00:00
}
basic support for AS friend requests
2011-10-02 23:18:01 +00:00
fangs bared
2014-08-20 05:47:11 +00:00
function handle_feed($uid,$abook_id,$url) {
require_once('include/Contact.php');
$channel = channelx_by_n($uid);
if(! $channel)
return;
$x = q("select * from abook left join xchan on abook_xchan = xchan_hash where abook_id = %d and abook_channel = %d limit 1",
dbesc($abook_id),
intval($uid)
);
$recurse = 0;
$z = z_fetch_url($url,false,$recurse,array('novalidate' => true));
various rss fixes
2014-09-01 03:51:05 +00:00
//logger('handle_feed:' . print_r($z,true));
fangs bared
2014-08-20 05:47:11 +00:00
if($z['success']) {
consume_feed($z['body'],$channel,$x[0],0);
consume_feed($z['body'],$channel,$x[0],1);
}
}
the rest of pubsubhubbub except for the UI that actually turns it on
2010-10-01 04:38:45 +00:00
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
function atom_author($tag,$name,$uri,$h,$w,$type,$photo) {
build feeds rather than template them
2010-11-02 00:56:36 +00:00
$o = '';
if(! $tag)
return $o;
$name = xmlify($name);
$uri = xmlify($uri);
$h = intval($h);
$w = intval($w);
$photo = xmlify($photo);
$o .= "<$tag>\r\n";
$o .= "<name>$name</name>\r\n";
$o .= "<uri>$uri</uri>\r\n";
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
$o .= '<link rel="photo" type="' . $type . '" media:width="' . $w . '" media:height="' . $h . '" href="' . $photo . '" />' . "\r\n";
$o .= '<link rel="avatar" type="' . $type . '" media:width="' . $w . '" media:height="' . $h . '" href="' . $photo . '" />' . "\r\n";
more plugin hooks
2010-12-25 02:32:23 +00:00
call_hooks('atom_author', $o);
build feeds rather than template them
2010-11-02 00:56:36 +00:00
$o .= "</$tag>\r\n";
return $o;
}
rework the way private photos are embedded to avoid url differences and also check the permissions if possible to make sure that nothing sneaks by.
2012-05-28 04:01:58 +00:00
function atom_entry($item,$type,$author,$owner,$comment = false,$cid = 0) {
build feeds rather than template them
2010-11-02 00:56:36 +00:00
fix private images
2011-05-01 00:24:37 +00:00
$a = get_app();
working on long-term solution to relay issue
2011-10-15 10:26:37 +00:00
if(! $item['parent'])
return;
build feeds rather than template them
2010-11-02 00:56:36 +00:00
if($item['deleted'])
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
return '<at:deleted-entry ref="' . xmlify($item['mid']) . '" when="' . xmlify(datetime_convert('UTC','UTC',$item['edited'] . '+00:00',ATOM_TIME)) . '" />' . "\r\n";
build feeds rather than template them
2010-11-02 00:56:36 +00:00
fix private images
2011-05-01 00:24:37 +00:00
if($item['allow_cid'] || $item['allow_gid'] || $item['deny_cid'] || $item['deny_gid'])
rework the way private photos are embedded to avoid url differences and also check the permissions if possible to make sure that nothing sneaks by.
2012-05-28 04:01:58 +00:00
$body = fix_private_photos($item['body'],$owner['uid'],$item,$cid);
fix private images
2011-05-01 00:24:37 +00:00
else
$body = $item['body'];
build feeds rather than template them
2010-11-02 00:56:36 +00:00
infrastructure for swat0 (#fsw)
2010-11-02 06:42:26 +00:00
$o = "\r\n\r\n<entry>\r\n";
build feeds rather than template them
2010-11-02 00:56:36 +00:00
if(is_array($author))
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
$o .= atom_author('author',$author['xchan_name'],$author['xchan_url'],80,80,$author['xchan_photo_mimetype'],$author['xchan_photo_m']);
build feeds rather than template them
2010-11-02 00:56:36 +00:00
else
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
$o .= atom_author('author',$item['author']['xchan_name'],$item['author']['xchan_url'],80,80,$item['author']['xchan_photo_mimetype'], $item['author']['xchan_photo_m']);
$o .= atom_author('zot:owner',$item['owner']['xchan_name'],$item['owner']['xchan_url'],80,80,$item['owner']['xchan_photo_mimetype'],$item['owner']['xchan_photo_m']);
build feeds rather than template them
2010-11-02 00:56:36 +00:00
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
if(($item['parent'] != $item['id']) || ($item['parent_mid'] !== $item['mid']) || (($item['thr_parent'] !== '') && ($item['thr_parent'] !== $item['mid']))) {
$parent_item = (($item['thr_parent']) ? $item['thr_parent'] : $item['parent_mid']);
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
$o .= '<thr:in-reply-to ref="' . xmlify($parent_item) . '" type="text/html" href="' . xmlify($item['plink']) . '" />' . "\r\n";
typo
2012-06-23 13:04:56 +00:00
}
build feeds rather than template them
2010-11-02 00:56:36 +00:00
rename 'uri' (and parent_uri) to 'mid' (and parent_mid) since these no longer remotely resemble uri's and are actually message_id's. This change is potentially destabilising because it touches a lot of code and structure. But it has to get done and there's no better time than the present.
2013-03-22 01:25:41 +00:00
$o .= '<id>' . xmlify($item['mid']) . '</id>' . "\r\n";
build feeds rather than template them
2010-11-02 00:56:36 +00:00
$o .= '<title>' . xmlify($item['title']) . '</title>' . "\r\n";
$o .= '<published>' . xmlify(datetime_convert('UTC','UTC',$item['created'] . '+00:00',ATOM_TIME)) . '</published>' . "\r\n";
$o .= '<updated>' . xmlify(datetime_convert('UTC','UTC',$item['edited'] . '+00:00',ATOM_TIME)) . '</updated>' . "\r\n";
pages feed
2013-09-06 02:31:26 +00:00
$o .= '<content type="' . $type . '" >' . xmlify(prepare_text($body,$item['mimetype'])) . '</content>' . "\r\n";
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
$o .= '<link rel="alternate" type="text/html" href="' . xmlify($item['plink']) . '" />' . "\r\n";
lint
2010-11-10 04:38:24 +00:00
if($item['location']) {
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
$o .= '<zot:location>' . xmlify($item['location']) . '</zot:location>' . "\r\n";
lint
2010-11-10 04:38:24 +00:00
$o .= '<poco:address><poco:formatted>' . xmlify($item['location']) . '</poco:formatted></poco:address>' . "\r\n";
}
build feeds rather than template them
2010-11-02 00:56:36 +00:00
if($item['coord'])
$o .= '<georss:point>' . xmlify($item['coord']) . '</georss:point>' . "\r\n";
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
if(($item['item_private']) || strlen($item['allow_cid']) || strlen($item['allow_gid']) || strlen($item['deny_cid']) || strlen($item['deny_gid']))
$o .= '<zot:private>' . (($item['item_private']) ? $item['item_private'] : 1) . '</zot:private>' . "\r\n";
display the fact that a conversation is private without disclosing the details of who can/cannot see it.
2010-12-08 04:47:53 +00:00
more improvements in twitter handling, reduce duplicates from FB when double friended
2011-05-23 01:40:00 +00:00
item source
2011-06-21 02:08:40 +00:00
if($item['app'])
bring Diaspora message signing back to the source author - whether they like it or not.
2011-08-29 02:22:27 +00:00
$o .= '<statusnet:notice_info local_id="' . $item['id'] . '" source="' . xmlify($item['app']) . '" ></statusnet:notice_info>' . "\r\n";
build feeds rather than template them
2010-11-02 00:56:36 +00:00
$verb = construct_verb($item);
$o .= '<as:verb>' . xmlify($verb) . '</as:verb>' . "\r\n";
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
$actobj = construct_activity_object($item);
build feeds rather than template them
2010-11-02 00:56:36 +00:00
if(strlen($actobj))
$o .= $actobj;
add activity target to items, allow multiple link relations
2010-11-03 23:48:21 +00:00
$actarg = construct_activity_target($item);
if(strlen($actarg))
$o .= $actarg;
build feeds rather than template them
2010-11-02 00:56:36 +00:00
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
// FIXME
// $tags = item_getfeedtags($item);
// if(count($tags)) {
// foreach($tags as $t) {
// $o .= '<category scheme="X-DFRN:' . xmlify($t[0]) . ':' . xmlify($t[1]) . '" term="' . xmlify($t[2]) . '" />' . "\r\n";
// }
// }
atom categories and tag import
2011-04-06 00:41:02 +00:00
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
// FIXME
// $o .= item_getfeedattach($item);
support feed attachments/enclosures, and fix for bug #30
2011-04-07 02:41:16 +00:00
feeds are bit more standards compliant and a bit less broken now
2013-06-13 04:12:27 +00:00
// $mentioned = get_mentions($item,$tags);
// if($mentioned)
// $o .= $mentioned;
build feeds rather than template them
2010-11-02 00:56:36 +00:00
more plugin hooks
2010-12-25 02:32:23 +00:00
call_hooks('atom_entry', $o);
build feeds rather than template them
2010-11-02 00:56:36 +00:00
$o .= '</entry>' . "\r\n";
return $o;
}
atom categories and tag import
2011-04-06 00:41:02 +00:00
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
function fix_private_photos($s, $uid, $item = null, $cid = 0) {
fix private images
2011-05-01 00:24:37 +00:00
$a = get_app();
rework the way private photos are embedded to avoid url differences and also check the permissions if possible to make sure that nothing sneaks by.
2012-05-28 04:01:58 +00:00
logger('fix_private_photos', LOGGER_DEBUG);
bugfixes: private photo embeds and search for strings with %
2012-05-29 23:44:02 +00:00
$site = substr($a->get_baseurl(),strpos($a->get_baseurl(),'://'));
fix private images
2011-05-01 00:24:37 +00:00
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
$orig_body = $s;
$new_body = '';
convert most red photos to zmg
2013-05-28 11:50:16 +00:00
$img_start = strpos($orig_body, '[zmg');
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
$img_st_close = ($img_start !== false ? strpos(substr($orig_body, $img_start), ']') : false);
convert most red photos to zmg
2013-05-28 11:50:16 +00:00
$img_len = ($img_start !== false ? strpos(substr($orig_body, $img_start + $img_st_close + 1), '[/zmg]') : false);
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
while( ($img_st_close !== false) && ($img_len !== false) ) {
$img_st_close++; // make it point to AFTER the closing bracket
$image = substr($orig_body, $img_start + $img_st_close, $img_len);
rework the way private photos are embedded to avoid url differences and also check the permissions if possible to make sure that nothing sneaks by.
2012-05-28 04:01:58 +00:00
logger('fix_private_photos: found photo ' . $image, LOGGER_DEBUG);
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
rework the way private photos are embedded to avoid url differences and also check the permissions if possible to make sure that nothing sneaks by.
2012-05-28 04:01:58 +00:00
if(stristr($image , $site . '/photo/')) {
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
// Only embed locally hosted photos
rework the way private photos are embedded to avoid url differences and also check the permissions if possible to make sure that nothing sneaks by.
2012-05-28 04:01:58 +00:00
$replace = false;
fix private images
2011-05-01 00:24:37 +00:00
$i = basename($image);
$x = strpos($i,'-');
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
fix private images
2011-05-01 00:24:37 +00:00
if($x) {
$res = substr($i,$x+1);
$i = substr($i,0,$x);
here's where the heavy lifting begins - everything is likely to be broken for quite some time as we add location and db independence to items and conversations and work through the rest of the permissions and how to federate the buggers.
2012-10-02 01:02:11 +00:00
$r = q("SELECT * FROM `photo` WHERE `resource_id` = '%s' AND `scale` = %d AND `uid` = %d",
fix private images
2011-05-01 00:24:37 +00:00
dbesc($i),
intval($res),
intval($uid)
);
if(count($r)) {
rework the way private photos are embedded to avoid url differences and also check the permissions if possible to make sure that nothing sneaks by.
2012-05-28 04:01:58 +00:00
// Check to see if we should replace this photo link with an embedded image
// 1. No need to do so if the photo is public
// 2. If there's a contact-id provided, see if they're in the access list
// for the photo. If so, embed it.
// 3. Otherwise, if we have an item, see if the item permissions match the photo
// permissions, regardless of order but first check to see if they're an exact
// match to save some processing overhead.
if(has_permissions($r[0])) {
if($cid) {
$recips = enumerate_permissions($r[0]);
if(in_array($cid, $recips)) {
$replace = true;
}
}
elseif($item) {
if(compare_permissions($item,$r[0]))
$replace = true;
}
}
if($replace) {
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
$data = $r[0]['data'];
$type = $r[0]['type'];
// If a custom width and height were specified, apply before embedding
convert most red photos to zmg
2013-05-28 11:50:16 +00:00
if(preg_match("/\[zmg\=([0-9]*)x([0-9]*)\]/is", substr($orig_body, $img_start, $img_st_close), $match)) {
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
logger('fix_private_photos: scaling photo', LOGGER_DEBUG);
$width = intval($match[1]);
$height = intval($match[2]);
photo driver abstraction
2013-04-26 03:01:24 +00:00
$ph = photo_factory($data, $type);
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
if($ph->is_valid()) {
$ph->scaleImage(max($width, $height));
$data = $ph->imageString();
$type = $ph->getType();
}
}
rev update
2012-05-28 23:51:52 +00:00
logger('fix_private_photos: replacing photo', LOGGER_DEBUG);
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
$image = 'data:' . $type . ';base64,' . base64_encode($data);
logger('fix_private_photos: replaced: ' . $image, LOGGER_DATA);
rework the way private photos are embedded to avoid url differences and also check the permissions if possible to make sure that nothing sneaks by.
2012-05-28 04:01:58 +00:00
}
fix private images
2011-05-01 00:24:37 +00:00
}
}
}
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
convert most red photos to zmg
2013-05-28 11:50:16 +00:00
$new_body = $new_body . substr($orig_body, 0, $img_start + $img_st_close) . $image . '[/zmg]';
$orig_body = substr($orig_body, $img_start + $img_st_close + $img_len + strlen('[/zmg]'));
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
if($orig_body === false)
$orig_body = '';
convert most red photos to zmg
2013-05-28 11:50:16 +00:00
$img_start = strpos($orig_body, '[zmg');
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
$img_st_close = ($img_start !== false ? strpos(substr($orig_body, $img_start), ']') : false);
convert most red photos to zmg
2013-05-28 11:50:16 +00:00
$img_len = ($img_start !== false ? strpos(substr($orig_body, $img_start + $img_st_close + 1), '[/zmg]') : false);
fix private images
2011-05-01 00:24:37 +00:00
}
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
fix bugs
2012-07-08 00:47:13 +00:00
$new_body = $new_body . $orig_body;
allow more than one embedded private photo
2012-07-07 22:20:24 +00:00
return($new_body);
fix private images
2011-05-01 00:24:37 +00:00
}
rework the way private photos are embedded to avoid url differences and also check the permissions if possible to make sure that nothing sneaks by.
2012-05-28 04:01:58 +00:00
function has_permissions($obj) {
if(($obj['allow_cid'] != '') || ($obj['allow_gid'] != '') || ($obj['deny_cid'] != '') || ($obj['deny_gid'] != ''))
return true;
return false;
}
function compare_permissions($obj1,$obj2) {
// first part is easy. Check that these are exactly the same.
if(($obj1['allow_cid'] == $obj2['allow_cid'])
&& ($obj1['allow_gid'] == $obj2['allow_gid'])
&& ($obj1['deny_cid'] == $obj2['deny_cid'])
&& ($obj1['deny_gid'] == $obj2['deny_gid']))
return true;
// This is harder. Parse all the permissions and compare the resulting set.
$recipients1 = enumerate_permissions($obj1);
$recipients2 = enumerate_permissions($obj2);
sort($recipients1);
sort($recipients2);
if($recipients1 == $recipients2)
return true;
return false;
}
// returns an array of contact-ids that are allowed to see this object
function enumerate_permissions($obj) {
require_once('include/group.php');
$allow_people = expand_acl($obj['allow_cid']);
$allow_groups = expand_groups(expand_acl($obj['allow_gid']));
$deny_people = expand_acl($obj['deny_cid']);
$deny_groups = expand_groups(expand_acl($obj['deny_gid']));
$recipients = array_unique(array_merge($allow_people,$allow_groups));
$deny = array_unique(array_merge($deny_people,$deny_groups));
$recipients = array_diff($recipients,$deny);
return $recipients;
}
fix private images
2011-05-01 00:24:37 +00:00
atom categories and tag import
2011-04-06 00:41:02 +00:00
function item_getfeedtags($item) {
catchup
2012-07-11 11:29:47 +00:00
$terms = get_terms_oftype($item['term'],array(TERM_HASHTAG,TERM_MENTION));
atom categories and tag import
2011-04-06 00:41:02 +00:00
$ret = array();
catchup
2012-07-11 11:29:47 +00:00
if(count($terms)) {
foreach($terms as $term) {
if($term['type'] == TERM_HASHTAG)
$ret[] = array('#',$term['url'],$term['term']);
else
$ret[] = array('@',$term['url'],$term['term']);
atom categories and tag import
2011-04-06 00:41:02 +00:00
}
}
return $ret;
}
support feed attachments/enclosures, and fix for bug #30
2011-04-07 02:41:16 +00:00
function item_getfeedattach($item) {
some fixes to attachment output in feed
2011-04-07 03:03:06 +00:00
$ret = '';
support feed attachments/enclosures, and fix for bug #30
2011-04-07 02:41:16 +00:00
$arr = explode(',',$item['attach']);
if(count($arr)) {
foreach($arr as $r) {
$matches = false;
several fixes for attachments
2011-08-04 02:18:58 +00:00
$cnt = preg_match('|\[attach\]href=\"(.*?)\" length=\"(.*?)\" type=\"(.*?)\" title=\"(.*?)\"\[\/attach\]|',$r,$matches);
support feed attachments/enclosures, and fix for bug #30
2011-04-07 02:41:16 +00:00
if($cnt) {
xmlify enclosure items
2011-04-07 03:36:24 +00:00
$ret .= '<link rel="enclosure" href="' . xmlify($matches[1]) . '" type="' . xmlify($matches[3]) . '" ';
support feed attachments/enclosures, and fix for bug #30
2011-04-07 02:41:16 +00:00
if(intval($matches[2]))
several fixes for attachments
2011-08-04 02:18:58 +00:00
$ret .= 'length="' . intval($matches[2]) . '" ';
support feed attachments/enclosures, and fix for bug #30
2011-04-07 02:41:16 +00:00
if($matches[4] !== ' ')
convert feed attach to bbcode
2011-04-13 08:53:40 +00:00
$ret .= 'title="' . xmlify(trim($matches[4])) . '" ';
support feed attachments/enclosures, and fix for bug #30
2011-04-07 02:41:16 +00:00
$ret .= ' />' . "\r\n";
}
}
}
return $ret;
}
atom categories and tag import
2011-04-06 00:41:02 +00:00
import items from feeds in inverse date order
2011-01-03 09:04:54 +00:00
content expiration
2011-03-16 00:31:49 +00:00
function item_expire($uid,$days) {
Provide backend option to only expire other peoples' conversations.
2012-05-29 06:24:10 +00:00
if((! $uid) || ($days < 1))
content expiration
2011-03-16 00:31:49 +00:00
return;
Provide backend option to only expire other peoples' conversations.
2012-05-29 06:24:10 +00:00
// $expire_network_only = save your own wall posts
// and just expire conversations started by others
enable network/matrix expiration, this should be functional but the options have been reduced/restricted so we're only looking at network posts and ignore anything that is filed, starred, or is resource_type 'photo' (which should not be possible in non-wall posts, but we just want to be sure). Will require the adventurous tester(s) to set 'channel_expire_days' in their channel record.
2014-03-04 03:15:11 +00:00
// do not enable this until we can pass bulk delete messages through zot
// $expire_network_only = get_pconfig($uid,'expire','network_only');
$expire_network_only = 1;
Provide backend option to only expire other peoples' conversations.
2012-05-29 06:24:10 +00:00
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$sql_extra = ((intval($expire_network_only)) ? " AND not (item_flags & " . intval(ITEM_WALL) . ")>0 " : "");
Provide backend option to only expire other peoples' conversations.
2012-05-29 06:24:10 +00:00
content expiration
2011-03-16 00:31:49 +00:00
$r = q("SELECT * FROM `item`
WHERE `uid` = %d
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
AND `created` < %s - INTERVAL %s
content expiration
2011-03-16 00:31:49 +00:00
AND `id` = `parent`
Provide backend option to only expire other peoples' conversations.
2012-05-29 06:24:10 +00:00
$sql_extra
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
AND NOT ( item_flags & %d )>0
enable network/matrix expiration, this should be functional but the options have been reduced/restricted so we're only looking at network posts and ignore anything that is filed, starred, or is resource_type 'photo' (which should not be possible in non-wall posts, but we just want to be sure). Will require the adventurous tester(s) to set 'channel_expire_days' in their channel record.
2014-03-04 03:15:11 +00:00
AND (item_restrict = 0 ) ",
content expiration
2011-03-16 00:31:49 +00:00
intval($uid),
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
db_utcnow(), db_quoteinterval(intval($days).' DAY'),
enable network/matrix expiration, this should be functional but the options have been reduced/restricted so we're only looking at network posts and ignore anything that is filed, starred, or is resource_type 'photo' (which should not be possible in non-wall posts, but we just want to be sure). Will require the adventurous tester(s) to set 'channel_expire_days' in their channel record.
2014-03-04 03:15:11 +00:00
intval(ITEM_RETAINED)
content expiration
2011-03-16 00:31:49 +00:00
);
fix expire so it actually should expire stuff; and mostly the correct stuff - though it DOESN'T currently send out notifications to connections
2013-08-14 12:26:17 +00:00
if(! $r)
content expiration
2011-03-16 00:31:49 +00:00
return;
expire settings for items, notes and photos. show default privacy acl in popup
2011-11-17 14:53:59 +00:00
fix expire so it actually should expire stuff; and mostly the correct stuff - though it DOESN'T currently send out notifications to connections
2013-08-14 12:26:17 +00:00
$r = fetch_post_tags($r,true);
content expiration
2011-03-16 00:31:49 +00:00
foreach($r as $item) {
don't expire filed items
2012-03-27 07:54:34 +00:00
// don't expire filed items
fix expire so it actually should expire stuff; and mostly the correct stuff - though it DOESN'T currently send out notifications to connections
2013-08-14 12:26:17 +00:00
$terms = get_terms_oftype($item['term'],TERM_FILE);
enable network/matrix expiration, this should be functional but the options have been reduced/restricted so we're only looking at network posts and ignore anything that is filed, starred, or is resource_type 'photo' (which should not be possible in non-wall posts, but we just want to be sure). Will require the adventurous tester(s) to set 'channel_expire_days' in their channel record.
2014-03-04 03:15:11 +00:00
if($terms) {
retain_item($item['id']);
don't expire filed items
2012-03-27 07:54:34 +00:00
continue;
enable network/matrix expiration, this should be functional but the options have been reduced/restricted so we're only looking at network posts and ignore anything that is filed, starred, or is resource_type 'photo' (which should not be possible in non-wall posts, but we just want to be sure). Will require the adventurous tester(s) to set 'channel_expire_days' in their channel record.
2014-03-04 03:15:11 +00:00
}
don't expire filed items
2012-03-27 07:54:34 +00:00
content expiration
2011-03-16 00:31:49 +00:00
// Only expire posts, not photos and photo comments
enable network/matrix expiration, this should be functional but the options have been reduced/restricted so we're only looking at network posts and ignore anything that is filed, starred, or is resource_type 'photo' (which should not be possible in non-wall posts, but we just want to be sure). Will require the adventurous tester(s) to set 'channel_expire_days' in their channel record.
2014-03-04 03:15:11 +00:00
if($item['resource_type'] === 'photo') {
retain_item($item['id']);
content expiration
2011-03-16 00:31:49 +00:00
continue;
enable network/matrix expiration, this should be functional but the options have been reduced/restricted so we're only looking at network posts and ignore anything that is filed, starred, or is resource_type 'photo' (which should not be possible in non-wall posts, but we just want to be sure). Will require the adventurous tester(s) to set 'channel_expire_days' in their channel record.
2014-03-04 03:15:11 +00:00
}
if($item['item_flags'] & ITEM_STARRED) {
retain_item($item['id']);
expire settings for items, notes and photos. show default privacy acl in popup
2011-11-17 14:53:59 +00:00
continue;
enable network/matrix expiration, this should be functional but the options have been reduced/restricted so we're only looking at network posts and ignore anything that is filed, starred, or is resource_type 'photo' (which should not be possible in non-wall posts, but we just want to be sure). Will require the adventurous tester(s) to set 'channel_expire_days' in their channel record.
2014-03-04 03:15:11 +00:00
}
expire settings for items, notes and photos. show default privacy acl in popup
2011-11-17 14:53:59 +00:00
consolidate item deletion so we can fix orphaned categories and file_as tags.
2012-05-05 14:05:32 +00:00
drop_item($item['id'],false);
content expiration
2011-03-16 00:31:49 +00:00
}
enable network/matrix expiration, this should be functional but the options have been reduced/restricted so we're only looking at network posts and ignore anything that is filed, starred, or is resource_type 'photo' (which should not be possible in non-wall posts, but we just want to be sure). Will require the adventurous tester(s) to set 'channel_expire_days' in their channel record.
2014-03-04 03:15:11 +00:00
// proc_run('php',"include/notifier.php","expire","$uid");
expire settings for items, notes and photos. show default privacy acl in popup
2011-11-17 14:53:59 +00:00
reduce visibility of lang-select flag (duepuntozero)
2011-06-11 03:39:46 +00:00
}
enable network/matrix expiration, this should be functional but the options have been reduced/restricted so we're only looking at network posts and ignore anything that is filed, starred, or is resource_type 'photo' (which should not be possible in non-wall posts, but we just want to be sure). Will require the adventurous tester(s) to set 'channel_expire_days' in their channel record.
2014-03-04 03:15:11 +00:00
function retain_item($id) {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("update item set item_flags = (item_flags | %d ) where id = %d",
enable network/matrix expiration, this should be functional but the options have been reduced/restricted so we're only looking at network posts and ignore anything that is filed, starred, or is resource_type 'photo' (which should not be possible in non-wall posts, but we just want to be sure). Will require the adventurous tester(s) to set 'channel_expire_days' in their channel record.
2014-03-04 03:15:11 +00:00
intval(ITEM_RETAINED),
intval($id)
);
}
delete selected items
2011-06-16 03:43:39 +00:00
function drop_items($items) {
$uid = 0;
Some Bugfixes, and variable checks
2012-04-17 11:33:50 +00:00
if(! local_user() && ! remote_user())
redirect to a useful location after deleting or ignoring a friend request, begin comment moderation logic
2012-01-25 02:59:55 +00:00
return;
delete selected items
2011-06-16 03:43:39 +00:00
if(count($items)) {
foreach($items as $item) {
$owner = drop_item($item,false);
if($owner && ! $uid)
$uid = $owner;
}
}
// multiple threads may have been deleted, send an expire notification
if($uid)
proc_run('php',"include/notifier.php","expire","$uid");
}
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
// Delete item with given item $id. $interactive means we're running interactively, and must check
// permissions to carry out this act. If it is non-interactive, we are deleting something at the
// system's request and do not check permission. This is very important to know.
try and sort out the item delete mess
2014-03-28 03:28:48 +00:00
// Some deletion requests (those coming from remote sites) must be staged.
// $stage = 0 => unstaged
// $stage = 1 => set deleted flag on the item and perform intial notifications
// $stage = 2 => perform low level delete at a later stage
change deletion behaviour of linked photo items. Previously deleting the visible item in the conversation deleted the photo as well. Now photos must be deleted in the photos module. Deleting the linked item removes any attached conversation elements (likes, etc.) and sets the conversation item to hidden. This may create an issue in the future if we move the photo tags, title, or other photo elements to the linked item rather than the photo. Noting here so this can potentially be discovered and remembered at that time.
2015-01-13 00:27:27 +00:00
function drop_item($id,$interactive = true,$stage = DROPITEM_NORMAL,$force = false) {
delete selected items
2011-06-16 03:43:39 +00:00
notification being sent for comments from blocked connections on same site.
2014-06-26 23:24:53 +00:00
delete selected items
2011-06-16 03:43:39 +00:00
$a = get_app();
// locate item to be deleted
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
$r = q("SELECT * FROM item WHERE id = %d LIMIT 1",
delete selected items
2011-06-16 03:43:39 +00:00
intval($id)
);
try and sort out the item delete mess
2014-03-28 03:28:48 +00:00
if((! $r) || (($r[0]['item_restrict'] & ITEM_DELETED) && ($stage === DROPITEM_NORMAL))) {
delete selected items
2011-06-16 03:43:39 +00:00
if(! $interactive)
return 0;
notice( t('Item not found.') . EOL);
goaway($a->get_baseurl() . '/' . $_SESSION['return_url']);
}
$item = $r[0];
change deletion behaviour of linked photo items. Previously deleting the visible item in the conversation deleted the photo as well. Now photos must be deleted in the photos module. Deleting the linked item removes any attached conversation elements (likes, etc.) and sets the conversation item to hidden. This may create an issue in the future if we move the photo tags, title, or other photo elements to the linked item rather than the photo. Noting here so this can potentially be discovered and remembered at that time.
2015-01-13 00:27:27 +00:00
$linked_item = (($item['resource_id']) ? true : false);
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
$ok_to_delete = false;
updates
2012-09-10 04:17:06 +00:00
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
// system deletion
if(! $interactive)
$ok_to_delete = true;
delete selected items
2011-06-16 03:43:39 +00:00
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
// owner deletion
if(local_user() && local_user() == $item['uid'])
$ok_to_delete = true;
delete of single items should now work again
2013-01-23 12:50:57 +00:00
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
// author deletion
$observer = $a->get_observer();
if($observer && $observer['xchan_hash'] && ($observer['xchan_hash'] === $item['author_xchan']))
$ok_to_delete = true;
delete of single items should now work again
2013-01-23 12:50:57 +00:00
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
if($ok_to_delete) {
consolidate item deletion so we can fix orphaned categories and file_as tags.
2012-05-05 14:05:32 +00:00
more loop prevention
2013-11-27 07:10:10 +00:00
// set the deleted flag immediately on this item just in case the
// hook calls a remote process which loops. We'll delete it properly in a second.
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("UPDATE item SET item_restrict = ( item_restrict | %d ) WHERE id = %d",
change deletion behaviour of linked photo items. Previously deleting the visible item in the conversation deleted the photo as well. Now photos must be deleted in the photos module. Deleting the linked item removes any attached conversation elements (likes, etc.) and sets the conversation item to hidden. This may create an issue in the future if we move the photo tags, title, or other photo elements to the linked item rather than the photo. Noting here so this can potentially be discovered and remembered at that time.
2015-01-13 00:27:27 +00:00
intval(($linked_item && ! $force) ? ITEM_HIDDEN : ITEM_DELETED),
more loop prevention
2013-11-27 07:10:10 +00:00
intval($item['id'])
);
change deletion behaviour of linked photo items. Previously deleting the visible item in the conversation deleted the photo as well. Now photos must be deleted in the photos module. Deleting the linked item removes any attached conversation elements (likes, etc.) and sets the conversation item to hidden. This may create an issue in the future if we move the photo tags, title, or other photo elements to the linked item rather than the photo. Noting here so this can potentially be discovered and remembered at that time.
2015-01-13 00:27:27 +00:00
try and sort out the item delete mess
2014-03-28 03:28:48 +00:00
$arr = array('item' => $item, 'interactive' => $interactive, 'stage' => $stage);
add drop_item hook
2013-11-26 23:26:11 +00:00
call_hooks('drop_item', $arr );
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
$notify_id = intval($item['id']);
consolidate item deletion so we can fix orphaned categories and file_as tags.
2012-05-05 14:05:32 +00:00
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
$items = q("select * from item where parent = %d and uid = %d",
consolidate item deletion so we can fix orphaned categories and file_as tags.
2012-05-05 14:05:32 +00:00
intval($item['id']),
intval($item['uid'])
);
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
if($items) {
foreach($items as $i)
change deletion behaviour of linked photo items. Previously deleting the visible item in the conversation deleted the photo as well. Now photos must be deleted in the photos module. Deleting the linked item removes any attached conversation elements (likes, etc.) and sets the conversation item to hidden. This may create an issue in the future if we move the photo tags, title, or other photo elements to the linked item rather than the photo. Noting here so this can potentially be discovered and remembered at that time.
2015-01-13 00:27:27 +00:00
delete_item_lowlevel($i,$stage,$force);
delete selected items
2011-06-16 03:43:39 +00:00
}
help deletions along
2013-01-27 21:43:43 +00:00
else
change deletion behaviour of linked photo items. Previously deleting the visible item in the conversation deleted the photo as well. Now photos must be deleted in the photos module. Deleting the linked item removes any attached conversation elements (likes, etc.) and sets the conversation item to hidden. This may create an issue in the future if we move the photo tags, title, or other photo elements to the linked item rather than the photo. Noting here so this can potentially be discovered and remembered at that time.
2015-01-13 00:27:27 +00:00
delete_item_lowlevel($item,$stage,$force);
Implement relaying of relayable_retractions Also: some whitespace cleanup, fix Diaspora parent DB query
2012-06-01 01:40:12 +00:00
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
if(! $interactive)
return 1;
consolidate item deletion so we can fix orphaned categories and file_as tags.
2012-05-05 14:05:32 +00:00
delete selected items
2011-06-16 03:43:39 +00:00
// send the notification upstream/downstream as the case may be
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
// only send notifications to others if this is the owner's wall item.
delete selected items
2011-06-16 03:43:39 +00:00
content deletion issue
2014-10-12 08:40:48 +00:00
// This isn't optimal. We somehow need to pass to this function whether or not
// to call the notifier, or we need to call the notifier from the calling function.
// We'll rely on the undocumented behaviour that DROPITEM_PHASE1 is (hopefully) only
// set if we know we're going to send delete notifications out to others.
if((($item['item_flags'] & ITEM_WALL) && ($stage != DROPITEM_PHASE2)) || ($stage == DROPITEM_PHASE1))
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
proc_run('php','include/notifier.php','drop',$notify_id);
delete selected items
2011-06-16 03:43:39 +00:00
goaway($a->get_baseurl() . '/' . $_SESSION['return_url']);
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
delete selected items
2011-06-16 03:43:39 +00:00
}
else {
if(! $interactive)
return 0;
notice( t('Permission denied.') . EOL);
goaway($a->get_baseurl() . '/' . $_SESSION['return_url']);
}
tagging
2011-10-23 07:24:37 +00:00
}
archive widget
2012-06-13 03:46:30 +00:00
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
// This function does not check for permission and does not send notifications and does not check recursion.
// It merely destroys all resources associated with an item.
// Please do not use without a suitable wrapper.
change deletion behaviour of linked photo items. Previously deleting the visible item in the conversation deleted the photo as well. Now photos must be deleted in the photos module. Deleting the linked item removes any attached conversation elements (likes, etc.) and sets the conversation item to hidden. This may create an issue in the future if we move the photo tags, title, or other photo elements to the linked item rather than the photo. Noting here so this can potentially be discovered and remembered at that time.
2015-01-13 00:27:27 +00:00
function delete_item_lowlevel($item,$stage = DROPITEM_NORMAL,$force = false) {
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
change deletion behaviour of linked photo items. Previously deleting the visible item in the conversation deleted the photo as well. Now photos must be deleted in the photos module. Deleting the linked item removes any attached conversation elements (likes, etc.) and sets the conversation item to hidden. This may create an issue in the future if we move the photo tags, title, or other photo elements to the linked item rather than the photo. Noting here so this can potentially be discovered and remembered at that time.
2015-01-13 00:27:27 +00:00
$linked_item = (($item['resource_id']) ? true : false);
try and sort out the item delete mess
2014-03-28 03:28:48 +00:00
switch($stage) {
case DROPITEM_PHASE2:
$r = q("UPDATE item SET item_restrict = ( item_restrict | %d ), body = '', title = '',
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
changed = '%s', edited = '%s' WHERE id = %d",
try and sort out the item delete mess
2014-03-28 03:28:48 +00:00
intval(ITEM_PENDING_REMOVE),
dbesc(datetime_convert()),
dbesc(datetime_convert()),
intval($item['id'])
);
break;
case DROPITEM_PHASE1:
$r = q("UPDATE item SET item_restrict = ( item_restrict | %d ),
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
changed = '%s', edited = '%s' WHERE id = %d",
change deletion behaviour of linked photo items. Previously deleting the visible item in the conversation deleted the photo as well. Now photos must be deleted in the photos module. Deleting the linked item removes any attached conversation elements (likes, etc.) and sets the conversation item to hidden. This may create an issue in the future if we move the photo tags, title, or other photo elements to the linked item rather than the photo. Noting here so this can potentially be discovered and remembered at that time.
2015-01-13 00:27:27 +00:00
intval(($linked_item && ! $force) ? ITEM_HIDDEN : ITEM_DELETED),
try and sort out the item delete mess
2014-03-28 03:28:48 +00:00
dbesc(datetime_convert()),
dbesc(datetime_convert()),
intval($item['id'])
);
break;
case DROPITEM_NORMAL:
default:
don't wipe out the item body of a linked item to a photo when doing a normal delete and not forced
2015-01-13 00:36:10 +00:00
if($linked_item && ! $force) {
$r = q("UPDATE item SET item_restrict = ( item_restrict | %d ),
changed = '%s', edited = '%s' WHERE id = %d",
intval(ITEM_HIDDEN),
dbesc(datetime_convert()),
dbesc(datetime_convert()),
intval($item['id'])
);
}
else {
$r = q("UPDATE item SET item_restrict = ( item_restrict | %d ), body = '', title = '',
changed = '%s', edited = '%s' WHERE id = %d",
intval(ITEM_DELETED),
dbesc(datetime_convert()),
dbesc(datetime_convert()),
intval($item['id'])
);
}
try and sort out the item delete mess
2014-03-28 03:28:48 +00:00
break;
}
process profile like removals with normal low_level drop_item
2014-06-23 08:22:58 +00:00
// immediately remove any undesired profile likes.
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
q("delete from likes where iid = %d and channel_id = %d",
process profile like removals with normal low_level drop_item
2014-06-23 08:22:58 +00:00
intval($item['id']),
intval($item['uid'])
);
when deleting items, delete linked events and photos during stage1
2014-07-05 03:34:54 +00:00
// network deletion request. Keep the message structure so that we can deliver delete notifications.
// Come back after several days (or perhaps a month) to do the lowlevel delete (DROPITEM_PHASE2).
if($stage == DROPITEM_PHASE1)
return true;
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("delete from term where otype = %d and oid = %d",
when deleting items, delete linked events and photos during stage1
2014-07-05 03:34:54 +00:00
intval(TERM_OBJ_POST),
intval($item['id'])
);
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
q("delete from item_id where iid = %d and uid = %d",
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
intval($item['id']),
intval($item['uid'])
);
moving on
2013-02-18 23:15:55 +00:00
q("delete from term where oid = %d and otype = %d",
intval($item['id']),
intval(TERM_OBJ_POST)
);
change deletion behaviour of linked photo items. Previously deleting the visible item in the conversation deleted the photo as well. Now photos must be deleted in the photos module. Deleting the linked item removes any attached conversation elements (likes, etc.) and sets the conversation item to hidden. This may create an issue in the future if we move the photo tags, title, or other photo elements to the linked item rather than the photo. Noting here so this can potentially be discovered and remembered at that time.
2015-01-13 00:27:27 +00:00
// FIXME remove notifications for this item
moving on
2013-02-18 23:15:55 +00:00
uncomplicate item deletion and fixup all the other linked tables
2013-01-25 00:44:10 +00:00
return true;
}
archive widget
2012-06-13 03:46:30 +00:00
function first_post_date($uid,$wall = false) {
start fixing all the item queries
2012-10-08 01:44:06 +00:00
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$wall_sql = (($wall) ? sprintf(" and (item_flags & %d)>0 ", ITEM_WALL) : "" );
start fixing all the item queries
2012-10-08 01:44:06 +00:00
$r = q("select id, created from item
where item_restrict = %d and uid = %d and id = parent $wall_sql
archive widget
2012-06-13 03:46:30 +00:00
order by created asc limit 1",
start fixing all the item queries
2012-10-08 01:44:06 +00:00
intval(ITEM_VISIBLE),
intval($uid)
archive widget
2012-06-13 03:46:30 +00:00
);
posted_dates() needs a re-write so we can make a much prettier archive widget
2014-05-01 01:45:46 +00:00
if($r) {
that was it
2012-06-13 04:39:39 +00:00
// logger('first_post_date: ' . $r[0]['id'] . ' ' . $r[0]['created'], LOGGER_DATA);
this should catch any weirdness caused by being in different timezones and making your first post on the 31st of March
2012-06-13 05:52:34 +00:00
return substr(datetime_convert('',date_default_timezone_get(),$r[0]['created']),0,10);
tracking another edge case
2012-06-13 04:34:28 +00:00
}
archive widget
2012-06-13 03:46:30 +00:00
return false;
}
posted_dates() needs a re-write so we can make a much prettier archive widget
2014-05-01 01:45:46 +00:00
/**
* modified posted_dates() {below} to arrange the list in years, which we'll eventually
* use to make a menu of years with collapsible sub-menus for the months instead of the
* current flat list of all representative dates.
*/
various UI issues related to having an enormous number of entries in the posted_date selector including years/months that had no posts
2014-11-03 22:49:03 +00:00
function list_post_dates($uid,$wall,$mindate) {
posted_dates() needs a re-write so we can make a much prettier archive widget
2014-05-01 01:45:46 +00:00
$dnow = datetime_convert('',date_default_timezone_get(),'now','Y-m-d');
various UI issues related to having an enormous number of entries in the posted_date selector including years/months that had no posts
2014-11-03 22:49:03 +00:00
if($mindate)
$dthen = datetime_convert('',date_default_timezone_get(),$mindate);
else
$dthen = first_post_date($uid,$wall);
posted_dates() needs a re-write so we can make a much prettier archive widget
2014-05-01 01:45:46 +00:00
if(! $dthen)
return array();
// If it's near the end of a long month, backup to the 28th so that in
// consecutive loops we'll always get a whole month difference.
if(intval(substr($dnow,8)) > 28)
$dnow = substr($dnow,0,8) . '28';
if(intval(substr($dthen,8)) > 28)
you've gotta' be on your toes to catch this bug... though it's been here for about 2 1/2 years and even survived a rewrite. Symptoms are that the archive widget only lists one month (the month when you first posted), and only if your first post was written between the 28th and 31st of whatever month that was.
2014-10-22 23:41:06 +00:00
$dthen = substr($dthen,0,8) . '28';
posted_dates() needs a re-write so we can make a much prettier archive widget
2014-05-01 01:45:46 +00:00
$ret = array();
// Starting with the current month, get the first and last days of every
// month down to and including the month of the first post
while(substr($dnow, 0, 7) >= substr($dthen, 0, 7)) {
$dyear = intval(substr($dnow,0,4));
$dstart = substr($dnow,0,8) . '01';
$dend = substr($dnow,0,8) . get_dim(intval($dnow),intval(substr($dnow,5)));
$start_month = datetime_convert('','',$dstart,'Y-m-d');
$end_month = datetime_convert('','',$dend,'Y-m-d');
$str = day_translate(datetime_convert('','',$dnow,'F'));
if(! $ret[$dyear])
$ret[$dyear] = array();
$ret[$dyear][] = array($str,$end_month,$start_month);
$dnow = datetime_convert('','',$dnow . ' -1 month', 'Y-m-d');
}
return $ret;
}
archive widget
2012-06-13 03:46:30 +00:00
function posted_dates($uid,$wall) {
this should catch any weirdness caused by being in different timezones and making your first post on the 31st of March
2012-06-13 05:52:34 +00:00
$dnow = datetime_convert('',date_default_timezone_get(),'now','Y-m-d');
archive widget
2012-06-13 03:46:30 +00:00
$dthen = first_post_date($uid,$wall);
if(! $dthen)
return array();
this should catch any weirdness caused by being in different timezones and making your first post on the 31st of March
2012-06-13 05:52:34 +00:00
// If it's near the end of a long month, backup to the 28th so that in
// consecutive loops we'll always get a whole month difference.
if(intval(substr($dnow,8)) > 28)
$dnow = substr($dnow,0,8) . '28';
if(intval(substr($dthen,8)) > 28)
you've gotta' be on your toes to catch this bug... though it's been here for about 2 1/2 years and even survived a rewrite. Symptoms are that the archive widget only lists one month (the month when you first posted), and only if your first post was written between the 28th and 31st of whatever month that was.
2014-10-22 23:41:06 +00:00
$dthen = substr($dthen,0,8) . '28';
this should catch any weirdness caused by being in different timezones and making your first post on the 31st of March
2012-06-13 05:52:34 +00:00
archive widget
2012-06-13 03:46:30 +00:00
$ret = array();
upstream fixes and a lot of taxonomy stuff
2012-07-10 13:28:02 +00:00
// Starting with the current month, get the first and last days of every
// month down to and including the month of the first post
while(substr($dnow, 0, 7) >= substr($dthen, 0, 7)) {
remove php version restriction on archive widget
2012-06-14 03:59:19 +00:00
$dstart = substr($dnow,0,8) . '01';
$dend = substr($dnow,0,8) . get_dim(intval($dnow),intval(substr($dnow,5)));
$start_month = datetime_convert('','',$dstart,'Y-m-d');
$end_month = datetime_convert('','',$dend,'Y-m-d');
archive widget
2012-06-13 03:46:30 +00:00
$str = day_translate(datetime_convert('','',$dnow,'F Y'));
$ret[] = array($str,$end_month,$start_month);
$dnow = datetime_convert('','',$dnow . ' -1 month', 'Y-m-d');
}
return $ret;
}
progress on tag delivery
2013-02-11 08:20:14 +00:00
function fetch_post_tags($items,$link = false) {
create fetch_tags function, make search work again, change logo
2012-07-11 02:28:02 +00:00
$tag_finder = array();
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
if($items) {
foreach($items as $item) {
zome issues registering new hubs
2013-01-03 00:28:47 +00:00
if(is_array($item)) {
if(array_key_exists('item_id',$item)) {
if(! in_array($item['item_id'],$tag_finder))
$tag_finder[] = $item['item_id'];
}
else {
if(! in_array($item['id'],$tag_finder))
$tag_finder[] = $item['id'];
}
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
}
}
}
create fetch_tags function, make search work again, change logo
2012-07-11 02:28:02 +00:00
$tag_finder_str = implode(', ', $tag_finder);
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
error cleanup
2012-07-17 12:30:32 +00:00
if(strlen($tag_finder_str)) {
$tags = q("select * from term where oid in ( %s ) and otype = %d",
dbesc($tag_finder_str),
intval(TERM_OBJ_POST)
);
}
create fetch_tags function, make search work again, change logo
2012-07-11 02:28:02 +00:00
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
create fetch_tags function, make search work again, change logo
2012-07-11 02:28:02 +00:00
for($x = 0; $x < count($items); $x ++) {
zome issues registering new hubs
2013-01-03 00:28:47 +00:00
if($tags) {
create fetch_tags function, make search work again, change logo
2012-07-11 02:28:02 +00:00
foreach($tags as $t) {
progress on tag delivery
2013-02-11 08:20:14 +00:00
if(($link) && ($t['type'] == TERM_MENTION))
$t['url'] = chanlink_url($t['url']);
more progress on notifier cleanup and encoding items for transit.
2012-11-16 05:52:05 +00:00
if(array_key_exists('item_id',$items[$x])) {
if($t['oid'] == $items[$x]['item_id']) {
if(! is_array($items[$x]['term']))
$items[$x]['term'] = array();
$items[$x]['term'][] = $t;
}
}
else {
if($t['oid'] == $items[$x]['id']) {
if(! is_array($items[$x]['term']))
$items[$x]['term'] = array();
$items[$x]['term'][] = $t;
}
create fetch_tags function, make search work again, change logo
2012-07-11 02:28:02 +00:00
}
}
}
}
return $items;
}
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
What this checkin does is catch the case where a comment arrived and there's a missing top-level post to match it with. So we'll send a request back to the sender that you've never seen this thread and please send a fresh copy of the entire conversation to date. We could soon have posts in the matrix from different platforms from days gone by, which have been migrated into the modern world. We'll be polite and not deliver these to everybody. However, if someone comments on one of these antique threads we wouldn't be able to see it in our own matrix because we won't have a copy of the parent post. So this rectifies that situation. Be aware that item deletion may need to change to keep "hard deleted" items indefinitely so that they don't keep coming back. We'll have to null out the important data of the former item to accomplish the deletion aspect.
2014-10-31 02:15:03 +00:00
function zot_feed($uid,$observer_xchan,$arr) {
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
populate some posts when somebody is granted "read stream" permission
2013-11-18 00:50:32 +00:00
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
$result = array();
What this checkin does is catch the case where a comment arrived and there's a missing top-level post to match it with. So we'll send a request back to the sender that you've never seen this thread and please send a fresh copy of the entire conversation to date. We could soon have posts in the matrix from different platforms from days gone by, which have been migrated into the modern world. We'll be polite and not deliver these to everybody. However, if someone comments on one of these antique threads we wouldn't be able to see it in our own matrix because we won't have a copy of the parent post. So this rectifies that situation. Be aware that item deletion may need to change to keep "hard deleted" items indefinitely so that they don't keep coming back. We'll have to null out the important data of the former item to accomplish the deletion aspect.
2014-10-31 02:15:03 +00:00
$mindate = null;
$message_id = null;
if(array_key_exists('mindate',$arr)) {
$mindate = datetime_convert('UTC','UTC',$arr['mindate']);
}
if(array_key_exists('message_id',$arr)) {
$message_id = $arr['message_id'];
}
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
if(! $mindate)
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
$mindate = NULL_DATE;
zot feed - getting cut off too early
2013-07-30 03:39:12 +00:00
$mindate = dbesc($mindate);
change logging of some troublesome or noisy bits so we can figure out what is really going on
2014-11-20 03:22:22 +00:00
logger('zot_feed: requested for uid ' . $uid . ' from observer ' . $observer_xchan, LOGGER_DEBUG);
more logging for zot_feed anomalies
2014-11-21 02:44:16 +00:00
if($message_id)
logger('message_id: ' . $message_id,LOGGER_DEBUG);
populate some posts when somebody is granted "read stream" permission
2013-11-18 00:50:32 +00:00
why oh why is poco_load not getting called? oh - that's why.
2013-01-24 22:31:57 +00:00
if(! perm_is_allowed($uid,$observer_xchan,'view_stream')) {
populate some posts when somebody is granted "read stream" permission
2013-11-18 00:50:32 +00:00
logger('zot_feed: permission denied.');
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
return $result;
}
there's the problem
2014-03-27 02:13:26 +00:00
if(! is_sys_channel($uid)) {
require_once('include/security.php');
$sql_extra = item_permissions_sql($uid);
}
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
This is long overdue - use a symblic constant NULL_DATE instead of the easily mis-typed sequence '0000-00-00 00:00:00'
2014-09-09 03:35:15 +00:00
if($mindate != NULL_DATE) {
issue #578 - this won't fix old items but should work going forward. The extra time slop shouldn't be needed and will in fact result in possibly undesired redundancy.
2014-09-02 02:55:43 +00:00
$sql_extra .= " and ( created > '$mindate' or edited > '$mindate' ) ";
zot feed - getting cut off too early
2013-07-30 03:39:12 +00:00
$limit = "";
}
else
$limit = " limit 0, 50 ";
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
What this checkin does is catch the case where a comment arrived and there's a missing top-level post to match it with. So we'll send a request back to the sender that you've never seen this thread and please send a fresh copy of the entire conversation to date. We could soon have posts in the matrix from different platforms from days gone by, which have been migrated into the modern world. We'll be polite and not deliver these to everybody. However, if someone comments on one of these antique threads we wouldn't be able to see it in our own matrix because we won't have a copy of the parent post. So this rectifies that situation. Be aware that item deletion may need to change to keep "hard deleted" items indefinitely so that they don't keep coming back. We'll have to null out the important data of the former item to accomplish the deletion aspect.
2014-10-31 02:15:03 +00:00
if($message_id) {
$sql_extra .= " and mid = '" . dbesc($message_id) . "' ";
$limit = '';
}
fix some weirdness on the zot feed
2013-07-30 03:31:02 +00:00
$items = array();
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
allow zotfeed to create a firehose of a site.
2014-03-26 23:09:07 +00:00
if(is_sys_channel($uid)) {
need include
2014-03-27 03:22:57 +00:00
require_once('include/security.php');
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("SELECT distinct parent, created from item
improve zotfeed for discover channel so it will send updates to posts you've already received. Incidentally I discovered why we had the meltdown replying to discover channel items the other day - but can't fix it easily.
2014-04-18 03:03:28 +00:00
WHERE uid != %d
and uid in (" . stream_perms_api_uids(PERMS_PUBLIC) . ") AND item_restrict = 0
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
AND (item_flags & %d)>0
there's the problem
2014-03-27 02:13:26 +00:00
and item_private = 0 $sql_extra ORDER BY created ASC $limit",
improve zotfeed for discover channel so it will send updates to posts you've already received. Incidentally I discovered why we had the meltdown replying to discover channel items the other day - but can't fix it easily.
2014-04-18 03:03:28 +00:00
intval($uid),
allow zotfeed to create a firehose of a site.
2014-03-26 23:09:07 +00:00
intval(ITEM_WALL)
);
}
else {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("SELECT distinct parent, created from item
improve zotfeed for discover channel so it will send updates to posts you've already received. Incidentally I discovered why we had the meltdown replying to discover channel items the other day - but can't fix it easily.
2014-04-18 03:03:28 +00:00
WHERE uid = %d AND item_restrict = 0
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
AND (item_flags & %d)>0
allow zotfeed to create a firehose of a site.
2014-03-26 23:09:07 +00:00
$sql_extra ORDER BY created ASC $limit",
intval($uid),
intval(ITEM_WALL)
);
}
fix some weirdness on the zot feed
2013-07-30 03:31:02 +00:00
allow zotfeed to create a firehose of a site.
2014-03-26 23:09:07 +00:00
if($r) {
improve zotfeed for discover channel so it will send updates to posts you've already received. Incidentally I discovered why we had the meltdown replying to discover channel items the other day - but can't fix it easily.
2014-04-18 03:03:28 +00:00
$parents_str = ids_to_querystr($r,'parent');
more optimisations for discovery mode of zotfeed
2014-04-18 05:06:35 +00:00
$sys_query = ((is_sys_channel($uid)) ? $sql_extra : '');
fix some weirdness on the zot feed
2013-07-30 03:31:02 +00:00
$items = q("SELECT `item`.*, `item`.`id` AS `item_id` FROM `item`
allow zotfeed to create a firehose of a site.
2014-03-26 23:09:07 +00:00
WHERE `item`.`item_restrict` = 0
more optimisations for discovery mode of zotfeed
2014-04-18 05:06:35 +00:00
AND `item`.`parent` IN ( %s ) $sys_query ",
fix some weirdness on the zot feed
2013-07-30 03:31:02 +00:00
dbesc($parents_str)
);
}
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
if($items) {
xchan_query($items);
$items = fetch_post_tags($items);
ensure conversation is included
2013-07-30 03:31:58 +00:00
require_once('include/conversation.php');
fix some weirdness on the zot feed
2013-07-30 03:31:02 +00:00
$items = conv_sort($items,'ascending');
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
}
fix some weirdness on the zot feed
2013-07-30 03:31:02 +00:00
else
$items = array();
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
more optimisations for discovery mode of zotfeed
2014-04-18 05:06:35 +00:00
logger('zot_feed: number items: ' . count($items),LOGGER_DEBUG);
temporary feed import/export (public only for the moment) until this is integrated with remote permissions
2013-01-04 03:34:04 +00:00
foreach($items as $item)
$result[] = encode_item($item);
return $result;
}
first cut at generic item search
2013-05-23 04:54:02 +00:00
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
function items_fetch($arr,$channel = null,$observer_hash = null,$client_mode = CLIENT_MODE_NORMAL,$module = 'network') {
first cut at generic item search
2013-05-23 04:54:02 +00:00
$result = array('success' => false);
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
$a = get_app();
first cut at generic item search
2013-05-23 04:54:02 +00:00
$sql_extra = '';
$sql_nets = '';
$sql_options = '';
$sql_extra2 = '';
$sql_extra3 = '';
suggestion widget tweaked to make it comanche capable. Remove old versions of specs that are so obsolete it isn't funny. Zot protocol reference is in red's github wiki, and in the code. We should move the github copy to /doc once it is updated to match the code. There's no point in documenting dfrn in the red code base.
2013-12-09 04:08:50 +00:00
$def_acl = '';
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
$item_uids = ' true ';
API My wall posts
2014-01-25 02:50:47 +00:00
if ($arr['uid']) $uid= $arr['uid'];
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
if($channel) {
$uid = $channel['channel_id'];
$uidhash = $channel['channel_hash'];
$item_uids = " item.uid = " . intval($uid) . " ";
}
Add $since_id to items_fetch
2014-01-22 22:15:42 +00:00
first cut at generic item search
2013-05-23 04:54:02 +00:00
if($arr['star'])
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$sql_options .= " and (item_flags & " . intval(ITEM_STARRED) . ")>0 ";
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
if($arr['wall'])
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$sql_options .= " and (item_flags & " . intval(ITEM_WALL) . ")>0 ";
API My wall posts
2014-01-25 02:50:47 +00:00
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$sql_extra = " AND item.parent IN ( SELECT parent FROM item WHERE (item_flags & " . intval(ITEM_THREAD_TOP) . ")>0 $sql_options ) ";
Add $since_id to items_fetch
2014-01-22 22:15:42 +00:00
if($arr['since_id'])
$sql_extra .= " and item.id > " . $since_id . " ";
suggestion widget tweaked to make it comanche capable. Remove old versions of specs that are so obsolete it isn't funny. Zot protocol reference is in red's github wiki, and in the code. We should move the github copy to /doc once it is updated to match the code. There's no point in documenting dfrn in the red code base.
2013-12-09 04:08:50 +00:00
if($arr['gid'] && $uid) {
remove a couple of mysql reserved words from being used as table or row names. For this round we're getting 'group' and 'desc'. Warning: potentially destabilising as this touches a lot of code.
2013-12-23 02:37:39 +00:00
$r = q("SELECT * FROM `groups` WHERE id = %d AND uid = %d LIMIT 1",
first cut at generic item search
2013-05-23 04:54:02 +00:00
intval($arr['group']),
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
intval($uid)
first cut at generic item search
2013-05-23 04:54:02 +00:00
);
if(! $r) {
$result['message'] = t('Collection not found.');
return $result;
}
sync item_search with yesterday's network fix for collections. Add ud_addr to update table to store the target address since it's possible the mirroring directory won't yet have an xchan or hubloc they can link the ud_hash to and therefore mayn't know how to contact them.
2013-09-19 08:01:51 +00:00
$contact_str = '';
$contacts = group_get_members($group);
if($contacts) {
foreach($contacts as $c) {
if($contact_str)
$contact_str .= ',';
$contact_str .= "'" . $c['xchan'] . "'";
}
first cut at generic item search
2013-05-23 04:54:02 +00:00
}
else {
sync item_search with yesterday's network fix for collections. Add ud_addr to update table to store the target address since it's possible the mirroring directory won't yet have an xchan or hubloc they can link the ud_hash to and therefore mayn't know how to contact them.
2013-09-19 08:01:51 +00:00
$contact_str = ' 0 ';
suggestion widget tweaked to make it comanche capable. Remove old versions of specs that are so obsolete it isn't funny. Zot protocol reference is in red's github wiki, and in the code. We should move the github copy to /doc once it is updated to match the code. There's no point in documenting dfrn in the red code base.
2013-12-09 04:08:50 +00:00
$result['message'] = t('Collection is empty.');
return $result;
first cut at generic item search
2013-05-23 04:54:02 +00:00
}
hopefully this will go over a bit better - check post owner permissions as well as local owner permissions for comments during submission, since the comment was being blindly accepted on the local system due to only checking the profile owner. Also change collections query to include only top-level posts by a member of the collection.
2013-09-21 00:58:11 +00:00
$sql_extra = " AND item.parent IN ( SELECT DISTINCT parent FROM item WHERE true $sql_options AND (( author_xchan IN ( $contact_str ) OR owner_xchan in ( $contact_str)) or allow_gid like '" . protect_sprintf('%<' . dbesc($r[0]['hash']) . '>%') . "' ) and id = parent and item_restrict = 0 ) ";
first cut at generic item search
2013-05-23 04:54:02 +00:00
suggestion widget tweaked to make it comanche capable. Remove old versions of specs that are so obsolete it isn't funny. Zot protocol reference is in red's github wiki, and in the code. We should move the github copy to /doc once it is updated to match the code. There's no point in documenting dfrn in the red code base.
2013-12-09 04:08:50 +00:00
$x = group_rec_byhash($uid,$r[0]['hash']);
$result['headline'] = sprintf( t('Collection: %s'),$x['name']);
first cut at generic item search
2013-05-23 04:54:02 +00:00
}
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
elseif($arr['cid'] && $uid) {
first cut at generic item search
2013-05-23 04:54:02 +00:00
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("SELECT abook.*, xchan.* from abook left join xchan on abook_xchan = xchan_hash where abook_id = %d and abook_channel = %d and not ( abook_flags & " . intval(ABOOK_FLAG_BLOCKED) . ")>0 limit 1",
first cut at generic item search
2013-05-23 04:54:02 +00:00
intval($arr['cid']),
Issue #158
2013-09-29 09:47:36 +00:00
intval(local_user())
first cut at generic item search
2013-05-23 04:54:02 +00:00
);
if($r) {
fix cid query
2013-06-28 04:40:56 +00:00
$sql_extra = " AND item.parent IN ( SELECT DISTINCT parent FROM item WHERE true $sql_options AND uid = " . intval($arr['uid']) . " AND ( author_xchan = '" . dbesc($r[0]['abook_xchan']) . "' or owner_xchan = '" . dbesc($r[0]['abook_xchan']) . "' ) and item_restrict = 0 ) ";
suggestion widget tweaked to make it comanche capable. Remove old versions of specs that are so obsolete it isn't funny. Zot protocol reference is in red's github wiki, and in the code. We should move the github copy to /doc once it is updated to match the code. There's no point in documenting dfrn in the red code base.
2013-12-09 04:08:50 +00:00
$result['headline'] = sprintf( t('Connection: %s'),$r[0]['xchan_name']);
first cut at generic item search
2013-05-23 04:54:02 +00:00
}
else {
$result['message'] = t('Connection not found.');
return $result;
}
}
if($arr['datequery']) {
$sql_extra3 .= protect_sprintf(sprintf(" AND item.created <= '%s' ", dbesc(datetime_convert(date_default_timezone_get(),'',$arr['datequery']))));
}
if($arr['datequery2']) {
$sql_extra3 .= protect_sprintf(sprintf(" AND item.created >= '%s' ", dbesc(datetime_convert(date_default_timezone_get(),'',$arr['datequery2']))));
}
if(! array_key_exists('nouveau',$arr)) {
Implements a new permission level PERMS_PENDING With this a user can allow some action to any user which connects to them, even before they've connected back. Ref. https://mobiliza.org.br/display/478d9e71eaf55748dc646d3990651d6d34cfb7db5c38360538ec730ca3ccf908@zothub.com Also some code cleanup and an alternative logic for handling notifications of permission changes in zot.php. This assumes that private posts are still restricted to people in your addressbook. Regardless of your global permissions, a pending channel won't get private posts, even if the post only has a deny clause not matching the pending channel.
2014-06-17 07:26:18 +00:00
$sql_extra2 = " AND item.parent = item.id ";
first cut at generic item search
2013-05-23 04:54:02 +00:00
$sql_extra3 = '';
}
if($arr['search']) {
if(strpos($arr['search'],'#') === 0)
$sql_extra .= term_query('item',substr($arr['search'],1),TERM_HASHTAG);
else
$sql_extra .= sprintf(" AND item.body like '%s' ",
dbesc(protect_sprintf('%' . $arr['search'] . '%'))
);
}
if(strlen($arr['file'])) {
$sql_extra .= term_query('item',$arr['files'],TERM_FILE);
}
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
if($arr['conv'] && $channel) {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$sql_extra .= sprintf(" AND parent IN (SELECT distinct parent from item where ( author_xchan like '%s' or ( item_flags & %d )>0)) ",
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
dbesc(protect_sprintf($uidhash)),
first cut at generic item search
2013-05-23 04:54:02 +00:00
intval(ITEM_MENTIONSME)
);
}
some tweaks to items_fetch for the api
2014-01-25 23:51:10 +00:00
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
if(($client_mode & CLIENT_MODE_UPDATE) && (! ($client_mode & CLIENT_MODE_LOAD))) {
first cut at generic item search
2013-05-23 04:54:02 +00:00
// only setup pagination on initial page view
$pager_sql = '';
}
else {
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
$itemspage = (($channel) ? get_pconfig($uid,'system','itemspage') : 20);
first cut at generic item search
2013-05-23 04:54:02 +00:00
$a->set_pager_itemspage(((intval($itemspage)) ? $itemspage : 20));
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$pager_sql = sprintf(" LIMIT %d OFFSET %d ", intval(get_app()->pager['itemspage']), intval(get_app()->pager['start']));
first cut at generic item search
2013-05-23 04:54:02 +00:00
}
some tweaks to items_fetch for the api
2014-01-25 23:51:10 +00:00
if(isset($arr['start']) && isset($arr['records']))
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$pager_sql = sprintf(" LIMIT %d OFFSET %d ", intval($arr['records']), intval($arr['start']));
first cut at generic item search
2013-05-23 04:54:02 +00:00
cleanup
2014-03-18 23:50:46 +00:00
if(array_key_exists('cmin',$arr) || array_key_exists('cmax',$arr)) {
Implements a new permission level PERMS_PENDING With this a user can allow some action to any user which connects to them, even before they've connected back. Ref. https://mobiliza.org.br/display/478d9e71eaf55748dc646d3990651d6d34cfb7db5c38360538ec730ca3ccf908@zothub.com Also some code cleanup and an alternative logic for handling notifications of permission changes in zot.php. This assumes that private posts are still restricted to people in your addressbook. Regardless of your global permissions, a pending channel won't get private posts, even if the post only has a deny clause not matching the pending channel.
2014-06-17 07:26:18 +00:00
if(($arr['cmin'] != 0) || ($arr['cmax'] != 99)) {
first cut at generic item search
2013-05-23 04:54:02 +00:00
Implements a new permission level PERMS_PENDING With this a user can allow some action to any user which connects to them, even before they've connected back. Ref. https://mobiliza.org.br/display/478d9e71eaf55748dc646d3990651d6d34cfb7db5c38360538ec730ca3ccf908@zothub.com Also some code cleanup and an alternative logic for handling notifications of permission changes in zot.php. This assumes that private posts are still restricted to people in your addressbook. Regardless of your global permissions, a pending channel won't get private posts, even if the post only has a deny clause not matching the pending channel.
2014-06-17 07:26:18 +00:00
// Not everybody who shows up in the network stream will be in your address book.
// By default those that aren't are assumed to have closeness = 99; but this isn't
// recorded anywhere. So if cmax is 99, we'll open the search up to anybody in
// the stream with a NULL address book entry.
first cut at generic item search
2013-05-23 04:54:02 +00:00
cleanup
2014-03-18 23:50:46 +00:00
$sql_nets .= " AND ";
first cut at generic item search
2013-05-23 04:54:02 +00:00
cleanup
2014-03-18 23:50:46 +00:00
if($arr['cmax'] == 99)
$sql_nets .= " ( ";
first cut at generic item search
2013-05-23 04:54:02 +00:00
cleanup
2014-03-18 23:50:46 +00:00
$sql_nets .= "( abook.abook_closeness >= " . intval($arr['cmin']) . " ";
$sql_nets .= " AND abook.abook_closeness <= " . intval($arr['cmax']) . " ) ";
if($cmax == 99)
$sql_nets .= " OR abook.abook_closeness IS NULL ) ";
}
}
first cut at generic item search
2013-05-23 04:54:02 +00:00
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$simple_update = (($client_mode & CLIENT_MODE_UPDATE) ? " and ( item.item_flags & " . intval(ITEM_UNSEEN) . " )>0 " : '');
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
if($client_mode & CLIENT_MODE_LOAD)
first cut at generic item search
2013-05-23 04:54:02 +00:00
$simple_update = '';
$start = dba_timer();
missing include
2013-06-14 00:53:16 +00:00
require_once('include/security.php');
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
$sql_extra .= item_permissions_sql($channel['channel_id']);
pages feed
2013-09-06 02:31:26 +00:00
if($arr['pages'])
$item_restrict = " AND (item_restrict & " . ITEM_WEBPAGE . ") ";
else
$item_restrict = " AND item_restrict = 0 ";
some tweaks to items_fetch for the api
2014-01-25 23:51:10 +00:00
if($arr['nouveau'] && ($client_mode & CLIENT_MODE_LOAD) && $channel) {
first cut at generic item search
2013-05-23 04:54:02 +00:00
// "New Item View" - show all items unthreaded in reverse created date order
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
$items = q("SELECT item.*, item.id AS item_id FROM item
pages feed
2013-09-06 02:31:26 +00:00
WHERE $item_uids $item_restrict
first cut at generic item search
2013-05-23 04:54:02 +00:00
$simple_update
$sql_extra $sql_nets
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
ORDER BY item.received DESC $pager_sql "
first cut at generic item search
2013-05-23 04:54:02 +00:00
);
require_once('include/items.php');
xchan_query($items);
$items = fetch_post_tags($items,true);
}
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
else {
first cut at generic item search
2013-05-23 04:54:02 +00:00
// Normal conversation view
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
if($arr['order'] === 'post')
$ordering = "created";
first cut at generic item search
2013-05-23 04:54:02 +00:00
else
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
$ordering = "commented";
first cut at generic item search
2013-05-23 04:54:02 +00:00
some tweaks to items_fetch for the api
2014-01-25 23:51:10 +00:00
if(($client_mode & CLIENT_MODE_LOAD) || ($client_mode == CLIENT_MODE_NORMAL)) {
first cut at generic item search
2013-05-23 04:54:02 +00:00
// Fetch a page full of parent items for this page
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("SELECT distinct item.id AS item_id, item.$ordering FROM item
first cut at generic item search
2013-05-23 04:54:02 +00:00
left join abook on item.author_xchan = abook.abook_xchan
pages feed
2013-09-06 02:31:26 +00:00
WHERE $item_uids $item_restrict
first cut at generic item search
2013-05-23 04:54:02 +00:00
AND item.parent = item.id
and ((abook.abook_flags & %d) = 0 or abook.abook_flags is null)
$sql_extra3 $sql_extra $sql_nets
ORDER BY item.$ordering DESC $pager_sql ",
intval(ABOOK_FLAG_BLOCKED)
);
API My wall posts
2014-01-25 02:50:47 +00:00
first cut at generic item search
2013-05-23 04:54:02 +00:00
}
else {
// update
$r = q("SELECT item.parent AS item_id FROM item
left join abook on item.author_xchan = abook.abook_xchan
pages feed
2013-09-06 02:31:26 +00:00
WHERE $item_uids $item_restrict $simple_update
first cut at generic item search
2013-05-23 04:54:02 +00:00
and ((abook.abook_flags & %d) = 0 or abook.abook_flags is null)
$sql_extra3 $sql_extra $sql_nets ",
intval(ABOOK_FLAG_BLOCKED)
);
}
$first = dba_timer();
// Then fetch all the children of the parents that are on this page
if($r) {
$parents_str = ids_to_querystr($r,'item_id');
issue #571 - use feed/nickname?top=1
2014-08-22 22:51:48 +00:00
if($arr['top'])
$sql_extra = ' and id = parent ' . $sql_extra;
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
$items = q("SELECT item.*, item.id AS item_id FROM item
pages feed
2013-09-06 02:31:26 +00:00
WHERE $item_uids $item_restrict
more progress on items_fetch (new name)
2013-05-24 01:50:27 +00:00
AND item.parent IN ( %s )
first cut at generic item search
2013-05-23 04:54:02 +00:00
$sql_extra ",
dbesc($parents_str)
);
$second = dba_timer();
xchan_query($items);
$third = dba_timer();
$items = fetch_post_tags($items,true);
$fourth = dba_timer();
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
require_once('include/conversation.php');
first cut at generic item search
2013-05-23 04:54:02 +00:00
$items = conv_sort($items,$ordering);
//logger('items: ' . print_r($items,true));
}
else {
$items = array();
}
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
if($parents_str && $arr['mark_seen'])
first cut at generic item search
2013-05-23 04:54:02 +00:00
$update_unseen = ' AND parent IN ( ' . dbesc($parents_str) . ' )';
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
// FIXME finish mark unseen sql
first cut at generic item search
2013-05-23 04:54:02 +00:00
}
very minimal Atom feed - there will be a lot of bugs and unparseable content if you've got elaborately structured objects, but I got the thing to at least render in firefox given some quick test posts. If you find bugs (as you certainly will), feel free to fix them.
2013-06-13 03:03:04 +00:00
return $items;
Add explanation why public comments are disabled.
2013-09-25 19:43:21 +00:00
}
a few fixes to webpage editing
2014-01-04 21:44:43 +00:00
function update_remote_id($channel,$post_id,$webpage,$pagetitle,$namespace,$remote_id,$mid) {
$page_type = '';
if($webpage & ITEM_WEBPAGE)
$page_type = 'WEBPAGE';
elseif($webpage & ITEM_BUILDBLOCK)
$page_type = 'BUILDBLOCK';
elseif($webpage & ITEM_PDL)
$page_type = 'PDL';
elseif($namespace && $remote_id) {
$page_type = $namespace;
$pagetitle = $remote_id;
}
if($page_type) {
// store page info as an alternate message_id so we can access it via
// https://sitename/page/$channelname/$pagetitle
// if no pagetitle was given or it couldn't be transliterated into a url, use the first
// sixteen bytes of the mid - which makes the link portable and not quite as daunting
// as the entire mid. If it were the post_id the link would be less portable.
$r = q("select * from item_id where iid = %d and uid = %d and service = '%s' limit 1",
intval($post_id),
intval($channel['channel_id']),
dbesc($page_type)
);
if($r) {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
q("update item_id set sid = '%s' where id = %d",
a few fixes to webpage editing
2014-01-04 21:44:43 +00:00
dbesc(($pagetitle) ? $pagetitle : substr($mid,0,16)),
intval($r[0]['id'])
);
}
else {
q("insert into item_id ( iid, uid, sid, service ) values ( %d, %d, '%s','%s' )",
intval($post_id),
intval($channel['channel_id']),
dbesc(($pagetitle) ? $pagetitle : substr($mid,0,16)),
dbesc($page_type)
);
}
}
}
access control modification functions (item specific - we may need these for several data types)
2014-06-18 02:21:46 +00:00
/**
* change access control for item with message_id $mid and channel_id $uid
*/
function item_add_cid($xchan_hash,$mid,$uid) {
$r = q("select id from item where mid = '%s' and uid = %d and allow_cid like '%s'",
dbesc($mid),
intval($uid),
dbesc('<' . $xchan_hash . '>')
);
if(! $r) {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$r = q("update item set allow_cid = concat(allow_cid,'%s') where mid = '%s' and uid = %d",
access control modification functions (item specific - we may need these for several data types)
2014-06-18 02:21:46 +00:00
dbesc('<' . $xchan_hash . '>'),
dbesc($mid),
intval($uid)
);
}
}
function item_remove_cid($xchan_hash,$mid,$uid) {
$r = q("select allow_cid from item where mid = '%s' and uid = %d and allow_cid like '%s'",
dbesc($mid),
intval($uid),
dbesc('<' . $xchan_hash . '>')
);
if($r) {
PostgreSQL support initial commit There were 11 main types of changes: - UPDATE's and DELETE's sometimes had LIMIT 1 at the end of them. This is not only non-compliant but it would certainly not do what whoever wrote it thought it would. It is likely this mistake was just copied from Friendica. All of these instances, the LIMIT 1 was simply removed. - Bitwise operations (and even some non-zero int checks) erroneously rely on MySQL implicit integer-boolean conversion in the WHERE clauses. This is non-compliant (and bad programming practice to boot). Proper explicit boolean conversions were added. New queries should use proper conventions. - MySQL has a different operator for bitwise XOR than postgres. Rather than add yet another dba_ func, I converted them to "& ~" ("AND NOT") when turning off, and "|" ("OR") when turning on. There were no true toggles (XOR). New queries should refrain from using XOR when not necessary. - There are several fields which the schema has marked as NOT NULL, but the inserts don't specify them. The reason this works is because mysql totally ignores the constraint and adds an empty text default automatically. Again, non-compliant, obviously. In these cases a default of empty text was added. - Several statements rely on a non-standard MySQL feature (http://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html). These queries can all be rewritten to be standards compliant. Interestingly enough, the newly rewritten standards compliant queries run a zillion times faster, even on MySQL. - A couple of function/operator name translations were needed (RAND/RANDOM, GROUP_CONCAT/STRING_AGG, UTC_NOW, REGEXP/~, ^/#) -- assist functions added in the dba_ - INTERVALs: postgres requires quotes around the value, mysql requires that there are not quotes around the value -- assist functions added in the dba_ - NULL_DATE's -- Postgres does not allow the invalid date '0000-00-00 00:00:00' (there is no such thing as year 0 or month 0 or day 0). We use '0001-01-01 00:00:00' for postgres. Conversions are handled in Zot/item packets automagically by quoting all dates with dbescdate(). - char(##) specifications in the schema creates fields with blank spaces that aren't trimmed in the code. MySQL apparently treats char(##) as varchar(##), again, non-compliant. Since postgres works better with text fields anyway, this ball of bugs was simply side-stepped by using 'text' datatype for all text fields in the postgres schema. varchar was used in a couple of places where it actually seemed appropriate (size constraint), but without rigorously vetting that all of the PHP code actually validates data, new bugs might come out from under the rug. - postgres doesn't store nul bytes and a few other non-printables in text fields, even when quoted. bytea fields were used when storing binary data (photo.data, attach.data). A new dbescbin() function was added to handle this transparently. - postgres does not support LIMIT #,# syntax. All databases support LIMIT # OFFSET # syntax. Statements were updated to be standard. These changes require corresponding changes in the coding standards. Please review those before adding any code going forward. Still on my TODO list: - remove quotes from non-reserved identifiers and make reserved identifiers use dba func for quoting - Rewrite search queries for better results (both MySQL and Postgres)
2014-11-13 20:21:58 +00:00
$x = q("update item set allow_cid = '%s' where mid = '%s' and uid = %d",
access control modification functions (item specific - we may need these for several data types)
2014-06-18 02:21:46 +00:00
dbesc(str_replace('<' . $xchan_hash . '>','',$r[0]['allow_cid'])),
dbesc($mid),
intval($uid)
);
}
}
Refactor mention code to make it more reusable
2015-01-12 19:01:07 +00:00
// Set item permissions based on results obtained from linkify_tags()
function set_linkified_perms($linkified, &$str_contact_allow, &$str_group_allow, $profile_uid, $parent_item = false) {
$first_access_tag = true;
foreach($linkified as $x) {
$access_tag = $x['access_tag'];
if(($access_tag) && (! $parent_item)) {
logger('access_tag: ' . $tag . ' ' . print_r($access_tag,true), LOGGER_DATA);
if ($first_access_tag && (! get_pconfig($profile_uid,'system','no_private_mention_acl_override'))) {
// This is a tough call, hence configurable. The issue is that one can type in a @!privacy mention
// and also have a default ACL (perhaps from viewing a collection) and could be suprised that the
// privacy mention wasn't the only recipient. So the default is to wipe out the existing ACL if a
// private mention is found. This can be over-ridden if you wish private mentions to be in
// addition to the current ACL settings.
$str_contact_allow = '';
$str_group_allow = '';
$first_access_tag = false;
}
if(strpos($access_tag,'cid:') === 0) {
$str_contact_allow .= '<' . substr($access_tag,4) . '>';
$access_tag = '';
}
elseif(strpos($access_tag,'gid:') === 0) {
$str_group_allow .= '<' . substr($access_tag,4) . '>';
$access_tag = '';
}
}
}
}
Reference in a new issue Copy permalink