friendica-github/include/Scrape.php

843 lines
24 KiB
PHP
Raw Normal View History

2010-07-01 23:48:07 +00:00
<?php
require_once('library/HTML5/Parser.php');
2011-08-12 10:01:11 +00:00
require_once('include/crypto.php');
2010-07-01 23:48:07 +00:00
if(! function_exists('scrape_dfrn')) {
function scrape_dfrn($url) {
$a = get_app();
2010-07-01 23:48:07 +00:00
$ret = array();
logger('scrape_dfrn: url=' . $url);
2010-07-01 23:48:07 +00:00
$s = fetch_url($url);
if(! $s)
2010-07-01 23:48:07 +00:00
return $ret;
$headers = $a->get_curl_headers();
logger('scrape_dfrn: headers=' . $headers, LOGGER_DEBUG);
$lines = explode("\n",$headers);
if(count($lines)) {
foreach($lines as $line) {
// don't try and run feeds through the html5 parser
if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
return ret;
}
}
2011-10-20 23:48:07 +00:00
try {
$dom = HTML5_Parser::parse($s);
} catch (DOMException $e) {
logger('scrape_dfrn: parse error: ' . $e);
}
2010-07-01 23:48:07 +00:00
if(! $dom)
return $ret;
$items = $dom->getElementsByTagName('link');
// get DFRN link elements
foreach($items as $item) {
$x = $item->getAttribute('rel');
2011-01-24 04:09:34 +00:00
if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
$ret['feed_atom'] = $item->getAttribute('href');
2011-08-18 06:01:44 +00:00
if(substr($x,0,5) == "dfrn-") {
2010-07-01 23:48:07 +00:00
$ret[$x] = $item->getAttribute('href');
2011-08-18 06:01:44 +00:00
}
if($x === 'lrdd') {
$decoded = urldecode($item->getAttribute('href'));
if(preg_match('/acct:([^@]*)@/',$decoded,$matches))
$ret['nick'] = $matches[1];
}
2010-07-01 23:48:07 +00:00
}
// Pull out hCard profile elements
2011-08-18 06:01:44 +00:00
$largest_photo = 0;
2010-07-01 23:48:07 +00:00
$items = $dom->getElementsByTagName('*');
foreach($items as $item) {
if(attribute_contains($item->getAttribute('class'), 'vcard')) {
$level2 = $item->getElementsByTagName('*');
foreach($level2 as $x) {
2011-08-18 06:01:44 +00:00
if(attribute_contains($x->getAttribute('class'),'fn')) {
2010-07-01 23:48:07 +00:00
$ret['fn'] = $x->textContent;
2011-08-18 06:01:44 +00:00
}
if((attribute_contains($x->getAttribute('class'),'photo'))
|| (attribute_contains($x->getAttribute('class'),'avatar'))) {
$size = intval($x->getAttribute('width'));
// dfrn prefers 175, so if we find this, we set largest_size so it can't be topped.
if(($size > $largest_photo) || ($size == 175) || (! $largest_photo)) {
$ret['photo'] = $x->getAttribute('src');
$largest_photo = (($size == 175) ? 9999 : $size);
}
}
if(attribute_contains($x->getAttribute('class'),'key')) {
2010-07-01 23:48:07 +00:00
$ret['key'] = $x->textContent;
2011-08-18 06:01:44 +00:00
}
}
2010-07-01 23:48:07 +00:00
}
}
return $ret;
}}
if(! function_exists('validate_dfrn')) {
function validate_dfrn($a) {
$errors = 0;
if(! x($a,'key'))
$errors ++;
if(! x($a,'dfrn-request'))
$errors ++;
if(! x($a,'dfrn-confirm'))
$errors ++;
if(! x($a,'dfrn-notify'))
$errors ++;
if(! x($a,'dfrn-poll'))
$errors ++;
return $errors;
}}
if(! function_exists('scrape_meta')) {
function scrape_meta($url) {
2010-07-01 23:48:07 +00:00
$a = get_app();
$ret = array();
logger('scrape_meta: url=' . $url);
$s = fetch_url($url);
if(! $s)
return $ret;
$headers = $a->get_curl_headers();
logger('scrape_meta: headers=' . $headers, LOGGER_DEBUG);
$lines = explode("\n",$headers);
if(count($lines)) {
foreach($lines as $line) {
// don't try and run feeds through the html5 parser
if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
return ret;
}
}
2011-10-20 23:48:07 +00:00
try {
$dom = HTML5_Parser::parse($s);
} catch (DOMException $e) {
logger('scrape_meta: parse error: ' . $e);
}
if(! $dom)
return $ret;
2010-07-01 23:48:07 +00:00
$items = $dom->getElementsByTagName('meta');
// get DFRN link elements
foreach($items as $item) {
$x = $item->getAttribute('name');
if(substr($x,0,5) == "dfrn-")
$ret[$x] = $item->getAttribute('content');
}
return $ret;
}}
if(! function_exists('scrape_vcard')) {
function scrape_vcard($url) {
$a = get_app();
$ret = array();
logger('scrape_vcard: url=' . $url);
$s = fetch_url($url);
if(! $s)
return $ret;
$headers = $a->get_curl_headers();
$lines = explode("\n",$headers);
if(count($lines)) {
foreach($lines as $line) {
// don't try and run feeds through the html5 parser
if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
return ret;
}
}
2011-10-20 23:48:07 +00:00
try {
$dom = HTML5_Parser::parse($s);
} catch (DOMException $e) {
logger('scrape_vcard: parse error: ' . $e);
}
if(! $dom)
return $ret;
// Pull out hCard profile elements
2011-08-12 10:01:11 +00:00
$largest_photo = 0;
$items = $dom->getElementsByTagName('*');
foreach($items as $item) {
if(attribute_contains($item->getAttribute('class'), 'vcard')) {
$level2 = $item->getElementsByTagName('*');
foreach($level2 as $x) {
if(attribute_contains($x->getAttribute('class'),'fn'))
$ret['fn'] = $x->textContent;
if((attribute_contains($x->getAttribute('class'),'photo'))
2011-08-12 10:01:11 +00:00
|| (attribute_contains($x->getAttribute('class'),'avatar'))) {
$size = intval($x->getAttribute('width'));
if(($size > $largest_photo) || (! $largest_photo)) {
$ret['photo'] = $x->getAttribute('src');
$largest_photo = $size;
}
}
if((attribute_contains($x->getAttribute('class'),'nickname'))
2011-08-18 06:01:44 +00:00
|| (attribute_contains($x->getAttribute('class'),'uid'))) {
$ret['nick'] = $x->textContent;
2011-08-18 06:01:44 +00:00
}
}
}
}
return $ret;
}}
2011-01-24 04:09:34 +00:00
if(! function_exists('scrape_feed')) {
function scrape_feed($url) {
$a = get_app();
2011-01-24 04:09:34 +00:00
$ret = array();
$s = fetch_url($url);
2012-02-26 22:34:01 +00:00
$headers = $a->get_curl_headers();
$code = $a->get_curl_code();
logger('scrape_feed: returns: ' . $code . ' headers=' . $headers, LOGGER_DEBUG);
if(! $s) {
logger('scrape_feed: no data returned for ' . $url);
2011-01-24 04:09:34 +00:00
return $ret;
2012-02-26 22:34:01 +00:00
}
2011-01-24 04:09:34 +00:00
$lines = explode("\n",$headers);
if(count($lines)) {
foreach($lines as $line) {
if(stristr($line,'content-type:')) {
if(stristr($line,'application/atom+xml') || stristr($s,'<feed')) {
$ret['feed_atom'] = $url;
return $ret;
}
2015-07-22 09:51:37 +00:00
if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) {
$ret['feed_rss'] = $url;
2011-04-05 02:36:18 +00:00
return $ret;
}
}
}
// perhaps an RSS version 1 feed with a generic or incorrect content-type?
if(stristr($s,'</item>')) {
$ret['feed_rss'] = $url;
return $ret;
}
}
2015-07-22 09:51:37 +00:00
$basename = implode('/', array_slice(explode('/',$url),0,3)) . '/';
2011-01-24 04:09:34 +00:00
2015-07-22 09:51:37 +00:00
$doc = new DOMDocument();
@$doc->loadHTML($s);
$xpath = new DomXPath($doc);
2011-01-24 04:09:34 +00:00
2015-07-22 09:51:37 +00:00
$base = $xpath->query("//base");
foreach ($base as $node) {
$attr = array();
2011-04-05 03:40:31 +00:00
2015-07-22 09:51:37 +00:00
if ($node->attributes->length)
foreach ($node->attributes as $attribute)
$attr[$attribute->name] = $attribute->value;
if ($attr["href"] != "")
$basename = $attr["href"] ;
}
2015-07-22 09:51:37 +00:00
$list = $xpath->query("//link");
foreach ($list as $node) {
$attr = array();
2011-01-24 04:09:34 +00:00
2015-07-22 09:51:37 +00:00
if ($node->attributes->length)
foreach ($node->attributes as $attribute)
$attr[$attribute->name] = $attribute->value;
2011-01-24 04:09:34 +00:00
2015-07-22 09:51:37 +00:00
if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/atom+xml"))
$ret["feed_atom"] = $attr["href"];
if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/rss+xml"))
$ret["feed_rss"] = $attr["href"];
2011-01-24 04:09:34 +00:00
}
2015-07-22 09:51:37 +00:00
// Drupal and perhaps others only provide relative URLs. Turn them into absolute.
if(x($ret,'feed_atom') && (! strstr($ret['feed_atom'],'://')))
$ret['feed_atom'] = $basename . $ret['feed_atom'];
if(x($ret,'feed_rss') && (! strstr($ret['feed_rss'],'://')))
$ret['feed_rss'] = $basename . $ret['feed_rss'];
2011-01-24 04:09:34 +00:00
return $ret;
}}
2011-08-18 06:10:55 +00:00
/**
*
* Probe a network address to discover what kind of protocols we need to communicate with it.
*
* Warning: this function is a bit touchy and there are some subtle dependencies within the logic flow.
* Edit with care.
*
*/
/**
*
* PROBE_DIASPORA has a bias towards returning Diaspora information
* while PROBE_NORMAL has a bias towards dfrn/zot - in the case where
2012-02-19 18:27:54 +00:00
* an address (such as a Friendica address) supports more than one type
2011-08-18 06:10:55 +00:00
* of network.
*
*/
2011-08-18 06:01:44 +00:00
define ( 'PROBE_NORMAL', 0);
define ( 'PROBE_DIASPORA', 1);
function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
2011-04-16 15:45:08 +00:00
require_once('include/email.php');
$result = array();
if(! $url)
return $result;
$result = Cache::get("probe_url:".$mode.":".$url);
if (!is_null($result)) {
$result = unserialize($result);
return $result;
}
$network = null;
2011-08-12 10:01:11 +00:00
$diaspora = false;
$diaspora_base = '';
$diaspora_guid = '';
2011-08-12 10:01:11 +00:00
$diaspora_key = '';
$has_lrdd = false;
$email_conversant = false;
$connectornetworks = false;
$appnet = false;
if (strpos($url,'twitter.com')) {
$connectornetworks = true;
$network = NETWORK_TWITTER;
}
// Twitter is deactivated since twitter closed its old API
//$twitter = ((strpos($url,'twitter.com') !== false) ? true : false);
2012-06-21 05:59:01 +00:00
$lastfm = ((strpos($url,'last.fm/user') !== false) ? true : false);
$at_addr = ((strpos($url,'@') !== false) ? true : false);
if((!$appnet) && (!$lastfm) && !$connectornetworks) {
if(strpos($url,'mailto:') !== false && $at_addr) {
$url = str_replace('mailto:','',$url);
$links = array();
}
else
$links = lrdd($url);
if(count($links)) {
$has_lrdd = true;
logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
foreach($links as $link) {
2011-07-12 01:28:13 +00:00
if($link['@attributes']['rel'] === NAMESPACE_ZOT)
$zot = unamp($link['@attributes']['href']);
if($link['@attributes']['rel'] === NAMESPACE_DFRN)
$dfrn = unamp($link['@attributes']['href']);
if($link['@attributes']['rel'] === 'salmon')
$notify = unamp($link['@attributes']['href']);
if($link['@attributes']['rel'] === NAMESPACE_FEED)
$poll = unamp($link['@attributes']['href']);
if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
$hcard = unamp($link['@attributes']['href']);
if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
$profile = unamp($link['@attributes']['href']);
if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
$poco = unamp($link['@attributes']['href']);
2011-08-12 10:01:11 +00:00
if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
$diaspora_base = unamp($link['@attributes']['href']);
$diaspora = true;
}
if($link['@attributes']['rel'] === 'http://joindiaspora.com/guid') {
$diaspora_guid = unamp($link['@attributes']['href']);
$diaspora = true;
2011-08-12 10:01:11 +00:00
}
if($link['@attributes']['rel'] === 'diaspora-public-key') {
$diaspora_key = base64_decode(unamp($link['@attributes']['href']));
if(strstr($diaspora_key,'RSA '))
$pubkey = rsatopem($diaspora_key);
else
$pubkey = $diaspora_key;
2011-08-12 10:01:11 +00:00
$diaspora = true;
}
if($link['@attributes']['rel'] === 'http://ostatus.org/schema/1.0/subscribe') {
$diaspora = false;
}
}
// Status.Net can have more than one profile URL. We need to match the profile URL
// to a contact on incoming messages to prevent spam, and we won't know which one
// to match. So in case of two, one of them is stored as an alias. Only store URL's
// and not webfinger user@host aliases. If they've got more than two non-email style
// aliases, let's hope we're lucky and get one that matches the feed author-uri because
// otherwise we're screwed.
foreach($links as $link) {
if($link['@attributes']['rel'] === 'alias') {
if(strpos($link['@attributes']['href'],'@') === false) {
if(isset($profile)) {
if($link['@attributes']['href'] !== $profile)
$alias = unamp($link['@attributes']['href']);
}
else
$profile = unamp($link['@attributes']['href']);
}
}
}
// If the profile is different from the url then the url is abviously an alias
if (($alias == "") AND ($profile != "") AND !$at_addr AND (normalise_link($profile) != normalise_link($url)))
$alias = $url;
}
2011-08-18 06:01:44 +00:00
elseif($mode == PROBE_NORMAL) {
2011-04-16 15:45:08 +00:00
// Check email
$orig_url = $url;
if((strpos($orig_url,'@')) && validate_email($orig_url)) {
2011-04-16 15:45:08 +00:00
$x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
intval(local_user())
);
2011-04-18 06:27:11 +00:00
$r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1",
2011-04-16 15:45:08 +00:00
intval(local_user())
);
if(count($x) && count($r)) {
$mailbox = construct_mailbox_name($r[0]);
2011-04-16 15:45:08 +00:00
$password = '';
openssl_private_decrypt(hex2bin($r[0]['pass']),$password,$x[0]['prvkey']);
$mbox = email_connect($mailbox,$r[0]['user'],$password);
if(! $mbox)
logger('probe_url: email_connect failed.');
2011-04-16 15:45:08 +00:00
unset($password);
}
if($mbox) {
$msgs = email_poll($mbox,$orig_url);
logger('probe_url: searching ' . $orig_url . ', ' . count($msgs) . ' messages found.', LOGGER_DEBUG);
2011-04-16 15:45:08 +00:00
if(count($msgs)) {
$addr = $orig_url;
$network = NETWORK_MAIL;
$name = substr($url,0,strpos($url,'@'));
2011-09-13 05:24:36 +00:00
$phost = substr($url,strpos($url,'@')+1);
$profile = 'http://' . $phost;
2011-04-17 22:24:24 +00:00
// fix nick character range
$vcard = array('fn' => $name, 'nick' => $name, 'photo' => avatar_img($url));
2011-04-18 06:27:11 +00:00
$notify = 'smtp ' . random_string();
$poll = 'email ' . random_string();
2011-04-16 15:45:08 +00:00
$priority = 0;
2011-04-17 22:24:24 +00:00
$x = email_msg_meta($mbox,$msgs[0]);
if(stristr($x[0]->from,$orig_url))
$adr = imap_rfc822_parse_adrlist($x[0]->from,'');
elseif(stristr($x[0]->to,$orig_url))
$adr = imap_rfc822_parse_adrlist($x[0]->to,'');
2011-09-13 05:24:36 +00:00
if(isset($adr)) {
foreach($adr as $feadr) {
if((strcasecmp($feadr->mailbox,$name) == 0)
&&(strcasecmp($feadr->host,$phost) == 0)
2011-09-13 05:24:36 +00:00
&& (strlen($feadr->personal))) {
$personal = imap_mime_header_decode($feadr->personal);
$vcard['fn'] = "";
foreach($personal as $perspart)
if ($perspart->charset != "default")
$vcard['fn'] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
else
$vcard['fn'] .= $perspart->text;
$vcard['fn'] = notags($vcard['fn']);
2011-09-13 05:24:36 +00:00
}
}
}
2011-04-16 15:45:08 +00:00
}
imap_close($mbox);
}
}
}
}
2011-08-18 06:01:44 +00:00
if($mode == PROBE_NORMAL) {
2015-07-22 09:51:37 +00:00
2011-08-18 06:01:44 +00:00
if(strlen($zot)) {
$s = fetch_url($zot);
if($s) {
$j = json_decode($s);
if($j) {
$network = NETWORK_ZOT;
$vcard = array(
'fn' => $j->fullname,
'nick' => $j->nickname,
2011-08-18 06:01:44 +00:00
'photo' => $j->photo
);
$profile = $j->url;
$notify = $j->post;
$pubkey = $j->pubkey;
$poll = 'N/A';
}
2011-07-12 01:28:13 +00:00
}
}
2015-07-22 09:51:37 +00:00
2011-08-18 06:01:44 +00:00
if(strlen($dfrn)) {
$ret = scrape_dfrn(($hcard) ? $hcard : $dfrn);
2011-08-18 06:01:44 +00:00
if(is_array($ret) && x($ret,'dfrn-request')) {
$network = NETWORK_DFRN;
$request = $ret['dfrn-request'];
$confirm = $ret['dfrn-confirm'];
$notify = $ret['dfrn-notify'];
$poll = $ret['dfrn-poll'];
$vcard = array();
$vcard['fn'] = $ret['fn'];
$vcard['nick'] = $ret['nick'];
$vcard['photo'] = $ret['photo'];
}
}
}
2011-08-12 10:01:11 +00:00
if($diaspora && $diaspora_base && $diaspora_guid) {
if($mode == PROBE_DIASPORA || ! $notify) {
2011-08-19 11:11:06 +00:00
$notify = $diaspora_base . 'receive/users/' . $diaspora_guid;
$batch = $diaspora_base . 'receive/public' ;
}
2011-08-12 10:01:11 +00:00
if(strpos($url,'@'))
$addr = str_replace('acct:', '', $url);
}
2011-08-12 10:01:11 +00:00
2011-07-12 01:28:13 +00:00
if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) {
2011-08-12 10:01:11 +00:00
if($diaspora)
$network = NETWORK_DIASPORA;
elseif($has_lrdd)
2011-08-12 10:01:11 +00:00
$network = NETWORK_OSTATUS;
if(strpos($url,'@'))
$addr = str_replace('acct:', '', $url);
$priority = 0;
2011-08-18 06:01:44 +00:00
if($hcard && ! $vcard) {
$vcard = scrape_vcard($hcard);
// Google doesn't use absolute url in profile photos
if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') {
$h = @parse_url($hcard);
if($h)
$vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo'];
}
logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA);
}
if($diaspora && $addr) {
// Diaspora returns the name as the nick. As the nick will never be updated,
// let's use the Diaspora nickname (the first part of the handle) as the nick instead
$addr_parts = explode('@', $addr);
$vcard['nick'] = $addr_parts[0];
}
/* if($twitter) {
logger('twitter: setup');
$tid = basename($url);
$tapi = 'https://api.twitter.com/1/statuses/user_timeline.rss';
if(intval($tid))
$poll = $tapi . '?user_id=' . $tid;
else
$poll = $tapi . '?screen_name=' . $tid;
2011-06-28 03:51:33 +00:00
$profile = 'http://twitter.com/#!/' . $tid;
2012-09-06 23:31:05 +00:00
//$vcard['photo'] = 'https://api.twitter.com/1/users/profile_image/' . $tid;
$vcard['photo'] = 'https://api.twitter.com/1/users/profile_image?screen_name=' . $tid . '&size=bigger';
2011-09-21 00:56:44 +00:00
$vcard['nick'] = $tid;
2012-09-06 23:31:05 +00:00
$vcard['fn'] = $tid;
} */
2012-06-21 05:59:01 +00:00
if($lastfm) {
$profile = $url;
$poll = str_replace(array('www.','last.fm/'),array('','ws.audioscrobbler.com/1.0/'),$url) . '/recenttracks.rss';
$vcard['nick'] = basename($url);
2012-06-21 06:04:55 +00:00
$vcard['fn'] = $vcard['nick'] . t(' on Last.fm');
2012-06-21 05:59:01 +00:00
$network = NETWORK_FEED;
}
if(! x($vcard,'fn'))
if(x($vcard,'nick'))
$vcard['fn'] = $vcard['nick'];
2011-08-15 01:13:52 +00:00
$check_feed = false;
if(stristr($url,'tumblr.com') && (! stristr($url,'/rss'))) {
$poll = $url . '/rss';
$check_feed = true;
// Will leave it to others to figure out how to grab the avatar, which is on the $url page in the open graph meta links
}
if($appnet || ! $poll)
2011-08-15 01:13:52 +00:00
$check_feed = true;
if((! isset($vcard)) || (! x($vcard,'fn')) || (! $profile))
2011-08-15 01:13:52 +00:00
$check_feed = true;
if(($at_addr) && (! count($links)))
$check_feed = false;
if ($connectornetworks)
$check_feed = false;
2011-08-15 01:13:52 +00:00
if($check_feed) {
2011-08-15 12:27:24 +00:00
$feedret = scrape_feed(($poll) ? $poll : $url);
2015-07-22 09:51:37 +00:00
2012-02-26 22:34:01 +00:00
logger('probe_url: scrape_feed ' . (($poll)? $poll : $url) . ' returns: ' . print_r($feedret,true), LOGGER_DATA);
if(count($feedret) && ($feedret['feed_atom'] || $feedret['feed_rss'])) {
$poll = ((x($feedret,'feed_atom')) ? unamp($feedret['feed_atom']) : unamp($feedret['feed_rss']));
if(! x($vcard))
2011-08-18 04:14:21 +00:00
$vcard = array();
}
2011-08-18 04:14:21 +00:00
if(x($feedret,'photo') && (! x($vcard,'photo')))
$vcard['photo'] = $feedret['photo'];
require_once('library/simplepie/simplepie.inc');
$feed = new SimplePie();
$xml = fetch_url($poll);
logger('probe_url: fetch feed: ' . $poll . ' returns: ' . $xml, LOGGER_DATA);
$a = get_app();
2012-04-17 11:33:50 +00:00
logger('probe_url: scrape_feed: headers: ' . $a->get_curl_headers(), LOGGER_DATA);
// Don't try and parse an empty string
2015-07-22 09:51:37 +00:00
$feed->set_raw_data(($xml) ? $xml : '<?xml version="1.0" encoding="utf-8" ?><xml></xml>');
$feed->init();
if($feed->error())
logger('probe_url: scrape_feed: Error parsing XML: ' . $feed->error());
2011-08-18 04:14:21 +00:00
if(! x($vcard,'photo'))
$vcard['photo'] = $feed->get_image_url();
$author = $feed->get_author();
2011-08-18 04:14:21 +00:00
if($author) {
$vcard['fn'] = unxmlify(trim($author->get_name()));
if(! $vcard['fn'])
$vcard['fn'] = trim(unxmlify($author->get_email()));
if(strpos($vcard['fn'],'@') !== false)
$vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
$email = unxmlify($author->get_email());
2011-08-15 01:13:52 +00:00
if(! $profile && $author->get_link())
$profile = trim(unxmlify($author->get_link()));
if(! $vcard['photo']) {
$rawtags = $feed->get_feed_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
if($rawtags) {
$elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
$vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
}
}
// Fetch fullname via poco:displayName
$pocotags = $feed->get_feed_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
if ($pocotags) {
$elems = $pocotags[0]['child']['http://portablecontacts.net/spec/1.0'];
if (isset($elems["displayName"]))
$vcard['fn'] = $elems["displayName"][0]["data"];
if (isset($elems["preferredUsername"]))
$vcard['nick'] = $elems["preferredUsername"][0]["data"];
}
}
else {
$item = $feed->get_item(0);
if($item) {
$author = $item->get_author();
if($author) {
$vcard['fn'] = trim(unxmlify($author->get_name()));
if(! $vcard['fn'])
$vcard['fn'] = trim(unxmlify($author->get_email()));
if(strpos($vcard['fn'],'@') !== false)
$vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
$email = unxmlify($author->get_email());
2011-08-15 01:13:52 +00:00
if(! $profile && $author->get_link())
$profile = trim(unxmlify($author->get_link()));
}
if(! $vcard['photo']) {
$rawmedia = $item->get_item_tags('http://search.yahoo.com/mrss/','thumbnail');
if($rawmedia && $rawmedia[0]['attribs']['']['url'])
$vcard['photo'] = unxmlify($rawmedia[0]['attribs']['']['url']);
}
if(! $vcard['photo']) {
2011-07-06 03:32:43 +00:00
$rawtags = $item->get_item_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
if($rawtags) {
$elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
$vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
}
}
}
}
2011-08-18 04:14:21 +00:00
if((! $vcard['photo']) && strlen($email))
$vcard['photo'] = avatar_img($email);
if($poll === $profile)
$lnk = $feed->get_permalink();
if(isset($lnk) && strlen($lnk))
$profile = $lnk;
if(! $network) {
$network = NETWORK_FEED;
// If it is a feed, don't take the author name as feed name
unset($vcard['fn']);
}
if(! (x($vcard,'fn')))
$vcard['fn'] = notags($feed->get_title());
if(! (x($vcard,'fn')))
$vcard['fn'] = notags($feed->get_description());
if(strpos($vcard['fn'],'Twitter / ') !== false) {
$vcard['fn'] = substr($vcard['fn'],strpos($vcard['fn'],'/')+1);
$vcard['fn'] = trim($vcard['fn']);
}
if(! x($vcard,'nick')) {
$vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
if(strpos($vcard['nick'],' '))
$vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
}
2011-08-15 01:13:52 +00:00
if(! $priority)
$priority = 2;
}
}
if(! x($vcard,'photo')) {
$a = get_app();
$vcard['photo'] = $a->get_baseurl() . '/images/person-175.jpg' ;
}
2011-08-15 01:13:52 +00:00
if(! $profile)
$profile = $url;
// No human could be associated with this link, use the URL as the contact name
if(($network === NETWORK_FEED) && ($poll) && (! x($vcard,'fn')))
$vcard['fn'] = $url;
2011-04-16 15:45:08 +00:00
if (($notify != "") AND ($poll != "")) {
$baseurl = matching(normalise_link($notify), normalise_link($poll));
$baseurl2 = matching($baseurl, normalise_link($profile));
if ($baseurl2 != "")
$baseurl = $baseurl2;
}
if (($baseurl == "") AND ($notify != ""))
$baseurl = matching(normalise_link($profile), normalise_link($notify));
if (($baseurl == "") AND ($poll != ""))
$baseurl = matching(normalise_link($profile), normalise_link($poll));
$baseurl = rtrim($baseurl, "/");
$vcard['fn'] = notags($vcard['fn']);
$vcard['nick'] = str_replace(' ','',notags($vcard['nick']));
$result['name'] = $vcard['fn'];
$result['nick'] = $vcard['nick'];
$result['url'] = $profile;
2011-04-16 15:45:08 +00:00
$result['addr'] = $addr;
$result['batch'] = $batch;
$result['notify'] = $notify;
$result['poll'] = $poll;
$result['request'] = $request;
$result['confirm'] = $confirm;
$result['poco'] = $poco;
$result['photo'] = $vcard['photo'];
$result['priority'] = $priority;
$result['network'] = $network;
$result['alias'] = $alias;
2011-07-15 10:08:43 +00:00
$result['pubkey'] = $pubkey;
$result['baseurl'] = $baseurl;
logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG);
if ($level == 1) {
// Trying if it maybe a diaspora account
if (($result['network'] == NETWORK_FEED) OR ($result['addr'] == "")) {
require_once('include/bbcode.php');
$address = GetProfileUsername($url, "", true);
$result2 = probe_url($address, $mode, ++$level);
if ($result2['network'] != "")
$result = $result2;
}
// Maybe it's some non standard GNU Social installation (Single user, subfolder or no uri rewrite)
if (($result['network'] == NETWORK_FEED) AND ($result['baseurl'] != "") AND ($result['nick'] != "")) {
$addr = $result['nick'].'@'.str_replace("http://", "", $result['baseurl']);
$result2 = probe_url($addr, $mode, ++$level);
if (($result2['network'] != "") AND ($result2['network'] != NETWORK_FEED))
$result = $result2;
}
}
Cache::set("probe_url:".$mode.":".$url,serialize($result));
return $result;
}
function matching($part1, $part2) {
$len = min(strlen($part1), strlen($part2));
$match = "";
$matching = true;
$i = 0;
while (($i <= $len) AND $matching) {
if (substr($part1, $i, 1) == substr($part2, $i, 1))
$match .= substr($part1, $i, 1);
else
$matching = false;
$i++;
}
return($match);
}