Poco: New fields for last contact and last failure of a contact.

boot.php

@@ -19,7 +19,7 @@ define ( 'FRIENDICA_PLATFORM',     'Friendica');
 define ( 'FRIENDICA_CODENAME',     'Lily of the valley');
 define ( 'FRIENDICA_VERSION',      '3.4.1' );
 define ( 'DFRN_PROTOCOL_VERSION',  '2.23'    );
-define ( 'DB_UPDATE_VERSION',      1185      );
+define ( 'DB_UPDATE_VERSION',      1186      );
 define ( 'EOL',                    "<br />\r\n"     );
 define ( 'ATOM_TIME',              'Y-m-d\TH:i:s\Z' );
 

include/dbstructure.php

@@ -630,6 +630,8 @@ function db_definition() {
 					"photo" => array("type" => "varchar(255)", "not null" => "1", "default" => ""),
 					"connect" => array("type" => "varchar(255)", "not null" => "1", "default" => ""),
 					"updated" => array("type" => "datetime", "default" => "0000-00-00 00:00:00"),
+					"last_contact" => array("type" => "datetime", "default" => "0000-00-00 00:00:00"),
+					"last_failure" => array("type" => "datetime", "default" => "0000-00-00 00:00:00"),
 					"location" => array("type" => "varchar(255)", "not null" => "1", "default" => ""),
 					"about" => array("type" => "text", "not null" => "1"),
 					"keywords" => array("type" => "text", "not null" => "1"),

include/plaintext.php

@@ -52,7 +52,7 @@ function get_attached_data($body) {
 		if (preg_match_all("(\[url=([$URLSearchString]*)\]\s*\[img\]([$URLSearchString]*)\[\/img\]\s*\[\/url\])ism", $body, $pictures,  PREG_SET_ORDER)) {
 			if (count($pictures) == 1) {
 				// Checking, if the link goes to a picture
-				$data = parseurl_getsiteinfo($pictures[0][1], true);
+				$data = parseurl_getsiteinfo_cached($pictures[0][1], true);
 				if ($data["type"] == "photo") {
 					$post["type"] = "photo";
 					if (isset($data["images"][0]))
@@ -95,7 +95,7 @@ function get_attached_data($body) {
 		}
 	} elseif (isset($post["url"]) AND ($post["type"] == "video")) {
 		require_once("mod/parse_url.php");
-		$data = parseurl_getsiteinfo($post["url"], true);
+		$data = parseurl_getsiteinfo_cached($post["url"], true);
 
 		if (isset($data["images"][0]))
 			$post["image"] = $data["images"][0]["src"];

include/socgraph.php

@@ -195,8 +195,20 @@ function poco_check($profile_url, $name, $network, $profile_photo, $about, $loca
 	$x = q("SELECT * FROM `gcontact` WHERE `nurl` = '%s' LIMIT 1",
 		dbesc(normalise_link($profile_url))
 	);
-	if(count($x) AND ($network == "") AND ($x[0]["network"] != NETWORK_STATUSNET))
+
-		$network = $x[0]["network"];
+	if (count($x)) {
+		if (($network == "") AND ($x[0]["network"] != NETWORK_STATUSNET))
+			$network = $x[0]["network"];
+
+		if ($updated = "0000-00-00 00:00:00")
+			$updated = $x[0]["updated"];
+
+		$last_contact = $x[0]["last_contact"];
+		$last_failure = $x[0]["last_failure"];
+	} else {
+		$last_contact = "0000-00-00 00:00:00";
+		$last_failure = "0000-00-00 00:00:00";
+	}
 
 	if (($network == "") OR ($name == "") OR ($profile_photo == "")) {
 		require_once("include/Scrape.php");
@@ -223,6 +235,23 @@ function poco_check($profile_url, $name, $network, $profile_photo, $about, $loca
 
 	logger("profile-check generation: ".$generation." Network: ".$network." URL: ".$profile_url." name: ".$name." avatar: ".$profile_photo, LOGGER_DEBUG);
 
+	if (poco_do_update($updated, $last_contact, $last_failure)) {
+		$last_updated = poco_last_updated($profile_url);
+		if ($last_updated) {
+			$updated = $last_updated;
+			$last_contact = datetime_convert();
+			logger("Last updated for profile ".$profile_url.": ".$updated, LOGGER_DEBUG);
+
+			if (count($x))
+				q("UPDATE `gcontact` SET `last_contact` = '%s' WHERE `nurl` = '%s'", dbesc($last_contact), dbesc(normalise_link($profile_url)));
+		} else {
+			$last_failure = datetime_convert();
+
+			if (count($x))
+				q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'", dbesc($last_failure), dbesc(normalise_link($profile_url)));
+		}
+	}
+
 	if(count($x)) {
 		$gcid = $x[0]['id'];
 
@@ -241,6 +270,9 @@ function poco_check($profile_url, $name, $network, $profile_photo, $about, $loca
 		if (($generation == 0) AND ($x[0]['generation'] > 0))
 			$generation = $x[0]['generation'];
 
+		if ($last_contact < $x[0]['last_contact'])
+			$last_contact = $x[0]['last_contact'];
+
 		if($x[0]['name'] != $name || $x[0]['photo'] != $profile_photo || $x[0]['updated'] < $updated) {
 			q("UPDATE `gcontact` SET `name` = '%s', `network` = '%s', `photo` = '%s', `connect` = '%s', `url` = '%s',
 				`updated` = '%s', `location` = '%s', `about` = '%s', `keywords` = '%s', `gender` = '%s', `generation` = %d
@@ -261,7 +293,7 @@ function poco_check($profile_url, $name, $network, $profile_photo, $about, $loca
 			);
 		}
 	} else {
-		q("INSERT INTO `gcontact` (`name`,`network`, `url`,`nurl`,`photo`,`connect`, `updated`, `location`, `about`, `keywords`, `gender`, `generation`)
+		q("INSERT INTO `gcontact` (`name`,`network`, `url`,`nurl`,`photo`,`connect`, `updated`, `last_contact`, `last_failure`, `location`, `about`, `keywords`, `gender`, `generation`)
 			VALUES ('%s', '%s', '%s', '%s', '%s','%s', '%s', '%s', '%s', '%s', '%s', %d)",
 			dbesc($name),
 			dbesc($network),
@@ -270,6 +302,8 @@ function poco_check($profile_url, $name, $network, $profile_photo, $about, $loca
 			dbesc($profile_photo),
 			dbesc($connect_url),
 			dbesc($updated),
+			dbesc($last_contact),
+			dbesc($last_failure),
 			dbesc($location),
 			dbesc($about),
 			dbesc($keywords),
@@ -320,6 +354,79 @@ function poco_check($profile_url, $name, $network, $profile_photo, $about, $loca
 	return $gcid;
 }
 
+function poco_last_updated($profile) {
+	$data = probe_url($profile);
+
+	if (($data["poll"] == "") OR ($data["network"] == NETWORK_FEED))
+		return false;
+
+	$feedret = z_fetch_url($data["poll"]);
+
+	if (!$feedret["success"])
+		return false;
+
+	$doc = new DOMDocument();
+	@$doc->loadXML($feedret["body"]);
+
+	$xpath = new DomXPath($doc);
+	$xpath->registerNamespace('atom', "http://www.w3.org/2005/Atom");
+
+	$entries = $xpath->query('/atom:feed/atom:entry');
+
+	$last_updated = "";
+
+	foreach ($entries AS $entry) {
+		$published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
+		$updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
+
+		if ($last_updated < $published)
+			$last_updated = $published;
+
+		if ($last_updated < $updated)
+			$last_updated = $updated;
+	}
+
+	// Maybe there aren't any entries. Then check if it is a valid feed
+	if ($last_updated == "")
+		if ($xpath->query('/atom:feed')->length > 0)
+			$last_updated = "0000-00-00 00:00:00";
+
+	return($last_updated);
+}
+
+function poco_do_update($updated, $last_contact, $last_failure) {
+	$now = strtotime(datetime_convert());
+
+	if ($updated > $last_contact)
+		$contact_time = strtotime($updated);
+	else
+		$contact_time = strtotime($last_contact);
+
+	$failure_time = strtotime($last_failure);
+
+	// If the last contact was less than 24 hours then don't update
+	if (($now - $contact_time) < (60 * 60 * 24))
+		return false;
+
+	// If the last failure was less than 24 hours then don't update
+	if (($now - $failure_time) < (60 * 60 * 24))
+		return false;
+
+	// If the last contact was less than a week ago and the last failure is older than a week then don't update
+	if ((($now - $contact_time) < (60 * 60 * 24 * 7)) AND ($contact_time > $failure_time))
+		return false;
+
+	// If the last contact time was more than a week ago, then only try once a week
+	if (($now - $contact_time) > (60 * 60 * 24 * 7) AND ($now - $failure_time) < (60 * 60 * 24 * 7))
+		return false;
+
+	// If the last contact time was more than a month ago, then only try once a month
+	if (($now - $contact_time) > (60 * 60 * 24 * 30) AND ($now - $failure_time) < (60 * 60 * 24 * 30))
+		return false;
+
+	return true;
+}
+
 function poco_contact_from_body($body, $created, $cid, $uid) {
 	preg_replace_callback("/\[share(.*?)\].*?\[\/share\]/ism",
 		function ($match) use ($created, $cid, $uid){
@@ -328,20 +435,20 @@ function poco_contact_from_body($body, $created, $cid, $uid) {
 }
 
 function sub_poco_from_share($share, $created, $cid, $uid) {
-        $profile = "";
+	$profile = "";
-        preg_match("/profile='(.*?)'/ism", $share[1], $matches);
+	preg_match("/profile='(.*?)'/ism", $share[1], $matches);
-        if ($matches[1] != "")
+	if ($matches[1] != "")
-                $profile = $matches[1];
+		$profile = $matches[1];
 
-        preg_match('/profile="(.*?)"/ism', $share[1], $matches);
+	preg_match('/profile="(.*?)"/ism', $share[1], $matches);
-        if ($matches[1] != "")
+	if ($matches[1] != "")
-                $profile = $matches[1];
+		$profile = $matches[1];
 
 	if ($profile == "")
 		return;
 
 	logger("prepare poco_check for profile ".$profile, LOGGER_DEBUG);
-        poco_check($profile, "", "", "", "", "", "", "", "", $created, 3, $cid, $uid);
+	poco_check($profile, "", "", "", "", "", "", "", "", $created, 3, $cid, $uid);
 }
 
 function poco_store($item) {

mod/parse_url.php

@@ -50,6 +50,21 @@ function completeurl($url, $scheme) {
 	return($complete);
 }
 
+function parseurl_getsiteinfo_cached($url, $no_guessing = false, $do_oembed = true) {
+
+	$data = Cache::get("parse_url:".$no_guessing.":".$do_oembed.":".$url);
+	if (!is_null($data)) {
+		$data = unserialize($data);
+		return $data;
+	}
+
+	$data = parseurl_getsiteinfo($url, $no_guessing, $do_oembed);
+
+	Cache::set("parse_url:".$no_guessing.":".$do_oembed.":".$url,serialize($data));
+
+	return $data;
+}
+
 function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1) {
 	require_once("include/network.php");
 

update.php

@@ -1,6 +1,6 @@
 <?php
 
-define( 'UPDATE_VERSION' , 1185 );
+define( 'UPDATE_VERSION' , 1186 );
 
 /**
  *