2011-12-13 12:06:25 +00:00
|
|
|
<?php
|
2017-05-05 15:10:07 +00:00
|
|
|
class RSSUtils {
|
|
|
|
static function calculate_article_hash($article, $pluginhost) {
|
2014-08-18 17:55:48 +00:00
|
|
|
$tmp = "";
|
|
|
|
|
|
|
|
foreach ($article as $k => $v) {
|
|
|
|
if ($k != "feed" && isset($v)) {
|
2015-12-05 00:10:57 +00:00
|
|
|
$x = strip_tags(is_array($v) ? implode(",", $v) : $v);
|
|
|
|
|
|
|
|
$tmp .= sha1("$k:" . sha1($x));
|
2014-08-18 17:55:48 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-08-18 19:11:29 +00:00
|
|
|
return sha1(implode(",", $pluginhost->get_plugin_names()) . $tmp);
|
2014-08-18 14:32:16 +00:00
|
|
|
}
|
|
|
|
|
2018-01-29 20:24:11 +00:00
|
|
|
// Strips utf8mb4 characters (i.e. emoji) for mysql
|
|
|
|
static function strip_utf8mb4($str) {
|
2018-01-30 07:44:31 +00:00
|
|
|
return preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $str);
|
|
|
|
}
|
2018-01-29 20:24:11 +00:00
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
static function update_feedbrowser_cache() {
|
2011-12-13 12:23:31 +00:00
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
$pdo = Db::pdo();
|
|
|
|
|
|
|
|
$sth = $pdo->query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
|
2017-02-16 11:05:32 +00:00
|
|
|
FROM ttrss_feeds WHERE feed_url NOT IN (SELECT feed_url FROM ttrss_feeds
|
|
|
|
WHERE private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%')
|
2011-12-13 12:23:31 +00:00
|
|
|
GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
|
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
$pdo->beginTransaction();
|
2011-12-13 12:23:31 +00:00
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
$pdo->query("DELETE FROM ttrss_feedbrowser_cache");
|
2011-12-13 12:23:31 +00:00
|
|
|
|
|
|
|
$count = 0;
|
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
while ($line = $sth->fetch()) {
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$subscribers = $line["subscribers"];
|
|
|
|
$feed_url = $line["feed_url"];
|
|
|
|
$title = $line["title"];
|
|
|
|
$site_url = $line["site_url"];
|
2011-12-13 12:23:31 +00:00
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
$tmph = $pdo->prepare("SELECT subscribers FROM
|
|
|
|
ttrss_feedbrowser_cache WHERE feed_url = ?");
|
|
|
|
$tmph->execute([$feed_url]);
|
|
|
|
|
|
|
|
if (!$tmph->fetch()) {
|
2011-12-13 12:23:31 +00:00
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
$tmph = $pdo->prepare("INSERT INTO ttrss_feedbrowser_cache
|
2018-11-03 12:00:58 +00:00
|
|
|
(feed_url, site_url, title, subscribers)
|
|
|
|
VALUES
|
2017-12-02 05:38:57 +00:00
|
|
|
(?, ?, ?, ?)");
|
2011-12-13 12:23:31 +00:00
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
$tmph->execute([$feed_url, $site_url, $title, $subscribers]);
|
2011-12-13 12:23:31 +00:00
|
|
|
|
|
|
|
++$count;
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
$pdo->commit();
|
2011-12-13 12:23:31 +00:00
|
|
|
|
|
|
|
return $count;
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
static function update_daemon_common($limit = DAEMON_FEED_LIMIT) {
|
2013-04-17 12:23:15 +00:00
|
|
|
$schema_version = get_schema_version();
|
2013-04-17 08:10:35 +00:00
|
|
|
|
|
|
|
if ($schema_version != SCHEMA_VERSION) {
|
|
|
|
die("Schema version is wrong, please upgrade the database.\n");
|
|
|
|
}
|
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
$pdo = Db::pdo();
|
|
|
|
|
2011-12-13 15:20:26 +00:00
|
|
|
if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
|
2011-12-13 12:06:25 +00:00
|
|
|
if (DB_TYPE == "pgsql") {
|
|
|
|
$login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
|
|
|
|
} else {
|
|
|
|
$login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
$login_thresh_qpart = "";
|
|
|
|
}
|
|
|
|
|
|
|
|
if (DB_TYPE == "pgsql") {
|
|
|
|
$update_limit_qpart = "AND ((
|
|
|
|
ttrss_feeds.update_interval = 0
|
2013-03-31 10:36:06 +00:00
|
|
|
AND ttrss_user_prefs.value != '-1'
|
2011-12-13 12:06:25 +00:00
|
|
|
AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
|
|
|
|
) OR (
|
|
|
|
ttrss_feeds.update_interval > 0
|
|
|
|
AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
|
2013-12-18 09:55:50 +00:00
|
|
|
) OR (ttrss_feeds.last_updated IS NULL
|
|
|
|
AND ttrss_user_prefs.value != '-1')
|
|
|
|
OR (last_updated = '1970-01-01 00:00:00'
|
|
|
|
AND ttrss_user_prefs.value != '-1'))";
|
2011-12-13 12:06:25 +00:00
|
|
|
} else {
|
|
|
|
$update_limit_qpart = "AND ((
|
|
|
|
ttrss_feeds.update_interval = 0
|
2013-03-31 10:36:06 +00:00
|
|
|
AND ttrss_user_prefs.value != '-1'
|
2011-12-13 12:06:25 +00:00
|
|
|
AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
|
|
|
|
) OR (
|
|
|
|
ttrss_feeds.update_interval > 0
|
|
|
|
AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
|
2013-12-18 09:55:50 +00:00
|
|
|
) OR (ttrss_feeds.last_updated IS NULL
|
|
|
|
AND ttrss_user_prefs.value != '-1')
|
|
|
|
OR (last_updated = '1970-01-01 00:00:00'
|
|
|
|
AND ttrss_user_prefs.value != '-1'))";
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
// Test if feed is currently being updated by another process.
|
|
|
|
if (DB_TYPE == "pgsql") {
|
2013-05-01 22:31:32 +00:00
|
|
|
$updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
|
2011-12-13 12:06:25 +00:00
|
|
|
} else {
|
2013-05-01 22:31:32 +00:00
|
|
|
$updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2017-06-13 17:09:33 +00:00
|
|
|
$query_limit = $limit ? sprintf("LIMIT %d", $limit) : "";
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2015-12-31 11:54:52 +00:00
|
|
|
// Update the least recently updated feeds first
|
|
|
|
$query_order = "ORDER BY last_updated";
|
|
|
|
if (DB_TYPE == "pgsql") $query_order .= " NULLS FIRST";
|
|
|
|
|
2013-04-24 09:43:40 +00:00
|
|
|
$query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
|
2011-12-13 12:06:25 +00:00
|
|
|
FROM
|
|
|
|
ttrss_feeds, ttrss_users, ttrss_user_prefs
|
2013-05-02 06:26:32 +00:00
|
|
|
WHERE
|
2011-12-13 12:06:25 +00:00
|
|
|
ttrss_feeds.owner_uid = ttrss_users.id
|
2013-12-18 09:55:50 +00:00
|
|
|
AND ttrss_user_prefs.profile IS NULL
|
2011-12-13 12:06:25 +00:00
|
|
|
AND ttrss_users.id = ttrss_user_prefs.owner_uid
|
|
|
|
AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
|
|
|
|
$login_thresh_qpart $update_limit_qpart
|
2013-03-29 08:21:40 +00:00
|
|
|
$updstart_thresh_qpart
|
2015-12-31 11:54:52 +00:00
|
|
|
$query_order $query_limit";
|
2013-04-24 09:43:40 +00:00
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
$res = $pdo->query($query);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
|
|
|
$feeds_to_update = array();
|
2017-12-02 05:38:57 +00:00
|
|
|
while ($line = $res->fetch()) {
|
2017-06-13 17:09:33 +00:00
|
|
|
array_push($feeds_to_update, $line['feed_url']);
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log(sprintf("Scheduled %d feeds to update...", count($feeds_to_update)));
|
2017-12-02 05:38:57 +00:00
|
|
|
|
2017-06-13 17:09:33 +00:00
|
|
|
// Update last_update_started before actually starting the batch
|
|
|
|
// in order to minimize collision risk for parallel daemon tasks
|
|
|
|
if (count($feeds_to_update) > 0) {
|
2017-12-02 05:38:57 +00:00
|
|
|
$feeds_qmarks = arr_qmarks($feeds_to_update);
|
2013-03-29 08:21:40 +00:00
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
$tmph = $pdo->prepare("UPDATE ttrss_feeds SET last_update_started = NOW()
|
|
|
|
WHERE feed_url IN ($feeds_qmarks)");
|
|
|
|
$tmph->execute($feeds_to_update);
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2013-03-30 12:23:34 +00:00
|
|
|
$nf = 0;
|
2014-08-19 06:32:01 +00:00
|
|
|
$bstarted = microtime(true);
|
2013-03-30 12:23:34 +00:00
|
|
|
|
2016-08-11 12:01:01 +00:00
|
|
|
$batch_owners = array();
|
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
// since we have the data cached, we can deal with other feeds with the same url
|
|
|
|
$usth = $pdo->prepare("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
|
2013-03-31 10:36:06 +00:00
|
|
|
FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
|
|
|
|
ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
|
|
|
|
ttrss_users.id = ttrss_user_prefs.owner_uid AND
|
|
|
|
ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
|
2013-12-18 09:55:50 +00:00
|
|
|
ttrss_user_prefs.profile IS NULL AND
|
2017-12-02 05:38:57 +00:00
|
|
|
feed_url = ?
|
2016-04-01 15:54:53 +00:00
|
|
|
$update_limit_qpart
|
2013-03-29 08:21:40 +00:00
|
|
|
$login_thresh_qpart
|
2013-04-04 09:20:44 +00:00
|
|
|
ORDER BY ttrss_feeds.id $query_limit");
|
2013-03-29 08:21:40 +00:00
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
foreach ($feeds_to_update as $feed) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("Base feed: $feed");
|
2017-12-02 05:38:57 +00:00
|
|
|
|
|
|
|
$usth->execute([$feed]);
|
|
|
|
//update_rss_feed($line["id"], true);
|
|
|
|
|
|
|
|
if ($tline = $usth->fetch()) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
|
2013-12-18 09:55:50 +00:00
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
if (array_search($tline["owner_uid"], $batch_owners) === FALSE)
|
|
|
|
array_push($batch_owners, $tline["owner_uid"]);
|
2016-08-11 12:01:01 +00:00
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
$fstarted = microtime(true);
|
2018-09-03 10:03:35 +00:00
|
|
|
|
|
|
|
try {
|
|
|
|
RSSUtils::update_rss_feed($tline["id"], true, false);
|
|
|
|
} catch (PDOException $e) {
|
|
|
|
Logger::get()->log_error(E_USER_NOTICE, $e->getMessage(), $e->getFile(), $e->getLine(), $e->getTraceAsString());
|
2018-11-03 12:00:58 +00:00
|
|
|
|
|
|
|
try {
|
|
|
|
$pdo->rollback();
|
|
|
|
} catch (PDOException $e) {
|
|
|
|
// it doesn't matter if there wasn't actually anything to rollback, PDO Exception can be
|
|
|
|
// thrown outside of an active transaction during feed update
|
|
|
|
}
|
2018-09-03 10:03:35 +00:00
|
|
|
}
|
2014-08-19 06:32:01 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
|
2014-08-19 06:32:01 +00:00
|
|
|
|
2017-12-02 05:38:57 +00:00
|
|
|
++$nf;
|
2013-03-29 08:21:40 +00:00
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2014-08-19 06:32:01 +00:00
|
|
|
if ($nf > 0) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
|
2014-08-19 06:32:01 +00:00
|
|
|
microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
|
|
|
|
}
|
|
|
|
|
2016-08-11 12:01:01 +00:00
|
|
|
foreach ($batch_owners as $owner_uid) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("Running housekeeping tasks for user $owner_uid...");
|
2016-08-11 12:01:01 +00:00
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
RSSUtils::housekeeping_user($owner_uid);
|
2016-08-11 12:01:01 +00:00
|
|
|
}
|
|
|
|
|
2011-12-13 12:06:25 +00:00
|
|
|
// Send feed digests by email if needed.
|
2018-11-30 05:34:29 +00:00
|
|
|
Digest::send_headlines_digests();
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2013-03-30 12:23:34 +00:00
|
|
|
return $nf;
|
2017-04-26 12:29:22 +00:00
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2015-06-17 20:11:47 +00:00
|
|
|
// this is used when subscribing
|
2017-05-05 15:10:07 +00:00
|
|
|
static function set_basic_feed_info($feed) {
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$pdo = Db::pdo();
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2018-08-13 12:59:24 +00:00
|
|
|
$sth = $pdo->prepare("SELECT owner_uid,feed_url,auth_pass,auth_login
|
2017-12-02 06:55:53 +00:00
|
|
|
FROM ttrss_feeds WHERE id = ?");
|
|
|
|
$sth->execute([$feed]);
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if ($row = $sth->fetch()) {
|
2017-09-25 00:37:49 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$owner_uid = $row["owner_uid"];
|
|
|
|
$auth_login = $row["auth_login"];
|
|
|
|
$auth_pass = $row["auth_pass"];
|
|
|
|
$fetch_url = $row["feed_url"];
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$pluginhost = new PluginHost();
|
|
|
|
$user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
|
|
|
|
$pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
|
|
|
|
$pluginhost->load_data();
|
|
|
|
|
|
|
|
$basic_info = array();
|
|
|
|
foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_BASIC_INFO) as $plugin) {
|
|
|
|
$basic_info = $plugin->hook_feed_basic_info($basic_info, $fetch_url, $owner_uid, $feed, $auth_login, $auth_pass);
|
|
|
|
}
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if (!$basic_info) {
|
|
|
|
$feed_data = fetch_file_contents($fetch_url, false,
|
|
|
|
$auth_login, $auth_pass, false,
|
|
|
|
FEED_FETCH_TIMEOUT,
|
|
|
|
0);
|
2017-09-25 00:37:49 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
global $fetch_curl_used;
|
2017-09-25 00:37:49 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if (!$fetch_curl_used) {
|
|
|
|
$tmp = @gzdecode($feed_data);
|
2017-09-25 00:37:49 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if ($tmp) $feed_data = $tmp;
|
|
|
|
}
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$feed_data = trim($feed_data);
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$rss = new FeedParser($feed_data);
|
|
|
|
$rss->init();
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if (!$rss->error()) {
|
|
|
|
$basic_info = array(
|
|
|
|
'title' => mb_substr($rss->get_title(), 0, 199),
|
|
|
|
'site_url' => mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245)
|
|
|
|
);
|
|
|
|
}
|
2017-09-27 01:37:05 +00:00
|
|
|
}
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if ($basic_info && is_array($basic_info)) {
|
|
|
|
$sth = $pdo->prepare("SELECT title, site_url FROM ttrss_feeds WHERE id = ?");
|
|
|
|
$sth->execute([$feed]);
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if ($row = $sth->fetch()) {
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$registered_title = $row["title"];
|
|
|
|
$orig_site_url = $row["site_url"];
|
|
|
|
|
|
|
|
if ($basic_info['title'] && (!$registered_title || $registered_title == "[Unknown]")) {
|
|
|
|
|
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_feeds SET
|
|
|
|
title = ? WHERE id = ?");
|
|
|
|
$sth->execute([$basic_info['title'], $feed]);
|
|
|
|
}
|
2015-06-17 20:11:47 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if ($basic_info['site_url'] && $orig_site_url != $basic_info['site_url']) {
|
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_feeds SET
|
|
|
|
site_url = ? WHERE id = ?");
|
|
|
|
$sth->execute([$basic_info['site_url'], $feed]);
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
2015-06-17 20:11:47 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
/**
|
2017-05-05 15:10:07 +00:00
|
|
|
* @SuppressWarnings(PHPMD.UnusedFormalParameter)
|
|
|
|
*/
|
|
|
|
static function update_rss_feed($feed, $no_cache = false) {
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("start", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$pdo = Db::pdo();
|
|
|
|
|
|
|
|
$sth = $pdo->prepare("SELECT title FROM ttrss_feeds WHERE id = ?");
|
|
|
|
$sth->execute([$feed]);
|
2016-11-30 11:03:21 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if (!$row = $sth->fetch()) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("feed $feed not found, skipping.");
|
2016-11-30 11:03:21 +00:00
|
|
|
user_error("Attempt to update unknown/invalid feed $feed", E_USER_WARNING);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$title = $row["title"];
|
2015-07-04 16:33:46 +00:00
|
|
|
|
|
|
|
// feed was batch-subscribed or something, we need to get basic info
|
|
|
|
// this is not optimal currently as it fetches stuff separately TODO: optimize
|
|
|
|
if ($title == "[Unknown]") {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("setting basic feed info for $feed...");
|
2017-05-05 15:10:07 +00:00
|
|
|
RSSUtils::set_basic_feed_info($feed);
|
2015-07-04 16:33:46 +00:00
|
|
|
}
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("SELECT id,update_interval,auth_login,
|
2014-02-19 11:17:29 +00:00
|
|
|
feed_url,auth_pass,cache_images,
|
2013-02-26 15:20:58 +00:00
|
|
|
mark_unread_on_update, owner_uid,
|
2018-11-03 12:00:58 +00:00
|
|
|
auth_pass_encrypted, feed_language,
|
|
|
|
last_modified,
|
|
|
|
".SUBSTRING_FOR_DATE."(last_unconditional, 1, 19) AS last_unconditional
|
2017-12-02 06:55:53 +00:00
|
|
|
FROM ttrss_feeds WHERE id = ?");
|
|
|
|
$sth->execute([$feed]);
|
2018-11-03 12:00:58 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if ($row = $sth->fetch()) {
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$owner_uid = $row["owner_uid"];
|
2017-12-03 06:35:59 +00:00
|
|
|
$mark_unread_on_update = $row["mark_unread_on_update"];
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_feeds SET last_update_started = NOW()
|
|
|
|
WHERE id = ?");
|
|
|
|
$sth->execute([$feed]);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$auth_login = $row["auth_login"];
|
|
|
|
$auth_pass = $row["auth_pass"];
|
|
|
|
$stored_last_modified = $row["last_modified"];
|
|
|
|
$last_unconditional = $row["last_unconditional"];
|
2017-12-03 06:35:59 +00:00
|
|
|
$cache_images = $row["cache_images"];
|
2017-12-02 06:55:53 +00:00
|
|
|
$fetch_url = $row["feed_url"];
|
2018-08-13 12:59:24 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$feed_language = mb_strtolower($row["feed_language"]);
|
|
|
|
if (!$feed_language) $feed_language = 'english';
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
} else {
|
|
|
|
return false;
|
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2013-03-29 07:31:10 +00:00
|
|
|
$date_feed_processed = date('Y-m-d H:i');
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-03 18:01:35 +00:00
|
|
|
$cache_filename = CACHE_DIR . "/feeds/" . sha1($fetch_url) . ".xml";
|
2013-03-29 07:31:10 +00:00
|
|
|
|
2013-05-20 11:26:53 +00:00
|
|
|
$pluginhost = new PluginHost();
|
|
|
|
$user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
|
|
|
|
|
|
|
|
$pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
|
|
|
|
$pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
|
|
|
|
$pluginhost->load_data();
|
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
$rss_hash = false;
|
2013-03-29 07:50:58 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
$force_refetch = isset($_REQUEST["force_refetch"]);
|
|
|
|
$feed_data = "";
|
2013-04-15 11:08:00 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
|
|
|
|
$feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
|
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
// try cache
|
|
|
|
if (!$feed_data &&
|
|
|
|
file_exists($cache_filename) &&
|
|
|
|
is_readable($cache_filename) &&
|
|
|
|
!$auth_login && !$auth_pass &&
|
|
|
|
filemtime($cache_filename) > time() - 30) {
|
2013-03-27 09:34:04 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("using local cache [$cache_filename].", Debug::$LOG_VERBOSE);
|
2013-05-01 21:36:17 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
@$feed_data = file_get_contents($cache_filename);
|
2013-03-29 07:31:10 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
if ($feed_data) {
|
|
|
|
$rss_hash = sha1($feed_data);
|
2013-10-25 10:42:43 +00:00
|
|
|
}
|
2013-05-20 11:26:53 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
} else {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("local cache will not be used for this feed", Debug::$LOG_VERBOSE);
|
2017-04-26 12:29:22 +00:00
|
|
|
}
|
2016-01-13 15:12:31 +00:00
|
|
|
|
2017-08-17 11:40:21 +00:00
|
|
|
global $fetch_last_modified;
|
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
// fetch feed from source
|
|
|
|
if (!$feed_data) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("last unconditional update request: $last_unconditional", Debug::$LOG_VERBOSE);
|
2016-01-13 15:12:31 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
if (ini_get("open_basedir") && function_exists("curl_init")) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("not using CURL due to open_basedir restrictions", Debug::$LOG_VERBOSE);
|
2017-04-26 12:29:22 +00:00
|
|
|
}
|
2013-04-18 07:35:58 +00:00
|
|
|
|
2018-01-30 07:44:31 +00:00
|
|
|
if (time() - strtotime($last_unconditional) > MAX_CONDITIONAL_INTERVAL) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("maximum allowed interval for conditional requests exceeded, forcing refetch", Debug::$LOG_VERBOSE);
|
2017-11-30 10:12:28 +00:00
|
|
|
|
2018-01-30 07:44:31 +00:00
|
|
|
$force_refetch = true;
|
|
|
|
} else {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("stored last modified for conditional request: $stored_last_modified", Debug::$LOG_VERBOSE);
|
2018-01-30 07:44:31 +00:00
|
|
|
}
|
2017-08-17 11:40:21 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("fetching [$fetch_url] (force_refetch: $force_refetch)...", Debug::$LOG_VERBOSE);
|
2017-08-17 11:40:21 +00:00
|
|
|
|
|
|
|
$feed_data = fetch_file_contents([
|
|
|
|
"url" => $fetch_url,
|
|
|
|
"login" => $auth_login,
|
|
|
|
"pass" => $auth_pass,
|
|
|
|
"timeout" => $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
|
|
|
|
"last_modified" => $force_refetch ? "" : $stored_last_modified
|
|
|
|
]);
|
2013-04-18 07:35:58 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
global $fetch_curl_used;
|
2013-04-18 07:35:58 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
if (!$fetch_curl_used) {
|
|
|
|
$tmp = @gzdecode($feed_data);
|
2013-04-19 08:35:30 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
if ($tmp) $feed_data = $tmp;
|
|
|
|
}
|
2013-02-23 14:38:50 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
$feed_data = trim($feed_data);
|
2013-05-08 15:22:33 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("fetch done.", Debug::$LOG_VERBOSE);
|
|
|
|
Debug::log("source last modified: " . $fetch_last_modified, Debug::$LOG_VERBOSE);
|
2017-08-17 11:40:21 +00:00
|
|
|
|
|
|
|
if ($feed_data && $fetch_last_modified != $stored_last_modified) {
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_feeds SET last_modified = ? WHERE id = ?");
|
|
|
|
$sth->execute([substr($fetch_last_modified, 0, 245), $feed]);
|
2017-08-17 11:40:21 +00:00
|
|
|
}
|
2014-03-14 07:29:58 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
// cache vanilla feed data for re-use
|
2017-12-03 18:01:35 +00:00
|
|
|
if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/feeds")) {
|
2017-04-26 12:29:22 +00:00
|
|
|
$new_rss_hash = sha1($feed_data);
|
|
|
|
|
|
|
|
if ($new_rss_hash != $rss_hash) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("saving $cache_filename", Debug::$LOG_VERBOSE);
|
2017-04-26 12:29:22 +00:00
|
|
|
@file_put_contents($cache_filename, $feed_data);
|
2014-03-14 07:29:58 +00:00
|
|
|
}
|
2013-03-29 07:50:58 +00:00
|
|
|
}
|
2017-04-26 12:29:22 +00:00
|
|
|
}
|
2013-02-23 14:38:50 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
if (!$feed_data) {
|
|
|
|
global $fetch_last_error;
|
|
|
|
global $fetch_last_error_code;
|
2013-03-29 07:31:10 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("unable to fetch: $fetch_last_error [$fetch_last_error_code]", Debug::$LOG_VERBOSE);
|
2013-03-29 07:31:10 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
// If-Modified-Since
|
|
|
|
if ($fetch_last_error_code != 304) {
|
2017-12-02 06:55:53 +00:00
|
|
|
$error_message = $fetch_last_error;
|
2017-04-26 12:29:22 +00:00
|
|
|
} else {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("source claims data not modified, nothing to do.", Debug::$LOG_VERBOSE);
|
2017-12-02 06:55:53 +00:00
|
|
|
$error_message = "";
|
2017-04-26 12:29:22 +00:00
|
|
|
}
|
2013-03-29 07:50:58 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_feeds SET last_error = ?,
|
|
|
|
last_updated = NOW() WHERE id = ?");
|
|
|
|
$sth->execute([$error_message, $feed]);
|
2013-03-29 07:50:58 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
return;
|
2013-03-29 07:31:10 +00:00
|
|
|
}
|
|
|
|
|
2013-04-18 08:27:34 +00:00
|
|
|
foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
|
2013-05-20 11:28:56 +00:00
|
|
|
$feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
|
2013-02-23 14:38:50 +00:00
|
|
|
}
|
|
|
|
|
2017-04-27 10:08:43 +00:00
|
|
|
$rss = new FeedParser($feed_data);
|
|
|
|
$rss->init();
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2012-12-24 09:45:34 +00:00
|
|
|
if (!$rss->error()) {
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2012-12-25 06:02:08 +00:00
|
|
|
// We use local pluginhost here because we need to load different per-user feed plugins
|
2013-04-18 08:27:34 +00:00
|
|
|
$pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
|
2012-12-24 10:13:03 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("language: $feed_language", Debug::$LOG_VERBOSE);
|
|
|
|
Debug::log("processing feed data...", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2012-05-24 12:20:03 +00:00
|
|
|
if (DB_TYPE == "pgsql") {
|
|
|
|
$favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
|
|
|
|
} else {
|
|
|
|
$favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
|
|
|
|
}
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("SELECT owner_uid,favicon_avg_color,
|
2012-05-24 12:20:03 +00:00
|
|
|
(favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
|
|
|
|
favicon_needs_check
|
2017-12-02 06:55:53 +00:00
|
|
|
FROM ttrss_feeds WHERE id = ?");
|
|
|
|
$sth->execute([$feed]);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if ($row = $sth->fetch()) {
|
2017-12-03 06:35:59 +00:00
|
|
|
$favicon_needs_check = $row["favicon_needs_check"];
|
2017-12-02 06:55:53 +00:00
|
|
|
$favicon_avg_color = $row["favicon_avg_color"];
|
|
|
|
$owner_uid = $row["owner_uid"];
|
|
|
|
} else {
|
|
|
|
return false;
|
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$site_url = mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("site_url: $site_url", Debug::$LOG_VERBOSE);
|
|
|
|
Debug::log("feed_title: " . $rss->get_title(), Debug::$LOG_VERBOSE);
|
2013-05-01 13:04:57 +00:00
|
|
|
|
2013-04-15 11:08:00 +00:00
|
|
|
if ($favicon_needs_check || $force_refetch) {
|
2013-04-23 18:30:06 +00:00
|
|
|
|
|
|
|
/* terrible hack: if we crash on floicon shit here, we won't check
|
2013-04-23 18:33:23 +00:00
|
|
|
* the icon avgcolor again (unless the icon got updated) */
|
2013-04-23 18:30:06 +00:00
|
|
|
|
2013-04-23 18:33:23 +00:00
|
|
|
$favicon_file = ICONS_DIR . "/$feed.ico";
|
|
|
|
$favicon_modified = @filemtime($favicon_file);
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("checking favicon...", Debug::$LOG_VERBOSE);
|
2013-04-15 11:08:00 +00:00
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
RSSUtils::check_feed_favicon($site_url, $feed);
|
2013-04-23 18:33:23 +00:00
|
|
|
$favicon_modified_new = @filemtime($favicon_file);
|
|
|
|
|
|
|
|
if ($favicon_modified_new > $favicon_modified)
|
|
|
|
$favicon_avg_color = '';
|
2013-04-15 11:08:00 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$favicon_colorstring = "";
|
2013-04-23 18:37:06 +00:00
|
|
|
if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
|
2017-05-05 15:10:07 +00:00
|
|
|
require_once "colors.php";
|
2013-04-15 11:08:00 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
|
|
|
|
id = ?");
|
|
|
|
$sth->execute([$feed]);
|
2013-04-24 15:01:51 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$favicon_color = calculate_avg_color($favicon_file);
|
|
|
|
|
|
|
|
$favicon_colorstring = ",favicon_avg_color = " . $pdo->quote($favicon_color);
|
2013-04-15 11:17:12 +00:00
|
|
|
|
2013-04-23 18:30:06 +00:00
|
|
|
} else if ($favicon_avg_color == 'fail') {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("floicon failed on this file, not trying to recalculate avg color", Debug::$LOG_VERBOSE);
|
2013-04-15 13:01:51 +00:00
|
|
|
}
|
2013-04-15 11:08:00 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
|
|
|
|
$favicon_colorstring WHERE id = ?");
|
|
|
|
$sth->execute([$feed]);
|
2012-05-24 12:45:54 +00:00
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("loading filters & labels...", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2013-04-17 14:34:18 +00:00
|
|
|
$filters = load_filters($feed, $owner_uid);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
if (Debug::get_loglevel() >= Debug::$LOG_EXTENDED) {
|
2018-01-30 07:44:31 +00:00
|
|
|
print_r($filters);
|
|
|
|
}
|
2017-07-02 17:37:52 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("" . count($filters) . " filters loaded.", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2012-12-24 09:45:34 +00:00
|
|
|
$items = $rss->get_items();
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2012-12-24 09:45:34 +00:00
|
|
|
if (!is_array($items)) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("no articles found.", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_feeds
|
|
|
|
SET last_updated = NOW(), last_unconditional = NOW(), last_error = '' WHERE id = ?");
|
|
|
|
$sth->execute([$feed]);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
return true; // no articles
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("processing articles...", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2015-06-17 20:25:21 +00:00
|
|
|
$tstart = time();
|
|
|
|
|
2012-12-24 09:45:34 +00:00
|
|
|
foreach ($items as $item) {
|
2017-12-02 08:16:33 +00:00
|
|
|
$pdo->beginTransaction();
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
if (Debug::get_loglevel() >= 3) {
|
2011-12-13 12:06:25 +00:00
|
|
|
print_r($item);
|
|
|
|
}
|
|
|
|
|
2015-06-17 20:25:21 +00:00
|
|
|
if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("looks like there's too many articles to process at once, breaking out", Debug::$LOG_VERBOSE);
|
2017-12-02 08:16:33 +00:00
|
|
|
$pdo->commit();
|
2015-06-17 20:25:21 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$entry_guid = strip_tags($item->get_id());
|
|
|
|
if (!$entry_guid) $entry_guid = strip_tags($item->get_link());
|
2017-05-05 15:10:07 +00:00
|
|
|
if (!$entry_guid) $entry_guid = RSSUtils::make_guid_from_title($item->get_title());
|
2017-12-02 08:16:33 +00:00
|
|
|
|
|
|
|
if (!$entry_guid) {
|
|
|
|
$pdo->commit();
|
|
|
|
continue;
|
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2013-01-18 13:42:26 +00:00
|
|
|
$entry_guid = "$owner_uid,$entry_guid";
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$entry_guid_hashed = 'SHA1:' . sha1($entry_guid);
|
2013-04-15 15:28:11 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("guid $entry_guid / $entry_guid_hashed", Debug::$LOG_VERBOSE);
|
2013-04-15 15:28:11 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$entry_timestamp = strip_tags($item->get_date());
|
2013-05-01 13:38:16 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("orig date: " . $item->get_date(), Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2013-03-25 16:49:11 +00:00
|
|
|
if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
|
2011-12-13 12:06:25 +00:00
|
|
|
$entry_timestamp = time();
|
|
|
|
}
|
|
|
|
|
|
|
|
$entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("date $entry_timestamp [$entry_timestamp_fmt]", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$entry_title = strip_tags($item->get_title());
|
2012-12-24 09:53:38 +00:00
|
|
|
|
2013-02-26 14:27:40 +00:00
|
|
|
$entry_link = rewrite_relative_url($site_url, $item->get_link());
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-08-12 15:12:34 +00:00
|
|
|
$entry_language = mb_substr(trim($item->get_language()), 0, 2);
|
2018-08-12 13:54:13 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("title $entry_title", Debug::$LOG_VERBOSE);
|
|
|
|
Debug::log("link $entry_link", Debug::$LOG_VERBOSE);
|
|
|
|
Debug::log("language $entry_language", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
|
|
|
if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
|
|
|
|
|
2012-12-24 09:45:34 +00:00
|
|
|
$entry_content = $item->get_content();
|
|
|
|
if (!$entry_content) $entry_content = $item->get_description();
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
if (Debug::get_loglevel() >= 3) {
|
2013-04-19 09:24:51 +00:00
|
|
|
print "content: ";
|
2016-01-22 23:17:24 +00:00
|
|
|
print htmlspecialchars($entry_content);
|
2011-12-26 09:11:08 +00:00
|
|
|
print "\n";
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$entry_comments = mb_substr(strip_tags($item->get_comments_url()), 0, 245);
|
2016-11-25 11:04:41 +00:00
|
|
|
$num_comments = (int) $item->get_comments_count();
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$entry_author = strip_tags($item->get_author());
|
|
|
|
$entry_guid = mb_substr($entry_guid, 0, 245);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("author $entry_author", Debug::$LOG_VERBOSE);
|
|
|
|
Debug::log("num_comments: $num_comments", Debug::$LOG_VERBOSE);
|
|
|
|
Debug::log("looking for tags...", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
|
|
|
// parse <category> entries into tags
|
|
|
|
|
|
|
|
$additional_tags = array();
|
|
|
|
|
2012-12-24 09:45:34 +00:00
|
|
|
$additional_tags_src = $item->get_categories();
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2012-12-24 09:45:34 +00:00
|
|
|
if (is_array($additional_tags_src)) {
|
|
|
|
foreach ($additional_tags_src as $tobj) {
|
2013-05-01 13:04:57 +00:00
|
|
|
array_push($additional_tags, $tobj);
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
2012-12-24 09:45:34 +00:00
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2012-12-24 09:56:36 +00:00
|
|
|
$entry_tags = array_unique($additional_tags);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-02-11 07:24:24 +00:00
|
|
|
for ($i = 0; $i < count($entry_tags); $i++) {
|
2011-12-13 12:06:25 +00:00
|
|
|
$entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
|
|
|
|
|
2018-02-11 07:24:24 +00:00
|
|
|
// we don't support numeric tags, let's prefix them
|
|
|
|
if (is_numeric($entry_tags[$i])) $entry_tags[$i] = 't:' . $entry_tags[$i];
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("tags found: " . join(",", $entry_tags), Debug::$LOG_VERBOSE);
|
2013-05-01 16:33:59 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("done collecting data.", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("SELECT id, content_hash, lang FROM ttrss_entries
|
|
|
|
WHERE guid = ? OR guid = ?");
|
|
|
|
$sth->execute([$entry_guid, $entry_guid_hashed]);
|
2013-02-21 18:24:06 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if ($row = $sth->fetch()) {
|
|
|
|
$base_entry_id = $row["id"];
|
|
|
|
$entry_stored_hash = $row["content_hash"];
|
2017-05-04 12:26:21 +00:00
|
|
|
$article_labels = Article::get_article_labels($base_entry_id, $owner_uid);
|
2015-12-07 08:54:24 +00:00
|
|
|
|
2017-05-04 12:22:57 +00:00
|
|
|
$existing_tags = Article::get_article_tags($base_entry_id, $owner_uid);
|
2015-12-07 08:54:24 +00:00
|
|
|
$entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
|
2013-02-21 18:24:06 +00:00
|
|
|
} else {
|
2014-08-18 14:32:16 +00:00
|
|
|
$base_entry_id = false;
|
|
|
|
$entry_stored_hash = "";
|
2015-03-03 21:26:52 +00:00
|
|
|
$article_labels = array();
|
2013-02-21 18:24:06 +00:00
|
|
|
}
|
|
|
|
|
2012-12-30 07:31:34 +00:00
|
|
|
$article = array("owner_uid" => $owner_uid, // read only
|
2013-02-21 18:24:06 +00:00
|
|
|
"guid" => $entry_guid, // read only
|
2015-06-17 12:15:04 +00:00
|
|
|
"guid_hashed" => $entry_guid_hashed, // read only
|
2012-12-24 09:45:34 +00:00
|
|
|
"title" => $entry_title,
|
|
|
|
"content" => $entry_content,
|
|
|
|
"link" => $entry_link,
|
2015-03-03 21:26:52 +00:00
|
|
|
"labels" => $article_labels, // current limitation: can add labels to article, can't remove them
|
2012-12-24 09:45:34 +00:00
|
|
|
"tags" => $entry_tags,
|
2013-02-23 08:07:46 +00:00
|
|
|
"author" => $entry_author,
|
2015-01-19 09:51:59 +00:00
|
|
|
"force_catchup" => false, // ugly hack for the time being
|
2015-06-15 17:01:07 +00:00
|
|
|
"score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
|
2015-06-19 07:12:47 +00:00
|
|
|
"language" => $entry_language,
|
2017-08-06 17:20:39 +00:00
|
|
|
"num_comments" => $num_comments, // read only
|
2013-09-02 06:03:04 +00:00
|
|
|
"feed" => array("id" => $feed,
|
|
|
|
"fetch_url" => $fetch_url,
|
2016-08-06 18:16:39 +00:00
|
|
|
"site_url" => $site_url,
|
|
|
|
"cache_images" => $cache_images)
|
2017-05-05 15:10:07 +00:00
|
|
|
);
|
2012-11-26 10:33:18 +00:00
|
|
|
|
2014-08-18 14:32:16 +00:00
|
|
|
$entry_plugin_data = "";
|
2017-05-05 15:10:07 +00:00
|
|
|
$entry_current_hash = RSSUtils::calculate_article_hash($article, $pluginhost);
|
2014-08-18 14:32:16 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("article hash: $entry_current_hash [stored=$entry_stored_hash]", Debug::$LOG_VERBOSE);
|
2014-08-18 14:32:16 +00:00
|
|
|
|
2014-08-21 09:20:41 +00:00
|
|
|
if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("stored article seems up to date [IID: $base_entry_id], updating timestamp only", Debug::$LOG_VERBOSE);
|
2014-08-18 14:32:16 +00:00
|
|
|
|
|
|
|
// we keep encountering the entry in feeds, so we need to
|
|
|
|
// update date_updated column so that we don't get horrible
|
|
|
|
// dupes when the entry gets purged and reinserted again e.g.
|
|
|
|
// in the case of SLOW SLOW OMG SLOW updating feeds
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_entries SET date_updated = NOW()
|
|
|
|
WHERE id = ?");
|
|
|
|
$sth->execute([$base_entry_id]);
|
2014-08-18 14:32:16 +00:00
|
|
|
|
2017-12-02 08:16:33 +00:00
|
|
|
$pdo->commit();
|
2016-01-07 19:01:35 +00:00
|
|
|
continue;
|
2014-08-18 14:32:16 +00:00
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("hash differs, applying plugin filters:", Debug::$LOG_VERBOSE);
|
2014-08-18 14:32:16 +00:00
|
|
|
|
2013-04-18 08:27:34 +00:00
|
|
|
foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("... " . get_class($plugin), Debug::$LOG_VERBOSE);
|
2014-08-18 14:32:16 +00:00
|
|
|
|
|
|
|
$start = microtime(true);
|
2012-12-24 09:45:34 +00:00
|
|
|
$article = $plugin->hook_article_filter($article);
|
2014-08-12 15:36:29 +00:00
|
|
|
|
2018-11-30 06:19:00 +00:00
|
|
|
Debug::log(sprintf("=== %.4f (sec)", microtime(true) - $start), Debug::$LOG_VERBOSE);
|
2014-08-18 14:32:16 +00:00
|
|
|
|
|
|
|
$entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
|
2012-11-26 10:33:18 +00:00
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
if (Debug::get_loglevel() >= 3) {
|
2016-01-22 23:17:24 +00:00
|
|
|
print "processed content: ";
|
|
|
|
print htmlspecialchars($article["content"]);
|
|
|
|
print "\n";
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("plugin data: $entry_plugin_data", Debug::$LOG_VERBOSE);
|
2014-08-18 14:32:16 +00:00
|
|
|
|
2015-06-16 05:17:49 +00:00
|
|
|
// Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
|
2018-08-04 13:07:32 +00:00
|
|
|
if (DB_TYPE == "mysql" && MYSQL_CHARSET != "UTF8MB4") {
|
2015-06-16 05:17:49 +00:00
|
|
|
foreach ($article as $k => $v) {
|
2015-07-06 12:32:26 +00:00
|
|
|
// i guess we'll have to take the risk of 4byte unicode labels & tags here
|
2016-07-26 07:51:22 +00:00
|
|
|
if (is_string($article[$k])) {
|
2018-01-29 20:24:11 +00:00
|
|
|
$article[$k] = RSSUtils::strip_utf8mb4($v);
|
2015-07-06 12:32:26 +00:00
|
|
|
}
|
2015-06-16 05:17:49 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-08-11 20:28:41 +00:00
|
|
|
/* Collect article tags here so we could filter by them: */
|
|
|
|
|
2018-12-06 16:15:00 +00:00
|
|
|
$matched_rules = [];
|
|
|
|
$matched_filters = [];
|
2016-08-26 11:45:34 +00:00
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
$article_filters = RSSUtils::get_article_filters($filters, $article["title"],
|
2017-04-26 12:29:22 +00:00
|
|
|
$article["content"], $article["link"], $article["author"],
|
2018-12-06 16:15:00 +00:00
|
|
|
$article["tags"], $matched_rules, $matched_filters);
|
|
|
|
|
|
|
|
// $article_filters should be renamed to something like $filter_actions; actual filter objects are in $matched_filters
|
|
|
|
foreach ($pluginhost->get_hooks(PluginHost::HOOK_FILTER_TRIGGERED) as $plugin) {
|
|
|
|
$plugin->hook_filter_triggered($feed, $owner_uid, $article, $matched_filters, $matched_rules, $article_filters);
|
|
|
|
}
|
2015-08-11 20:28:41 +00:00
|
|
|
|
2018-12-06 20:12:20 +00:00
|
|
|
$matched_filter_ids = array_map(function($f) { return $f['id']; }, $matched_filters);
|
2018-12-06 16:37:20 +00:00
|
|
|
|
2018-12-06 20:12:20 +00:00
|
|
|
if (count($matched_filter_ids) > 0) {
|
|
|
|
$filter_ids_qmarks = arr_qmarks($matched_filter_ids);
|
2018-12-06 20:05:02 +00:00
|
|
|
|
|
|
|
$fsth = $pdo->prepare("UPDATE ttrss_filters2 SET last_triggered = NOW() WHERE
|
2018-12-06 20:12:20 +00:00
|
|
|
id IN ($filter_ids_qmarks) AND owner_uid = ?");
|
|
|
|
|
|
|
|
$fsth->execute(array_merge($matched_filter_ids, [$owner_uid]));
|
2018-12-06 20:05:02 +00:00
|
|
|
}
|
2018-12-06 16:37:20 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
if (Debug::get_loglevel() >= Debug::$LOG_EXTENDED) {
|
2018-12-06 16:15:00 +00:00
|
|
|
Debug::log("matched filters: ", Debug::$LOG_VERBOSE);
|
|
|
|
|
|
|
|
if (count($matched_filters != 0)) {
|
|
|
|
print_r($matched_filters);
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("matched filter rules: ", Debug::$LOG_VERBOSE);
|
2016-08-26 11:45:34 +00:00
|
|
|
|
|
|
|
if (count($matched_rules) != 0) {
|
|
|
|
print_r($matched_rules);
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("filter actions: ", Debug::$LOG_VERBOSE);
|
2016-08-26 11:45:34 +00:00
|
|
|
|
2015-08-11 20:28:41 +00:00
|
|
|
if (count($article_filters) != 0) {
|
|
|
|
print_r($article_filters);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
$plugin_filter_names = RSSUtils::find_article_filters($article_filters, "plugin");
|
2015-08-11 20:28:41 +00:00
|
|
|
$plugin_filter_actions = $pluginhost->get_filter_actions();
|
|
|
|
|
|
|
|
if (count($plugin_filter_names) > 0) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("applying plugin filter actions...", Debug::$LOG_VERBOSE);
|
2015-08-11 20:28:41 +00:00
|
|
|
|
|
|
|
foreach ($plugin_filter_names as $pfn) {
|
|
|
|
list($pfclass,$pfaction) = explode(":", $pfn["param"]);
|
|
|
|
|
|
|
|
if (isset($plugin_filter_actions[$pfclass])) {
|
|
|
|
$plugin = $pluginhost->get_plugin($pfclass);
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("... $pfclass: $pfaction", Debug::$LOG_VERBOSE);
|
2015-08-11 20:28:41 +00:00
|
|
|
|
|
|
|
if ($plugin) {
|
|
|
|
$start = microtime(true);
|
|
|
|
$article = $plugin->hook_article_filter_action($article, $pfaction);
|
|
|
|
|
2018-12-13 02:30:16 +00:00
|
|
|
Debug::log(sprintf("=== %.4f (sec)", microtime(true) - $start), Debug::$LOG_VERBOSE);
|
2015-08-11 20:28:41 +00:00
|
|
|
} else {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("??? $pfclass: plugin object not found.", Debug::$LOG_VERBOSE);
|
2015-08-11 20:28:41 +00:00
|
|
|
}
|
|
|
|
} else {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("??? $pfclass: filter plugin not registered.", Debug::$LOG_VERBOSE);
|
2015-08-11 20:28:41 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2012-12-24 09:45:34 +00:00
|
|
|
$entry_tags = $article["tags"];
|
2017-12-02 06:55:53 +00:00
|
|
|
$entry_title = strip_tags($article["title"]);
|
|
|
|
$entry_author = mb_substr(strip_tags($article["author"]), 0, 245);
|
|
|
|
$entry_link = strip_tags($article["link"]);
|
2013-03-18 20:42:59 +00:00
|
|
|
$entry_content = $article["content"]; // escaped below
|
2015-01-19 09:51:59 +00:00
|
|
|
$entry_force_catchup = $article["force_catchup"];
|
2015-03-03 21:26:52 +00:00
|
|
|
$article_labels = $article["labels"];
|
2015-06-15 17:01:07 +00:00
|
|
|
$entry_score_modifier = (int) $article["score_modifier"];
|
2017-12-02 06:55:53 +00:00
|
|
|
$entry_language = $article["language"];
|
2015-03-03 21:26:52 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
if (Debug::get_loglevel() >= Debug::$LOG_EXTENDED) {
|
|
|
|
Debug::log("article labels:", Debug::$LOG_VERBOSE);
|
2016-08-26 11:45:34 +00:00
|
|
|
|
|
|
|
if (count($article_labels) != 0) {
|
|
|
|
print_r($article_labels);
|
|
|
|
}
|
2015-03-03 21:26:52 +00:00
|
|
|
}
|
2015-01-19 09:51:59 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("force catchup: $entry_force_catchup", Debug::$LOG_VERBOSE);
|
2013-03-18 20:42:59 +00:00
|
|
|
|
2013-02-04 09:00:50 +00:00
|
|
|
if ($cache_images && is_writable(CACHE_DIR . '/images'))
|
2018-11-30 05:34:29 +00:00
|
|
|
RSSUtils::cache_media($entry_content, $site_url);
|
2013-02-04 09:00:50 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$csth = $pdo->prepare("SELECT id FROM ttrss_entries
|
|
|
|
WHERE guid = ? OR guid = ?");
|
|
|
|
$csth->execute([$entry_guid, $entry_guid_hashed]);
|
2012-12-30 08:22:29 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if (!$row = $csth->fetch()) {
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("base guid [$entry_guid or $entry_guid_hashed] not found, creating...", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
|
|
|
// base post entry does not exist, create it
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$usth = $pdo->prepare(
|
2011-12-13 12:06:25 +00:00
|
|
|
"INSERT INTO ttrss_entries
|
2018-11-03 12:00:58 +00:00
|
|
|
(title,
|
2011-12-13 12:06:25 +00:00
|
|
|
guid,
|
|
|
|
link,
|
|
|
|
updated,
|
|
|
|
content,
|
|
|
|
content_hash,
|
|
|
|
no_orig_date,
|
|
|
|
date_updated,
|
|
|
|
date_entered,
|
|
|
|
comments,
|
|
|
|
num_comments,
|
2013-02-21 18:24:06 +00:00
|
|
|
plugin_data,
|
2013-07-31 06:30:14 +00:00
|
|
|
lang,
|
2011-12-13 12:06:25 +00:00
|
|
|
author)
|
|
|
|
VALUES
|
2017-12-02 06:55:53 +00:00
|
|
|
(?, ?, ?, ?, ?, ?,
|
2014-02-19 11:17:29 +00:00
|
|
|
false,
|
2011-12-13 12:06:25 +00:00
|
|
|
NOW(),
|
2017-12-02 06:55:53 +00:00
|
|
|
?, ?, ?, ?, ?, ?)");
|
|
|
|
|
|
|
|
$usth->execute([$entry_title,
|
|
|
|
$entry_guid_hashed,
|
|
|
|
$entry_link,
|
|
|
|
$entry_timestamp_fmt,
|
2017-12-03 10:32:24 +00:00
|
|
|
"$entry_content",
|
2017-12-02 06:55:53 +00:00
|
|
|
$entry_current_hash,
|
|
|
|
$date_feed_processed,
|
|
|
|
$entry_comments,
|
2017-12-03 06:35:59 +00:00
|
|
|
(int)$num_comments,
|
2017-12-02 06:55:53 +00:00
|
|
|
$entry_plugin_data,
|
2017-12-03 10:32:24 +00:00
|
|
|
"$entry_language",
|
|
|
|
"$entry_author"]);
|
2012-10-30 11:10:52 +00:00
|
|
|
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$csth->execute([$entry_guid, $entry_guid_hashed]);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
|
|
|
$entry_ref_id = 0;
|
|
|
|
$entry_int_id = 0;
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if ($row = $csth->fetch()) {
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("base guid found, checking for user record", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$ref_id = $row['id'];
|
2011-12-13 12:06:25 +00:00
|
|
|
$entry_ref_id = $ref_id;
|
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
if (RSSUtils::find_article_filter($article_filters, "filter")) {
|
2018-12-08 14:01:30 +00:00
|
|
|
Debug::log("article is filtered out, nothing to do.");
|
2017-12-02 08:16:33 +00:00
|
|
|
$pdo->commit();
|
2011-12-13 12:06:25 +00:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
$score = RSSUtils::calculate_article_score($article_filters) + $entry_score_modifier;
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("initial score: $score [including plugin modifier: $entry_score_modifier]", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2016-01-07 18:49:31 +00:00
|
|
|
// check for user post link to main table
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("SELECT ref_id, int_id FROM ttrss_user_entries WHERE
|
|
|
|
ref_id = ? AND owner_uid = ?");
|
|
|
|
$sth->execute([$ref_id, $owner_uid]);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
|
|
|
// okay it doesn't exist - create user entry
|
2017-12-02 06:55:53 +00:00
|
|
|
if ($row = $sth->fetch()) {
|
|
|
|
$entry_ref_id = $row["ref_id"];
|
|
|
|
$entry_int_id = $row["int_id"];
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("user record FOUND: RID: $entry_ref_id, IID: $entry_int_id", Debug::$LOG_VERBOSE);
|
2017-12-02 06:55:53 +00:00
|
|
|
} else {
|
2018-11-03 12:00:58 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("user record not found, creating...", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
if ($score >= -500 && !RSSUtils::find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
|
2017-12-02 06:55:53 +00:00
|
|
|
$unread = 1;
|
|
|
|
$last_read_qpart = null;
|
2011-12-13 12:06:25 +00:00
|
|
|
} else {
|
2017-12-02 06:55:53 +00:00
|
|
|
$unread = 0;
|
2017-12-18 15:27:23 +00:00
|
|
|
$last_read_qpart = date("Y-m-d H:i"); // we can't use NOW() here because it gets quoted
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
if (RSSUtils::find_article_filter($article_filters, 'mark') || $score > 1000) {
|
2017-12-02 06:55:53 +00:00
|
|
|
$marked = 1;
|
2011-12-13 12:06:25 +00:00
|
|
|
} else {
|
2017-12-02 06:55:53 +00:00
|
|
|
$marked = 0;
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
if (RSSUtils::find_article_filter($article_filters, 'publish')) {
|
2017-12-02 06:55:53 +00:00
|
|
|
$published = 1;
|
2011-12-13 12:06:25 +00:00
|
|
|
} else {
|
2017-12-02 06:55:53 +00:00
|
|
|
$published = 0;
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2018-01-22 22:41:24 +00:00
|
|
|
$last_marked = ($marked == 1) ? 'NOW()' : 'NULL';
|
|
|
|
$last_published = ($published == 1) ? 'NOW()' : 'NULL';
|
2013-03-17 11:32:44 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare(
|
2011-12-13 12:06:25 +00:00
|
|
|
"INSERT INTO ttrss_user_entries
|
|
|
|
(ref_id, owner_uid, feed_id, unread, last_read, marked,
|
2013-03-17 11:32:44 +00:00
|
|
|
published, score, tag_cache, label_cache, uuid,
|
|
|
|
last_marked, last_published)
|
2017-12-10 23:05:03 +00:00
|
|
|
VALUES (?, ?, ?, ?, ?, ?, ?, ?, '', '', '', ".$last_marked.", ".$last_published.")");
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth->execute([$ref_id, $owner_uid, $feed, $unread, $last_read_qpart, $marked,
|
2017-12-10 23:05:03 +00:00
|
|
|
$published, $score]);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("SELECT int_id FROM ttrss_user_entries WHERE
|
|
|
|
ref_id = ? AND owner_uid = ? AND
|
|
|
|
feed_id = ? LIMIT 1");
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth->execute([$ref_id, $owner_uid, $feed]);
|
|
|
|
|
|
|
|
if ($row = $sth->fetch())
|
|
|
|
$entry_int_id = $row['int_id'];
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("resulting RID: $entry_ref_id, IID: $entry_int_id", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-04-03 10:57:27 +00:00
|
|
|
if (DB_TYPE == "pgsql")
|
|
|
|
$tsvector_qpart = "tsvector_combined = to_tsvector(:ts_lang, :ts_content),";
|
|
|
|
else
|
2015-08-04 09:52:49 +00:00
|
|
|
$tsvector_qpart = "";
|
2017-12-03 07:10:01 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_entries
|
2017-12-03 07:10:01 +00:00
|
|
|
SET title = :title,
|
2015-08-04 09:52:49 +00:00
|
|
|
$tsvector_qpart
|
2017-12-03 07:10:01 +00:00
|
|
|
content = :content,
|
|
|
|
content_hash = :content_hash,
|
|
|
|
updated = :updated,
|
2018-05-20 07:41:08 +00:00
|
|
|
date_updated = NOW(),
|
2017-12-03 07:10:01 +00:00
|
|
|
num_comments = :num_comments,
|
|
|
|
plugin_data = :plugin_data,
|
|
|
|
author = :author,
|
2018-11-03 12:00:58 +00:00
|
|
|
lang = :lang
|
2017-12-03 07:10:01 +00:00
|
|
|
WHERE id = :id");
|
|
|
|
|
2018-04-03 10:57:27 +00:00
|
|
|
$params = [":title" => $entry_title,
|
2017-12-03 10:32:24 +00:00
|
|
|
":content" => "$entry_content",
|
2017-12-03 07:10:01 +00:00
|
|
|
":content_hash" => $entry_current_hash,
|
|
|
|
":updated" => $entry_timestamp_fmt,
|
|
|
|
":num_comments" => (int)$num_comments,
|
|
|
|
":plugin_data" => $entry_plugin_data,
|
2017-12-03 10:32:24 +00:00
|
|
|
":author" => "$entry_author",
|
2017-12-03 07:10:01 +00:00
|
|
|
":lang" => $entry_language,
|
2018-04-03 10:57:27 +00:00
|
|
|
":id" => $ref_id];
|
|
|
|
|
|
|
|
if (DB_TYPE == "pgsql") {
|
|
|
|
$params[":ts_lang"] = $feed_language;
|
2018-07-10 11:08:55 +00:00
|
|
|
$params[":ts_content"] = mb_substr(strip_tags($entry_title . " " . $entry_content), 0, 900000);
|
2018-04-03 10:57:27 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
$sth->execute($params);
|
2014-08-18 14:32:16 +00:00
|
|
|
|
2015-06-17 12:15:04 +00:00
|
|
|
// update aux data
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_user_entries
|
|
|
|
SET score = ? WHERE ref_id = ?");
|
|
|
|
$sth->execute([$score, $ref_id]);
|
2015-06-17 12:15:04 +00:00
|
|
|
|
2014-08-18 14:32:16 +00:00
|
|
|
if ($mark_unread_on_update) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("article updated, marking unread as requested.", Debug::$LOG_VERBOSE);
|
2015-12-05 00:10:57 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_user_entries
|
|
|
|
SET last_read = null, unread = true WHERE ref_id = ?");
|
|
|
|
$sth->execute([$ref_id]);
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("assigning labels [other]...", Debug::$LOG_VERBOSE);
|
2015-03-03 21:26:52 +00:00
|
|
|
|
|
|
|
foreach ($article_labels as $label) {
|
2017-05-04 12:57:40 +00:00
|
|
|
Labels::add_article($entry_ref_id, $label[1], $owner_uid);
|
2015-03-03 21:26:52 +00:00
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("assigning labels [filters]...", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
RSSUtils::assign_article_to_label_filters($entry_ref_id, $article_filters,
|
2012-10-30 08:58:35 +00:00
|
|
|
$owner_uid, $article_labels);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("looking for enclosures...", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
|
|
|
// enclosures
|
|
|
|
|
|
|
|
$enclosures = array();
|
|
|
|
|
2012-12-24 09:45:34 +00:00
|
|
|
$encs = $item->get_enclosures();
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2012-12-24 09:45:34 +00:00
|
|
|
if (is_array($encs)) {
|
|
|
|
foreach ($encs as $e) {
|
|
|
|
$e_item = array(
|
2017-03-27 16:20:46 +00:00
|
|
|
rewrite_relative_url($site_url, $e->link),
|
|
|
|
$e->type, $e->length, $e->title, $e->width, $e->height);
|
2018-01-29 20:24:11 +00:00
|
|
|
|
|
|
|
// Yet another episode of "mysql utf8_general_ci is gimped"
|
2018-08-04 13:07:32 +00:00
|
|
|
if (DB_TYPE == "mysql" && MYSQL_CHARSET != "UTF8MB4") {
|
2018-01-30 07:44:31 +00:00
|
|
|
for ($i = 0; $i < count($e_item); $i++) {
|
|
|
|
if (is_string($e_item[$i])) {
|
|
|
|
$e_item[$i] = RSSUtils::strip_utf8mb4($e_item[$i]);
|
|
|
|
}
|
|
|
|
}
|
2018-01-29 20:24:11 +00:00
|
|
|
}
|
|
|
|
|
2018-01-30 07:44:31 +00:00
|
|
|
array_push($enclosures, $e_item);
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-03-23 12:19:25 +00:00
|
|
|
if ($cache_images && is_writable(CACHE_DIR . '/images'))
|
2018-11-30 05:34:29 +00:00
|
|
|
RSSUtils::cache_enclosures($enclosures, $site_url);
|
2017-03-23 12:19:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
if (Debug::get_loglevel() >= Debug::$LOG_EXTENDED) {
|
|
|
|
Debug::log("article enclosures:", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
print_r($enclosures);
|
|
|
|
}
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$esth = $pdo->prepare("SELECT id FROM ttrss_enclosures
|
2018-08-18 16:17:30 +00:00
|
|
|
WHERE content_url = ? AND content_type = ? AND post_id = ?");
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$usth = $pdo->prepare("INSERT INTO ttrss_enclosures
|
|
|
|
(content_url, content_type, title, duration, post_id, width, height) VALUES
|
|
|
|
(?, ?, ?, ?, ?, ?, ?)");
|
2013-08-05 08:26:09 +00:00
|
|
|
|
2011-12-13 12:06:25 +00:00
|
|
|
foreach ($enclosures as $enc) {
|
2017-12-02 06:55:53 +00:00
|
|
|
$enc_url = $enc[0];
|
|
|
|
$enc_type = $enc[1];
|
2017-12-02 08:16:33 +00:00
|
|
|
$enc_dur = (int)$enc[2];
|
2017-12-02 06:55:53 +00:00
|
|
|
$enc_title = $enc[3];
|
2014-07-15 13:32:28 +00:00
|
|
|
$enc_width = intval($enc[4]);
|
|
|
|
$enc_height = intval($enc[5]);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-08-18 16:17:30 +00:00
|
|
|
$esth->execute([$enc_url, $enc_type, $entry_ref_id]);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if (!$esth->fetch()) {
|
|
|
|
$usth->execute([$enc_url, $enc_type, (string)$enc_title, $enc_dur, $entry_ref_id, $enc_width, $enc_height]);
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// check for manual tags (we have to do it here since they're loaded from filters)
|
|
|
|
|
|
|
|
foreach ($article_filters as $f) {
|
2012-08-30 14:50:56 +00:00
|
|
|
if ($f["type"] == "tag") {
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2012-08-30 14:50:56 +00:00
|
|
|
$manual_tags = trim_array(explode(",", $f["param"]));
|
2011-12-13 12:06:25 +00:00
|
|
|
|
|
|
|
foreach ($manual_tags as $tag) {
|
|
|
|
if (tag_is_valid($tag)) {
|
|
|
|
array_push($entry_tags, $tag);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Skip boring tags
|
|
|
|
|
2013-04-17 12:23:15 +00:00
|
|
|
$boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
|
2011-12-13 12:06:25 +00:00
|
|
|
'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
|
|
|
|
|
|
|
|
$filtered_tags = array();
|
|
|
|
$tags_to_cache = array();
|
|
|
|
|
|
|
|
if ($entry_tags && is_array($entry_tags)) {
|
|
|
|
foreach ($entry_tags as $tag) {
|
|
|
|
if (array_search($tag, $boring_tags) === false) {
|
|
|
|
array_push($filtered_tags, $tag);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
$filtered_tags = array_unique($filtered_tags);
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
if (Debug::get_loglevel() >= Debug::$LOG_EXTENDED) {
|
|
|
|
Debug::log("filtered article tags:", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
print_r($filtered_tags);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Save article tags in the database
|
|
|
|
|
|
|
|
if (count($filtered_tags) > 0) {
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$tsth = $pdo->prepare("SELECT id FROM ttrss_tags
|
|
|
|
WHERE tag_name = ? AND post_int_id = ? AND
|
|
|
|
owner_uid = ? LIMIT 1");
|
|
|
|
|
|
|
|
$usth = $pdo->prepare("INSERT INTO ttrss_tags
|
|
|
|
(owner_uid,tag_name,post_int_id)
|
|
|
|
VALUES (?, ?, ?)");
|
2011-12-13 12:06:25 +00:00
|
|
|
|
|
|
|
foreach ($filtered_tags as $tag) {
|
|
|
|
|
|
|
|
$tag = sanitize_tag($tag);
|
|
|
|
|
|
|
|
if (!tag_is_valid($tag)) continue;
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$tsth->execute([$tag, $entry_int_id, $owner_uid]);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
if (!$tsth->fetch()) {
|
|
|
|
$usth->execute([$owner_uid, $tag, $entry_int_id]);
|
2017-05-05 15:10:07 +00:00
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
|
|
|
|
array_push($tags_to_cache, $tag);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* update the cache */
|
|
|
|
|
|
|
|
$tags_to_cache = array_unique($tags_to_cache);
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$tags_str = join(",", $tags_to_cache);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$tsth = $pdo->prepare("UPDATE ttrss_user_entries
|
|
|
|
SET tag_cache = ? WHERE ref_id = ?
|
|
|
|
AND owner_uid = ?");
|
|
|
|
$tsth->execute([$tags_str, $entry_ref_id, $owner_uid]);
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("article processed", Debug::$LOG_VERBOSE);
|
2017-12-02 08:16:33 +00:00
|
|
|
|
|
|
|
$pdo->commit();
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("purging feed...", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
purge_feed($feed, 0);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_feeds
|
|
|
|
SET last_updated = NOW(), last_unconditional = NOW(), last_error = '' WHERE id = ?");
|
|
|
|
$sth->execute([$feed]);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
|
|
|
} else {
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$error_msg = mb_substr($rss->error(), 0, 245);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("fetch error: $error_msg", Debug::$LOG_VERBOSE);
|
2013-10-29 08:15:26 +00:00
|
|
|
|
|
|
|
if (count($rss->errors()) > 1) {
|
|
|
|
foreach ($rss->errors() as $error) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("+ $error", Debug::$LOG_VERBOSE);
|
2013-10-29 08:15:26 +00:00
|
|
|
}
|
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$sth = $pdo->prepare("UPDATE ttrss_feeds SET last_error = ?,
|
|
|
|
last_updated = NOW(), last_unconditional = NOW() WHERE id = ?");
|
|
|
|
$sth->execute([$error_msg, $feed]);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2013-10-25 10:42:43 +00:00
|
|
|
unset($rss);
|
2017-12-02 06:55:53 +00:00
|
|
|
return false;
|
2013-10-25 10:42:43 +00:00
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("done", Debug::$LOG_VERBOSE);
|
2013-10-25 10:42:43 +00:00
|
|
|
|
2017-04-26 12:29:22 +00:00
|
|
|
return true;
|
2011-12-13 12:06:25 +00:00
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
static function cache_enclosures($enclosures, $site_url) {
|
2017-03-23 12:19:25 +00:00
|
|
|
foreach ($enclosures as $enc) {
|
|
|
|
|
|
|
|
if (preg_match("/(image|audio|video)/", $enc[1])) {
|
|
|
|
|
|
|
|
$src = rewrite_relative_url($site_url, $enc[0]);
|
|
|
|
|
|
|
|
$local_filename = CACHE_DIR . "/images/" . sha1($src);
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("cache_enclosures: downloading: $src to $local_filename", Debug::$LOG_VERBOSE);
|
2017-03-23 12:19:25 +00:00
|
|
|
|
|
|
|
if (!file_exists($local_filename)) {
|
|
|
|
$file_content = fetch_file_contents($src);
|
|
|
|
|
2017-05-29 20:14:42 +00:00
|
|
|
if ($file_content && strlen($file_content) > MIN_CACHE_FILE_SIZE) {
|
2017-03-23 12:19:25 +00:00
|
|
|
file_put_contents($local_filename, $file_content);
|
|
|
|
}
|
2018-09-10 13:17:12 +00:00
|
|
|
} else if (is_writable($local_filename)) {
|
2017-03-23 12:19:25 +00:00
|
|
|
touch($local_filename);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
static function cache_media($html, $site_url) {
|
2011-12-26 09:11:08 +00:00
|
|
|
libxml_use_internal_errors(true);
|
|
|
|
|
|
|
|
$charset_hack = '<head>
|
|
|
|
<meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
|
|
|
|
</head>';
|
|
|
|
|
|
|
|
$doc = new DOMDocument();
|
|
|
|
$doc->loadHTML($charset_hack . $html);
|
|
|
|
$xpath = new DOMXPath($doc);
|
|
|
|
|
2017-03-23 12:19:25 +00:00
|
|
|
$entries = $xpath->query('(//img[@src])|(//video/source[@src])|(//audio/source[@src])');
|
2011-12-26 09:11:08 +00:00
|
|
|
|
|
|
|
foreach ($entries as $entry) {
|
2017-02-04 08:50:01 +00:00
|
|
|
if ($entry->hasAttribute('src') && strpos($entry->getAttribute('src'), "data:") !== 0) {
|
2011-12-26 09:11:08 +00:00
|
|
|
$src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
|
|
|
|
|
2017-03-23 11:55:40 +00:00
|
|
|
$local_filename = CACHE_DIR . "/images/" . sha1($src);
|
2011-12-26 09:11:08 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("cache_media: checking $src", Debug::$LOG_VERBOSE);
|
2011-12-26 09:11:08 +00:00
|
|
|
|
|
|
|
if (!file_exists($local_filename)) {
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("cache_media: downloading: $src to $local_filename", Debug::$LOG_VERBOSE);
|
2018-08-16 08:56:51 +00:00
|
|
|
|
2011-12-26 09:11:08 +00:00
|
|
|
$file_content = fetch_file_contents($src);
|
|
|
|
|
2017-05-29 20:14:42 +00:00
|
|
|
if ($file_content && strlen($file_content) > MIN_CACHE_FILE_SIZE) {
|
2011-12-26 09:11:08 +00:00
|
|
|
file_put_contents($local_filename, $file_content);
|
|
|
|
}
|
2018-09-10 13:17:12 +00:00
|
|
|
} else if (is_writable($local_filename)) {
|
2015-08-14 23:49:41 +00:00
|
|
|
touch($local_filename);
|
2011-12-26 09:11:08 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
static function expire_error_log() {
|
|
|
|
Debug::log("Removing old error log entries...");
|
2013-04-16 16:34:20 +00:00
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$pdo = Db::pdo();
|
|
|
|
|
2013-04-16 16:34:20 +00:00
|
|
|
if (DB_TYPE == "pgsql") {
|
2017-12-02 06:55:53 +00:00
|
|
|
$pdo->query("DELETE FROM ttrss_error_log
|
2013-04-16 16:34:20 +00:00
|
|
|
WHERE created_at < NOW() - INTERVAL '7 days'");
|
|
|
|
} else {
|
2017-12-02 06:55:53 +00:00
|
|
|
$pdo->query("DELETE FROM ttrss_error_log
|
2013-04-16 16:34:20 +00:00
|
|
|
WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
static function expire_lock_files() {
|
|
|
|
Debug::log("Removing old lock files...", Debug::$LOG_VERBOSE);
|
2012-12-12 05:46:02 +00:00
|
|
|
|
|
|
|
$num_deleted = 0;
|
|
|
|
|
|
|
|
if (is_writable(LOCK_DIRECTORY)) {
|
|
|
|
$files = glob(LOCK_DIRECTORY . "/*.lock");
|
|
|
|
|
|
|
|
if ($files) {
|
|
|
|
foreach ($files as $file) {
|
2013-04-19 19:25:26 +00:00
|
|
|
if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
|
2012-12-12 05:46:02 +00:00
|
|
|
unlink($file);
|
|
|
|
++$num_deleted;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("Removed $num_deleted old lock files.");
|
2012-12-12 05:46:02 +00:00
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
static function expire_cached_files() {
|
2018-11-23 09:31:47 +00:00
|
|
|
foreach (array("feeds", "images", "export", "upload") as $dir) {
|
2011-12-26 09:11:08 +00:00
|
|
|
$cache_dir = CACHE_DIR . "/$dir";
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("Expiring $cache_dir", Debug::$LOG_VERBOSE);
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2011-12-26 09:11:08 +00:00
|
|
|
$num_deleted = 0;
|
|
|
|
|
|
|
|
if (is_writable($cache_dir)) {
|
|
|
|
$files = glob("$cache_dir/*");
|
|
|
|
|
2012-12-12 05:46:02 +00:00
|
|
|
if ($files) {
|
2012-02-24 09:43:50 +00:00
|
|
|
foreach ($files as $file) {
|
2017-05-29 20:14:42 +00:00
|
|
|
if (time() - filemtime($file) > 86400*CACHE_MAX_DAYS) {
|
2012-02-24 09:43:50 +00:00
|
|
|
unlink($file);
|
2011-12-26 09:11:08 +00:00
|
|
|
|
2012-02-24 09:43:50 +00:00
|
|
|
++$num_deleted;
|
|
|
|
}
|
2011-12-26 09:11:08 +00:00
|
|
|
}
|
|
|
|
}
|
2012-12-12 05:46:02 +00:00
|
|
|
}
|
2011-12-26 09:11:08 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("$cache_dir: removed $num_deleted files.");
|
2011-12-26 09:11:08 +00:00
|
|
|
}
|
|
|
|
}
|
2011-12-13 12:06:25 +00:00
|
|
|
|
2012-01-10 08:49:10 +00:00
|
|
|
/**
|
2017-05-05 15:10:07 +00:00
|
|
|
* Source: http://www.php.net/manual/en/function.parse-url.php#104527
|
|
|
|
* Returns the url query as associative array
|
|
|
|
*
|
|
|
|
* @param string query
|
|
|
|
* @return array params
|
|
|
|
*/
|
|
|
|
static function convertUrlQuery($query) {
|
2012-01-10 08:49:10 +00:00
|
|
|
$queryParts = explode('&', $query);
|
|
|
|
|
|
|
|
$params = array();
|
|
|
|
|
|
|
|
foreach ($queryParts as $param) {
|
|
|
|
$item = explode('=', $param);
|
|
|
|
$params[$item[0]] = $item[1];
|
|
|
|
}
|
|
|
|
|
|
|
|
return $params;
|
|
|
|
}
|
2012-10-30 08:11:46 +00:00
|
|
|
|
2018-12-06 16:15:00 +00:00
|
|
|
static function get_article_filters($filters, $title, $content, $link, $author, $tags, &$matched_rules = false, &$matched_filters = false) {
|
2012-10-30 08:11:46 +00:00
|
|
|
$matches = array();
|
|
|
|
|
|
|
|
foreach ($filters as $filter) {
|
|
|
|
$match_any_rule = $filter["match_any_rule"];
|
2013-03-25 15:46:43 +00:00
|
|
|
$inverse = $filter["inverse"];
|
2012-10-30 08:11:46 +00:00
|
|
|
$filter_match = false;
|
|
|
|
|
|
|
|
foreach ($filter["rules"] as $rule) {
|
|
|
|
$match = false;
|
2013-04-26 11:46:48 +00:00
|
|
|
$reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
|
2013-03-25 15:46:43 +00:00
|
|
|
$rule_inverse = $rule["inverse"];
|
2012-10-30 08:11:46 +00:00
|
|
|
|
|
|
|
if (!$reg_exp)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
switch ($rule["type"]) {
|
2017-05-05 15:10:07 +00:00
|
|
|
case "title":
|
|
|
|
$match = @preg_match("/$reg_exp/iu", $title);
|
|
|
|
break;
|
|
|
|
case "content":
|
|
|
|
// we don't need to deal with multiline regexps
|
|
|
|
$content = preg_replace("/[\r\n\t]/", "", $content);
|
2012-11-20 17:27:53 +00:00
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
$match = @preg_match("/$reg_exp/iu", $content);
|
|
|
|
break;
|
|
|
|
case "both":
|
|
|
|
// we don't need to deal with multiline regexps
|
|
|
|
$content = preg_replace("/[\r\n\t]/", "", $content);
|
2012-11-20 17:27:53 +00:00
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
$match = (@preg_match("/$reg_exp/iu", $title) || @preg_match("/$reg_exp/iu", $content));
|
|
|
|
break;
|
|
|
|
case "link":
|
|
|
|
$match = @preg_match("/$reg_exp/iu", $link);
|
|
|
|
break;
|
|
|
|
case "author":
|
|
|
|
$match = @preg_match("/$reg_exp/iu", $author);
|
|
|
|
break;
|
|
|
|
case "tag":
|
|
|
|
foreach ($tags as $tag) {
|
|
|
|
if (@preg_match("/$reg_exp/iu", $tag)) {
|
|
|
|
$match = true;
|
|
|
|
break;
|
|
|
|
}
|
2013-04-26 11:54:04 +00:00
|
|
|
}
|
2017-05-05 15:10:07 +00:00
|
|
|
break;
|
2012-10-30 08:11:46 +00:00
|
|
|
}
|
|
|
|
|
2013-03-25 15:46:43 +00:00
|
|
|
if ($rule_inverse) $match = !$match;
|
|
|
|
|
2012-10-30 08:11:46 +00:00
|
|
|
if ($match_any_rule) {
|
|
|
|
if ($match) {
|
|
|
|
$filter_match = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
$filter_match = $match;
|
|
|
|
if (!$match) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-03-25 15:46:43 +00:00
|
|
|
if ($inverse) $filter_match = !$filter_match;
|
|
|
|
|
2012-10-30 08:11:46 +00:00
|
|
|
if ($filter_match) {
|
2016-08-26 11:45:34 +00:00
|
|
|
if (is_array($matched_rules)) array_push($matched_rules, $rule);
|
2018-12-06 16:15:00 +00:00
|
|
|
if (is_array($matched_filters)) array_push($matched_filters, $filter);
|
2016-08-26 11:45:34 +00:00
|
|
|
|
2012-10-30 08:11:46 +00:00
|
|
|
foreach ($filter["actions"] AS $action) {
|
|
|
|
array_push($matches, $action);
|
2013-03-31 15:28:36 +00:00
|
|
|
|
|
|
|
// if Stop action encountered, perform no further processing
|
2015-12-03 12:33:47 +00:00
|
|
|
if (isset($action["type"]) && $action["type"] == "stop") return $matches;
|
2012-10-30 08:11:46 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return $matches;
|
|
|
|
}
|
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
static function find_article_filter($filters, $filter_name) {
|
2012-10-30 08:11:46 +00:00
|
|
|
foreach ($filters as $f) {
|
|
|
|
if ($f["type"] == $filter_name) {
|
|
|
|
return $f;
|
|
|
|
};
|
|
|
|
}
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
static function find_article_filters($filters, $filter_name) {
|
2012-10-30 08:11:46 +00:00
|
|
|
$results = array();
|
|
|
|
|
|
|
|
foreach ($filters as $f) {
|
|
|
|
if ($f["type"] == $filter_name) {
|
|
|
|
array_push($results, $f);
|
|
|
|
};
|
|
|
|
}
|
|
|
|
return $results;
|
|
|
|
}
|
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
static function calculate_article_score($filters) {
|
2012-10-30 08:11:46 +00:00
|
|
|
$score = 0;
|
|
|
|
|
|
|
|
foreach ($filters as $f) {
|
|
|
|
if ($f["type"] == "score") {
|
|
|
|
$score += $f["param"];
|
|
|
|
};
|
|
|
|
}
|
|
|
|
return $score;
|
|
|
|
}
|
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
static function labels_contains_caption($labels, $caption) {
|
2012-10-30 08:58:35 +00:00
|
|
|
foreach ($labels as $label) {
|
|
|
|
if ($label[1] == $caption) {
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
static function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
|
2012-10-30 08:11:46 +00:00
|
|
|
foreach ($filters as $f) {
|
|
|
|
if ($f["type"] == "label") {
|
2017-05-05 15:10:07 +00:00
|
|
|
if (!RSSUtils::labels_contains_caption($article_labels, $f["param"])) {
|
2017-05-04 12:57:40 +00:00
|
|
|
Labels::add_article($id, $f["param"], $owner_uid);
|
2012-10-30 08:58:35 +00:00
|
|
|
}
|
|
|
|
}
|
2012-10-30 08:11:46 +00:00
|
|
|
}
|
|
|
|
}
|
2012-11-23 09:22:34 +00:00
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
static function make_guid_from_title($title) {
|
2013-01-22 18:32:17 +00:00
|
|
|
return preg_replace("/[ \"\',.:;]/", "-",
|
|
|
|
mb_strtolower(strip_tags($title), 'utf-8'));
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
static function cleanup_counters_cache() {
|
2017-12-02 06:55:53 +00:00
|
|
|
$pdo = Db::pdo();
|
|
|
|
|
|
|
|
$res = $pdo->query("DELETE FROM ttrss_counters_cache
|
2015-03-06 11:14:51 +00:00
|
|
|
WHERE feed_id > 0 AND
|
|
|
|
(SELECT COUNT(id) FROM ttrss_feeds WHERE
|
|
|
|
id = feed_id AND
|
|
|
|
ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
|
|
|
|
|
2017-12-02 06:55:53 +00:00
|
|
|
$frows = $res->rowCount();
|
|
|
|
|
|
|
|
$res = $pdo->query("DELETE FROM ttrss_cat_counters_cache
|
2015-03-06 11:14:51 +00:00
|
|
|
WHERE feed_id > 0 AND
|
|
|
|
(SELECT COUNT(id) FROM ttrss_feed_categories WHERE
|
|
|
|
id = feed_id AND
|
|
|
|
ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
|
2017-12-02 06:55:53 +00:00
|
|
|
|
|
|
|
$crows = $res->rowCount();
|
2015-03-06 11:14:51 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
|
2015-03-06 11:14:51 +00:00
|
|
|
}
|
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
static function housekeeping_user($owner_uid) {
|
2016-08-11 12:01:01 +00:00
|
|
|
$tmph = new PluginHost();
|
|
|
|
|
|
|
|
load_user_plugins($owner_uid, $tmph);
|
|
|
|
|
|
|
|
$tmph->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
|
|
|
|
}
|
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
static function housekeeping_common() {
|
|
|
|
RSSUtils::expire_cached_files();
|
|
|
|
RSSUtils::expire_lock_files();
|
|
|
|
RSSUtils::expire_error_log();
|
2013-04-24 10:54:59 +00:00
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
$count = RSSUtils::update_feedbrowser_cache();
|
2018-11-30 05:34:29 +00:00
|
|
|
Debug::log("Feedbrowser updated, $count feeds processed.");
|
2013-04-24 10:54:59 +00:00
|
|
|
|
2018-11-30 05:34:29 +00:00
|
|
|
Article::purge_orphans();
|
|
|
|
RSSUtils::cleanup_counters_cache();
|
2013-08-02 10:06:18 +00:00
|
|
|
|
2013-08-02 10:47:34 +00:00
|
|
|
PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
|
2013-04-24 10:54:59 +00:00
|
|
|
}
|
2017-04-26 17:24:18 +00:00
|
|
|
|
2017-05-05 15:10:07 +00:00
|
|
|
static function check_feed_favicon($site_url, $feed) {
|
|
|
|
# print "FAVICON [$site_url]: $favicon_url\n";
|
2017-05-04 12:00:21 +00:00
|
|
|
|
|
|
|
$icon_file = ICONS_DIR . "/$feed.ico";
|
|
|
|
|
|
|
|
if (!file_exists($icon_file)) {
|
|
|
|
$favicon_url = get_favicon_url($site_url);
|
|
|
|
|
|
|
|
if ($favicon_url) {
|
|
|
|
// Limiting to "image" type misses those served with text/plain
|
|
|
|
$contents = fetch_file_contents($favicon_url); // , "image");
|
|
|
|
|
|
|
|
if ($contents) {
|
|
|
|
// Crude image type matching.
|
|
|
|
// Patterns gleaned from the file(1) source code.
|
|
|
|
if (preg_match('/^\x00\x00\x01\x00/', $contents)) {
|
|
|
|
// 0 string \000\000\001\000 MS Windows icon resource
|
|
|
|
//error_log("check_feed_favicon: favicon_url=$favicon_url isa MS Windows icon resource");
|
|
|
|
}
|
|
|
|
elseif (preg_match('/^GIF8/', $contents)) {
|
|
|
|
// 0 string GIF8 GIF image data
|
|
|
|
//error_log("check_feed_favicon: favicon_url=$favicon_url isa GIF image");
|
|
|
|
}
|
|
|
|
elseif (preg_match('/^\x89PNG\x0d\x0a\x1a\x0a/', $contents)) {
|
|
|
|
// 0 string \x89PNG\x0d\x0a\x1a\x0a PNG image data
|
|
|
|
//error_log("check_feed_favicon: favicon_url=$favicon_url isa PNG image");
|
|
|
|
}
|
|
|
|
elseif (preg_match('/^\xff\xd8/', $contents)) {
|
|
|
|
// 0 beshort 0xffd8 JPEG image data
|
|
|
|
//error_log("check_feed_favicon: favicon_url=$favicon_url isa JPG image");
|
|
|
|
}
|
2017-10-09 15:25:24 +00:00
|
|
|
elseif (preg_match('/^BM/', $contents)) {
|
|
|
|
// 0 string BM PC bitmap (OS2, Windows BMP files)
|
|
|
|
//error_log("check_feed_favicon, favicon_url=$favicon_url isa BMP image");
|
|
|
|
}
|
2017-05-04 12:00:21 +00:00
|
|
|
else {
|
|
|
|
//error_log("check_feed_favicon: favicon_url=$favicon_url isa UNKNOWN type");
|
|
|
|
$contents = "";
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if ($contents) {
|
|
|
|
$fp = @fopen($icon_file, "w");
|
|
|
|
|
|
|
|
if ($fp) {
|
|
|
|
fwrite($fp, $contents);
|
|
|
|
fclose($fp);
|
|
|
|
chmod($icon_file, 0644);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return $icon_file;
|
|
|
|
}
|
|
|
|
}
|
2017-05-05 15:10:07 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
2017-09-25 00:37:49 +00:00
|
|
|
}
|