3 static function calculate_article_hash($article, $pluginhost) {
6 foreach ($article as $k => $v) {
7 if ($k != "feed" && isset($v)) {
8 $x = strip_tags(is_array($v) ?
implode(",", $v) : $v);
10 //_debug("$k:" . sha1($x) . ":" . htmlspecialchars($x), true);
12 $tmp .= sha1("$k:" . sha1($x));
16 return sha1(implode(",", $pluginhost->get_plugin_names()) . $tmp);
19 static function update_feedbrowser_cache() {
23 $sth = $pdo->query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
24 FROM ttrss_feeds WHERE feed_url NOT IN (SELECT feed_url FROM ttrss_feeds
25 WHERE private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%')
26 GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
28 $pdo->beginTransaction();
30 $pdo->query("DELETE FROM ttrss_feedbrowser_cache");
34 while ($line = $sth->fetch()) {
36 $subscribers = $line["subscribers"];
37 $feed_url = $line["feed_url"];
38 $title = $line["title"];
39 $site_url = $line["site_url"];
41 $tmph = $pdo->prepare("SELECT subscribers FROM
42 ttrss_feedbrowser_cache WHERE feed_url = ?");
43 $tmph->execute([$feed_url]);
45 if (!$tmph->fetch()) {
47 $tmph = $pdo->prepare("INSERT INTO ttrss_feedbrowser_cache
48 (feed_url, site_url, title, subscribers)
52 $tmph->execute([$feed_url, $site_url, $title, $subscribers]);
66 static function update_daemon_common($limit = DAEMON_FEED_LIMIT
, $debug = true) {
67 $schema_version = get_schema_version();
69 if ($schema_version != SCHEMA_VERSION
) {
70 die("Schema version is wrong, please upgrade the database.\n");
75 if (!SINGLE_USER_MODE
&& DAEMON_UPDATE_LOGIN_LIMIT
> 0) {
76 if (DB_TYPE
== "pgsql") {
77 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT
." days'";
79 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT
." DAY)";
82 $login_thresh_qpart = "";
85 if (DB_TYPE
== "pgsql") {
86 $update_limit_qpart = "AND ((
87 ttrss_feeds.update_interval = 0
88 AND ttrss_user_prefs.value != '-1'
89 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
91 ttrss_feeds.update_interval > 0
92 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
93 ) OR (ttrss_feeds.last_updated IS NULL
94 AND ttrss_user_prefs.value != '-1')
95 OR (last_updated = '1970-01-01 00:00:00'
96 AND ttrss_user_prefs.value != '-1'))";
98 $update_limit_qpart = "AND ((
99 ttrss_feeds.update_interval = 0
100 AND ttrss_user_prefs.value != '-1'
101 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
103 ttrss_feeds.update_interval > 0
104 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
105 ) OR (ttrss_feeds.last_updated IS NULL
106 AND ttrss_user_prefs.value != '-1')
107 OR (last_updated = '1970-01-01 00:00:00'
108 AND ttrss_user_prefs.value != '-1'))";
111 // Test if feed is currently being updated by another process.
112 if (DB_TYPE
== "pgsql") {
113 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
115 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
118 $query_limit = $limit ?
sprintf("LIMIT %d", $limit) : "";
120 // Update the least recently updated feeds first
121 $query_order = "ORDER BY last_updated";
122 if (DB_TYPE
== "pgsql") $query_order .= " NULLS FIRST";
124 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
126 ttrss_feeds, ttrss_users, ttrss_user_prefs
128 ttrss_feeds.owner_uid = ttrss_users.id
129 AND ttrss_user_prefs.profile IS NULL
130 AND ttrss_users.id = ttrss_user_prefs.owner_uid
131 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
132 $login_thresh_qpart $update_limit_qpart
133 $updstart_thresh_qpart
134 $query_order $query_limit";
136 $res = $pdo->query($query);
138 $feeds_to_update = array();
139 while ($line = $res->fetch()) {
140 array_push($feeds_to_update, $line['feed_url']);
143 if ($debug) _debug(sprintf("Scheduled %d feeds to update...", count($feeds_to_update)));
145 // Update last_update_started before actually starting the batch
146 // in order to minimize collision risk for parallel daemon tasks
147 if (count($feeds_to_update) > 0) {
148 $feeds_qmarks = arr_qmarks($feeds_to_update);
150 $tmph = $pdo->prepare("UPDATE ttrss_feeds SET last_update_started = NOW()
151 WHERE feed_url IN ($feeds_qmarks)");
152 $tmph->execute($feeds_to_update);
156 $bstarted = microtime(true);
158 $batch_owners = array();
160 // since we have the data cached, we can deal with other feeds with the same url
161 $usth = $pdo->prepare("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
162 FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
163 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
164 ttrss_users.id = ttrss_user_prefs.owner_uid AND
165 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
166 ttrss_user_prefs.profile IS NULL AND
170 ORDER BY ttrss_feeds.id $query_limit");
172 foreach ($feeds_to_update as $feed) {
173 if($debug) _debug("Base feed: $feed");
175 $usth->execute([$feed]);
176 //update_rss_feed($line["id"], true);
178 if ($tline = $usth->fetch()) {
179 if ($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
181 if (array_search($tline["owner_uid"], $batch_owners) === FALSE)
182 array_push($batch_owners, $tline["owner_uid"]);
184 $fstarted = microtime(true);
185 RSSUtils
::update_rss_feed($tline["id"], true, false);
186 _debug_suppress(false);
188 _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
195 _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
196 microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
199 foreach ($batch_owners as $owner_uid) {
200 _debug("Running housekeeping tasks for user $owner_uid...");
202 RSSUtils
::housekeeping_user($owner_uid);
205 // Send feed digests by email if needed.
206 Digest
::send_headlines_digests($debug);
211 // this is used when subscribing
212 static function set_basic_feed_info($feed) {
216 $sth = $pdo->prepare("SELECT owner_uid,feed_url,auth_pass,auth_login,auth_pass_encrypted
217 FROM ttrss_feeds WHERE id = ?");
218 $sth->execute([$feed]);
220 if ($row = $sth->fetch()) {
222 $owner_uid = $row["owner_uid"];
224 $auth_pass_encrypted = $row["auth_pass_encrypted"];
226 $auth_login = $row["auth_login"];
227 $auth_pass = $row["auth_pass"];
229 if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
230 require_once "crypt.php";
231 $auth_pass = decrypt_string($auth_pass);
234 $fetch_url = $row["feed_url"];
236 $pluginhost = new PluginHost();
237 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
239 $pluginhost->load(PLUGINS
, PluginHost
::KIND_ALL
);
240 $pluginhost->load($user_plugins, PluginHost
::KIND_USER
, $owner_uid);
241 $pluginhost->load_data();
243 $basic_info = array();
244 foreach ($pluginhost->get_hooks(PluginHost
::HOOK_FEED_BASIC_INFO
) as $plugin) {
245 $basic_info = $plugin->hook_feed_basic_info($basic_info, $fetch_url, $owner_uid, $feed, $auth_login, $auth_pass);
249 $feed_data = fetch_file_contents($fetch_url, false,
250 $auth_login, $auth_pass, false,
254 global $fetch_curl_used;
256 if (!$fetch_curl_used) {
257 $tmp = @gzdecode
($feed_data);
259 if ($tmp) $feed_data = $tmp;
262 $feed_data = trim($feed_data);
264 $rss = new FeedParser($feed_data);
267 if (!$rss->error()) {
269 'title' => mb_substr($rss->get_title(), 0, 199),
270 'site_url' => mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245)
275 if ($basic_info && is_array($basic_info)) {
276 $sth = $pdo->prepare("SELECT title, site_url FROM ttrss_feeds WHERE id = ?");
277 $sth->execute([$feed]);
279 if ($row = $sth->fetch()) {
281 $registered_title = $row["title"];
282 $orig_site_url = $row["site_url"];
284 if ($basic_info['title'] && (!$registered_title ||
$registered_title == "[Unknown]")) {
286 $sth = $pdo->prepare("UPDATE ttrss_feeds SET
287 title = ? WHERE id = ?");
288 $sth->execute([$basic_info['title'], $feed]);
291 if ($basic_info['site_url'] && $orig_site_url != $basic_info['site_url']) {
292 $sth = $pdo->prepare("UPDATE ttrss_feeds SET
293 site_url = ? WHERE id = ?");
294 $sth->execute([$basic_info['site_url'], $feed]);
303 * @SuppressWarnings(PHPMD.UnusedFormalParameter)
305 static function update_rss_feed($feed, $no_cache = false) {
307 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') ||
clean($_REQUEST['xdebug']);
309 _debug_suppress(!$debug_enabled);
310 _debug("start", $debug_enabled);
314 $sth = $pdo->prepare("SELECT title FROM ttrss_feeds WHERE id = ?");
315 $sth->execute([$feed]);
317 if (!$row = $sth->fetch()) {
318 _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
319 user_error("Attempt to update unknown/invalid feed $feed", E_USER_WARNING
);
323 $title = $row["title"];
325 // feed was batch-subscribed or something, we need to get basic info
326 // this is not optimal currently as it fetches stuff separately TODO: optimize
327 if ($title == "[Unknown]") {
328 _debug("setting basic feed info for $feed...");
329 RSSUtils
::set_basic_feed_info($feed);
332 $sth = $pdo->prepare("SELECT id,update_interval,auth_login,
333 feed_url,auth_pass,cache_images,
334 mark_unread_on_update, owner_uid,
335 auth_pass_encrypted, feed_language,
337 ".SUBSTRING_FOR_DATE
."(last_unconditional, 1, 19) AS last_unconditional
338 FROM ttrss_feeds WHERE id = ?");
339 $sth->execute([$feed]);
341 if ($row = $sth->fetch()) {
343 $owner_uid = $row["owner_uid"];
344 $mark_unread_on_update = $row["mark_unread_on_update"];
345 $auth_pass_encrypted = $row["auth_pass_encrypted"];
347 $sth = $pdo->prepare("UPDATE ttrss_feeds SET last_update_started = NOW()
349 $sth->execute([$feed]);
351 $auth_login = $row["auth_login"];
352 $auth_pass = $row["auth_pass"];
354 if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
355 require_once "crypt.php";
356 $auth_pass = decrypt_string($auth_pass);
359 $stored_last_modified = $row["last_modified"];
360 $last_unconditional = $row["last_unconditional"];
361 $cache_images = $row["cache_images"];
362 $fetch_url = $row["feed_url"];
363 $feed_language = mb_strtolower($row["feed_language"]);
364 if (!$feed_language) $feed_language = 'english';
370 $date_feed_processed = date('Y-m-d H:i');
372 $cache_filename = CACHE_DIR
. "/feeds/" . sha1($fetch_url) . ".xml";
374 $pluginhost = new PluginHost();
375 $pluginhost->set_debug($debug_enabled);
376 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
378 $pluginhost->load(PLUGINS
, PluginHost
::KIND_ALL
);
379 $pluginhost->load($user_plugins, PluginHost
::KIND_USER
, $owner_uid);
380 $pluginhost->load_data();
384 $force_refetch = isset($_REQUEST["force_refetch"]);
387 foreach ($pluginhost->get_hooks(PluginHost
::HOOK_FETCH_FEED
) as $plugin) {
388 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
393 file_exists($cache_filename) &&
394 is_readable($cache_filename) &&
395 !$auth_login && !$auth_pass &&
396 filemtime($cache_filename) > time() - 30) {
398 _debug("using local cache [$cache_filename].", $debug_enabled);
400 @$feed_data = file_get_contents($cache_filename);
403 $rss_hash = sha1($feed_data);
407 _debug("local cache will not be used for this feed", $debug_enabled);
410 global $fetch_last_modified;
412 // fetch feed from source
414 _debug("last unconditional update request: $last_unconditional");
416 if (ini_get("open_basedir") && function_exists("curl_init")) {
417 _debug("not using CURL due to open_basedir restrictions");
420 if (time() - strtotime($last_unconditional) > MAX_CONDITIONAL_INTERVAL
) {
421 _debug("maximum allowed interval for conditional requests exceeded, forcing refetch");
423 $force_refetch = true;
425 _debug("stored last modified for conditional request: $stored_last_modified", $debug_enabled);
428 _debug("fetching [$fetch_url] (force_refetch: $force_refetch)...", $debug_enabled);
430 $feed_data = fetch_file_contents([
432 "login" => $auth_login,
433 "pass" => $auth_pass,
434 "timeout" => $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT
: FEED_FETCH_TIMEOUT
,
435 "last_modified" => $force_refetch ?
"" : $stored_last_modified
438 global $fetch_curl_used;
440 if (!$fetch_curl_used) {
441 $tmp = @gzdecode
($feed_data);
443 if ($tmp) $feed_data = $tmp;
446 $feed_data = trim($feed_data);
448 _debug("fetch done.", $debug_enabled);
449 _debug("source last modified: " . $fetch_last_modified, $debug_enabled);
451 if ($feed_data && $fetch_last_modified != $stored_last_modified) {
452 $sth = $pdo->prepare("UPDATE ttrss_feeds SET last_modified = ? WHERE id = ?");
453 $sth->execute([substr($fetch_last_modified, 0, 245), $feed]);
456 // cache vanilla feed data for re-use
457 if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR
. "/feeds")) {
458 $new_rss_hash = sha1($feed_data);
460 if ($new_rss_hash != $rss_hash) {
461 _debug("saving $cache_filename", $debug_enabled);
462 @file_put_contents
($cache_filename, $feed_data);
468 global $fetch_last_error;
469 global $fetch_last_error_code;
471 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
474 if ($fetch_last_error_code != 304) {
475 $error_message = $fetch_last_error;
477 _debug("source claims data not modified, nothing to do.", $debug_enabled);
481 $sth = $pdo->prepare("UPDATE ttrss_feeds SET last_error = ?,
482 last_updated = NOW() WHERE id = ?");
483 $sth->execute([$error_message, $feed]);
488 foreach ($pluginhost->get_hooks(PluginHost
::HOOK_FEED_FETCHED
) as $plugin) {
489 $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
492 $rss = new FeedParser($feed_data);
495 if (!$rss->error()) {
497 // We use local pluginhost here because we need to load different per-user feed plugins
498 $pluginhost->run_hooks(PluginHost
::HOOK_FEED_PARSED
, "hook_feed_parsed", $rss);
500 _debug("language: $feed_language", $debug_enabled);
501 _debug("processing feed data...", $debug_enabled);
503 if (DB_TYPE
== "pgsql") {
504 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
506 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
509 $sth = $pdo->prepare("SELECT owner_uid,favicon_avg_color,
510 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
512 FROM ttrss_feeds WHERE id = ?");
513 $sth->execute([$feed]);
515 if ($row = $sth->fetch()) {
516 $favicon_needs_check = $row["favicon_needs_check"];
517 $favicon_avg_color = $row["favicon_avg_color"];
518 $owner_uid = $row["owner_uid"];
523 $site_url = mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245);
525 _debug("site_url: $site_url", $debug_enabled);
526 _debug("feed_title: " . $rss->get_title(), $debug_enabled);
528 if ($favicon_needs_check ||
$force_refetch) {
530 /* terrible hack: if we crash on floicon shit here, we won't check
531 * the icon avgcolor again (unless the icon got updated) */
533 $favicon_file = ICONS_DIR
. "/$feed.ico";
534 $favicon_modified = @filemtime
($favicon_file);
536 _debug("checking favicon...", $debug_enabled);
538 RSSUtils
::check_feed_favicon($site_url, $feed);
539 $favicon_modified_new = @filemtime
($favicon_file);
541 if ($favicon_modified_new > $favicon_modified)
542 $favicon_avg_color = '';
544 $favicon_colorstring = "";
545 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
546 require_once "colors.php";
548 $sth = $pdo->prepare("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
550 $sth->execute([$feed]);
552 $favicon_color = calculate_avg_color($favicon_file);
554 $favicon_colorstring = ",favicon_avg_color = " . $pdo->quote($favicon_color);
556 } else if ($favicon_avg_color == 'fail') {
557 _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
560 $sth = $pdo->prepare("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
561 $favicon_colorstring WHERE id = ?");
562 $sth->execute([$feed]);
565 _debug("loading filters & labels...", $debug_enabled);
567 $filters = load_filters($feed, $owner_uid);
569 if ($debug_enabled) {
573 _debug("" . count($filters) . " filters loaded.", $debug_enabled);
575 $items = $rss->get_items();
577 if (!is_array($items)) {
578 _debug("no articles found.", $debug_enabled);
580 $sth = $pdo->prepare("UPDATE ttrss_feeds
581 SET last_updated = NOW(), last_unconditional = NOW(), last_error = '' WHERE id = ?");
582 $sth->execute([$feed]);
584 return true; // no articles
587 _debug("processing articles...", $debug_enabled);
591 foreach ($items as $item) {
592 $pdo->beginTransaction();
594 if (clean($_REQUEST['xdebug']) == 3) {
598 if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
599 _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
604 $entry_guid = strip_tags($item->get_id());
605 if (!$entry_guid) $entry_guid = strip_tags($item->get_link());
606 if (!$entry_guid) $entry_guid = RSSUtils
::make_guid_from_title($item->get_title());
613 $entry_guid = "$owner_uid,$entry_guid";
615 $entry_guid_hashed = 'SHA1:' . sha1($entry_guid);
617 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
619 $entry_timestamp = strip_tags($item->get_date());
621 _debug("orig date: " . $item->get_date(), $debug_enabled);
623 if ($entry_timestamp == -1 ||
!$entry_timestamp ||
$entry_timestamp > time()) {
624 $entry_timestamp = time();
627 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
629 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
631 $entry_title = strip_tags($item->get_title());
633 $entry_link = rewrite_relative_url($site_url, $item->get_link());
635 _debug("title $entry_title", $debug_enabled);
636 _debug("link $entry_link", $debug_enabled);
638 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
640 $entry_content = $item->get_content();
641 if (!$entry_content) $entry_content = $item->get_description();
643 if (clean($_REQUEST["xdebug"]) == 2) {
645 print htmlspecialchars($entry_content);
649 $entry_comments = mb_substr(strip_tags($item->get_comments_url()), 0, 245);
650 $num_comments = (int) $item->get_comments_count();
652 $entry_author = strip_tags($item->get_author());
653 $entry_guid = mb_substr($entry_guid, 0, 245);
655 _debug("author $entry_author", $debug_enabled);
656 _debug("num_comments: $num_comments", $debug_enabled);
657 _debug("looking for tags...", $debug_enabled);
659 // parse <category> entries into tags
661 $additional_tags = array();
663 $additional_tags_src = $item->get_categories();
665 if (is_array($additional_tags_src)) {
666 foreach ($additional_tags_src as $tobj) {
667 array_push($additional_tags, $tobj);
671 $entry_tags = array_unique($additional_tags);
673 for ($i = 0; $i < count($entry_tags); $i++
)
674 $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
676 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
678 _debug("done collecting data.", $debug_enabled);
680 $sth = $pdo->prepare("SELECT id, content_hash, lang FROM ttrss_entries
681 WHERE guid = ? OR guid = ?");
682 $sth->execute([$entry_guid, $entry_guid_hashed]);
684 if ($row = $sth->fetch()) {
685 $base_entry_id = $row["id"];
686 $entry_stored_hash = $row["content_hash"];
687 $article_labels = Article
::get_article_labels($base_entry_id, $owner_uid);
688 $entry_language = $row["lang"];
690 $existing_tags = Article
::get_article_tags($base_entry_id, $owner_uid);
691 $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
693 $base_entry_id = false;
694 $entry_stored_hash = "";
695 $article_labels = array();
696 $entry_language = "";
699 $article = array("owner_uid" => $owner_uid, // read only
700 "guid" => $entry_guid, // read only
701 "guid_hashed" => $entry_guid_hashed, // read only
702 "title" => $entry_title,
703 "content" => $entry_content,
704 "link" => $entry_link,
705 "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
706 "tags" => $entry_tags,
707 "author" => $entry_author,
708 "force_catchup" => false, // ugly hack for the time being
709 "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
710 "language" => $entry_language,
711 "num_comments" => $num_comments, // read only
712 "feed" => array("id" => $feed,
713 "fetch_url" => $fetch_url,
714 "site_url" => $site_url,
715 "cache_images" => $cache_images)
718 $entry_plugin_data = "";
719 $entry_current_hash = RSSUtils
::calculate_article_hash($article, $pluginhost);
721 _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
723 if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
724 _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
726 // we keep encountering the entry in feeds, so we need to
727 // update date_updated column so that we don't get horrible
728 // dupes when the entry gets purged and reinserted again e.g.
729 // in the case of SLOW SLOW OMG SLOW updating feeds
731 $sth = $pdo->prepare("UPDATE ttrss_entries SET date_updated = NOW()
733 $sth->execute([$base_entry_id]);
739 _debug("hash differs, applying plugin filters:", $debug_enabled);
741 foreach ($pluginhost->get_hooks(PluginHost
::HOOK_ARTICLE_FILTER
) as $plugin) {
742 _debug("... " . get_class($plugin), $debug_enabled);
744 $start = microtime(true);
745 $article = $plugin->hook_article_filter($article);
747 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
749 $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
752 if (clean($_REQUEST["xdebug"]) == 2) {
753 print "processed content: ";
754 print htmlspecialchars($article["content"]);
758 _debug("plugin data: $entry_plugin_data", $debug_enabled);
760 // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
761 if (DB_TYPE
== "mysql") {
762 foreach ($article as $k => $v) {
764 // i guess we'll have to take the risk of 4byte unicode labels & tags here
765 if (is_string($article[$k])) {
766 $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
771 /* Collect article tags here so we could filter by them: */
773 $matched_rules = array();
775 $article_filters = RSSUtils
::get_article_filters($filters, $article["title"],
776 $article["content"], $article["link"], $article["author"],
777 $article["tags"], $matched_rules);
779 if ($debug_enabled) {
780 _debug("matched filter rules: ", $debug_enabled);
782 if (count($matched_rules) != 0) {
783 print_r($matched_rules);
786 _debug("filter actions: ", $debug_enabled);
788 if (count($article_filters) != 0) {
789 print_r($article_filters);
793 $plugin_filter_names = RSSUtils
::find_article_filters($article_filters, "plugin");
794 $plugin_filter_actions = $pluginhost->get_filter_actions();
796 if (count($plugin_filter_names) > 0) {
797 _debug("applying plugin filter actions...", $debug_enabled);
799 foreach ($plugin_filter_names as $pfn) {
800 list($pfclass,$pfaction) = explode(":", $pfn["param"]);
802 if (isset($plugin_filter_actions[$pfclass])) {
803 $plugin = $pluginhost->get_plugin($pfclass);
805 _debug("... $pfclass: $pfaction", $debug_enabled);
808 $start = microtime(true);
809 $article = $plugin->hook_article_filter_action($article, $pfaction);
811 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
813 _debug("??? $pfclass: plugin object not found.");
816 _debug("??? $pfclass: filter plugin not registered.");
821 $entry_tags = $article["tags"];
822 $entry_title = strip_tags($article["title"]);
823 $entry_author = mb_substr(strip_tags($article["author"]), 0, 245);
824 $entry_link = strip_tags($article["link"]);
825 $entry_content = $article["content"]; // escaped below
826 $entry_force_catchup = $article["force_catchup"];
827 $article_labels = $article["labels"];
828 $entry_score_modifier = (int) $article["score_modifier"];
829 $entry_language = $article["language"];
831 if ($debug_enabled) {
832 _debug("article labels:", $debug_enabled);
834 if (count($article_labels) != 0) {
835 print_r($article_labels);
839 _debug("force catchup: $entry_force_catchup");
841 if ($cache_images && is_writable(CACHE_DIR
. '/images'))
842 RSSUtils
::cache_media($entry_content, $site_url, $debug_enabled);
844 $csth = $pdo->prepare("SELECT id FROM ttrss_entries
845 WHERE guid = ? OR guid = ?");
846 $csth->execute([$entry_guid, $entry_guid_hashed]);
848 if (!$row = $csth->fetch()) {
850 _debug("base guid [$entry_guid or $entry_guid_hashed] not found, creating...", $debug_enabled);
852 // base post entry does not exist, create it
854 $usth = $pdo->prepare(
855 "INSERT INTO ttrss_entries
876 $usth->execute([$entry_title,
879 $entry_timestamp_fmt,
882 $date_feed_processed,
891 $csth->execute([$entry_guid, $entry_guid_hashed]);
896 if ($row = $csth->fetch()) {
898 _debug("base guid found, checking for user record", $debug_enabled);
900 $ref_id = $row['id'];
901 $entry_ref_id = $ref_id;
903 if (RSSUtils
::find_article_filter($article_filters, "filter")) {
908 $score = RSSUtils
::calculate_article_score($article_filters) +
$entry_score_modifier;
910 _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
912 // check for user post link to main table
914 $sth = $pdo->prepare("SELECT ref_id, int_id FROM ttrss_user_entries WHERE
915 ref_id = ? AND owner_uid = ?");
916 $sth->execute([$ref_id, $owner_uid]);
918 // okay it doesn't exist - create user entry
919 if ($row = $sth->fetch()) {
920 $entry_ref_id = $row["ref_id"];
921 $entry_int_id = $row["int_id"];
923 _debug("user record FOUND: RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
926 _debug("user record not found, creating...", $debug_enabled);
928 if ($score >= -500 && !RSSUtils
::find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
930 $last_read_qpart = null;
933 $last_read_qpart = 'NOW()';
936 if (RSSUtils
::find_article_filter($article_filters, 'mark') ||
$score > 1000) {
942 if (RSSUtils
::find_article_filter($article_filters, 'publish')) {
948 $last_marked = ($marked == 'true') ?
'NOW()' : null;
949 $last_published = ($published == 'true') ?
'NOW()' : null;
951 $sth = $pdo->prepare(
952 "INSERT INTO ttrss_user_entries
953 (ref_id, owner_uid, feed_id, unread, last_read, marked,
954 published, score, tag_cache, label_cache, uuid,
955 last_marked, last_published)
956 VALUES (?, ?, ?, ?, ?, ?, ?, ?, '', '', '', ".$last_marked.", ".$last_published.")");
958 $sth->execute([$ref_id, $owner_uid, $feed, $unread, $last_read_qpart, $marked,
959 $published, $score]);
961 $sth = $pdo->prepare("SELECT int_id FROM ttrss_user_entries WHERE
962 ref_id = ? AND owner_uid = ? AND
963 feed_id = ? LIMIT 1");
965 $sth->execute([$ref_id, $owner_uid, $feed]);
967 if ($row = $sth->fetch())
968 $entry_int_id = $row['int_id'];
971 _debug("resulting RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
973 if (DB_TYPE
== "pgsql") {
974 $tsvector_combined = mb_substr($entry_title . ' ' .
975 preg_replace('/[<\?\:]/', ' ', strip_tags($entry_content)),
978 $tsvector_qpart = "tsvector_combined = to_tsvector(".$pdo->quote($feed_language).", ".$pdo->quote($tsvector_combined)."),";
981 $tsvector_qpart = "";
984 //_debug($tsvector_qpart);
986 $sth = $pdo->prepare("UPDATE ttrss_entries
990 content_hash = :content_hash,
992 num_comments = :num_comments,
993 plugin_data = :plugin_data,
998 $sth->execute([":title" => $entry_title,
999 ":content" => "$entry_content",
1000 ":content_hash" => $entry_current_hash,
1001 ":updated" => $entry_timestamp_fmt,
1002 ":num_comments" => (int)$num_comments,
1003 ":plugin_data" => $entry_plugin_data,
1004 ":author" => "$entry_author",
1005 ":lang" => $entry_language,
1009 $sth = $pdo->prepare("UPDATE ttrss_user_entries
1010 SET score = ? WHERE ref_id = ?");
1011 $sth->execute([$score, $ref_id]);
1013 if ($mark_unread_on_update) {
1014 _debug("article updated, marking unread as requested.", $debug_enabled);
1016 $sth = $pdo->prepare("UPDATE ttrss_user_entries
1017 SET last_read = null, unread = true WHERE ref_id = ?");
1018 $sth->execute([$ref_id]);
1022 _debug("assigning labels [other]...", $debug_enabled);
1024 foreach ($article_labels as $label) {
1025 Labels
::add_article($entry_ref_id, $label[1], $owner_uid);
1028 _debug("assigning labels [filters]...", $debug_enabled);
1030 RSSUtils
::assign_article_to_label_filters($entry_ref_id, $article_filters,
1031 $owner_uid, $article_labels);
1033 _debug("looking for enclosures...", $debug_enabled);
1037 $enclosures = array();
1039 $encs = $item->get_enclosures();
1041 if (is_array($encs)) {
1042 foreach ($encs as $e) {
1044 rewrite_relative_url($site_url, $e->link
),
1045 $e->type
, $e->length
, $e->title
, $e->width
, $e->height
);
1046 array_push($enclosures, $e_item);
1050 if ($cache_images && is_writable(CACHE_DIR
. '/images'))
1051 RSSUtils
::cache_enclosures($enclosures, $site_url, $debug_enabled);
1053 if ($debug_enabled) {
1054 _debug("article enclosures:", $debug_enabled);
1055 print_r($enclosures);
1058 $esth = $pdo->prepare("SELECT id FROM ttrss_enclosures
1059 WHERE content_url = ? AND post_id = ?");
1061 $usth = $pdo->prepare("INSERT INTO ttrss_enclosures
1062 (content_url, content_type, title, duration, post_id, width, height) VALUES
1063 (?, ?, ?, ?, ?, ?, ?)");
1065 foreach ($enclosures as $enc) {
1067 $enc_type = $enc[1];
1068 $enc_dur = (int)$enc[2];
1069 $enc_title = $enc[3];
1070 $enc_width = intval($enc[4]);
1071 $enc_height = intval($enc[5]);
1073 $esth->execute([$enc_url, $entry_ref_id]);
1075 if (!$esth->fetch()) {
1076 $usth->execute([$enc_url, $enc_type, (string)$enc_title, $enc_dur, $entry_ref_id, $enc_width, $enc_height]);
1080 // check for manual tags (we have to do it here since they're loaded from filters)
1082 foreach ($article_filters as $f) {
1083 if ($f["type"] == "tag") {
1085 $manual_tags = trim_array(explode(",", $f["param"]));
1087 foreach ($manual_tags as $tag) {
1088 if (tag_is_valid($tag)) {
1089 array_push($entry_tags, $tag);
1097 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
1098 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1100 $filtered_tags = array();
1101 $tags_to_cache = array();
1103 if ($entry_tags && is_array($entry_tags)) {
1104 foreach ($entry_tags as $tag) {
1105 if (array_search($tag, $boring_tags) === false) {
1106 array_push($filtered_tags, $tag);
1111 $filtered_tags = array_unique($filtered_tags);
1113 if ($debug_enabled) {
1114 _debug("filtered article tags:", $debug_enabled);
1115 print_r($filtered_tags);
1118 // Save article tags in the database
1120 if (count($filtered_tags) > 0) {
1122 $tsth = $pdo->prepare("SELECT id FROM ttrss_tags
1123 WHERE tag_name = ? AND post_int_id = ? AND
1124 owner_uid = ? LIMIT 1");
1126 $usth = $pdo->prepare("INSERT INTO ttrss_tags
1127 (owner_uid,tag_name,post_int_id)
1130 foreach ($filtered_tags as $tag) {
1132 $tag = sanitize_tag($tag);
1134 if (!tag_is_valid($tag)) continue;
1136 $tsth->execute([$tag, $entry_int_id, $owner_uid]);
1138 if (!$tsth->fetch()) {
1139 $usth->execute([$owner_uid, $tag, $entry_int_id]);
1142 array_push($tags_to_cache, $tag);
1145 /* update the cache */
1147 $tags_to_cache = array_unique($tags_to_cache);
1149 $tags_str = join(",", $tags_to_cache);
1151 $tsth = $pdo->prepare("UPDATE ttrss_user_entries
1152 SET tag_cache = ? WHERE ref_id = ?
1153 AND owner_uid = ?");
1154 $tsth->execute([$tags_str, $entry_ref_id, $owner_uid]);
1157 _debug("article processed", $debug_enabled);
1162 _debug("purging feed...", $debug_enabled);
1164 purge_feed($feed, 0, $debug_enabled);
1166 $sth = $pdo->prepare("UPDATE ttrss_feeds
1167 SET last_updated = NOW(), last_unconditional = NOW(), last_error = '' WHERE id = ?");
1168 $sth->execute([$feed]);
1172 $error_msg = mb_substr($rss->error(), 0, 245);
1174 _debug("fetch error: $error_msg", $debug_enabled);
1176 if (count($rss->errors()) > 1) {
1177 foreach ($rss->errors() as $error) {
1182 $sth = $pdo->prepare("UPDATE ttrss_feeds SET last_error = ?,
1183 last_updated = NOW(), last_unconditional = NOW() WHERE id = ?");
1184 $sth->execute([$error_msg, $feed]);
1190 _debug("done", $debug_enabled);
1195 static function cache_enclosures($enclosures, $site_url, $debug) {
1196 foreach ($enclosures as $enc) {
1198 if (preg_match("/(image|audio|video)/", $enc[1])) {
1200 $src = rewrite_relative_url($site_url, $enc[0]);
1202 $local_filename = CACHE_DIR
. "/images/" . sha1($src);
1204 if ($debug) _debug("cache_enclosures: downloading: $src to $local_filename");
1206 if (!file_exists($local_filename)) {
1207 $file_content = fetch_file_contents($src);
1209 if ($file_content && strlen($file_content) > MIN_CACHE_FILE_SIZE
) {
1210 file_put_contents($local_filename, $file_content);
1213 touch($local_filename);
1219 static function cache_media($html, $site_url, $debug) {
1220 libxml_use_internal_errors(true);
1222 $charset_hack = '<head>
1223 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1226 $doc = new DOMDocument();
1227 $doc->loadHTML($charset_hack . $html);
1228 $xpath = new DOMXPath($doc);
1230 $entries = $xpath->query('(//img[@src])|(//video/source[@src])|(//audio/source[@src])');
1232 foreach ($entries as $entry) {
1233 if ($entry->hasAttribute('src') && strpos($entry->getAttribute('src'), "data:") !== 0) {
1234 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1236 $local_filename = CACHE_DIR
. "/images/" . sha1($src);
1238 if ($debug) _debug("cache_media: downloading: $src to $local_filename");
1240 if (!file_exists($local_filename)) {
1241 $file_content = fetch_file_contents($src);
1243 if ($file_content && strlen($file_content) > MIN_CACHE_FILE_SIZE
) {
1244 file_put_contents($local_filename, $file_content);
1247 touch($local_filename);
1253 static function expire_error_log($debug) {
1254 if ($debug) _debug("Removing old error log entries...");
1258 if (DB_TYPE
== "pgsql") {
1259 $pdo->query("DELETE FROM ttrss_error_log
1260 WHERE created_at < NOW() - INTERVAL '7 days'");
1262 $pdo->query("DELETE FROM ttrss_error_log
1263 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1267 static function expire_lock_files($debug) {
1268 //if ($debug) _debug("Removing old lock files...");
1272 if (is_writable(LOCK_DIRECTORY
)) {
1273 $files = glob(LOCK_DIRECTORY
. "/*.lock");
1276 foreach ($files as $file) {
1277 if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
1285 if ($debug) _debug("Removed $num_deleted old lock files.");
1288 static function expire_cached_files($debug) {
1289 foreach (array("simplepie", "feeds", "images", "export", "upload") as $dir) {
1290 $cache_dir = CACHE_DIR
. "/$dir";
1292 // if ($debug) _debug("Expiring $cache_dir");
1296 if (is_writable($cache_dir)) {
1297 $files = glob("$cache_dir/*");
1300 foreach ($files as $file) {
1301 if (time() - filemtime($file) > 86400*CACHE_MAX_DAYS
) {
1310 if ($debug) _debug("$cache_dir: removed $num_deleted files.");
1315 * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1316 * Returns the url query as associative array
1318 * @param string query
1319 * @return array params
1321 static function convertUrlQuery($query) {
1322 $queryParts = explode('&', $query);
1326 foreach ($queryParts as $param) {
1327 $item = explode('=', $param);
1328 $params[$item[0]] = $item[1];
1334 static function get_article_filters($filters, $title, $content, $link, $author, $tags, &$matched_rules = false) {
1337 foreach ($filters as $filter) {
1338 $match_any_rule = $filter["match_any_rule"];
1339 $inverse = $filter["inverse"];
1340 $filter_match = false;
1342 foreach ($filter["rules"] as $rule) {
1344 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
1345 $rule_inverse = $rule["inverse"];
1350 switch ($rule["type"]) {
1352 $match = @preg_match
("/$reg_exp/iu", $title);
1355 // we don't need to deal with multiline regexps
1356 $content = preg_replace("/[\r\n\t]/", "", $content);
1358 $match = @preg_match
("/$reg_exp/iu", $content);
1361 // we don't need to deal with multiline regexps
1362 $content = preg_replace("/[\r\n\t]/", "", $content);
1364 $match = (@preg_match
("/$reg_exp/iu", $title) || @preg_match
("/$reg_exp/iu", $content));
1367 $match = @preg_match
("/$reg_exp/iu", $link);
1370 $match = @preg_match
("/$reg_exp/iu", $author);
1373 foreach ($tags as $tag) {
1374 if (@preg_match
("/$reg_exp/iu", $tag)) {
1382 if ($rule_inverse) $match = !$match;
1384 if ($match_any_rule) {
1386 $filter_match = true;
1390 $filter_match = $match;
1397 if ($inverse) $filter_match = !$filter_match;
1399 if ($filter_match) {
1400 if (is_array($matched_rules)) array_push($matched_rules, $rule);
1402 foreach ($filter["actions"] AS $action) {
1403 array_push($matches, $action);
1405 // if Stop action encountered, perform no further processing
1406 if (isset($action["type"]) && $action["type"] == "stop") return $matches;
1414 static function find_article_filter($filters, $filter_name) {
1415 foreach ($filters as $f) {
1416 if ($f["type"] == $filter_name) {
1423 static function find_article_filters($filters, $filter_name) {
1426 foreach ($filters as $f) {
1427 if ($f["type"] == $filter_name) {
1428 array_push($results, $f);
1434 static function calculate_article_score($filters) {
1437 foreach ($filters as $f) {
1438 if ($f["type"] == "score") {
1439 $score +
= $f["param"];
1445 static function labels_contains_caption($labels, $caption) {
1446 foreach ($labels as $label) {
1447 if ($label[1] == $caption) {
1455 static function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
1456 foreach ($filters as $f) {
1457 if ($f["type"] == "label") {
1458 if (!RSSUtils
::labels_contains_caption($article_labels, $f["param"])) {
1459 Labels
::add_article($id, $f["param"], $owner_uid);
1465 static function make_guid_from_title($title) {
1466 return preg_replace("/[ \"\',.:;]/", "-",
1467 mb_strtolower(strip_tags($title), 'utf-8'));
1470 static function cleanup_counters_cache($debug) {
1473 $res = $pdo->query("DELETE FROM ttrss_counters_cache
1474 WHERE feed_id > 0 AND
1475 (SELECT COUNT(id) FROM ttrss_feeds WHERE
1477 ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
1479 $frows = $res->rowCount();
1481 $res = $pdo->query("DELETE FROM ttrss_cat_counters_cache
1482 WHERE feed_id > 0 AND
1483 (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
1485 ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
1487 $crows = $res->rowCount();
1489 if ($debug) _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
1492 static function housekeeping_user($owner_uid) {
1493 $tmph = new PluginHost();
1495 load_user_plugins($owner_uid, $tmph);
1497 $tmph->run_hooks(PluginHost
::HOOK_HOUSE_KEEPING
, "hook_house_keeping", "");
1500 static function housekeeping_common($debug) {
1501 RSSUtils
::expire_cached_files($debug);
1502 RSSUtils
::expire_lock_files($debug);
1503 RSSUtils
::expire_error_log($debug);
1505 $count = RSSUtils
::update_feedbrowser_cache();
1506 _debug("Feedbrowser updated, $count feeds processed.");
1508 Article
::purge_orphans( true);
1509 RSSUtils
::cleanup_counters_cache($debug);
1511 //$rc = cleanup_tags( 14, 50000);
1512 //_debug("Cleaned $rc cached tags.");
1514 PluginHost
::getInstance()->run_hooks(PluginHost
::HOOK_HOUSE_KEEPING
, "hook_house_keeping", "");
1517 static function check_feed_favicon($site_url, $feed) {
1518 # print "FAVICON [$site_url]: $favicon_url\n";
1520 $icon_file = ICONS_DIR
. "/$feed.ico";
1522 if (!file_exists($icon_file)) {
1523 $favicon_url = get_favicon_url($site_url);
1526 // Limiting to "image" type misses those served with text/plain
1527 $contents = fetch_file_contents($favicon_url); // , "image");
1530 // Crude image type matching.
1531 // Patterns gleaned from the file(1) source code.
1532 if (preg_match('/^\x00\x00\x01\x00/', $contents)) {
1533 // 0 string \000\000\001\000 MS Windows icon resource
1534 //error_log("check_feed_favicon: favicon_url=$favicon_url isa MS Windows icon resource");
1536 elseif (preg_match('/^GIF8/', $contents)) {
1537 // 0 string GIF8 GIF image data
1538 //error_log("check_feed_favicon: favicon_url=$favicon_url isa GIF image");
1540 elseif (preg_match('/^\x89PNG\x0d\x0a\x1a\x0a/', $contents)) {
1541 // 0 string \x89PNG\x0d\x0a\x1a\x0a PNG image data
1542 //error_log("check_feed_favicon: favicon_url=$favicon_url isa PNG image");
1544 elseif (preg_match('/^\xff\xd8/', $contents)) {
1545 // 0 beshort 0xffd8 JPEG image data
1546 //error_log("check_feed_favicon: favicon_url=$favicon_url isa JPG image");
1548 elseif (preg_match('/^BM/', $contents)) {
1549 // 0 string BM PC bitmap (OS2, Windows BMP files)
1550 //error_log("check_feed_favicon, favicon_url=$favicon_url isa BMP image");
1553 //error_log("check_feed_favicon: favicon_url=$favicon_url isa UNKNOWN type");
1559 $fp = @fopen
($icon_file, "w");
1562 fwrite($fp, $contents);
1564 chmod($icon_file, 0644);