rssfuncs.php 44 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514
  1. <?php
  2. define_default('DAEMON_UPDATE_LOGIN_LIMIT', 30);
  3. define_default('DAEMON_FEED_LIMIT', 500);
  4. define_default('DAEMON_SLEEP_INTERVAL', 120);
  5. define_default('_MIN_CACHE_IMAGE_SIZE', 1024);
  6. function calculate_article_hash($article, $pluginhost) {
  7. $tmp = "";
  8. foreach ($article as $k => $v) {
  9. if ($k != "feed" && isset($v)) {
  10. $tmp .= sha1("$k:" . (is_array($v) ? implode(",", $v) : $v));
  11. }
  12. }
  13. return sha1(implode(",", $pluginhost->get_plugin_names()) . $tmp);
  14. }
  15. function update_feedbrowser_cache() {
  16. $result = db_query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
  17. FROM ttrss_feeds WHERE (SELECT COUNT(id) = 0 FROM ttrss_feeds AS tf
  18. WHERE tf.feed_url = ttrss_feeds.feed_url
  19. AND (private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%'))
  20. GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
  21. db_query("BEGIN");
  22. db_query("DELETE FROM ttrss_feedbrowser_cache");
  23. $count = 0;
  24. while ($line = db_fetch_assoc($result)) {
  25. $subscribers = db_escape_string($line["subscribers"]);
  26. $feed_url = db_escape_string($line["feed_url"]);
  27. $title = db_escape_string($line["title"]);
  28. $site_url = db_escape_string($line["site_url"]);
  29. $tmp_result = db_query("SELECT subscribers FROM
  30. ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
  31. if (db_num_rows($tmp_result) == 0) {
  32. db_query("INSERT INTO ttrss_feedbrowser_cache
  33. (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
  34. '$site_url', '$title', '$subscribers')");
  35. ++$count;
  36. }
  37. }
  38. db_query("COMMIT");
  39. return $count;
  40. }
  41. /**
  42. * Update a feed batch.
  43. * Used by daemons to update n feeds by run.
  44. * Only update feed needing a update, and not being processed
  45. * by another process.
  46. *
  47. * @param mixed $link Database link
  48. * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
  49. * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
  50. * @param boolean $debug Set to false to disable debug output. Default to true.
  51. * @return void
  52. */
  53. function update_daemon_common($limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
  54. // Process all other feeds using last_updated and interval parameters
  55. $schema_version = get_schema_version();
  56. if ($schema_version != SCHEMA_VERSION) {
  57. die("Schema version is wrong, please upgrade the database.\n");
  58. }
  59. define('PREFS_NO_CACHE', true);
  60. // Test if the user has loggued in recently. If not, it does not update its feeds.
  61. if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
  62. if (DB_TYPE == "pgsql") {
  63. $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
  64. } else {
  65. $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
  66. }
  67. } else {
  68. $login_thresh_qpart = "";
  69. }
  70. // Test if the feed need a update (update interval exceeded).
  71. if (DB_TYPE == "pgsql") {
  72. $update_limit_qpart = "AND ((
  73. ttrss_feeds.update_interval = 0
  74. AND ttrss_user_prefs.value != '-1'
  75. AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
  76. ) OR (
  77. ttrss_feeds.update_interval > 0
  78. AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
  79. ) OR (ttrss_feeds.last_updated IS NULL
  80. AND ttrss_user_prefs.value != '-1')
  81. OR (last_updated = '1970-01-01 00:00:00'
  82. AND ttrss_user_prefs.value != '-1'))";
  83. } else {
  84. $update_limit_qpart = "AND ((
  85. ttrss_feeds.update_interval = 0
  86. AND ttrss_user_prefs.value != '-1'
  87. AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
  88. ) OR (
  89. ttrss_feeds.update_interval > 0
  90. AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
  91. ) OR (ttrss_feeds.last_updated IS NULL
  92. AND ttrss_user_prefs.value != '-1')
  93. OR (last_updated = '1970-01-01 00:00:00'
  94. AND ttrss_user_prefs.value != '-1'))";
  95. }
  96. // Test if feed is currently being updated by another process.
  97. if (DB_TYPE == "pgsql") {
  98. $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
  99. } else {
  100. $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
  101. }
  102. // Test if there is a limit to number of updated feeds
  103. $query_limit = "";
  104. if($limit) $query_limit = sprintf("LIMIT %d", $limit);
  105. $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
  106. FROM
  107. ttrss_feeds, ttrss_users, ttrss_user_prefs
  108. WHERE
  109. ttrss_feeds.owner_uid = ttrss_users.id
  110. AND ttrss_user_prefs.profile IS NULL
  111. AND ttrss_users.id = ttrss_user_prefs.owner_uid
  112. AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
  113. $login_thresh_qpart $update_limit_qpart
  114. $updstart_thresh_qpart
  115. ORDER BY last_updated $query_limit";
  116. // We search for feed needing update.
  117. $result = db_query($query);
  118. if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
  119. // Here is a little cache magic in order to minimize risk of double feed updates.
  120. $feeds_to_update = array();
  121. while ($line = db_fetch_assoc($result)) {
  122. array_push($feeds_to_update, db_escape_string($line['feed_url']));
  123. }
  124. // We update the feed last update started date before anything else.
  125. // There is no lag due to feed contents downloads
  126. // It prevent an other process to update the same feed.
  127. if(count($feeds_to_update) > 0) {
  128. $feeds_quoted = array();
  129. foreach ($feeds_to_update as $feed) {
  130. array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
  131. }
  132. db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
  133. WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
  134. }
  135. $nf = 0;
  136. $bstarted = microtime(true);
  137. // For each feed, we call the feed update function.
  138. foreach ($feeds_to_update as $feed) {
  139. if($debug) _debug("Base feed: $feed");
  140. //update_rss_feed($line["id"], true);
  141. // since we have the data cached, we can deal with other feeds with the same url
  142. $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
  143. FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
  144. ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
  145. ttrss_users.id = ttrss_user_prefs.owner_uid AND
  146. ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
  147. ttrss_user_prefs.profile IS NULL AND
  148. feed_url = '".db_escape_string($feed)."' AND
  149. (ttrss_feeds.update_interval > 0 OR
  150. ttrss_user_prefs.value != '-1')
  151. $login_thresh_qpart
  152. ORDER BY ttrss_feeds.id $query_limit");
  153. if (db_num_rows($tmp_result) > 0) {
  154. $rss = false;
  155. while ($tline = db_fetch_assoc($tmp_result)) {
  156. if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
  157. $fstarted = microtime(true);
  158. $rss = update_rss_feed($tline["id"], true, false);
  159. _debug_suppress(false);
  160. _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
  161. ++$nf;
  162. }
  163. }
  164. }
  165. if ($nf > 0) {
  166. _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
  167. microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
  168. }
  169. require_once "digest.php";
  170. // Send feed digests by email if needed.
  171. send_headlines_digests($debug);
  172. return $nf;
  173. } // function update_daemon_common
  174. // this is used when subscribing
  175. function set_basic_feed_info($feed) {
  176. $feed = db_escape_string($feed);
  177. $result = db_query("SELECT feed_url,auth_pass,auth_pass_encrypted
  178. FROM ttrss_feeds WHERE id = '$feed'");
  179. $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
  180. 0, "auth_pass_encrypted"));
  181. $auth_login = db_fetch_result($result, 0, "auth_login");
  182. $auth_pass = db_fetch_result($result, 0, "auth_pass");
  183. if ($auth_pass_encrypted) {
  184. require_once "crypt.php";
  185. $auth_pass = decrypt_string($auth_pass);
  186. }
  187. $fetch_url = db_fetch_result($result, 0, "feed_url");
  188. $feed_data = fetch_file_contents($fetch_url, false,
  189. $auth_login, $auth_pass, false,
  190. FEED_FETCH_TIMEOUT_TIMEOUT,
  191. 0);
  192. global $fetch_curl_used;
  193. if (!$fetch_curl_used) {
  194. $tmp = @gzdecode($feed_data);
  195. if ($tmp) $feed_data = $tmp;
  196. }
  197. $feed_data = trim($feed_data);
  198. $rss = new FeedParser($feed_data);
  199. $rss->init();
  200. if (!$rss->error()) {
  201. $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
  202. $registered_title = db_fetch_result($result, 0, "title");
  203. $orig_site_url = db_fetch_result($result, 0, "site_url");
  204. $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
  205. $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
  206. if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
  207. db_query("UPDATE ttrss_feeds SET
  208. title = '$feed_title' WHERE id = '$feed'");
  209. }
  210. if ($site_url && $orig_site_url != $site_url) {
  211. db_query("UPDATE ttrss_feeds SET
  212. site_url = '$site_url' WHERE id = '$feed'");
  213. }
  214. }
  215. }
  216. // ignore_daemon is not used
  217. function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
  218. $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
  219. _debug_suppress(!$debug_enabled);
  220. _debug("start", $debug_enabled);
  221. $result = db_query("SELECT title FROM ttrss_feeds
  222. WHERE id = '$feed'");
  223. $title = db_fetch_result($result, 0, "title");
  224. // feed was batch-subscribed or something, we need to get basic info
  225. // this is not optimal currently as it fetches stuff separately TODO: optimize
  226. if ($title == "[Unknown]") {
  227. _debug("setting basic feed info for $feed...");
  228. set_basic_feed_info($feed);
  229. }
  230. $result = db_query("SELECT id,update_interval,auth_login,
  231. feed_url,auth_pass,cache_images,
  232. mark_unread_on_update, owner_uid,
  233. pubsub_state, auth_pass_encrypted,
  234. feed_language,
  235. (SELECT max(date_entered) FROM
  236. ttrss_entries, ttrss_user_entries where ref_id = id AND feed_id = '$feed') AS last_article_timestamp
  237. FROM ttrss_feeds WHERE id = '$feed'");
  238. if (db_num_rows($result) == 0) {
  239. _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
  240. return false;
  241. }
  242. $last_article_timestamp = @strtotime(db_fetch_result($result, 0, "last_article_timestamp"));
  243. if (defined('_DISABLE_HTTP_304'))
  244. $last_article_timestamp = 0;
  245. $owner_uid = db_fetch_result($result, 0, "owner_uid");
  246. $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
  247. 0, "mark_unread_on_update"));
  248. $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
  249. $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
  250. 0, "auth_pass_encrypted"));
  251. db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
  252. WHERE id = '$feed'");
  253. $auth_login = db_fetch_result($result, 0, "auth_login");
  254. $auth_pass = db_fetch_result($result, 0, "auth_pass");
  255. if ($auth_pass_encrypted) {
  256. require_once "crypt.php";
  257. $auth_pass = decrypt_string($auth_pass);
  258. }
  259. $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
  260. $fetch_url = db_fetch_result($result, 0, "feed_url");
  261. $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
  262. if (!$feed_language) $feed_language = 'english';
  263. $feed = db_escape_string($feed);
  264. $date_feed_processed = date('Y-m-d H:i');
  265. $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
  266. $pluginhost = new PluginHost();
  267. $pluginhost->set_debug($debug_enabled);
  268. $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
  269. $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
  270. $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
  271. $pluginhost->load_data();
  272. if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
  273. _debug("using previously initialized parser object");
  274. } else {
  275. $rss_hash = false;
  276. $force_refetch = isset($_REQUEST["force_refetch"]);
  277. foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
  278. $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, $last_article_timestamp, $auth_login, $auth_pass);
  279. }
  280. // try cache
  281. if (!$feed_data &&
  282. file_exists($cache_filename) &&
  283. is_readable($cache_filename) &&
  284. !$auth_login && !$auth_pass &&
  285. filemtime($cache_filename) > time() - 30) {
  286. _debug("using local cache [$cache_filename].", $debug_enabled);
  287. @$feed_data = file_get_contents($cache_filename);
  288. if ($feed_data) {
  289. $rss_hash = sha1($feed_data);
  290. }
  291. } else {
  292. _debug("local cache will not be used for this feed", $debug_enabled);
  293. }
  294. // fetch feed from source
  295. if (!$feed_data) {
  296. _debug("fetching [$fetch_url]...", $debug_enabled);
  297. _debug("If-Modified-Since: ".gmdate('D, d M Y H:i:s \G\M\T', $last_article_timestamp), $debug_enabled);
  298. $feed_data = fetch_file_contents($fetch_url, false,
  299. $auth_login, $auth_pass, false,
  300. $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
  301. $force_refetch ? 0 : $last_article_timestamp);
  302. global $fetch_curl_used;
  303. if (!$fetch_curl_used) {
  304. $tmp = @gzdecode($feed_data);
  305. if ($tmp) $feed_data = $tmp;
  306. }
  307. $feed_data = trim($feed_data);
  308. _debug("fetch done.", $debug_enabled);
  309. // cache vanilla feed data for re-use
  310. if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
  311. $new_rss_hash = sha1($feed_data);
  312. if ($new_rss_hash != $rss_hash) {
  313. _debug("saving $cache_filename", $debug_enabled);
  314. @file_put_contents($cache_filename, $feed_data);
  315. }
  316. }
  317. }
  318. if (!$feed_data) {
  319. global $fetch_last_error;
  320. global $fetch_last_error_code;
  321. _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
  322. $error_escaped = '';
  323. // If-Modified-Since
  324. if ($fetch_last_error_code != 304) {
  325. $error_escaped = db_escape_string($fetch_last_error);
  326. } else {
  327. _debug("source claims data not modified, nothing to do.", $debug_enabled);
  328. }
  329. db_query(
  330. "UPDATE ttrss_feeds SET last_error = '$error_escaped',
  331. last_updated = NOW() WHERE id = '$feed'");
  332. return;
  333. }
  334. }
  335. foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
  336. $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
  337. }
  338. // set last update to now so if anything *simplepie* crashes later we won't be
  339. // continuously failing on the same feed
  340. //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
  341. if (!$rss) {
  342. $rss = new FeedParser($feed_data);
  343. $rss->init();
  344. }
  345. // print_r($rss);
  346. $feed = db_escape_string($feed);
  347. if (!$rss->error()) {
  348. // We use local pluginhost here because we need to load different per-user feed plugins
  349. $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
  350. _debug("language: $feed_language", $debug_enabled);
  351. _debug("processing feed data...", $debug_enabled);
  352. // db_query("BEGIN");
  353. if (DB_TYPE == "pgsql") {
  354. $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
  355. } else {
  356. $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
  357. }
  358. $result = db_query("SELECT owner_uid,favicon_avg_color,
  359. (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
  360. favicon_needs_check
  361. FROM ttrss_feeds WHERE id = '$feed'");
  362. $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
  363. "favicon_needs_check"));
  364. $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
  365. $owner_uid = db_fetch_result($result, 0, "owner_uid");
  366. $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
  367. _debug("site_url: $site_url", $debug_enabled);
  368. _debug("feed_title: " . $rss->get_title(), $debug_enabled);
  369. if ($favicon_needs_check || $force_refetch) {
  370. /* terrible hack: if we crash on floicon shit here, we won't check
  371. * the icon avgcolor again (unless the icon got updated) */
  372. $favicon_file = ICONS_DIR . "/$feed.ico";
  373. $favicon_modified = @filemtime($favicon_file);
  374. _debug("checking favicon...", $debug_enabled);
  375. check_feed_favicon($site_url, $feed);
  376. $favicon_modified_new = @filemtime($favicon_file);
  377. if ($favicon_modified_new > $favicon_modified)
  378. $favicon_avg_color = '';
  379. if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
  380. require_once "colors.php";
  381. db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
  382. id = '$feed'");
  383. $favicon_color = db_escape_string(
  384. calculate_avg_color($favicon_file));
  385. $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
  386. } else if ($favicon_avg_color == 'fail') {
  387. _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
  388. }
  389. db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
  390. $favicon_colorstring
  391. WHERE id = '$feed'");
  392. }
  393. _debug("loading filters & labels...", $debug_enabled);
  394. $filters = load_filters($feed, $owner_uid);
  395. _debug("" . count($filters) . " filters loaded.", $debug_enabled);
  396. $items = $rss->get_items();
  397. if (!is_array($items)) {
  398. _debug("no articles found.", $debug_enabled);
  399. db_query("UPDATE ttrss_feeds
  400. SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
  401. return; // no articles
  402. }
  403. if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
  404. _debug("checking for PUSH hub...", $debug_enabled);
  405. $feed_hub_url = false;
  406. $links = $rss->get_links('hub');
  407. if ($links && is_array($links)) {
  408. foreach ($links as $l) {
  409. $feed_hub_url = $l;
  410. break;
  411. }
  412. }
  413. _debug("feed hub url: $feed_hub_url", $debug_enabled);
  414. $feed_self_url = $fetch_url;
  415. $links = $rss->get_links('self');
  416. if ($links && is_array($links)) {
  417. foreach ($links as $l) {
  418. $feed_self_url = $l;
  419. break;
  420. }
  421. }
  422. _debug("feed self url = $feed_self_url");
  423. if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
  424. !ini_get("open_basedir")) {
  425. require_once 'lib/pubsubhubbub/subscriber.php';
  426. $callback_url = get_self_url_prefix() .
  427. "/public.php?op=pubsub&id=$feed";
  428. $s = new Subscriber($feed_hub_url, $callback_url);
  429. $rc = $s->subscribe($feed_self_url);
  430. _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
  431. db_query("UPDATE ttrss_feeds SET pubsub_state = 1
  432. WHERE id = '$feed'");
  433. }
  434. }
  435. _debug("processing articles...", $debug_enabled);
  436. $tstart = time();
  437. foreach ($items as $item) {
  438. if ($_REQUEST['xdebug'] == 3) {
  439. print_r($item);
  440. }
  441. if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
  442. _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
  443. break;
  444. }
  445. $entry_guid = $item->get_id();
  446. if (!$entry_guid) $entry_guid = $item->get_link();
  447. if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
  448. if (!$entry_guid) continue;
  449. $entry_guid = "$owner_uid,$entry_guid";
  450. $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
  451. _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
  452. $entry_timestamp = "";
  453. $entry_timestamp = $item->get_date();
  454. _debug("orig date: " . $item->get_date(), $debug_enabled);
  455. if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
  456. $entry_timestamp = time();
  457. }
  458. $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
  459. _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
  460. // $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
  461. // $entry_title = decode_numeric_entities($entry_title);
  462. $entry_title = $item->get_title();
  463. $entry_link = rewrite_relative_url($site_url, $item->get_link());
  464. _debug("title $entry_title", $debug_enabled);
  465. _debug("link $entry_link", $debug_enabled);
  466. if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
  467. $entry_content = $item->get_content();
  468. if (!$entry_content) $entry_content = $item->get_description();
  469. if ($_REQUEST["xdebug"] == 2) {
  470. print "content: ";
  471. print $entry_content;
  472. print "\n";
  473. }
  474. $entry_comments = $item->get_comments_url();
  475. $entry_author = $item->get_author();
  476. $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
  477. $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
  478. $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
  479. $num_comments = (int) $item->get_comments_count();
  480. _debug("author $entry_author", $debug_enabled);
  481. _debug("num_comments: $num_comments", $debug_enabled);
  482. _debug("looking for tags...", $debug_enabled);
  483. // parse <category> entries into tags
  484. $additional_tags = array();
  485. $additional_tags_src = $item->get_categories();
  486. if (is_array($additional_tags_src)) {
  487. foreach ($additional_tags_src as $tobj) {
  488. array_push($additional_tags, $tobj);
  489. }
  490. }
  491. $entry_tags = array_unique($additional_tags);
  492. for ($i = 0; $i < count($entry_tags); $i++)
  493. $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
  494. _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
  495. _debug("done collecting data.", $debug_enabled);
  496. $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
  497. WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
  498. if (db_num_rows($result) != 0) {
  499. $base_entry_id = db_fetch_result($result, 0, "id");
  500. $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
  501. $article_labels = get_article_labels($base_entry_id, $owner_uid);
  502. $entry_language = db_fetch_result($result, 0, "lang");
  503. } else {
  504. $base_entry_id = false;
  505. $entry_stored_hash = "";
  506. $article_labels = array();
  507. $entry_language = "";
  508. }
  509. $article = array("owner_uid" => $owner_uid, // read only
  510. "guid" => $entry_guid, // read only
  511. "guid_hashed" => $entry_guid_hashed, // read only
  512. "title" => $entry_title,
  513. "content" => $entry_content,
  514. "link" => $entry_link,
  515. "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
  516. "tags" => $entry_tags,
  517. "author" => $entry_author,
  518. "force_catchup" => false, // ugly hack for the time being
  519. "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
  520. "language" => $entry_language,
  521. "feed" => array("id" => $feed,
  522. "fetch_url" => $fetch_url,
  523. "site_url" => $site_url)
  524. );
  525. $entry_plugin_data = "";
  526. $entry_current_hash = calculate_article_hash($article, $pluginhost);
  527. _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
  528. if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
  529. _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
  530. // we keep encountering the entry in feeds, so we need to
  531. // update date_updated column so that we don't get horrible
  532. // dupes when the entry gets purged and reinserted again e.g.
  533. // in the case of SLOW SLOW OMG SLOW updating feeds
  534. $base_entry_id = db_fetch_result($result, 0, "id");
  535. db_query("UPDATE ttrss_entries SET date_updated = NOW()
  536. WHERE id = '$base_entry_id'");
  537. // if we allow duplicate posts, we have to continue to
  538. // create the user entries for this feed
  539. if (!get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
  540. continue;
  541. }
  542. }
  543. _debug("hash differs, applying plugin filters:", $debug_enabled);
  544. foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
  545. _debug("... " . get_class($plugin), $debug_enabled);
  546. $start = microtime(true);
  547. $article = $plugin->hook_article_filter($article);
  548. _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
  549. $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
  550. }
  551. $entry_plugin_data = db_escape_string($entry_plugin_data);
  552. _debug("plugin data: $entry_plugin_data", $debug_enabled);
  553. // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
  554. if (DB_TYPE == "mysql") {
  555. foreach ($article as $k => $v) {
  556. // i guess we'll have to take the risk of 4byte unicode labels & tags here
  557. if (!is_array($article[$k])) {
  558. $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
  559. }
  560. }
  561. }
  562. /* Collect article tags here so we could filter by them: */
  563. $article_filters = get_article_filters($filters, $article["title"],
  564. $article["content"], $article["link"], 0, $article["author"],
  565. $article["tags"]);
  566. if ($debug_enabled) {
  567. _debug("article filters: ", $debug_enabled);
  568. if (count($article_filters) != 0) {
  569. print_r($article_filters);
  570. }
  571. }
  572. $plugin_filter_names = find_article_filters($article_filters, "plugin");
  573. $plugin_filter_actions = $pluginhost->get_filter_actions();
  574. if (count($plugin_filter_names) > 0) {
  575. _debug("applying plugin filter actions...", $debug_enabled);
  576. foreach ($plugin_filter_names as $pfn) {
  577. list($pfclass,$pfaction) = explode(":", $pfn["param"]);
  578. if (isset($plugin_filter_actions[$pfclass])) {
  579. $plugin = $pluginhost->get_plugin($pfclass);
  580. _debug("... $pfclass: $pfaction", $debug_enabled);
  581. if ($plugin) {
  582. $start = microtime(true);
  583. $article = $plugin->hook_article_filter_action($article, $pfaction);
  584. _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
  585. } else {
  586. _debug("??? $pfclass: plugin object not found.");
  587. }
  588. } else {
  589. _debug("??? $pfclass: filter plugin not registered.");
  590. }
  591. }
  592. }
  593. $entry_tags = $article["tags"];
  594. $entry_guid = db_escape_string($entry_guid);
  595. $entry_title = db_escape_string($article["title"]);
  596. $entry_author = db_escape_string($article["author"]);
  597. $entry_link = db_escape_string($article["link"]);
  598. $entry_content = $article["content"]; // escaped below
  599. $entry_force_catchup = $article["force_catchup"];
  600. $article_labels = $article["labels"];
  601. $entry_score_modifier = (int) $article["score_modifier"];
  602. $entry_language = db_escape_string($article["language"]);
  603. if ($debug_enabled) {
  604. _debug("article labels:", $debug_enabled);
  605. print_r($article_labels);
  606. }
  607. _debug("force catchup: $entry_force_catchup");
  608. if ($cache_images && is_writable(CACHE_DIR . '/images'))
  609. cache_images($entry_content, $site_url, $debug_enabled);
  610. $entry_content = db_escape_string($entry_content, false);
  611. //db_query("BEGIN");
  612. $result = db_query("SELECT id FROM ttrss_entries
  613. WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
  614. if (db_num_rows($result) == 0) {
  615. _debug("base guid [$entry_guid] not found", $debug_enabled);
  616. // base post entry does not exist, create it
  617. $result = db_query(
  618. "INSERT INTO ttrss_entries
  619. (title,
  620. guid,
  621. link,
  622. updated,
  623. content,
  624. content_hash,
  625. no_orig_date,
  626. date_updated,
  627. date_entered,
  628. comments,
  629. num_comments,
  630. plugin_data,
  631. lang,
  632. author)
  633. VALUES
  634. ('$entry_title',
  635. '$entry_guid_hashed',
  636. '$entry_link',
  637. '$entry_timestamp_fmt',
  638. '$entry_content',
  639. '$entry_current_hash',
  640. false,
  641. NOW(),
  642. '$date_feed_processed',
  643. '$entry_comments',
  644. '$num_comments',
  645. '$entry_plugin_data',
  646. '$entry_language',
  647. '$entry_author')");
  648. } else {
  649. $base_entry_id = db_fetch_result($result, 0, "id");
  650. }
  651. // now it should exist, if not - bad luck then
  652. $result = db_query("SELECT id FROM ttrss_entries
  653. WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
  654. $entry_ref_id = 0;
  655. $entry_int_id = 0;
  656. if (db_num_rows($result) == 1) {
  657. _debug("base guid found, checking for user record", $debug_enabled);
  658. $ref_id = db_fetch_result($result, 0, "id");
  659. $entry_ref_id = $ref_id;
  660. /* $stored_guid = db_fetch_result($result, 0, "guid");
  661. if ($stored_guid != $entry_guid_hashed) {
  662. if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
  663. db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
  664. id = '$ref_id'");
  665. } */
  666. // check for user post link to main table
  667. // do we allow duplicate posts with same GUID in different feeds?
  668. if (get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
  669. $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
  670. } else {
  671. $dupcheck_qpart = "";
  672. }
  673. if (find_article_filter($article_filters, "filter")) {
  674. //db_query("COMMIT"); // close transaction in progress
  675. continue;
  676. }
  677. $score = calculate_article_score($article_filters) + $entry_score_modifier;
  678. _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
  679. $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
  680. ref_id = '$ref_id' AND owner_uid = '$owner_uid'
  681. $dupcheck_qpart";
  682. // if ($_REQUEST["xdebug"]) print "$query\n";
  683. $result = db_query($query);
  684. // okay it doesn't exist - create user entry
  685. if (db_num_rows($result) == 0) {
  686. _debug("user record not found, creating...", $debug_enabled);
  687. if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
  688. $unread = 'true';
  689. $last_read_qpart = 'NULL';
  690. } else {
  691. $unread = 'false';
  692. $last_read_qpart = 'NOW()';
  693. }
  694. if (find_article_filter($article_filters, 'mark') || $score > 1000) {
  695. $marked = 'true';
  696. } else {
  697. $marked = 'false';
  698. }
  699. if (find_article_filter($article_filters, 'publish')) {
  700. $published = 'true';
  701. } else {
  702. $published = 'false';
  703. }
  704. // N-grams
  705. /* if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
  706. $result = db_query("SELECT COUNT(*) AS similar FROM
  707. ttrss_entries,ttrss_user_entries
  708. WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
  709. AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
  710. AND owner_uid = $owner_uid");
  711. $ngram_similar = db_fetch_result($result, 0, "similar");
  712. _debug("N-gram similar results: $ngram_similar", $debug_enabled);
  713. if ($ngram_similar > 0) {
  714. $unread = 'false';
  715. }
  716. } */
  717. $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
  718. $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
  719. $result = db_query(
  720. "INSERT INTO ttrss_user_entries
  721. (ref_id, owner_uid, feed_id, unread, last_read, marked,
  722. published, score, tag_cache, label_cache, uuid,
  723. last_marked, last_published)
  724. VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
  725. $last_read_qpart, $marked, $published, '$score', '', '',
  726. '', $last_marked, $last_published)");
  727. if (PUBSUBHUBBUB_HUB && $published == 'true') {
  728. $rss_link = get_self_url_prefix() .
  729. "/public.php?op=rss&id=-2&key=" .
  730. get_feed_access_key(-2, false, $owner_uid);
  731. $p = new Publisher(PUBSUBHUBBUB_HUB);
  732. /* $pubsub_result = */ $p->publish_update($rss_link);
  733. }
  734. $result = db_query(
  735. "SELECT int_id FROM ttrss_user_entries WHERE
  736. ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
  737. feed_id = '$feed' LIMIT 1");
  738. if (db_num_rows($result) == 1) {
  739. $entry_int_id = db_fetch_result($result, 0, "int_id");
  740. }
  741. } else {
  742. _debug("user record FOUND", $debug_enabled);
  743. $entry_ref_id = db_fetch_result($result, 0, "ref_id");
  744. $entry_int_id = db_fetch_result($result, 0, "int_id");
  745. }
  746. _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
  747. if (DB_TYPE == "pgsql") {
  748. $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags($entry_content),
  749. 0, 1000000));
  750. $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
  751. } else {
  752. $tsvector_qpart = "";
  753. }
  754. db_query("UPDATE ttrss_entries
  755. SET title = '$entry_title',
  756. content = '$entry_content',
  757. content_hash = '$entry_current_hash',
  758. updated = '$entry_timestamp_fmt',
  759. $tsvector_qpart
  760. num_comments = '$num_comments',
  761. plugin_data = '$entry_plugin_data',
  762. author = '$entry_author',
  763. lang = '$entry_language'
  764. WHERE id = '$ref_id'");
  765. // update aux data
  766. db_query("UPDATE ttrss_user_entries
  767. SET score = '$score' WHERE ref_id = '$ref_id'");
  768. if ($mark_unread_on_update) {
  769. db_query("UPDATE ttrss_user_entries
  770. SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
  771. }
  772. }
  773. //db_query("COMMIT");
  774. _debug("assigning labels [other]...", $debug_enabled);
  775. foreach ($article_labels as $label) {
  776. label_add_article($entry_ref_id, $label[1], $owner_uid);
  777. }
  778. _debug("assigning labels [filters]...", $debug_enabled);
  779. assign_article_to_label_filters($entry_ref_id, $article_filters,
  780. $owner_uid, $article_labels);
  781. _debug("looking for enclosures...", $debug_enabled);
  782. // enclosures
  783. $enclosures = array();
  784. $encs = $item->get_enclosures();
  785. if (is_array($encs)) {
  786. foreach ($encs as $e) {
  787. $e_item = array(
  788. $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
  789. array_push($enclosures, $e_item);
  790. }
  791. }
  792. if ($debug_enabled) {
  793. _debug("article enclosures:", $debug_enabled);
  794. print_r($enclosures);
  795. }
  796. //db_query("BEGIN");
  797. // debugging
  798. // db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
  799. foreach ($enclosures as $enc) {
  800. $enc_url = db_escape_string($enc[0]);
  801. $enc_type = db_escape_string($enc[1]);
  802. $enc_dur = db_escape_string($enc[2]);
  803. $enc_title = db_escape_string($enc[3]);
  804. $enc_width = intval($enc[4]);
  805. $enc_height = intval($enc[5]);
  806. $result = db_query("SELECT id FROM ttrss_enclosures
  807. WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
  808. if (db_num_rows($result) == 0) {
  809. db_query("INSERT INTO ttrss_enclosures
  810. (content_url, content_type, title, duration, post_id, width, height) VALUES
  811. ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
  812. }
  813. }
  814. //db_query("COMMIT");
  815. // check for manual tags (we have to do it here since they're loaded from filters)
  816. foreach ($article_filters as $f) {
  817. if ($f["type"] == "tag") {
  818. $manual_tags = trim_array(explode(",", $f["param"]));
  819. foreach ($manual_tags as $tag) {
  820. if (tag_is_valid($tag)) {
  821. array_push($entry_tags, $tag);
  822. }
  823. }
  824. }
  825. }
  826. // Skip boring tags
  827. $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
  828. 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
  829. $filtered_tags = array();
  830. $tags_to_cache = array();
  831. if ($entry_tags && is_array($entry_tags)) {
  832. foreach ($entry_tags as $tag) {
  833. if (array_search($tag, $boring_tags) === false) {
  834. array_push($filtered_tags, $tag);
  835. }
  836. }
  837. }
  838. $filtered_tags = array_unique($filtered_tags);
  839. if ($debug_enabled) {
  840. _debug("filtered article tags:", $debug_enabled);
  841. print_r($filtered_tags);
  842. }
  843. // Save article tags in the database
  844. if (count($filtered_tags) > 0) {
  845. //db_query("BEGIN");
  846. foreach ($filtered_tags as $tag) {
  847. $tag = sanitize_tag($tag);
  848. $tag = db_escape_string($tag);
  849. if (!tag_is_valid($tag)) continue;
  850. $result = db_query("SELECT id FROM ttrss_tags
  851. WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
  852. owner_uid = '$owner_uid' LIMIT 1");
  853. if ($result && db_num_rows($result) == 0) {
  854. db_query("INSERT INTO ttrss_tags
  855. (owner_uid,tag_name,post_int_id)
  856. VALUES ('$owner_uid','$tag', '$entry_int_id')");
  857. }
  858. array_push($tags_to_cache, $tag);
  859. }
  860. /* update the cache */
  861. $tags_to_cache = array_unique($tags_to_cache);
  862. $tags_str = db_escape_string(join(",", $tags_to_cache));
  863. db_query("UPDATE ttrss_user_entries
  864. SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
  865. AND owner_uid = $owner_uid");
  866. //db_query("COMMIT");
  867. }
  868. _debug("article processed", $debug_enabled);
  869. }
  870. _debug("purging feed...", $debug_enabled);
  871. purge_feed($feed, 0, $debug_enabled);
  872. db_query("UPDATE ttrss_feeds
  873. SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
  874. // db_query("COMMIT");
  875. } else {
  876. $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
  877. _debug("fetch error: $error_msg", $debug_enabled);
  878. if (count($rss->errors()) > 1) {
  879. foreach ($rss->errors() as $error) {
  880. _debug("+ $error");
  881. }
  882. }
  883. db_query(
  884. "UPDATE ttrss_feeds SET last_error = '$error_msg',
  885. last_updated = NOW() WHERE id = '$feed'");
  886. unset($rss);
  887. }
  888. _debug("done", $debug_enabled);
  889. return $rss;
  890. }
  891. function cache_images($html, $site_url, $debug) {
  892. libxml_use_internal_errors(true);
  893. $charset_hack = '<head>
  894. <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
  895. </head>';
  896. $doc = new DOMDocument();
  897. $doc->loadHTML($charset_hack . $html);
  898. $xpath = new DOMXPath($doc);
  899. $entries = $xpath->query('(//img[@src])');
  900. foreach ($entries as $entry) {
  901. if ($entry->hasAttribute('src')) {
  902. $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
  903. $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
  904. if ($debug) _debug("cache_images: downloading: $src to $local_filename");
  905. if (!file_exists($local_filename)) {
  906. $file_content = fetch_file_contents($src);
  907. if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
  908. file_put_contents($local_filename, $file_content);
  909. }
  910. }
  911. }
  912. }
  913. }
  914. function expire_error_log($debug) {
  915. if ($debug) _debug("Removing old error log entries...");
  916. if (DB_TYPE == "pgsql") {
  917. db_query("DELETE FROM ttrss_error_log
  918. WHERE created_at < NOW() - INTERVAL '7 days'");
  919. } else {
  920. db_query("DELETE FROM ttrss_error_log
  921. WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
  922. }
  923. }
  924. function expire_lock_files($debug) {
  925. //if ($debug) _debug("Removing old lock files...");
  926. $num_deleted = 0;
  927. if (is_writable(LOCK_DIRECTORY)) {
  928. $files = glob(LOCK_DIRECTORY . "/*.lock");
  929. if ($files) {
  930. foreach ($files as $file) {
  931. if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
  932. unlink($file);
  933. ++$num_deleted;
  934. }
  935. }
  936. }
  937. }
  938. if ($debug) _debug("Removed $num_deleted old lock files.");
  939. }
  940. function expire_cached_files($debug) {
  941. foreach (array("simplepie", "images", "export", "upload") as $dir) {
  942. $cache_dir = CACHE_DIR . "/$dir";
  943. // if ($debug) _debug("Expiring $cache_dir");
  944. $num_deleted = 0;
  945. if (is_writable($cache_dir)) {
  946. $files = glob("$cache_dir/*");
  947. if ($files) {
  948. foreach ($files as $file) {
  949. if (time() - filemtime($file) > 86400*7) {
  950. unlink($file);
  951. ++$num_deleted;
  952. }
  953. }
  954. }
  955. }
  956. if ($debug) _debug("$cache_dir: removed $num_deleted files.");
  957. }
  958. }
  959. /**
  960. * Source: http://www.php.net/manual/en/function.parse-url.php#104527
  961. * Returns the url query as associative array
  962. *
  963. * @param string query
  964. * @return array params
  965. */
  966. function convertUrlQuery($query) {
  967. $queryParts = explode('&', $query);
  968. $params = array();
  969. foreach ($queryParts as $param) {
  970. $item = explode('=', $param);
  971. $params[$item[0]] = $item[1];
  972. }
  973. return $params;
  974. }
  975. function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
  976. $matches = array();
  977. foreach ($filters as $filter) {
  978. $match_any_rule = $filter["match_any_rule"];
  979. $inverse = $filter["inverse"];
  980. $filter_match = false;
  981. foreach ($filter["rules"] as $rule) {
  982. $match = false;
  983. $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
  984. $rule_inverse = $rule["inverse"];
  985. if (!$reg_exp)
  986. continue;
  987. switch ($rule["type"]) {
  988. case "title":
  989. $match = @preg_match("/$reg_exp/i", $title);
  990. break;
  991. case "content":
  992. // we don't need to deal with multiline regexps
  993. $content = preg_replace("/[\r\n\t]/", "", $content);
  994. $match = @preg_match("/$reg_exp/i", $content);
  995. break;
  996. case "both":
  997. // we don't need to deal with multiline regexps
  998. $content = preg_replace("/[\r\n\t]/", "", $content);
  999. $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
  1000. break;
  1001. case "link":
  1002. $match = @preg_match("/$reg_exp/i", $link);
  1003. break;
  1004. case "author":
  1005. $match = @preg_match("/$reg_exp/i", $author);
  1006. break;
  1007. case "tag":
  1008. foreach ($tags as $tag) {
  1009. if (@preg_match("/$reg_exp/i", $tag)) {
  1010. $match = true;
  1011. break;
  1012. }
  1013. }
  1014. break;
  1015. }
  1016. if ($rule_inverse) $match = !$match;
  1017. if ($match_any_rule) {
  1018. if ($match) {
  1019. $filter_match = true;
  1020. break;
  1021. }
  1022. } else {
  1023. $filter_match = $match;
  1024. if (!$match) {
  1025. break;
  1026. }
  1027. }
  1028. }
  1029. if ($inverse) $filter_match = !$filter_match;
  1030. if ($filter_match) {
  1031. foreach ($filter["actions"] AS $action) {
  1032. array_push($matches, $action);
  1033. // if Stop action encountered, perform no further processing
  1034. if ($action["type"] == "stop") return $matches;
  1035. }
  1036. }
  1037. }
  1038. return $matches;
  1039. }
  1040. function find_article_filter($filters, $filter_name) {
  1041. foreach ($filters as $f) {
  1042. if ($f["type"] == $filter_name) {
  1043. return $f;
  1044. };
  1045. }
  1046. return false;
  1047. }
  1048. function find_article_filters($filters, $filter_name) {
  1049. $results = array();
  1050. foreach ($filters as $f) {
  1051. if ($f["type"] == $filter_name) {
  1052. array_push($results, $f);
  1053. };
  1054. }
  1055. return $results;
  1056. }
  1057. function calculate_article_score($filters) {
  1058. $score = 0;
  1059. foreach ($filters as $f) {
  1060. if ($f["type"] == "score") {
  1061. $score += $f["param"];
  1062. };
  1063. }
  1064. return $score;
  1065. }
  1066. function labels_contains_caption($labels, $caption) {
  1067. foreach ($labels as $label) {
  1068. if ($label[1] == $caption) {
  1069. return true;
  1070. }
  1071. }
  1072. return false;
  1073. }
  1074. function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
  1075. foreach ($filters as $f) {
  1076. if ($f["type"] == "label") {
  1077. if (!labels_contains_caption($article_labels, $f["param"])) {
  1078. label_add_article($id, $f["param"], $owner_uid);
  1079. }
  1080. }
  1081. }
  1082. }
  1083. function make_guid_from_title($title) {
  1084. return preg_replace("/[ \"\',.:;]/", "-",
  1085. mb_strtolower(strip_tags($title), 'utf-8'));
  1086. }
  1087. /* function verify_feed_xml($feed_data) {
  1088. libxml_use_internal_errors(true);
  1089. $doc = new DOMDocument();
  1090. $doc->loadXML($feed_data);
  1091. $error = libxml_get_last_error();
  1092. libxml_clear_errors();
  1093. return $error;
  1094. } */
  1095. function cleanup_counters_cache($debug) {
  1096. $result = db_query("DELETE FROM ttrss_counters_cache
  1097. WHERE feed_id > 0 AND
  1098. (SELECT COUNT(id) FROM ttrss_feeds WHERE
  1099. id = feed_id AND
  1100. ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
  1101. $frows = db_affected_rows($result);
  1102. $result = db_query("DELETE FROM ttrss_cat_counters_cache
  1103. WHERE feed_id > 0 AND
  1104. (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
  1105. id = feed_id AND
  1106. ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
  1107. $crows = db_affected_rows($result);
  1108. _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
  1109. }
  1110. function housekeeping_common($debug) {
  1111. expire_cached_files($debug);
  1112. expire_lock_files($debug);
  1113. expire_error_log($debug);
  1114. $count = update_feedbrowser_cache();
  1115. _debug("Feedbrowser updated, $count feeds processed.");
  1116. purge_orphans( true);
  1117. cleanup_counters_cache($debug);
  1118. $rc = cleanup_tags( 14, 50000);
  1119. _debug("Cleaned $rc cached tags.");
  1120. PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
  1121. }
  1122. ?>