rssfuncs.php 44 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529
  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. $x = strip_tags(is_array($v) ? implode(",", $v) : $v);
  11. //_debug("$k:" . sha1($x) . ":" . htmlspecialchars($x), true);
  12. $tmp .= sha1("$k:" . sha1($x));
  13. }
  14. }
  15. return sha1(implode(",", $pluginhost->get_plugin_names()) . $tmp);
  16. }
  17. function update_feedbrowser_cache() {
  18. $result = db_query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
  19. FROM ttrss_feeds WHERE (SELECT COUNT(id) = 0 FROM ttrss_feeds AS tf
  20. WHERE tf.feed_url = ttrss_feeds.feed_url
  21. AND (private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%'))
  22. GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
  23. db_query("BEGIN");
  24. db_query("DELETE FROM ttrss_feedbrowser_cache");
  25. $count = 0;
  26. while ($line = db_fetch_assoc($result)) {
  27. $subscribers = db_escape_string($line["subscribers"]);
  28. $feed_url = db_escape_string($line["feed_url"]);
  29. $title = db_escape_string($line["title"]);
  30. $site_url = db_escape_string($line["site_url"]);
  31. $tmp_result = db_query("SELECT subscribers FROM
  32. ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
  33. if (db_num_rows($tmp_result) == 0) {
  34. db_query("INSERT INTO ttrss_feedbrowser_cache
  35. (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
  36. '$site_url', '$title', '$subscribers')");
  37. ++$count;
  38. }
  39. }
  40. db_query("COMMIT");
  41. return $count;
  42. }
  43. /**
  44. * Update a feed batch.
  45. * Used by daemons to update n feeds by run.
  46. * Only update feed needing a update, and not being processed
  47. * by another process.
  48. *
  49. * @param mixed $link Database link
  50. * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
  51. * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
  52. * @param boolean $debug Set to false to disable debug output. Default to true.
  53. * @return void
  54. */
  55. function update_daemon_common($limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
  56. // Process all other feeds using last_updated and interval parameters
  57. $schema_version = get_schema_version();
  58. if ($schema_version != SCHEMA_VERSION) {
  59. die("Schema version is wrong, please upgrade the database.\n");
  60. }
  61. define('PREFS_NO_CACHE', true);
  62. // Test if the user has loggued in recently. If not, it does not update its feeds.
  63. if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
  64. if (DB_TYPE == "pgsql") {
  65. $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
  66. } else {
  67. $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
  68. }
  69. } else {
  70. $login_thresh_qpart = "";
  71. }
  72. // Test if the feed need a update (update interval exceeded).
  73. if (DB_TYPE == "pgsql") {
  74. $update_limit_qpart = "AND ((
  75. ttrss_feeds.update_interval = 0
  76. AND ttrss_user_prefs.value != '-1'
  77. AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
  78. ) OR (
  79. ttrss_feeds.update_interval > 0
  80. AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
  81. ) OR (ttrss_feeds.last_updated IS NULL
  82. AND ttrss_user_prefs.value != '-1')
  83. OR (last_updated = '1970-01-01 00:00:00'
  84. AND ttrss_user_prefs.value != '-1'))";
  85. } else {
  86. $update_limit_qpart = "AND ((
  87. ttrss_feeds.update_interval = 0
  88. AND ttrss_user_prefs.value != '-1'
  89. AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
  90. ) OR (
  91. ttrss_feeds.update_interval > 0
  92. AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
  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'))";
  97. }
  98. // Test if feed is currently being updated by another process.
  99. if (DB_TYPE == "pgsql") {
  100. $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
  101. } else {
  102. $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
  103. }
  104. // Test if there is a limit to number of updated feeds
  105. $query_limit = "";
  106. if($limit) $query_limit = sprintf("LIMIT %d", $limit);
  107. // Update the least recently updated feeds first
  108. $query_order = "ORDER BY last_updated";
  109. if (DB_TYPE == "pgsql") $query_order .= " NULLS FIRST";
  110. $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
  111. FROM
  112. ttrss_feeds, ttrss_users, ttrss_user_prefs
  113. WHERE
  114. ttrss_feeds.owner_uid = ttrss_users.id
  115. AND ttrss_user_prefs.profile IS NULL
  116. AND ttrss_users.id = ttrss_user_prefs.owner_uid
  117. AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
  118. $login_thresh_qpart $update_limit_qpart
  119. $updstart_thresh_qpart
  120. $query_order $query_limit";
  121. // We search for feed needing update.
  122. $result = db_query($query);
  123. if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
  124. // Here is a little cache magic in order to minimize risk of double feed updates.
  125. $feeds_to_update = array();
  126. while ($line = db_fetch_assoc($result)) {
  127. array_push($feeds_to_update, db_escape_string($line['feed_url']));
  128. }
  129. // We update the feed last update started date before anything else.
  130. // There is no lag due to feed contents downloads
  131. // It prevent an other process to update the same feed.
  132. if(count($feeds_to_update) > 0) {
  133. $feeds_quoted = array();
  134. foreach ($feeds_to_update as $feed) {
  135. array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
  136. }
  137. db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
  138. WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
  139. }
  140. $nf = 0;
  141. $bstarted = microtime(true);
  142. // For each feed, we call the feed update function.
  143. foreach ($feeds_to_update as $feed) {
  144. if($debug) _debug("Base feed: $feed");
  145. //update_rss_feed($line["id"], true);
  146. // since we have the data cached, we can deal with other feeds with the same url
  147. $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
  148. FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
  149. ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
  150. ttrss_users.id = ttrss_user_prefs.owner_uid AND
  151. ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
  152. ttrss_user_prefs.profile IS NULL AND
  153. feed_url = '".db_escape_string($feed)."' AND
  154. (ttrss_feeds.update_interval > 0 OR
  155. ttrss_user_prefs.value != '-1')
  156. $login_thresh_qpart
  157. ORDER BY ttrss_feeds.id $query_limit");
  158. if (db_num_rows($tmp_result) > 0) {
  159. $rss = false;
  160. while ($tline = db_fetch_assoc($tmp_result)) {
  161. if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
  162. $fstarted = microtime(true);
  163. $rss = update_rss_feed($tline["id"], true, false);
  164. _debug_suppress(false);
  165. _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
  166. ++$nf;
  167. }
  168. }
  169. }
  170. if ($nf > 0) {
  171. _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
  172. microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
  173. }
  174. require_once "digest.php";
  175. // Send feed digests by email if needed.
  176. send_headlines_digests($debug);
  177. return $nf;
  178. } // function update_daemon_common
  179. // this is used when subscribing
  180. function set_basic_feed_info($feed) {
  181. $feed = db_escape_string($feed);
  182. $result = db_query("SELECT feed_url,auth_pass,auth_login,auth_pass_encrypted
  183. FROM ttrss_feeds WHERE id = '$feed'");
  184. $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
  185. 0, "auth_pass_encrypted"));
  186. $auth_login = db_fetch_result($result, 0, "auth_login");
  187. $auth_pass = db_fetch_result($result, 0, "auth_pass");
  188. if ($auth_pass_encrypted) {
  189. require_once "crypt.php";
  190. $auth_pass = decrypt_string($auth_pass);
  191. }
  192. $fetch_url = db_fetch_result($result, 0, "feed_url");
  193. $feed_data = fetch_file_contents($fetch_url, false,
  194. $auth_login, $auth_pass, false,
  195. FEED_FETCH_TIMEOUT_TIMEOUT,
  196. 0);
  197. global $fetch_curl_used;
  198. if (!$fetch_curl_used) {
  199. $tmp = @gzdecode($feed_data);
  200. if ($tmp) $feed_data = $tmp;
  201. }
  202. $feed_data = trim($feed_data);
  203. $rss = new FeedParser($feed_data);
  204. $rss->init();
  205. if (!$rss->error()) {
  206. $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
  207. $registered_title = db_fetch_result($result, 0, "title");
  208. $orig_site_url = db_fetch_result($result, 0, "site_url");
  209. $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
  210. $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
  211. if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
  212. db_query("UPDATE ttrss_feeds SET
  213. title = '$feed_title' WHERE id = '$feed'");
  214. }
  215. if ($site_url && $orig_site_url != $site_url) {
  216. db_query("UPDATE ttrss_feeds SET
  217. site_url = '$site_url' WHERE id = '$feed'");
  218. }
  219. }
  220. }
  221. // ignore_daemon is not used
  222. function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
  223. $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
  224. _debug_suppress(!$debug_enabled);
  225. _debug("start", $debug_enabled);
  226. $result = db_query("SELECT title FROM ttrss_feeds
  227. WHERE id = '$feed'");
  228. $title = db_fetch_result($result, 0, "title");
  229. // feed was batch-subscribed or something, we need to get basic info
  230. // this is not optimal currently as it fetches stuff separately TODO: optimize
  231. if ($title == "[Unknown]") {
  232. _debug("setting basic feed info for $feed...");
  233. set_basic_feed_info($feed);
  234. }
  235. $result = db_query("SELECT id,update_interval,auth_login,
  236. feed_url,auth_pass,cache_images,
  237. mark_unread_on_update, owner_uid,
  238. pubsub_state, auth_pass_encrypted,
  239. feed_language,
  240. (SELECT max(date_entered) FROM
  241. ttrss_entries, ttrss_user_entries where ref_id = id AND feed_id = '$feed') AS last_article_timestamp
  242. FROM ttrss_feeds WHERE id = '$feed'");
  243. if (db_num_rows($result) == 0) {
  244. _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
  245. return false;
  246. }
  247. $last_article_timestamp = @strtotime(db_fetch_result($result, 0, "last_article_timestamp"));
  248. if (defined('_DISABLE_HTTP_304'))
  249. $last_article_timestamp = 0;
  250. $owner_uid = db_fetch_result($result, 0, "owner_uid");
  251. $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
  252. 0, "mark_unread_on_update"));
  253. $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
  254. $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
  255. 0, "auth_pass_encrypted"));
  256. db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
  257. WHERE id = '$feed'");
  258. $auth_login = db_fetch_result($result, 0, "auth_login");
  259. $auth_pass = db_fetch_result($result, 0, "auth_pass");
  260. if ($auth_pass_encrypted) {
  261. require_once "crypt.php";
  262. $auth_pass = decrypt_string($auth_pass);
  263. }
  264. $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
  265. $fetch_url = db_fetch_result($result, 0, "feed_url");
  266. $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
  267. if (!$feed_language) $feed_language = 'english';
  268. $feed = db_escape_string($feed);
  269. $date_feed_processed = date('Y-m-d H:i');
  270. $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
  271. $pluginhost = new PluginHost();
  272. $pluginhost->set_debug($debug_enabled);
  273. $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
  274. $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
  275. $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
  276. $pluginhost->load_data();
  277. if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
  278. _debug("using previously initialized parser object");
  279. } else {
  280. $rss_hash = false;
  281. $force_refetch = isset($_REQUEST["force_refetch"]);
  282. foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
  283. $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, $last_article_timestamp, $auth_login, $auth_pass);
  284. }
  285. // try cache
  286. if (!$feed_data &&
  287. file_exists($cache_filename) &&
  288. is_readable($cache_filename) &&
  289. !$auth_login && !$auth_pass &&
  290. filemtime($cache_filename) > time() - 30) {
  291. _debug("using local cache [$cache_filename].", $debug_enabled);
  292. @$feed_data = file_get_contents($cache_filename);
  293. if ($feed_data) {
  294. $rss_hash = sha1($feed_data);
  295. }
  296. } else {
  297. _debug("local cache will not be used for this feed", $debug_enabled);
  298. }
  299. // fetch feed from source
  300. if (!$feed_data) {
  301. _debug("fetching [$fetch_url]...", $debug_enabled);
  302. if (ini_get("open_basedir") && function_exists("curl_init")) {
  303. _debug("not using CURL due to open_basedir restrictions");
  304. }
  305. _debug("If-Modified-Since: ".gmdate('D, d M Y H:i:s \G\M\T', $last_article_timestamp), $debug_enabled);
  306. $feed_data = fetch_file_contents($fetch_url, false,
  307. $auth_login, $auth_pass, false,
  308. $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
  309. $force_refetch ? 0 : $last_article_timestamp);
  310. global $fetch_curl_used;
  311. if (!$fetch_curl_used) {
  312. $tmp = @gzdecode($feed_data);
  313. if ($tmp) $feed_data = $tmp;
  314. }
  315. $feed_data = trim($feed_data);
  316. _debug("fetch done.", $debug_enabled);
  317. // cache vanilla feed data for re-use
  318. if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
  319. $new_rss_hash = sha1($feed_data);
  320. if ($new_rss_hash != $rss_hash) {
  321. _debug("saving $cache_filename", $debug_enabled);
  322. @file_put_contents($cache_filename, $feed_data);
  323. }
  324. }
  325. }
  326. if (!$feed_data) {
  327. global $fetch_last_error;
  328. global $fetch_last_error_code;
  329. _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
  330. $error_escaped = '';
  331. // If-Modified-Since
  332. if ($fetch_last_error_code != 304) {
  333. $error_escaped = db_escape_string($fetch_last_error);
  334. } else {
  335. _debug("source claims data not modified, nothing to do.", $debug_enabled);
  336. }
  337. db_query(
  338. "UPDATE ttrss_feeds SET last_error = '$error_escaped',
  339. last_updated = NOW() WHERE id = '$feed'");
  340. return;
  341. }
  342. }
  343. foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
  344. $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
  345. }
  346. // set last update to now so if anything *simplepie* crashes later we won't be
  347. // continuously failing on the same feed
  348. //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
  349. if (!$rss) {
  350. $rss = new FeedParser($feed_data);
  351. $rss->init();
  352. }
  353. // print_r($rss);
  354. $feed = db_escape_string($feed);
  355. if (!$rss->error()) {
  356. // We use local pluginhost here because we need to load different per-user feed plugins
  357. $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
  358. _debug("language: $feed_language", $debug_enabled);
  359. _debug("processing feed data...", $debug_enabled);
  360. // db_query("BEGIN");
  361. if (DB_TYPE == "pgsql") {
  362. $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
  363. } else {
  364. $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
  365. }
  366. $result = db_query("SELECT owner_uid,favicon_avg_color,
  367. (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
  368. favicon_needs_check
  369. FROM ttrss_feeds WHERE id = '$feed'");
  370. $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
  371. "favicon_needs_check"));
  372. $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
  373. $owner_uid = db_fetch_result($result, 0, "owner_uid");
  374. $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
  375. _debug("site_url: $site_url", $debug_enabled);
  376. _debug("feed_title: " . $rss->get_title(), $debug_enabled);
  377. if ($favicon_needs_check || $force_refetch) {
  378. /* terrible hack: if we crash on floicon shit here, we won't check
  379. * the icon avgcolor again (unless the icon got updated) */
  380. $favicon_file = ICONS_DIR . "/$feed.ico";
  381. $favicon_modified = @filemtime($favicon_file);
  382. _debug("checking favicon...", $debug_enabled);
  383. check_feed_favicon($site_url, $feed);
  384. $favicon_modified_new = @filemtime($favicon_file);
  385. if ($favicon_modified_new > $favicon_modified)
  386. $favicon_avg_color = '';
  387. if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
  388. require_once "colors.php";
  389. db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
  390. id = '$feed'");
  391. $favicon_color = db_escape_string(
  392. calculate_avg_color($favicon_file));
  393. $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
  394. } else if ($favicon_avg_color == 'fail') {
  395. _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
  396. }
  397. db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
  398. $favicon_colorstring
  399. WHERE id = '$feed'");
  400. }
  401. _debug("loading filters & labels...", $debug_enabled);
  402. $filters = load_filters($feed, $owner_uid);
  403. _debug("" . count($filters) . " filters loaded.", $debug_enabled);
  404. $items = $rss->get_items();
  405. if (!is_array($items)) {
  406. _debug("no articles found.", $debug_enabled);
  407. db_query("UPDATE ttrss_feeds
  408. SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
  409. return; // no articles
  410. }
  411. if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
  412. _debug("checking for PUSH hub...", $debug_enabled);
  413. $feed_hub_url = false;
  414. $links = $rss->get_links('hub');
  415. if ($links && is_array($links)) {
  416. foreach ($links as $l) {
  417. $feed_hub_url = $l;
  418. break;
  419. }
  420. }
  421. _debug("feed hub url: $feed_hub_url", $debug_enabled);
  422. $feed_self_url = $fetch_url;
  423. $links = $rss->get_links('self');
  424. if ($links && is_array($links)) {
  425. foreach ($links as $l) {
  426. $feed_self_url = $l;
  427. break;
  428. }
  429. }
  430. _debug("feed self url = $feed_self_url");
  431. if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
  432. !ini_get("open_basedir")) {
  433. require_once 'lib/pubsubhubbub/subscriber.php';
  434. $callback_url = get_self_url_prefix() .
  435. "/public.php?op=pubsub&id=$feed";
  436. $s = new Subscriber($feed_hub_url, $callback_url);
  437. $rc = $s->subscribe($feed_self_url);
  438. _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
  439. db_query("UPDATE ttrss_feeds SET pubsub_state = 1
  440. WHERE id = '$feed'");
  441. }
  442. }
  443. _debug("processing articles...", $debug_enabled);
  444. $tstart = time();
  445. foreach ($items as $item) {
  446. if ($_REQUEST['xdebug'] == 3) {
  447. print_r($item);
  448. }
  449. if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
  450. _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
  451. break;
  452. }
  453. $entry_guid = $item->get_id();
  454. if (!$entry_guid) $entry_guid = $item->get_link();
  455. if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
  456. if (!$entry_guid) continue;
  457. $entry_guid = "$owner_uid,$entry_guid";
  458. $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
  459. _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
  460. $entry_timestamp = "";
  461. $entry_timestamp = $item->get_date();
  462. _debug("orig date: " . $item->get_date(), $debug_enabled);
  463. if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
  464. $entry_timestamp = time();
  465. }
  466. $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
  467. _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
  468. // $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
  469. // $entry_title = decode_numeric_entities($entry_title);
  470. $entry_title = $item->get_title();
  471. $entry_link = rewrite_relative_url($site_url, $item->get_link());
  472. _debug("title $entry_title", $debug_enabled);
  473. _debug("link $entry_link", $debug_enabled);
  474. if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
  475. $entry_content = $item->get_content();
  476. if (!$entry_content) $entry_content = $item->get_description();
  477. if ($_REQUEST["xdebug"] == 2) {
  478. print "content: ";
  479. print htmlspecialchars($entry_content);
  480. print "\n";
  481. }
  482. $entry_comments = $item->get_comments_url();
  483. $entry_author = $item->get_author();
  484. $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
  485. $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
  486. $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
  487. $num_comments = (int) $item->get_comments_count();
  488. _debug("author $entry_author", $debug_enabled);
  489. _debug("num_comments: $num_comments", $debug_enabled);
  490. _debug("looking for tags...", $debug_enabled);
  491. // parse <category> entries into tags
  492. $additional_tags = array();
  493. $additional_tags_src = $item->get_categories();
  494. if (is_array($additional_tags_src)) {
  495. foreach ($additional_tags_src as $tobj) {
  496. array_push($additional_tags, $tobj);
  497. }
  498. }
  499. $entry_tags = array_unique($additional_tags);
  500. for ($i = 0; $i < count($entry_tags); $i++)
  501. $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
  502. _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
  503. _debug("done collecting data.", $debug_enabled);
  504. $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
  505. WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
  506. if (db_num_rows($result) != 0) {
  507. $base_entry_id = db_fetch_result($result, 0, "id");
  508. $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
  509. $article_labels = get_article_labels($base_entry_id, $owner_uid);
  510. $entry_language = db_fetch_result($result, 0, "lang");
  511. $existing_tags = get_article_tags($base_entry_id, $owner_uid);
  512. $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
  513. } else {
  514. $base_entry_id = false;
  515. $entry_stored_hash = "";
  516. $article_labels = array();
  517. $entry_language = "";
  518. }
  519. $article = array("owner_uid" => $owner_uid, // read only
  520. "guid" => $entry_guid, // read only
  521. "guid_hashed" => $entry_guid_hashed, // read only
  522. "title" => $entry_title,
  523. "content" => $entry_content,
  524. "link" => $entry_link,
  525. "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
  526. "tags" => $entry_tags,
  527. "author" => $entry_author,
  528. "force_catchup" => false, // ugly hack for the time being
  529. "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
  530. "language" => $entry_language,
  531. "feed" => array("id" => $feed,
  532. "fetch_url" => $fetch_url,
  533. "site_url" => $site_url)
  534. );
  535. $entry_plugin_data = "";
  536. $entry_current_hash = calculate_article_hash($article, $pluginhost);
  537. _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
  538. if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
  539. _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
  540. // we keep encountering the entry in feeds, so we need to
  541. // update date_updated column so that we don't get horrible
  542. // dupes when the entry gets purged and reinserted again e.g.
  543. // in the case of SLOW SLOW OMG SLOW updating feeds
  544. $base_entry_id = db_fetch_result($result, 0, "id");
  545. db_query("UPDATE ttrss_entries SET date_updated = NOW()
  546. WHERE id = '$base_entry_id'");
  547. continue;
  548. }
  549. _debug("hash differs, applying plugin filters:", $debug_enabled);
  550. foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
  551. _debug("... " . get_class($plugin), $debug_enabled);
  552. $start = microtime(true);
  553. $article = $plugin->hook_article_filter($article);
  554. _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
  555. $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
  556. }
  557. if ($_REQUEST["xdebug"] == 2) {
  558. print "processed content: ";
  559. print htmlspecialchars($article["content"]);
  560. print "\n";
  561. }
  562. $entry_plugin_data = db_escape_string($entry_plugin_data);
  563. _debug("plugin data: $entry_plugin_data", $debug_enabled);
  564. // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
  565. if (DB_TYPE == "mysql") {
  566. foreach ($article as $k => $v) {
  567. // i guess we'll have to take the risk of 4byte unicode labels & tags here
  568. if (!is_array($article[$k])) {
  569. $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
  570. }
  571. }
  572. }
  573. /* Collect article tags here so we could filter by them: */
  574. $article_filters = get_article_filters($filters, $article["title"],
  575. $article["content"], $article["link"], 0, $article["author"],
  576. $article["tags"]);
  577. if ($debug_enabled) {
  578. _debug("article filters: ", $debug_enabled);
  579. if (count($article_filters) != 0) {
  580. print_r($article_filters);
  581. }
  582. }
  583. $plugin_filter_names = find_article_filters($article_filters, "plugin");
  584. $plugin_filter_actions = $pluginhost->get_filter_actions();
  585. if (count($plugin_filter_names) > 0) {
  586. _debug("applying plugin filter actions...", $debug_enabled);
  587. foreach ($plugin_filter_names as $pfn) {
  588. list($pfclass,$pfaction) = explode(":", $pfn["param"]);
  589. if (isset($plugin_filter_actions[$pfclass])) {
  590. $plugin = $pluginhost->get_plugin($pfclass);
  591. _debug("... $pfclass: $pfaction", $debug_enabled);
  592. if ($plugin) {
  593. $start = microtime(true);
  594. $article = $plugin->hook_article_filter_action($article, $pfaction);
  595. _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
  596. } else {
  597. _debug("??? $pfclass: plugin object not found.");
  598. }
  599. } else {
  600. _debug("??? $pfclass: filter plugin not registered.");
  601. }
  602. }
  603. }
  604. $entry_tags = $article["tags"];
  605. $entry_guid = db_escape_string($entry_guid);
  606. $entry_title = db_escape_string($article["title"]);
  607. $entry_author = db_escape_string($article["author"]);
  608. $entry_link = db_escape_string($article["link"]);
  609. $entry_content = $article["content"]; // escaped below
  610. $entry_force_catchup = $article["force_catchup"];
  611. $article_labels = $article["labels"];
  612. $entry_score_modifier = (int) $article["score_modifier"];
  613. $entry_language = db_escape_string($article["language"]);
  614. if ($debug_enabled) {
  615. _debug("article labels:", $debug_enabled);
  616. print_r($article_labels);
  617. }
  618. _debug("force catchup: $entry_force_catchup");
  619. if ($cache_images && is_writable(CACHE_DIR . '/images'))
  620. cache_images($entry_content, $site_url, $debug_enabled);
  621. $entry_content = db_escape_string($entry_content, false);
  622. //db_query("BEGIN");
  623. $result = db_query("SELECT id FROM ttrss_entries
  624. WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
  625. if (db_num_rows($result) == 0) {
  626. _debug("base guid [$entry_guid] not found", $debug_enabled);
  627. // base post entry does not exist, create it
  628. $result = db_query(
  629. "INSERT INTO ttrss_entries
  630. (title,
  631. guid,
  632. link,
  633. updated,
  634. content,
  635. content_hash,
  636. no_orig_date,
  637. date_updated,
  638. date_entered,
  639. comments,
  640. num_comments,
  641. plugin_data,
  642. lang,
  643. author)
  644. VALUES
  645. ('$entry_title',
  646. '$entry_guid_hashed',
  647. '$entry_link',
  648. '$entry_timestamp_fmt',
  649. '$entry_content',
  650. '$entry_current_hash',
  651. false,
  652. NOW(),
  653. '$date_feed_processed',
  654. '$entry_comments',
  655. '$num_comments',
  656. '$entry_plugin_data',
  657. '$entry_language',
  658. '$entry_author')");
  659. } else {
  660. $base_entry_id = db_fetch_result($result, 0, "id");
  661. }
  662. // now it should exist, if not - bad luck then
  663. $result = db_query("SELECT id FROM ttrss_entries
  664. WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
  665. $entry_ref_id = 0;
  666. $entry_int_id = 0;
  667. if (db_num_rows($result) == 1) {
  668. _debug("base guid found, checking for user record", $debug_enabled);
  669. $ref_id = db_fetch_result($result, 0, "id");
  670. $entry_ref_id = $ref_id;
  671. /* $stored_guid = db_fetch_result($result, 0, "guid");
  672. if ($stored_guid != $entry_guid_hashed) {
  673. if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
  674. db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
  675. id = '$ref_id'");
  676. } */
  677. if (find_article_filter($article_filters, "filter")) {
  678. //db_query("COMMIT"); // close transaction in progress
  679. continue;
  680. }
  681. $score = calculate_article_score($article_filters) + $entry_score_modifier;
  682. _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
  683. // check for user post link to main table
  684. $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
  685. ref_id = '$ref_id' AND owner_uid = '$owner_uid'";
  686. // if ($_REQUEST["xdebug"]) print "$query\n";
  687. $result = db_query($query);
  688. // okay it doesn't exist - create user entry
  689. if (db_num_rows($result) == 0) {
  690. _debug("user record not found, creating...", $debug_enabled);
  691. if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
  692. $unread = 'true';
  693. $last_read_qpart = 'NULL';
  694. } else {
  695. $unread = 'false';
  696. $last_read_qpart = 'NOW()';
  697. }
  698. if (find_article_filter($article_filters, 'mark') || $score > 1000) {
  699. $marked = 'true';
  700. } else {
  701. $marked = 'false';
  702. }
  703. if (find_article_filter($article_filters, 'publish')) {
  704. $published = 'true';
  705. } else {
  706. $published = 'false';
  707. }
  708. // N-grams
  709. /* if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
  710. $result = db_query("SELECT COUNT(*) AS similar FROM
  711. ttrss_entries,ttrss_user_entries
  712. WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
  713. AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
  714. AND owner_uid = $owner_uid");
  715. $ngram_similar = db_fetch_result($result, 0, "similar");
  716. _debug("N-gram similar results: $ngram_similar", $debug_enabled);
  717. if ($ngram_similar > 0) {
  718. $unread = 'false';
  719. }
  720. } */
  721. $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
  722. $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
  723. $result = db_query(
  724. "INSERT INTO ttrss_user_entries
  725. (ref_id, owner_uid, feed_id, unread, last_read, marked,
  726. published, score, tag_cache, label_cache, uuid,
  727. last_marked, last_published)
  728. VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
  729. $last_read_qpart, $marked, $published, '$score', '', '',
  730. '', $last_marked, $last_published)");
  731. if (PUBSUBHUBBUB_HUB && $published == 'true') {
  732. $rss_link = get_self_url_prefix() .
  733. "/public.php?op=rss&id=-2&key=" .
  734. get_feed_access_key(-2, false, $owner_uid);
  735. $p = new Publisher(PUBSUBHUBBUB_HUB);
  736. /* $pubsub_result = */ $p->publish_update($rss_link);
  737. }
  738. $result = db_query(
  739. "SELECT int_id FROM ttrss_user_entries WHERE
  740. ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
  741. feed_id = '$feed' LIMIT 1");
  742. if (db_num_rows($result) == 1) {
  743. $entry_int_id = db_fetch_result($result, 0, "int_id");
  744. }
  745. } else {
  746. _debug("user record FOUND", $debug_enabled);
  747. $entry_ref_id = db_fetch_result($result, 0, "ref_id");
  748. $entry_int_id = db_fetch_result($result, 0, "int_id");
  749. }
  750. _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
  751. if (DB_TYPE == "pgsql") {
  752. $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
  753. 0, 1000000));
  754. $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
  755. } else {
  756. $tsvector_qpart = "";
  757. }
  758. db_query("UPDATE ttrss_entries
  759. SET title = '$entry_title',
  760. content = '$entry_content',
  761. content_hash = '$entry_current_hash',
  762. updated = '$entry_timestamp_fmt',
  763. $tsvector_qpart
  764. num_comments = '$num_comments',
  765. plugin_data = '$entry_plugin_data',
  766. author = '$entry_author',
  767. lang = '$entry_language'
  768. WHERE id = '$ref_id'");
  769. // update aux data
  770. db_query("UPDATE ttrss_user_entries
  771. SET score = '$score' WHERE ref_id = '$ref_id'");
  772. if ($mark_unread_on_update) {
  773. _debug("article updated, marking unread as requested.", $debug_enabled);
  774. db_query("UPDATE ttrss_user_entries
  775. SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
  776. }
  777. }
  778. //db_query("COMMIT");
  779. _debug("assigning labels [other]...", $debug_enabled);
  780. foreach ($article_labels as $label) {
  781. label_add_article($entry_ref_id, $label[1], $owner_uid);
  782. }
  783. _debug("assigning labels [filters]...", $debug_enabled);
  784. assign_article_to_label_filters($entry_ref_id, $article_filters,
  785. $owner_uid, $article_labels);
  786. _debug("looking for enclosures...", $debug_enabled);
  787. // enclosures
  788. $enclosures = array();
  789. $encs = $item->get_enclosures();
  790. if (is_array($encs)) {
  791. foreach ($encs as $e) {
  792. $e_item = array(
  793. $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
  794. array_push($enclosures, $e_item);
  795. }
  796. }
  797. if ($debug_enabled) {
  798. _debug("article enclosures:", $debug_enabled);
  799. print_r($enclosures);
  800. }
  801. //db_query("BEGIN");
  802. // debugging
  803. // db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
  804. foreach ($enclosures as $enc) {
  805. $enc_url = db_escape_string($enc[0]);
  806. $enc_type = db_escape_string($enc[1]);
  807. $enc_dur = db_escape_string($enc[2]);
  808. $enc_title = db_escape_string($enc[3]);
  809. $enc_width = intval($enc[4]);
  810. $enc_height = intval($enc[5]);
  811. $result = db_query("SELECT id FROM ttrss_enclosures
  812. WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
  813. if (db_num_rows($result) == 0) {
  814. db_query("INSERT INTO ttrss_enclosures
  815. (content_url, content_type, title, duration, post_id, width, height) VALUES
  816. ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
  817. }
  818. }
  819. //db_query("COMMIT");
  820. // check for manual tags (we have to do it here since they're loaded from filters)
  821. foreach ($article_filters as $f) {
  822. if ($f["type"] == "tag") {
  823. $manual_tags = trim_array(explode(",", $f["param"]));
  824. foreach ($manual_tags as $tag) {
  825. if (tag_is_valid($tag)) {
  826. array_push($entry_tags, $tag);
  827. }
  828. }
  829. }
  830. }
  831. // Skip boring tags
  832. $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
  833. 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
  834. $filtered_tags = array();
  835. $tags_to_cache = array();
  836. if ($entry_tags && is_array($entry_tags)) {
  837. foreach ($entry_tags as $tag) {
  838. if (array_search($tag, $boring_tags) === false) {
  839. array_push($filtered_tags, $tag);
  840. }
  841. }
  842. }
  843. $filtered_tags = array_unique($filtered_tags);
  844. if ($debug_enabled) {
  845. _debug("filtered article tags:", $debug_enabled);
  846. print_r($filtered_tags);
  847. }
  848. // Save article tags in the database
  849. if (count($filtered_tags) > 0) {
  850. //db_query("BEGIN");
  851. foreach ($filtered_tags as $tag) {
  852. $tag = sanitize_tag($tag);
  853. $tag = db_escape_string($tag);
  854. if (!tag_is_valid($tag)) continue;
  855. $result = db_query("SELECT id FROM ttrss_tags
  856. WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
  857. owner_uid = '$owner_uid' LIMIT 1");
  858. if ($result && db_num_rows($result) == 0) {
  859. db_query("INSERT INTO ttrss_tags
  860. (owner_uid,tag_name,post_int_id)
  861. VALUES ('$owner_uid','$tag', '$entry_int_id')");
  862. }
  863. array_push($tags_to_cache, $tag);
  864. }
  865. /* update the cache */
  866. $tags_to_cache = array_unique($tags_to_cache);
  867. $tags_str = db_escape_string(join(",", $tags_to_cache));
  868. db_query("UPDATE ttrss_user_entries
  869. SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
  870. AND owner_uid = $owner_uid");
  871. //db_query("COMMIT");
  872. }
  873. _debug("article processed", $debug_enabled);
  874. }
  875. _debug("purging feed...", $debug_enabled);
  876. purge_feed($feed, 0, $debug_enabled);
  877. db_query("UPDATE ttrss_feeds
  878. SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
  879. // db_query("COMMIT");
  880. } else {
  881. $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
  882. _debug("fetch error: $error_msg", $debug_enabled);
  883. if (count($rss->errors()) > 1) {
  884. foreach ($rss->errors() as $error) {
  885. _debug("+ $error");
  886. }
  887. }
  888. db_query(
  889. "UPDATE ttrss_feeds SET last_error = '$error_msg',
  890. last_updated = NOW() WHERE id = '$feed'");
  891. unset($rss);
  892. }
  893. _debug("done", $debug_enabled);
  894. return $rss;
  895. }
  896. function cache_images($html, $site_url, $debug) {
  897. libxml_use_internal_errors(true);
  898. $charset_hack = '<head>
  899. <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
  900. </head>';
  901. $doc = new DOMDocument();
  902. $doc->loadHTML($charset_hack . $html);
  903. $xpath = new DOMXPath($doc);
  904. $entries = $xpath->query('(//img[@src])');
  905. foreach ($entries as $entry) {
  906. if ($entry->hasAttribute('src')) {
  907. $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
  908. $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
  909. if ($debug) _debug("cache_images: downloading: $src to $local_filename");
  910. if (!file_exists($local_filename)) {
  911. $file_content = fetch_file_contents($src);
  912. if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
  913. file_put_contents($local_filename, $file_content);
  914. }
  915. } else {
  916. touch($local_filename);
  917. }
  918. }
  919. }
  920. }
  921. function expire_error_log($debug) {
  922. if ($debug) _debug("Removing old error log entries...");
  923. if (DB_TYPE == "pgsql") {
  924. db_query("DELETE FROM ttrss_error_log
  925. WHERE created_at < NOW() - INTERVAL '7 days'");
  926. } else {
  927. db_query("DELETE FROM ttrss_error_log
  928. WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
  929. }
  930. }
  931. function expire_lock_files($debug) {
  932. //if ($debug) _debug("Removing old lock files...");
  933. $num_deleted = 0;
  934. if (is_writable(LOCK_DIRECTORY)) {
  935. $files = glob(LOCK_DIRECTORY . "/*.lock");
  936. if ($files) {
  937. foreach ($files as $file) {
  938. if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
  939. unlink($file);
  940. ++$num_deleted;
  941. }
  942. }
  943. }
  944. }
  945. if ($debug) _debug("Removed $num_deleted old lock files.");
  946. }
  947. function expire_cached_files($debug) {
  948. foreach (array("simplepie", "images", "export", "upload") as $dir) {
  949. $cache_dir = CACHE_DIR . "/$dir";
  950. // if ($debug) _debug("Expiring $cache_dir");
  951. $num_deleted = 0;
  952. if (is_writable($cache_dir)) {
  953. $files = glob("$cache_dir/*");
  954. if ($files) {
  955. foreach ($files as $file) {
  956. if (time() - filemtime($file) > 86400*7) {
  957. unlink($file);
  958. ++$num_deleted;
  959. }
  960. }
  961. }
  962. }
  963. if ($debug) _debug("$cache_dir: removed $num_deleted files.");
  964. }
  965. }
  966. /**
  967. * Source: http://www.php.net/manual/en/function.parse-url.php#104527
  968. * Returns the url query as associative array
  969. *
  970. * @param string query
  971. * @return array params
  972. */
  973. function convertUrlQuery($query) {
  974. $queryParts = explode('&', $query);
  975. $params = array();
  976. foreach ($queryParts as $param) {
  977. $item = explode('=', $param);
  978. $params[$item[0]] = $item[1];
  979. }
  980. return $params;
  981. }
  982. function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
  983. $matches = array();
  984. foreach ($filters as $filter) {
  985. $match_any_rule = $filter["match_any_rule"];
  986. $inverse = $filter["inverse"];
  987. $filter_match = false;
  988. foreach ($filter["rules"] as $rule) {
  989. $match = false;
  990. $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
  991. $rule_inverse = $rule["inverse"];
  992. if (!$reg_exp)
  993. continue;
  994. switch ($rule["type"]) {
  995. case "title":
  996. $match = @preg_match("/$reg_exp/i", $title);
  997. break;
  998. case "content":
  999. // we don't need to deal with multiline regexps
  1000. $content = preg_replace("/[\r\n\t]/", "", $content);
  1001. $match = @preg_match("/$reg_exp/i", $content);
  1002. break;
  1003. case "both":
  1004. // we don't need to deal with multiline regexps
  1005. $content = preg_replace("/[\r\n\t]/", "", $content);
  1006. $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
  1007. break;
  1008. case "link":
  1009. $match = @preg_match("/$reg_exp/i", $link);
  1010. break;
  1011. case "author":
  1012. $match = @preg_match("/$reg_exp/i", $author);
  1013. break;
  1014. case "tag":
  1015. foreach ($tags as $tag) {
  1016. if (@preg_match("/$reg_exp/i", $tag)) {
  1017. $match = true;
  1018. break;
  1019. }
  1020. }
  1021. break;
  1022. }
  1023. if ($rule_inverse) $match = !$match;
  1024. if ($match_any_rule) {
  1025. if ($match) {
  1026. $filter_match = true;
  1027. break;
  1028. }
  1029. } else {
  1030. $filter_match = $match;
  1031. if (!$match) {
  1032. break;
  1033. }
  1034. }
  1035. }
  1036. if ($inverse) $filter_match = !$filter_match;
  1037. if ($filter_match) {
  1038. foreach ($filter["actions"] AS $action) {
  1039. array_push($matches, $action);
  1040. // if Stop action encountered, perform no further processing
  1041. if (isset($action["type"]) && $action["type"] == "stop") return $matches;
  1042. }
  1043. }
  1044. }
  1045. return $matches;
  1046. }
  1047. function find_article_filter($filters, $filter_name) {
  1048. foreach ($filters as $f) {
  1049. if ($f["type"] == $filter_name) {
  1050. return $f;
  1051. };
  1052. }
  1053. return false;
  1054. }
  1055. function find_article_filters($filters, $filter_name) {
  1056. $results = array();
  1057. foreach ($filters as $f) {
  1058. if ($f["type"] == $filter_name) {
  1059. array_push($results, $f);
  1060. };
  1061. }
  1062. return $results;
  1063. }
  1064. function calculate_article_score($filters) {
  1065. $score = 0;
  1066. foreach ($filters as $f) {
  1067. if ($f["type"] == "score") {
  1068. $score += $f["param"];
  1069. };
  1070. }
  1071. return $score;
  1072. }
  1073. function labels_contains_caption($labels, $caption) {
  1074. foreach ($labels as $label) {
  1075. if ($label[1] == $caption) {
  1076. return true;
  1077. }
  1078. }
  1079. return false;
  1080. }
  1081. function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
  1082. foreach ($filters as $f) {
  1083. if ($f["type"] == "label") {
  1084. if (!labels_contains_caption($article_labels, $f["param"])) {
  1085. label_add_article($id, $f["param"], $owner_uid);
  1086. }
  1087. }
  1088. }
  1089. }
  1090. function make_guid_from_title($title) {
  1091. return preg_replace("/[ \"\',.:;]/", "-",
  1092. mb_strtolower(strip_tags($title), 'utf-8'));
  1093. }
  1094. /* function verify_feed_xml($feed_data) {
  1095. libxml_use_internal_errors(true);
  1096. $doc = new DOMDocument();
  1097. $doc->loadXML($feed_data);
  1098. $error = libxml_get_last_error();
  1099. libxml_clear_errors();
  1100. return $error;
  1101. } */
  1102. function cleanup_counters_cache($debug) {
  1103. $result = db_query("DELETE FROM ttrss_counters_cache
  1104. WHERE feed_id > 0 AND
  1105. (SELECT COUNT(id) FROM ttrss_feeds WHERE
  1106. id = feed_id AND
  1107. ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
  1108. $frows = db_affected_rows($result);
  1109. $result = db_query("DELETE FROM ttrss_cat_counters_cache
  1110. WHERE feed_id > 0 AND
  1111. (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
  1112. id = feed_id AND
  1113. ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
  1114. $crows = db_affected_rows($result);
  1115. _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
  1116. }
  1117. function housekeeping_common($debug) {
  1118. expire_cached_files($debug);
  1119. expire_lock_files($debug);
  1120. expire_error_log($debug);
  1121. $count = update_feedbrowser_cache();
  1122. _debug("Feedbrowser updated, $count feeds processed.");
  1123. purge_orphans( true);
  1124. cleanup_counters_cache($debug);
  1125. //$rc = cleanup_tags( 14, 50000);
  1126. //_debug("Cleaned $rc cached tags.");
  1127. PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
  1128. }
  1129. ?>