rssfuncs.php 44 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520
  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)."'
  154. $update_limit_qpart
  155. $login_thresh_qpart
  156. ORDER BY ttrss_feeds.id $query_limit");
  157. if (db_num_rows($tmp_result) > 0) {
  158. $rss = false;
  159. while ($tline = db_fetch_assoc($tmp_result)) {
  160. if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
  161. $fstarted = microtime(true);
  162. $rss = update_rss_feed($tline["id"], true, false);
  163. _debug_suppress(false);
  164. _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
  165. ++$nf;
  166. }
  167. }
  168. }
  169. if ($nf > 0) {
  170. _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
  171. microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
  172. }
  173. require_once "digest.php";
  174. // Send feed digests by email if needed.
  175. send_headlines_digests($debug);
  176. return $nf;
  177. } // function update_daemon_common
  178. // this is used when subscribing
  179. function set_basic_feed_info($feed) {
  180. $feed = db_escape_string($feed);
  181. $result = db_query("SELECT feed_url,auth_pass,auth_login,auth_pass_encrypted
  182. FROM ttrss_feeds WHERE id = '$feed'");
  183. $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
  184. 0, "auth_pass_encrypted"));
  185. $auth_login = db_fetch_result($result, 0, "auth_login");
  186. $auth_pass = db_fetch_result($result, 0, "auth_pass");
  187. if ($auth_pass_encrypted) {
  188. require_once "crypt.php";
  189. $auth_pass = decrypt_string($auth_pass);
  190. }
  191. $fetch_url = db_fetch_result($result, 0, "feed_url");
  192. $feed_data = fetch_file_contents($fetch_url, false,
  193. $auth_login, $auth_pass, false,
  194. FEED_FETCH_TIMEOUT_TIMEOUT,
  195. 0);
  196. global $fetch_curl_used;
  197. if (!$fetch_curl_used) {
  198. $tmp = @gzdecode($feed_data);
  199. if ($tmp) $feed_data = $tmp;
  200. }
  201. $feed_data = trim($feed_data);
  202. $rss = new FeedParser($feed_data);
  203. $rss->init();
  204. if (!$rss->error()) {
  205. $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
  206. $registered_title = db_fetch_result($result, 0, "title");
  207. $orig_site_url = db_fetch_result($result, 0, "site_url");
  208. $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
  209. $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
  210. if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
  211. db_query("UPDATE ttrss_feeds SET
  212. title = '$feed_title' WHERE id = '$feed'");
  213. }
  214. if ($site_url && $orig_site_url != $site_url) {
  215. db_query("UPDATE ttrss_feeds SET
  216. site_url = '$site_url' WHERE id = '$feed'");
  217. }
  218. }
  219. }
  220. // ignore_daemon is not used
  221. function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
  222. $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
  223. _debug_suppress(!$debug_enabled);
  224. _debug("start", $debug_enabled);
  225. $result = db_query("SELECT title FROM ttrss_feeds
  226. WHERE id = '$feed'");
  227. $title = db_fetch_result($result, 0, "title");
  228. // feed was batch-subscribed or something, we need to get basic info
  229. // this is not optimal currently as it fetches stuff separately TODO: optimize
  230. if ($title == "[Unknown]") {
  231. _debug("setting basic feed info for $feed...");
  232. set_basic_feed_info($feed);
  233. }
  234. $result = db_query("SELECT id,update_interval,auth_login,
  235. feed_url,auth_pass,cache_images,
  236. mark_unread_on_update, owner_uid,
  237. pubsub_state, auth_pass_encrypted,
  238. feed_language
  239. FROM ttrss_feeds WHERE id = '$feed'");
  240. if (db_num_rows($result) == 0) {
  241. _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
  242. return false;
  243. }
  244. $owner_uid = db_fetch_result($result, 0, "owner_uid");
  245. $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
  246. 0, "mark_unread_on_update"));
  247. $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
  248. $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
  249. 0, "auth_pass_encrypted"));
  250. db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
  251. WHERE id = '$feed'");
  252. $auth_login = db_fetch_result($result, 0, "auth_login");
  253. $auth_pass = db_fetch_result($result, 0, "auth_pass");
  254. if ($auth_pass_encrypted) {
  255. require_once "crypt.php";
  256. $auth_pass = decrypt_string($auth_pass);
  257. }
  258. $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
  259. $fetch_url = db_fetch_result($result, 0, "feed_url");
  260. $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
  261. if (!$feed_language) $feed_language = 'english';
  262. $feed = db_escape_string($feed);
  263. $date_feed_processed = date('Y-m-d H:i');
  264. $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
  265. $pluginhost = new PluginHost();
  266. $pluginhost->set_debug($debug_enabled);
  267. $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
  268. $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
  269. $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
  270. $pluginhost->load_data();
  271. if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
  272. _debug("using previously initialized parser object");
  273. } else {
  274. $rss_hash = false;
  275. $force_refetch = isset($_REQUEST["force_refetch"]);
  276. $feed_data = "";
  277. foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
  278. $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $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. if (ini_get("open_basedir") && function_exists("curl_init")) {
  298. _debug("not using CURL due to open_basedir restrictions");
  299. }
  300. $feed_data = fetch_file_contents($fetch_url, false,
  301. $auth_login, $auth_pass, false,
  302. $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
  303. 0);
  304. global $fetch_curl_used;
  305. if (!$fetch_curl_used) {
  306. $tmp = @gzdecode($feed_data);
  307. if ($tmp) $feed_data = $tmp;
  308. }
  309. $feed_data = trim($feed_data);
  310. _debug("fetch done.", $debug_enabled);
  311. // cache vanilla feed data for re-use
  312. if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
  313. $new_rss_hash = sha1($feed_data);
  314. if ($new_rss_hash != $rss_hash) {
  315. _debug("saving $cache_filename", $debug_enabled);
  316. @file_put_contents($cache_filename, $feed_data);
  317. }
  318. }
  319. }
  320. if (!$feed_data) {
  321. global $fetch_last_error;
  322. global $fetch_last_error_code;
  323. _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
  324. $error_escaped = '';
  325. // If-Modified-Since
  326. if ($fetch_last_error_code != 304) {
  327. $error_escaped = db_escape_string($fetch_last_error);
  328. } else {
  329. _debug("source claims data not modified, nothing to do.", $debug_enabled);
  330. }
  331. db_query(
  332. "UPDATE ttrss_feeds SET last_error = '$error_escaped',
  333. last_updated = NOW() WHERE id = '$feed'");
  334. return;
  335. }
  336. }
  337. foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
  338. $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
  339. }
  340. // set last update to now so if anything *simplepie* crashes later we won't be
  341. // continuously failing on the same feed
  342. //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
  343. if (!$rss) {
  344. $rss = new FeedParser($feed_data);
  345. $rss->init();
  346. }
  347. // print_r($rss);
  348. $feed = db_escape_string($feed);
  349. if (!$rss->error()) {
  350. // We use local pluginhost here because we need to load different per-user feed plugins
  351. $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
  352. _debug("language: $feed_language", $debug_enabled);
  353. _debug("processing feed data...", $debug_enabled);
  354. // db_query("BEGIN");
  355. if (DB_TYPE == "pgsql") {
  356. $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
  357. } else {
  358. $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
  359. }
  360. $result = db_query("SELECT owner_uid,favicon_avg_color,
  361. (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
  362. favicon_needs_check
  363. FROM ttrss_feeds WHERE id = '$feed'");
  364. $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
  365. "favicon_needs_check"));
  366. $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
  367. $owner_uid = db_fetch_result($result, 0, "owner_uid");
  368. $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
  369. _debug("site_url: $site_url", $debug_enabled);
  370. _debug("feed_title: " . $rss->get_title(), $debug_enabled);
  371. if ($favicon_needs_check || $force_refetch) {
  372. /* terrible hack: if we crash on floicon shit here, we won't check
  373. * the icon avgcolor again (unless the icon got updated) */
  374. $favicon_file = ICONS_DIR . "/$feed.ico";
  375. $favicon_modified = @filemtime($favicon_file);
  376. _debug("checking favicon...", $debug_enabled);
  377. check_feed_favicon($site_url, $feed);
  378. $favicon_modified_new = @filemtime($favicon_file);
  379. if ($favicon_modified_new > $favicon_modified)
  380. $favicon_avg_color = '';
  381. if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
  382. require_once "colors.php";
  383. db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
  384. id = '$feed'");
  385. $favicon_color = db_escape_string(
  386. calculate_avg_color($favicon_file));
  387. $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
  388. } else if ($favicon_avg_color == 'fail') {
  389. _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
  390. }
  391. db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
  392. $favicon_colorstring
  393. WHERE id = '$feed'");
  394. }
  395. _debug("loading filters & labels...", $debug_enabled);
  396. $filters = load_filters($feed, $owner_uid);
  397. _debug("" . count($filters) . " filters loaded.", $debug_enabled);
  398. $items = $rss->get_items();
  399. if (!is_array($items)) {
  400. _debug("no articles found.", $debug_enabled);
  401. db_query("UPDATE ttrss_feeds
  402. SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
  403. return; // no articles
  404. }
  405. if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
  406. _debug("checking for PUSH hub...", $debug_enabled);
  407. $feed_hub_url = false;
  408. $links = $rss->get_links('hub');
  409. if ($links && is_array($links)) {
  410. foreach ($links as $l) {
  411. $feed_hub_url = $l;
  412. break;
  413. }
  414. }
  415. _debug("feed hub url: $feed_hub_url", $debug_enabled);
  416. $feed_self_url = $fetch_url;
  417. $links = $rss->get_links('self');
  418. if ($links && is_array($links)) {
  419. foreach ($links as $l) {
  420. $feed_self_url = $l;
  421. break;
  422. }
  423. }
  424. _debug("feed self url = $feed_self_url");
  425. if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
  426. !ini_get("open_basedir")) {
  427. require_once 'lib/pubsubhubbub/subscriber.php';
  428. $callback_url = get_self_url_prefix() .
  429. "/public.php?op=pubsub&id=$feed";
  430. $s = new Subscriber($feed_hub_url, $callback_url);
  431. $rc = $s->subscribe($feed_self_url);
  432. _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
  433. db_query("UPDATE ttrss_feeds SET pubsub_state = 1
  434. WHERE id = '$feed'");
  435. }
  436. }
  437. _debug("processing articles...", $debug_enabled);
  438. $tstart = time();
  439. foreach ($items as $item) {
  440. if ($_REQUEST['xdebug'] == 3) {
  441. print_r($item);
  442. }
  443. if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
  444. _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
  445. break;
  446. }
  447. $entry_guid = $item->get_id();
  448. if (!$entry_guid) $entry_guid = $item->get_link();
  449. if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
  450. if (!$entry_guid) continue;
  451. $entry_guid = "$owner_uid,$entry_guid";
  452. $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
  453. _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
  454. $entry_timestamp = "";
  455. $entry_timestamp = $item->get_date();
  456. _debug("orig date: " . $item->get_date(), $debug_enabled);
  457. if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
  458. $entry_timestamp = time();
  459. }
  460. $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
  461. _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
  462. // $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
  463. // $entry_title = decode_numeric_entities($entry_title);
  464. $entry_title = $item->get_title();
  465. $entry_link = rewrite_relative_url($site_url, $item->get_link());
  466. _debug("title $entry_title", $debug_enabled);
  467. _debug("link $entry_link", $debug_enabled);
  468. if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
  469. $entry_content = $item->get_content();
  470. if (!$entry_content) $entry_content = $item->get_description();
  471. if ($_REQUEST["xdebug"] == 2) {
  472. print "content: ";
  473. print htmlspecialchars($entry_content);
  474. print "\n";
  475. }
  476. $entry_comments = $item->get_comments_url();
  477. $entry_author = $item->get_author();
  478. $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
  479. $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
  480. $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
  481. $num_comments = (int) $item->get_comments_count();
  482. _debug("author $entry_author", $debug_enabled);
  483. _debug("num_comments: $num_comments", $debug_enabled);
  484. _debug("looking for tags...", $debug_enabled);
  485. // parse <category> entries into tags
  486. $additional_tags = array();
  487. $additional_tags_src = $item->get_categories();
  488. if (is_array($additional_tags_src)) {
  489. foreach ($additional_tags_src as $tobj) {
  490. array_push($additional_tags, $tobj);
  491. }
  492. }
  493. $entry_tags = array_unique($additional_tags);
  494. for ($i = 0; $i < count($entry_tags); $i++)
  495. $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
  496. _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
  497. _debug("done collecting data.", $debug_enabled);
  498. $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
  499. WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
  500. if (db_num_rows($result) != 0) {
  501. $base_entry_id = db_fetch_result($result, 0, "id");
  502. $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
  503. $article_labels = get_article_labels($base_entry_id, $owner_uid);
  504. $entry_language = db_fetch_result($result, 0, "lang");
  505. $existing_tags = get_article_tags($base_entry_id, $owner_uid);
  506. $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
  507. } else {
  508. $base_entry_id = false;
  509. $entry_stored_hash = "";
  510. $article_labels = array();
  511. $entry_language = "";
  512. }
  513. $article = array("owner_uid" => $owner_uid, // read only
  514. "guid" => $entry_guid, // read only
  515. "guid_hashed" => $entry_guid_hashed, // read only
  516. "title" => $entry_title,
  517. "content" => $entry_content,
  518. "link" => $entry_link,
  519. "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
  520. "tags" => $entry_tags,
  521. "author" => $entry_author,
  522. "force_catchup" => false, // ugly hack for the time being
  523. "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
  524. "language" => $entry_language,
  525. "feed" => array("id" => $feed,
  526. "fetch_url" => $fetch_url,
  527. "site_url" => $site_url)
  528. );
  529. $entry_plugin_data = "";
  530. $entry_current_hash = calculate_article_hash($article, $pluginhost);
  531. _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
  532. if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
  533. _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
  534. // we keep encountering the entry in feeds, so we need to
  535. // update date_updated column so that we don't get horrible
  536. // dupes when the entry gets purged and reinserted again e.g.
  537. // in the case of SLOW SLOW OMG SLOW updating feeds
  538. $base_entry_id = db_fetch_result($result, 0, "id");
  539. db_query("UPDATE ttrss_entries SET date_updated = NOW()
  540. WHERE id = '$base_entry_id'");
  541. continue;
  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. if ($_REQUEST["xdebug"] == 2) {
  552. print "processed content: ";
  553. print htmlspecialchars($article["content"]);
  554. print "\n";
  555. }
  556. $entry_plugin_data = db_escape_string($entry_plugin_data);
  557. _debug("plugin data: $entry_plugin_data", $debug_enabled);
  558. // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
  559. if (DB_TYPE == "mysql") {
  560. foreach ($article as $k => $v) {
  561. // i guess we'll have to take the risk of 4byte unicode labels & tags here
  562. if (!is_array($article[$k])) {
  563. $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
  564. }
  565. }
  566. }
  567. /* Collect article tags here so we could filter by them: */
  568. $article_filters = get_article_filters($filters, $article["title"],
  569. $article["content"], $article["link"], 0, $article["author"],
  570. $article["tags"]);
  571. if ($debug_enabled) {
  572. _debug("article filters: ", $debug_enabled);
  573. if (count($article_filters) != 0) {
  574. print_r($article_filters);
  575. }
  576. }
  577. $plugin_filter_names = find_article_filters($article_filters, "plugin");
  578. $plugin_filter_actions = $pluginhost->get_filter_actions();
  579. if (count($plugin_filter_names) > 0) {
  580. _debug("applying plugin filter actions...", $debug_enabled);
  581. foreach ($plugin_filter_names as $pfn) {
  582. list($pfclass,$pfaction) = explode(":", $pfn["param"]);
  583. if (isset($plugin_filter_actions[$pfclass])) {
  584. $plugin = $pluginhost->get_plugin($pfclass);
  585. _debug("... $pfclass: $pfaction", $debug_enabled);
  586. if ($plugin) {
  587. $start = microtime(true);
  588. $article = $plugin->hook_article_filter_action($article, $pfaction);
  589. _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
  590. } else {
  591. _debug("??? $pfclass: plugin object not found.");
  592. }
  593. } else {
  594. _debug("??? $pfclass: filter plugin not registered.");
  595. }
  596. }
  597. }
  598. $entry_tags = $article["tags"];
  599. $entry_guid = db_escape_string($entry_guid);
  600. $entry_title = db_escape_string($article["title"]);
  601. $entry_author = db_escape_string($article["author"]);
  602. $entry_link = db_escape_string($article["link"]);
  603. $entry_content = $article["content"]; // escaped below
  604. $entry_force_catchup = $article["force_catchup"];
  605. $article_labels = $article["labels"];
  606. $entry_score_modifier = (int) $article["score_modifier"];
  607. $entry_language = db_escape_string($article["language"]);
  608. if ($debug_enabled) {
  609. _debug("article labels:", $debug_enabled);
  610. print_r($article_labels);
  611. }
  612. _debug("force catchup: $entry_force_catchup");
  613. if ($cache_images && is_writable(CACHE_DIR . '/images'))
  614. cache_images($entry_content, $site_url, $debug_enabled);
  615. $entry_content = db_escape_string($entry_content, false);
  616. //db_query("BEGIN");
  617. $result = db_query("SELECT id FROM ttrss_entries
  618. WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
  619. if (db_num_rows($result) == 0) {
  620. _debug("base guid [$entry_guid] not found", $debug_enabled);
  621. // base post entry does not exist, create it
  622. $result = db_query(
  623. "INSERT INTO ttrss_entries
  624. (title,
  625. guid,
  626. link,
  627. updated,
  628. content,
  629. content_hash,
  630. no_orig_date,
  631. date_updated,
  632. date_entered,
  633. comments,
  634. num_comments,
  635. plugin_data,
  636. lang,
  637. author)
  638. VALUES
  639. ('$entry_title',
  640. '$entry_guid_hashed',
  641. '$entry_link',
  642. '$entry_timestamp_fmt',
  643. '$entry_content',
  644. '$entry_current_hash',
  645. false,
  646. NOW(),
  647. '$date_feed_processed',
  648. '$entry_comments',
  649. '$num_comments',
  650. '$entry_plugin_data',
  651. '$entry_language',
  652. '$entry_author')");
  653. } else {
  654. $base_entry_id = db_fetch_result($result, 0, "id");
  655. }
  656. // now it should exist, if not - bad luck then
  657. $result = db_query("SELECT id FROM ttrss_entries
  658. WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
  659. $entry_ref_id = 0;
  660. $entry_int_id = 0;
  661. if (db_num_rows($result) == 1) {
  662. _debug("base guid found, checking for user record", $debug_enabled);
  663. $ref_id = db_fetch_result($result, 0, "id");
  664. $entry_ref_id = $ref_id;
  665. /* $stored_guid = db_fetch_result($result, 0, "guid");
  666. if ($stored_guid != $entry_guid_hashed) {
  667. if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
  668. db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
  669. id = '$ref_id'");
  670. } */
  671. if (find_article_filter($article_filters, "filter")) {
  672. //db_query("COMMIT"); // close transaction in progress
  673. continue;
  674. }
  675. $score = calculate_article_score($article_filters) + $entry_score_modifier;
  676. _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
  677. // check for user post link to main table
  678. $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
  679. ref_id = '$ref_id' AND owner_uid = '$owner_uid'";
  680. // if ($_REQUEST["xdebug"]) print "$query\n";
  681. $result = db_query($query);
  682. // okay it doesn't exist - create user entry
  683. if (db_num_rows($result) == 0) {
  684. _debug("user record not found, creating...", $debug_enabled);
  685. if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
  686. $unread = 'true';
  687. $last_read_qpart = 'NULL';
  688. } else {
  689. $unread = 'false';
  690. $last_read_qpart = 'NOW()';
  691. }
  692. if (find_article_filter($article_filters, 'mark') || $score > 1000) {
  693. $marked = 'true';
  694. } else {
  695. $marked = 'false';
  696. }
  697. if (find_article_filter($article_filters, 'publish')) {
  698. $published = 'true';
  699. } else {
  700. $published = 'false';
  701. }
  702. // N-grams
  703. /* if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
  704. $result = db_query("SELECT COUNT(*) AS similar FROM
  705. ttrss_entries,ttrss_user_entries
  706. WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
  707. AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
  708. AND owner_uid = $owner_uid");
  709. $ngram_similar = db_fetch_result($result, 0, "similar");
  710. _debug("N-gram similar results: $ngram_similar", $debug_enabled);
  711. if ($ngram_similar > 0) {
  712. $unread = 'false';
  713. }
  714. } */
  715. $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
  716. $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
  717. $result = db_query(
  718. "INSERT INTO ttrss_user_entries
  719. (ref_id, owner_uid, feed_id, unread, last_read, marked,
  720. published, score, tag_cache, label_cache, uuid,
  721. last_marked, last_published)
  722. VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
  723. $last_read_qpart, $marked, $published, '$score', '', '',
  724. '', $last_marked, $last_published)");
  725. if (PUBSUBHUBBUB_HUB && $published == 'true') {
  726. $rss_link = get_self_url_prefix() .
  727. "/public.php?op=rss&id=-2&key=" .
  728. get_feed_access_key(-2, false, $owner_uid);
  729. $p = new Publisher(PUBSUBHUBBUB_HUB);
  730. /* $pubsub_result = */ $p->publish_update($rss_link);
  731. }
  732. $result = db_query(
  733. "SELECT int_id FROM ttrss_user_entries WHERE
  734. ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
  735. feed_id = '$feed' LIMIT 1");
  736. if (db_num_rows($result) == 1) {
  737. $entry_int_id = db_fetch_result($result, 0, "int_id");
  738. }
  739. } else {
  740. _debug("user record FOUND", $debug_enabled);
  741. $entry_ref_id = db_fetch_result($result, 0, "ref_id");
  742. $entry_int_id = db_fetch_result($result, 0, "int_id");
  743. }
  744. _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
  745. if (DB_TYPE == "pgsql") {
  746. $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
  747. 0, 1000000));
  748. $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
  749. } else {
  750. $tsvector_qpart = "";
  751. }
  752. db_query("UPDATE ttrss_entries
  753. SET title = '$entry_title',
  754. content = '$entry_content',
  755. content_hash = '$entry_current_hash',
  756. updated = '$entry_timestamp_fmt',
  757. $tsvector_qpart
  758. num_comments = '$num_comments',
  759. plugin_data = '$entry_plugin_data',
  760. author = '$entry_author',
  761. lang = '$entry_language'
  762. WHERE id = '$ref_id'");
  763. // update aux data
  764. db_query("UPDATE ttrss_user_entries
  765. SET score = '$score' WHERE ref_id = '$ref_id'");
  766. if ($mark_unread_on_update) {
  767. _debug("article updated, marking unread as requested.", $debug_enabled);
  768. db_query("UPDATE ttrss_user_entries
  769. SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
  770. }
  771. }
  772. //db_query("COMMIT");
  773. _debug("assigning labels [other]...", $debug_enabled);
  774. foreach ($article_labels as $label) {
  775. label_add_article($entry_ref_id, $label[1], $owner_uid);
  776. }
  777. _debug("assigning labels [filters]...", $debug_enabled);
  778. assign_article_to_label_filters($entry_ref_id, $article_filters,
  779. $owner_uid, $article_labels);
  780. _debug("looking for enclosures...", $debug_enabled);
  781. // enclosures
  782. $enclosures = array();
  783. $encs = $item->get_enclosures();
  784. if (is_array($encs)) {
  785. foreach ($encs as $e) {
  786. $e_item = array(
  787. $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
  788. array_push($enclosures, $e_item);
  789. }
  790. }
  791. if ($debug_enabled) {
  792. _debug("article enclosures:", $debug_enabled);
  793. print_r($enclosures);
  794. }
  795. //db_query("BEGIN");
  796. // debugging
  797. // db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
  798. foreach ($enclosures as $enc) {
  799. $enc_url = db_escape_string($enc[0]);
  800. $enc_type = db_escape_string($enc[1]);
  801. $enc_dur = db_escape_string($enc[2]);
  802. $enc_title = db_escape_string($enc[3]);
  803. $enc_width = intval($enc[4]);
  804. $enc_height = intval($enc[5]);
  805. $result = db_query("SELECT id FROM ttrss_enclosures
  806. WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
  807. if (db_num_rows($result) == 0) {
  808. db_query("INSERT INTO ttrss_enclosures
  809. (content_url, content_type, title, duration, post_id, width, height) VALUES
  810. ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
  811. }
  812. }
  813. //db_query("COMMIT");
  814. // check for manual tags (we have to do it here since they're loaded from filters)
  815. foreach ($article_filters as $f) {
  816. if ($f["type"] == "tag") {
  817. $manual_tags = trim_array(explode(",", $f["param"]));
  818. foreach ($manual_tags as $tag) {
  819. if (tag_is_valid($tag)) {
  820. array_push($entry_tags, $tag);
  821. }
  822. }
  823. }
  824. }
  825. // Skip boring tags
  826. $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
  827. 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
  828. $filtered_tags = array();
  829. $tags_to_cache = array();
  830. if ($entry_tags && is_array($entry_tags)) {
  831. foreach ($entry_tags as $tag) {
  832. if (array_search($tag, $boring_tags) === false) {
  833. array_push($filtered_tags, $tag);
  834. }
  835. }
  836. }
  837. $filtered_tags = array_unique($filtered_tags);
  838. if ($debug_enabled) {
  839. _debug("filtered article tags:", $debug_enabled);
  840. print_r($filtered_tags);
  841. }
  842. // Save article tags in the database
  843. if (count($filtered_tags) > 0) {
  844. //db_query("BEGIN");
  845. foreach ($filtered_tags as $tag) {
  846. $tag = sanitize_tag($tag);
  847. $tag = db_escape_string($tag);
  848. if (!tag_is_valid($tag)) continue;
  849. $result = db_query("SELECT id FROM ttrss_tags
  850. WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
  851. owner_uid = '$owner_uid' LIMIT 1");
  852. if ($result && db_num_rows($result) == 0) {
  853. db_query("INSERT INTO ttrss_tags
  854. (owner_uid,tag_name,post_int_id)
  855. VALUES ('$owner_uid','$tag', '$entry_int_id')");
  856. }
  857. array_push($tags_to_cache, $tag);
  858. }
  859. /* update the cache */
  860. $tags_to_cache = array_unique($tags_to_cache);
  861. $tags_str = db_escape_string(join(",", $tags_to_cache));
  862. db_query("UPDATE ttrss_user_entries
  863. SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
  864. AND owner_uid = $owner_uid");
  865. //db_query("COMMIT");
  866. }
  867. _debug("article processed", $debug_enabled);
  868. }
  869. _debug("purging feed...", $debug_enabled);
  870. purge_feed($feed, 0, $debug_enabled);
  871. db_query("UPDATE ttrss_feeds
  872. SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
  873. // db_query("COMMIT");
  874. } else {
  875. $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
  876. _debug("fetch error: $error_msg", $debug_enabled);
  877. if (count($rss->errors()) > 1) {
  878. foreach ($rss->errors() as $error) {
  879. _debug("+ $error");
  880. }
  881. }
  882. db_query(
  883. "UPDATE ttrss_feeds SET last_error = '$error_msg',
  884. last_updated = NOW() WHERE id = '$feed'");
  885. unset($rss);
  886. }
  887. _debug("done", $debug_enabled);
  888. return $rss;
  889. }
  890. function cache_images($html, $site_url, $debug) {
  891. libxml_use_internal_errors(true);
  892. $charset_hack = '<head>
  893. <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
  894. </head>';
  895. $doc = new DOMDocument();
  896. $doc->loadHTML($charset_hack . $html);
  897. $xpath = new DOMXPath($doc);
  898. $entries = $xpath->query('(//img[@src])');
  899. foreach ($entries as $entry) {
  900. if ($entry->hasAttribute('src')) {
  901. $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
  902. $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
  903. if ($debug) _debug("cache_images: downloading: $src to $local_filename");
  904. if (!file_exists($local_filename)) {
  905. $file_content = fetch_file_contents($src);
  906. if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
  907. file_put_contents($local_filename, $file_content);
  908. }
  909. } else {
  910. touch($local_filename);
  911. }
  912. }
  913. }
  914. }
  915. function expire_error_log($debug) {
  916. if ($debug) _debug("Removing old error log entries...");
  917. if (DB_TYPE == "pgsql") {
  918. db_query("DELETE FROM ttrss_error_log
  919. WHERE created_at < NOW() - INTERVAL '7 days'");
  920. } else {
  921. db_query("DELETE FROM ttrss_error_log
  922. WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
  923. }
  924. }
  925. function expire_lock_files($debug) {
  926. //if ($debug) _debug("Removing old lock files...");
  927. $num_deleted = 0;
  928. if (is_writable(LOCK_DIRECTORY)) {
  929. $files = glob(LOCK_DIRECTORY . "/*.lock");
  930. if ($files) {
  931. foreach ($files as $file) {
  932. if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
  933. unlink($file);
  934. ++$num_deleted;
  935. }
  936. }
  937. }
  938. }
  939. if ($debug) _debug("Removed $num_deleted old lock files.");
  940. }
  941. function expire_cached_files($debug) {
  942. foreach (array("simplepie", "images", "export", "upload") as $dir) {
  943. $cache_dir = CACHE_DIR . "/$dir";
  944. // if ($debug) _debug("Expiring $cache_dir");
  945. $num_deleted = 0;
  946. if (is_writable($cache_dir)) {
  947. $files = glob("$cache_dir/*");
  948. if ($files) {
  949. foreach ($files as $file) {
  950. if (time() - filemtime($file) > 86400*7) {
  951. unlink($file);
  952. ++$num_deleted;
  953. }
  954. }
  955. }
  956. }
  957. if ($debug) _debug("$cache_dir: removed $num_deleted files.");
  958. }
  959. }
  960. /**
  961. * Source: http://www.php.net/manual/en/function.parse-url.php#104527
  962. * Returns the url query as associative array
  963. *
  964. * @param string query
  965. * @return array params
  966. */
  967. function convertUrlQuery($query) {
  968. $queryParts = explode('&', $query);
  969. $params = array();
  970. foreach ($queryParts as $param) {
  971. $item = explode('=', $param);
  972. $params[$item[0]] = $item[1];
  973. }
  974. return $params;
  975. }
  976. function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
  977. $matches = array();
  978. foreach ($filters as $filter) {
  979. $match_any_rule = $filter["match_any_rule"];
  980. $inverse = $filter["inverse"];
  981. $filter_match = false;
  982. foreach ($filter["rules"] as $rule) {
  983. $match = false;
  984. $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
  985. $rule_inverse = $rule["inverse"];
  986. if (!$reg_exp)
  987. continue;
  988. switch ($rule["type"]) {
  989. case "title":
  990. $match = @preg_match("/$reg_exp/i", $title);
  991. break;
  992. case "content":
  993. // we don't need to deal with multiline regexps
  994. $content = preg_replace("/[\r\n\t]/", "", $content);
  995. $match = @preg_match("/$reg_exp/i", $content);
  996. break;
  997. case "both":
  998. // we don't need to deal with multiline regexps
  999. $content = preg_replace("/[\r\n\t]/", "", $content);
  1000. $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
  1001. break;
  1002. case "link":
  1003. $match = @preg_match("/$reg_exp/i", $link);
  1004. break;
  1005. case "author":
  1006. $match = @preg_match("/$reg_exp/i", $author);
  1007. break;
  1008. case "tag":
  1009. foreach ($tags as $tag) {
  1010. if (@preg_match("/$reg_exp/i", $tag)) {
  1011. $match = true;
  1012. break;
  1013. }
  1014. }
  1015. break;
  1016. }
  1017. if ($rule_inverse) $match = !$match;
  1018. if ($match_any_rule) {
  1019. if ($match) {
  1020. $filter_match = true;
  1021. break;
  1022. }
  1023. } else {
  1024. $filter_match = $match;
  1025. if (!$match) {
  1026. break;
  1027. }
  1028. }
  1029. }
  1030. if ($inverse) $filter_match = !$filter_match;
  1031. if ($filter_match) {
  1032. foreach ($filter["actions"] AS $action) {
  1033. array_push($matches, $action);
  1034. // if Stop action encountered, perform no further processing
  1035. if (isset($action["type"]) && $action["type"] == "stop") return $matches;
  1036. }
  1037. }
  1038. }
  1039. return $matches;
  1040. }
  1041. function find_article_filter($filters, $filter_name) {
  1042. foreach ($filters as $f) {
  1043. if ($f["type"] == $filter_name) {
  1044. return $f;
  1045. };
  1046. }
  1047. return false;
  1048. }
  1049. function find_article_filters($filters, $filter_name) {
  1050. $results = array();
  1051. foreach ($filters as $f) {
  1052. if ($f["type"] == $filter_name) {
  1053. array_push($results, $f);
  1054. };
  1055. }
  1056. return $results;
  1057. }
  1058. function calculate_article_score($filters) {
  1059. $score = 0;
  1060. foreach ($filters as $f) {
  1061. if ($f["type"] == "score") {
  1062. $score += $f["param"];
  1063. };
  1064. }
  1065. return $score;
  1066. }
  1067. function labels_contains_caption($labels, $caption) {
  1068. foreach ($labels as $label) {
  1069. if ($label[1] == $caption) {
  1070. return true;
  1071. }
  1072. }
  1073. return false;
  1074. }
  1075. function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
  1076. foreach ($filters as $f) {
  1077. if ($f["type"] == "label") {
  1078. if (!labels_contains_caption($article_labels, $f["param"])) {
  1079. label_add_article($id, $f["param"], $owner_uid);
  1080. }
  1081. }
  1082. }
  1083. }
  1084. function make_guid_from_title($title) {
  1085. return preg_replace("/[ \"\',.:;]/", "-",
  1086. mb_strtolower(strip_tags($title), 'utf-8'));
  1087. }
  1088. /* function verify_feed_xml($feed_data) {
  1089. libxml_use_internal_errors(true);
  1090. $doc = new DOMDocument();
  1091. $doc->loadXML($feed_data);
  1092. $error = libxml_get_last_error();
  1093. libxml_clear_errors();
  1094. return $error;
  1095. } */
  1096. function cleanup_counters_cache($debug) {
  1097. $result = db_query("DELETE FROM ttrss_counters_cache
  1098. WHERE feed_id > 0 AND
  1099. (SELECT COUNT(id) FROM ttrss_feeds WHERE
  1100. id = feed_id AND
  1101. ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
  1102. $frows = db_affected_rows($result);
  1103. $result = db_query("DELETE FROM ttrss_cat_counters_cache
  1104. WHERE feed_id > 0 AND
  1105. (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
  1106. id = feed_id AND
  1107. ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
  1108. $crows = db_affected_rows($result);
  1109. _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
  1110. }
  1111. function housekeeping_common($debug) {
  1112. expire_cached_files($debug);
  1113. expire_lock_files($debug);
  1114. expire_error_log($debug);
  1115. $count = update_feedbrowser_cache();
  1116. _debug("Feedbrowser updated, $count feeds processed.");
  1117. purge_orphans( true);
  1118. cleanup_counters_cache($debug);
  1119. //$rc = cleanup_tags( 14, 50000);
  1120. //_debug("Cleaned $rc cached tags.");
  1121. PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
  1122. }
  1123. ?>