rssfuncs.php 45 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558
  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_FILE_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 feed_url NOT IN (SELECT feed_url FROM ttrss_feeds
  20. WHERE private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%')
  21. GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
  22. db_query("BEGIN");
  23. db_query("DELETE FROM ttrss_feedbrowser_cache");
  24. $count = 0;
  25. while ($line = db_fetch_assoc($result)) {
  26. $subscribers = db_escape_string($line["subscribers"]);
  27. $feed_url = db_escape_string($line["feed_url"]);
  28. $title = db_escape_string($line["title"]);
  29. $site_url = db_escape_string($line["site_url"]);
  30. $tmp_result = db_query("SELECT subscribers FROM
  31. ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
  32. if (db_num_rows($tmp_result) == 0) {
  33. db_query("INSERT INTO ttrss_feedbrowser_cache
  34. (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
  35. '$site_url', '$title', '$subscribers')");
  36. ++$count;
  37. }
  38. }
  39. db_query("COMMIT");
  40. return $count;
  41. }
  42. /**
  43. * Update a feed batch.
  44. * Used by daemons to update n feeds by run.
  45. * Only update feed needing a update, and not being processed
  46. * by another process.
  47. *
  48. * @param mixed $link Database link
  49. * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
  50. * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
  51. * @param boolean $debug Set to false to disable debug output. Default to true.
  52. * @return void
  53. */
  54. function update_daemon_common($limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
  55. // Process all other feeds using last_updated and interval parameters
  56. $schema_version = get_schema_version();
  57. if ($schema_version != SCHEMA_VERSION) {
  58. die("Schema version is wrong, please upgrade the database.\n");
  59. }
  60. define('PREFS_NO_CACHE', true);
  61. // Test if the user has loggued in recently. If not, it does not update its feeds.
  62. if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
  63. if (DB_TYPE == "pgsql") {
  64. $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
  65. } else {
  66. $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
  67. }
  68. } else {
  69. $login_thresh_qpart = "";
  70. }
  71. // Test if the feed need a update (update interval exceeded).
  72. if (DB_TYPE == "pgsql") {
  73. $update_limit_qpart = "AND ((
  74. ttrss_feeds.update_interval = 0
  75. AND ttrss_user_prefs.value != '-1'
  76. AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
  77. ) OR (
  78. ttrss_feeds.update_interval > 0
  79. AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
  80. ) OR (ttrss_feeds.last_updated IS NULL
  81. AND ttrss_user_prefs.value != '-1')
  82. OR (last_updated = '1970-01-01 00:00:00'
  83. AND ttrss_user_prefs.value != '-1'))";
  84. } else {
  85. $update_limit_qpart = "AND ((
  86. ttrss_feeds.update_interval = 0
  87. AND ttrss_user_prefs.value != '-1'
  88. AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
  89. ) OR (
  90. ttrss_feeds.update_interval > 0
  91. AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
  92. ) OR (ttrss_feeds.last_updated IS NULL
  93. AND ttrss_user_prefs.value != '-1')
  94. OR (last_updated = '1970-01-01 00:00:00'
  95. AND ttrss_user_prefs.value != '-1'))";
  96. }
  97. // Test if feed is currently being updated by another process.
  98. if (DB_TYPE == "pgsql") {
  99. $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
  100. } else {
  101. $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
  102. }
  103. // Test if there is a limit to number of updated feeds
  104. $query_limit = "";
  105. if($limit) $query_limit = sprintf("LIMIT %d", $limit);
  106. // Update the least recently updated feeds first
  107. $query_order = "ORDER BY last_updated";
  108. if (DB_TYPE == "pgsql") $query_order .= " NULLS FIRST";
  109. $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
  110. FROM
  111. ttrss_feeds, ttrss_users, ttrss_user_prefs
  112. WHERE
  113. ttrss_feeds.owner_uid = ttrss_users.id
  114. AND ttrss_user_prefs.profile IS NULL
  115. AND ttrss_users.id = ttrss_user_prefs.owner_uid
  116. AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
  117. $login_thresh_qpart $update_limit_qpart
  118. $updstart_thresh_qpart
  119. $query_order $query_limit";
  120. // We search for feed needing update.
  121. $result = db_query($query);
  122. if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
  123. // Here is a little cache magic in order to minimize risk of double feed updates.
  124. $feeds_to_update = array();
  125. while ($line = db_fetch_assoc($result)) {
  126. array_push($feeds_to_update, db_escape_string($line['feed_url']));
  127. }
  128. // We update the feed last update started date before anything else.
  129. // There is no lag due to feed contents downloads
  130. // It prevent an other process to update the same feed.
  131. if(count($feeds_to_update) > 0) {
  132. $feeds_quoted = array();
  133. foreach ($feeds_to_update as $feed) {
  134. array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
  135. }
  136. db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
  137. WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
  138. }
  139. $nf = 0;
  140. $bstarted = microtime(true);
  141. $batch_owners = array();
  142. // For each feed, we call the feed update function.
  143. foreach ($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"] . " " . $tline["owner_uid"]);
  161. if (array_search($tline["owner_uid"], $batch_owners) === FALSE)
  162. array_push($batch_owners, $tline["owner_uid"]);
  163. $fstarted = microtime(true);
  164. $rss = update_rss_feed($tline["id"], true, false);
  165. _debug_suppress(false);
  166. _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
  167. ++$nf;
  168. }
  169. }
  170. }
  171. if ($nf > 0) {
  172. _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
  173. microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
  174. }
  175. foreach ($batch_owners as $owner_uid) {
  176. _debug("Running housekeeping tasks for user $owner_uid...");
  177. housekeeping_user($owner_uid);
  178. }
  179. require_once "digest.php";
  180. // Send feed digests by email if needed.
  181. send_headlines_digests($debug);
  182. return $nf;
  183. } // function update_daemon_common
  184. // this is used when subscribing
  185. function set_basic_feed_info($feed) {
  186. $feed = db_escape_string($feed);
  187. $result = db_query("SELECT feed_url,auth_pass,auth_login,auth_pass_encrypted
  188. FROM ttrss_feeds WHERE id = '$feed'");
  189. $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
  190. 0, "auth_pass_encrypted"));
  191. $auth_login = db_fetch_result($result, 0, "auth_login");
  192. $auth_pass = db_fetch_result($result, 0, "auth_pass");
  193. if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
  194. require_once "crypt.php";
  195. $auth_pass = decrypt_string($auth_pass);
  196. }
  197. $fetch_url = db_fetch_result($result, 0, "feed_url");
  198. $feed_data = fetch_file_contents($fetch_url, false,
  199. $auth_login, $auth_pass, false,
  200. FEED_FETCH_TIMEOUT,
  201. 0);
  202. global $fetch_curl_used;
  203. if (!$fetch_curl_used) {
  204. $tmp = @gzdecode($feed_data);
  205. if ($tmp) $feed_data = $tmp;
  206. }
  207. $feed_data = trim($feed_data);
  208. $rss = new FeedParser($feed_data);
  209. $rss->init();
  210. if (!$rss->error()) {
  211. $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
  212. $registered_title = db_fetch_result($result, 0, "title");
  213. $orig_site_url = db_fetch_result($result, 0, "site_url");
  214. $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
  215. $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
  216. if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
  217. db_query("UPDATE ttrss_feeds SET
  218. title = '$feed_title' WHERE id = '$feed'");
  219. }
  220. if ($site_url && $orig_site_url != $site_url) {
  221. db_query("UPDATE ttrss_feeds SET
  222. site_url = '$site_url' WHERE id = '$feed'");
  223. }
  224. }
  225. }
  226. // ignore_daemon is not used
  227. function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
  228. $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
  229. _debug_suppress(!$debug_enabled);
  230. _debug("start", $debug_enabled);
  231. $result = db_query("SELECT title FROM ttrss_feeds
  232. WHERE id = '$feed'");
  233. if (db_num_rows($result) == 0) {
  234. _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
  235. user_error("Attempt to update unknown/invalid feed $feed", E_USER_WARNING);
  236. return false;
  237. }
  238. $title = db_fetch_result($result, 0, "title");
  239. // feed was batch-subscribed or something, we need to get basic info
  240. // this is not optimal currently as it fetches stuff separately TODO: optimize
  241. if ($title == "[Unknown]") {
  242. _debug("setting basic feed info for $feed...");
  243. set_basic_feed_info($feed);
  244. }
  245. $result = db_query("SELECT id,update_interval,auth_login,
  246. feed_url,auth_pass,cache_images,
  247. mark_unread_on_update, owner_uid,
  248. pubsub_state, auth_pass_encrypted,
  249. feed_language
  250. FROM ttrss_feeds WHERE id = '$feed'");
  251. $owner_uid = db_fetch_result($result, 0, "owner_uid");
  252. $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
  253. 0, "mark_unread_on_update"));
  254. $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
  255. $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
  256. 0, "auth_pass_encrypted"));
  257. db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
  258. WHERE id = '$feed'");
  259. $auth_login = db_fetch_result($result, 0, "auth_login");
  260. $auth_pass = db_fetch_result($result, 0, "auth_pass");
  261. if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
  262. require_once "crypt.php";
  263. $auth_pass = decrypt_string($auth_pass);
  264. }
  265. $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
  266. $fetch_url = db_fetch_result($result, 0, "feed_url");
  267. $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
  268. if (!$feed_language) $feed_language = 'english';
  269. $feed = db_escape_string($feed);
  270. $date_feed_processed = date('Y-m-d H:i');
  271. $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
  272. $pluginhost = new PluginHost();
  273. $pluginhost->set_debug($debug_enabled);
  274. $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
  275. $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
  276. $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
  277. $pluginhost->load_data();
  278. if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
  279. _debug("using previously initialized parser object");
  280. } else {
  281. $rss_hash = false;
  282. $force_refetch = isset($_REQUEST["force_refetch"]);
  283. $feed_data = "";
  284. foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
  285. $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
  286. }
  287. // try cache
  288. if (!$feed_data &&
  289. file_exists($cache_filename) &&
  290. is_readable($cache_filename) &&
  291. !$auth_login && !$auth_pass &&
  292. filemtime($cache_filename) > time() - 30) {
  293. _debug("using local cache [$cache_filename].", $debug_enabled);
  294. @$feed_data = file_get_contents($cache_filename);
  295. if ($feed_data) {
  296. $rss_hash = sha1($feed_data);
  297. }
  298. } else {
  299. _debug("local cache will not be used for this feed", $debug_enabled);
  300. }
  301. // fetch feed from source
  302. if (!$feed_data) {
  303. _debug("fetching [$fetch_url]...", $debug_enabled);
  304. if (ini_get("open_basedir") && function_exists("curl_init")) {
  305. _debug("not using CURL due to open_basedir restrictions");
  306. }
  307. $feed_data = fetch_file_contents($fetch_url, false,
  308. $auth_login, $auth_pass, false,
  309. $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
  310. 0);
  311. global $fetch_curl_used;
  312. if (!$fetch_curl_used) {
  313. $tmp = @gzdecode($feed_data);
  314. if ($tmp) $feed_data = $tmp;
  315. }
  316. $feed_data = trim($feed_data);
  317. _debug("fetch done.", $debug_enabled);
  318. // cache vanilla feed data for re-use
  319. if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
  320. $new_rss_hash = sha1($feed_data);
  321. if ($new_rss_hash != $rss_hash) {
  322. _debug("saving $cache_filename", $debug_enabled);
  323. @file_put_contents($cache_filename, $feed_data);
  324. }
  325. }
  326. }
  327. if (!$feed_data) {
  328. global $fetch_last_error;
  329. global $fetch_last_error_code;
  330. _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
  331. $error_escaped = '';
  332. // If-Modified-Since
  333. if ($fetch_last_error_code != 304) {
  334. $error_escaped = db_escape_string($fetch_last_error);
  335. } else {
  336. _debug("source claims data not modified, nothing to do.", $debug_enabled);
  337. }
  338. db_query(
  339. "UPDATE ttrss_feeds SET last_error = '$error_escaped',
  340. last_updated = NOW() WHERE id = '$feed'");
  341. return;
  342. }
  343. }
  344. foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
  345. $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
  346. }
  347. // set last update to now so if anything *simplepie* crashes later we won't be
  348. // continuously failing on the same feed
  349. //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
  350. if (!$rss) {
  351. $rss = new FeedParser($feed_data);
  352. $rss->init();
  353. }
  354. // print_r($rss);
  355. $feed = db_escape_string($feed);
  356. if (!$rss->error()) {
  357. // We use local pluginhost here because we need to load different per-user feed plugins
  358. $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
  359. _debug("language: $feed_language", $debug_enabled);
  360. _debug("processing feed data...", $debug_enabled);
  361. // db_query("BEGIN");
  362. if (DB_TYPE == "pgsql") {
  363. $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
  364. } else {
  365. $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
  366. }
  367. $result = db_query("SELECT owner_uid,favicon_avg_color,
  368. (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
  369. favicon_needs_check
  370. FROM ttrss_feeds WHERE id = '$feed'");
  371. $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
  372. "favicon_needs_check"));
  373. $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
  374. $owner_uid = db_fetch_result($result, 0, "owner_uid");
  375. $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
  376. _debug("site_url: $site_url", $debug_enabled);
  377. _debug("feed_title: " . $rss->get_title(), $debug_enabled);
  378. if ($favicon_needs_check || $force_refetch) {
  379. /* terrible hack: if we crash on floicon shit here, we won't check
  380. * the icon avgcolor again (unless the icon got updated) */
  381. $favicon_file = ICONS_DIR . "/$feed.ico";
  382. $favicon_modified = @filemtime($favicon_file);
  383. _debug("checking favicon...", $debug_enabled);
  384. check_feed_favicon($site_url, $feed);
  385. $favicon_modified_new = @filemtime($favicon_file);
  386. if ($favicon_modified_new > $favicon_modified)
  387. $favicon_avg_color = '';
  388. if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
  389. require_once "colors.php";
  390. db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
  391. id = '$feed'");
  392. $favicon_color = db_escape_string(
  393. calculate_avg_color($favicon_file));
  394. $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
  395. } else if ($favicon_avg_color == 'fail') {
  396. _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
  397. }
  398. db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
  399. $favicon_colorstring
  400. WHERE id = '$feed'");
  401. }
  402. _debug("loading filters & labels...", $debug_enabled);
  403. $filters = load_filters($feed, $owner_uid);
  404. _debug("" . count($filters) . " filters loaded.", $debug_enabled);
  405. $items = $rss->get_items();
  406. if (!is_array($items)) {
  407. _debug("no articles found.", $debug_enabled);
  408. db_query("UPDATE ttrss_feeds
  409. SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
  410. return; // no articles
  411. }
  412. if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
  413. _debug("checking for PUSH hub...", $debug_enabled);
  414. $feed_hub_url = false;
  415. $links = $rss->get_links('hub');
  416. if ($links && is_array($links)) {
  417. foreach ($links as $l) {
  418. $feed_hub_url = $l;
  419. break;
  420. }
  421. }
  422. _debug("feed hub url: $feed_hub_url", $debug_enabled);
  423. $feed_self_url = $fetch_url;
  424. $links = $rss->get_links('self');
  425. if ($links && is_array($links)) {
  426. foreach ($links as $l) {
  427. $feed_self_url = $l;
  428. break;
  429. }
  430. }
  431. _debug("feed self url = $feed_self_url");
  432. if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
  433. !ini_get("open_basedir")) {
  434. require_once 'lib/pubsubhubbub/Subscriber.php';
  435. $callback_url = get_self_url_prefix() .
  436. "/public.php?op=pubsub&id=$feed";
  437. $s = new Pubsubhubbub\Subscriber\Subscriber($feed_hub_url, $callback_url);
  438. $rc = $s->subscribe($feed_self_url);
  439. _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
  440. db_query("UPDATE ttrss_feeds SET pubsub_state = 1
  441. WHERE id = '$feed'");
  442. }
  443. }
  444. _debug("processing articles...", $debug_enabled);
  445. $tstart = time();
  446. foreach ($items as $item) {
  447. if ($_REQUEST['xdebug'] == 3) {
  448. print_r($item);
  449. }
  450. if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
  451. _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
  452. break;
  453. }
  454. $entry_guid = $item->get_id();
  455. if (!$entry_guid) $entry_guid = $item->get_link();
  456. if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
  457. if (!$entry_guid) continue;
  458. $entry_guid = "$owner_uid,$entry_guid";
  459. $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
  460. _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
  461. $entry_timestamp = "";
  462. $entry_timestamp = $item->get_date();
  463. _debug("orig date: " . $item->get_date(), $debug_enabled);
  464. if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
  465. $entry_timestamp = time();
  466. }
  467. $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
  468. _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
  469. // $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
  470. // $entry_title = decode_numeric_entities($entry_title);
  471. $entry_title = $item->get_title();
  472. $entry_link = rewrite_relative_url($site_url, $item->get_link());
  473. _debug("title $entry_title", $debug_enabled);
  474. _debug("link $entry_link", $debug_enabled);
  475. if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
  476. $entry_content = $item->get_content();
  477. if (!$entry_content) $entry_content = $item->get_description();
  478. if ($_REQUEST["xdebug"] == 2) {
  479. print "content: ";
  480. print htmlspecialchars($entry_content);
  481. print "\n";
  482. }
  483. $entry_comments = db_escape_string(mb_substr($item->get_comments_url(), 0, 245));
  484. $num_comments = (int) $item->get_comments_count();
  485. $entry_author = $item->get_author(); // escaped later
  486. $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
  487. _debug("author $entry_author", $debug_enabled);
  488. _debug("num_comments: $num_comments", $debug_enabled);
  489. _debug("looking for tags...", $debug_enabled);
  490. // parse <category> entries into tags
  491. $additional_tags = array();
  492. $additional_tags_src = $item->get_categories();
  493. if (is_array($additional_tags_src)) {
  494. foreach ($additional_tags_src as $tobj) {
  495. array_push($additional_tags, $tobj);
  496. }
  497. }
  498. $entry_tags = array_unique($additional_tags);
  499. for ($i = 0; $i < count($entry_tags); $i++)
  500. $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
  501. _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
  502. _debug("done collecting data.", $debug_enabled);
  503. $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
  504. WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
  505. if (db_num_rows($result) != 0) {
  506. $base_entry_id = db_fetch_result($result, 0, "id");
  507. $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
  508. $article_labels = get_article_labels($base_entry_id, $owner_uid);
  509. $entry_language = db_fetch_result($result, 0, "lang");
  510. $existing_tags = get_article_tags($base_entry_id, $owner_uid);
  511. $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
  512. } else {
  513. $base_entry_id = false;
  514. $entry_stored_hash = "";
  515. $article_labels = array();
  516. $entry_language = "";
  517. }
  518. $article = array("owner_uid" => $owner_uid, // read only
  519. "guid" => $entry_guid, // read only
  520. "guid_hashed" => $entry_guid_hashed, // read only
  521. "title" => $entry_title,
  522. "content" => $entry_content,
  523. "link" => $entry_link,
  524. "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
  525. "tags" => $entry_tags,
  526. "author" => $entry_author,
  527. "force_catchup" => false, // ugly hack for the time being
  528. "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
  529. "language" => $entry_language,
  530. "feed" => array("id" => $feed,
  531. "fetch_url" => $fetch_url,
  532. "site_url" => $site_url,
  533. "cache_images" => $cache_images)
  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 => $v) {
  567. // i guess we'll have to take the risk of 4byte unicode labels & tags here
  568. if (is_string($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. $matched_rules = array();
  575. $article_filters = get_article_filters($filters, $article["title"],
  576. $article["content"], $article["link"], 0, $article["author"],
  577. $article["tags"], $matched_rules);
  578. if ($debug_enabled) {
  579. _debug("matched filter rules: ", $debug_enabled);
  580. if (count($matched_rules) != 0) {
  581. print_r($matched_rules);
  582. }
  583. _debug("filter actions: ", $debug_enabled);
  584. if (count($article_filters) != 0) {
  585. print_r($article_filters);
  586. }
  587. }
  588. $plugin_filter_names = find_article_filters($article_filters, "plugin");
  589. $plugin_filter_actions = $pluginhost->get_filter_actions();
  590. if (count($plugin_filter_names) > 0) {
  591. _debug("applying plugin filter actions...", $debug_enabled);
  592. foreach ($plugin_filter_names as $pfn) {
  593. list($pfclass,$pfaction) = explode(":", $pfn["param"]);
  594. if (isset($plugin_filter_actions[$pfclass])) {
  595. $plugin = $pluginhost->get_plugin($pfclass);
  596. _debug("... $pfclass: $pfaction", $debug_enabled);
  597. if ($plugin) {
  598. $start = microtime(true);
  599. $article = $plugin->hook_article_filter_action($article, $pfaction);
  600. _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
  601. } else {
  602. _debug("??? $pfclass: plugin object not found.");
  603. }
  604. } else {
  605. _debug("??? $pfclass: filter plugin not registered.");
  606. }
  607. }
  608. }
  609. $entry_tags = $article["tags"];
  610. $entry_guid = db_escape_string($entry_guid);
  611. $entry_title = db_escape_string($article["title"]);
  612. $entry_author = db_escape_string(mb_substr($article["author"], 0, 245));
  613. $entry_link = db_escape_string($article["link"]);
  614. $entry_content = $article["content"]; // escaped below
  615. $entry_force_catchup = $article["force_catchup"];
  616. $article_labels = $article["labels"];
  617. $entry_score_modifier = (int) $article["score_modifier"];
  618. $entry_language = db_escape_string($article["language"]);
  619. if ($debug_enabled) {
  620. _debug("article labels:", $debug_enabled);
  621. if (count($article_labels) != 0) {
  622. print_r($article_labels);
  623. }
  624. }
  625. _debug("force catchup: $entry_force_catchup");
  626. if ($cache_images && is_writable(CACHE_DIR . '/images'))
  627. cache_media($entry_content, $site_url, $debug_enabled);
  628. $entry_content = db_escape_string($entry_content, false);
  629. //db_query("BEGIN");
  630. $result = db_query("SELECT id FROM ttrss_entries
  631. WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
  632. if (db_num_rows($result) == 0) {
  633. _debug("base guid [$entry_guid] not found", $debug_enabled);
  634. // base post entry does not exist, create it
  635. $result = db_query(
  636. "INSERT INTO ttrss_entries
  637. (title,
  638. guid,
  639. link,
  640. updated,
  641. content,
  642. content_hash,
  643. no_orig_date,
  644. date_updated,
  645. date_entered,
  646. comments,
  647. num_comments,
  648. plugin_data,
  649. lang,
  650. author)
  651. VALUES
  652. ('$entry_title',
  653. '$entry_guid_hashed',
  654. '$entry_link',
  655. '$entry_timestamp_fmt',
  656. '$entry_content',
  657. '$entry_current_hash',
  658. false,
  659. NOW(),
  660. '$date_feed_processed',
  661. '$entry_comments',
  662. '$num_comments',
  663. '$entry_plugin_data',
  664. '$entry_language',
  665. '$entry_author')");
  666. } else {
  667. $base_entry_id = db_fetch_result($result, 0, "id");
  668. }
  669. // now it should exist, if not - bad luck then
  670. $result = db_query("SELECT id FROM ttrss_entries
  671. WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
  672. $entry_ref_id = 0;
  673. $entry_int_id = 0;
  674. if (db_num_rows($result) == 1) {
  675. _debug("base guid found, checking for user record", $debug_enabled);
  676. $ref_id = db_fetch_result($result, 0, "id");
  677. $entry_ref_id = $ref_id;
  678. /* $stored_guid = db_fetch_result($result, 0, "guid");
  679. if ($stored_guid != $entry_guid_hashed) {
  680. if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
  681. db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
  682. id = '$ref_id'");
  683. } */
  684. if (find_article_filter($article_filters, "filter")) {
  685. //db_query("COMMIT"); // close transaction in progress
  686. continue;
  687. }
  688. $score = calculate_article_score($article_filters) + $entry_score_modifier;
  689. _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
  690. // check for user post link to main table
  691. $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
  692. ref_id = '$ref_id' AND owner_uid = '$owner_uid'";
  693. // if ($_REQUEST["xdebug"]) print "$query\n";
  694. $result = db_query($query);
  695. // okay it doesn't exist - create user entry
  696. if (db_num_rows($result) == 0) {
  697. _debug("user record not found, creating...", $debug_enabled);
  698. if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
  699. $unread = 'true';
  700. $last_read_qpart = 'NULL';
  701. } else {
  702. $unread = 'false';
  703. $last_read_qpart = 'NOW()';
  704. }
  705. if (find_article_filter($article_filters, 'mark') || $score > 1000) {
  706. $marked = 'true';
  707. } else {
  708. $marked = 'false';
  709. }
  710. if (find_article_filter($article_filters, 'publish')) {
  711. $published = 'true';
  712. } else {
  713. $published = 'false';
  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 pubsubhubbub\publisher\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, $e_item);
  789. }
  790. }
  791. if ($cache_images && is_writable(CACHE_DIR . '/images'))
  792. cache_enclosures($enclosures, $site_url, $debug_enabled);
  793. if ($debug_enabled) {
  794. _debug("article enclosures:", $debug_enabled);
  795. print_r($enclosures);
  796. }
  797. //db_query("BEGIN");
  798. // debugging
  799. // db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
  800. foreach ($enclosures as $enc) {
  801. $enc_url = db_escape_string($enc[0]);
  802. $enc_type = db_escape_string($enc[1]);
  803. $enc_dur = db_escape_string($enc[2]);
  804. $enc_title = db_escape_string($enc[3]);
  805. $enc_width = intval($enc[4]);
  806. $enc_height = intval($enc[5]);
  807. $result = db_query("SELECT id FROM ttrss_enclosures
  808. WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
  809. if (db_num_rows($result) == 0) {
  810. db_query("INSERT INTO ttrss_enclosures
  811. (content_url, content_type, title, duration, post_id, width, height) VALUES
  812. ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
  813. }
  814. }
  815. //db_query("COMMIT");
  816. // check for manual tags (we have to do it here since they're loaded from filters)
  817. foreach ($article_filters as $f) {
  818. if ($f["type"] == "tag") {
  819. $manual_tags = trim_array(explode(",", $f["param"]));
  820. foreach ($manual_tags as $tag) {
  821. if (tag_is_valid($tag)) {
  822. array_push($entry_tags, $tag);
  823. }
  824. }
  825. }
  826. }
  827. // Skip boring tags
  828. $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
  829. 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
  830. $filtered_tags = array();
  831. $tags_to_cache = array();
  832. if ($entry_tags && is_array($entry_tags)) {
  833. foreach ($entry_tags as $tag) {
  834. if (array_search($tag, $boring_tags) === false) {
  835. array_push($filtered_tags, $tag);
  836. }
  837. }
  838. }
  839. $filtered_tags = array_unique($filtered_tags);
  840. if ($debug_enabled) {
  841. _debug("filtered article tags:", $debug_enabled);
  842. print_r($filtered_tags);
  843. }
  844. // Save article tags in the database
  845. if (count($filtered_tags) > 0) {
  846. //db_query("BEGIN");
  847. foreach ($filtered_tags as $tag) {
  848. $tag = sanitize_tag($tag);
  849. $tag = db_escape_string($tag);
  850. if (!tag_is_valid($tag)) continue;
  851. $result = db_query("SELECT id FROM ttrss_tags
  852. WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
  853. owner_uid = '$owner_uid' LIMIT 1");
  854. if ($result && db_num_rows($result) == 0) {
  855. db_query("INSERT INTO ttrss_tags
  856. (owner_uid,tag_name,post_int_id)
  857. VALUES ('$owner_uid','$tag', '$entry_int_id')");
  858. }
  859. array_push($tags_to_cache, $tag);
  860. }
  861. /* update the cache */
  862. $tags_to_cache = array_unique($tags_to_cache);
  863. $tags_str = db_escape_string(join(",", $tags_to_cache));
  864. db_query("UPDATE ttrss_user_entries
  865. SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
  866. AND owner_uid = $owner_uid");
  867. //db_query("COMMIT");
  868. }
  869. _debug("article processed", $debug_enabled);
  870. }
  871. _debug("purging feed...", $debug_enabled);
  872. purge_feed($feed, 0, $debug_enabled);
  873. db_query("UPDATE ttrss_feeds
  874. SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
  875. // db_query("COMMIT");
  876. } else {
  877. $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
  878. _debug("fetch error: $error_msg", $debug_enabled);
  879. if (count($rss->errors()) > 1) {
  880. foreach ($rss->errors() as $error) {
  881. _debug("+ $error");
  882. }
  883. }
  884. db_query(
  885. "UPDATE ttrss_feeds SET last_error = '$error_msg',
  886. last_updated = NOW() WHERE id = '$feed'");
  887. unset($rss);
  888. }
  889. _debug("done", $debug_enabled);
  890. return $rss;
  891. }
  892. function cache_enclosures($enclosures, $site_url, $debug) {
  893. foreach ($enclosures as $enc) {
  894. if (preg_match("/(image|audio|video)/", $enc[1])) {
  895. $src = rewrite_relative_url($site_url, $enc[0]);
  896. $local_filename = CACHE_DIR . "/images/" . sha1($src);
  897. if ($debug) _debug("cache_enclosures: downloading: $src to $local_filename");
  898. if (!file_exists($local_filename)) {
  899. $file_content = fetch_file_contents($src);
  900. if ($file_content && strlen($file_content) > _MIN_CACHE_FILE_SIZE) {
  901. file_put_contents($local_filename, $file_content);
  902. }
  903. } else {
  904. touch($local_filename);
  905. }
  906. }
  907. }
  908. }
  909. function cache_media($html, $site_url, $debug) {
  910. libxml_use_internal_errors(true);
  911. $charset_hack = '<head>
  912. <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
  913. </head>';
  914. $doc = new DOMDocument();
  915. $doc->loadHTML($charset_hack . $html);
  916. $xpath = new DOMXPath($doc);
  917. $entries = $xpath->query('(//img[@src])|(//video/source[@src])|(//audio/source[@src])');
  918. foreach ($entries as $entry) {
  919. if ($entry->hasAttribute('src') && strpos($entry->getAttribute('src'), "data:") !== 0) {
  920. $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
  921. $local_filename = CACHE_DIR . "/images/" . sha1($src);
  922. if ($debug) _debug("cache_media: downloading: $src to $local_filename");
  923. if (!file_exists($local_filename)) {
  924. $file_content = fetch_file_contents($src);
  925. if ($file_content && strlen($file_content) > _MIN_CACHE_FILE_SIZE) {
  926. file_put_contents($local_filename, $file_content);
  927. }
  928. } else {
  929. touch($local_filename);
  930. }
  931. }
  932. }
  933. }
  934. function expire_error_log($debug) {
  935. if ($debug) _debug("Removing old error log entries...");
  936. if (DB_TYPE == "pgsql") {
  937. db_query("DELETE FROM ttrss_error_log
  938. WHERE created_at < NOW() - INTERVAL '7 days'");
  939. } else {
  940. db_query("DELETE FROM ttrss_error_log
  941. WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
  942. }
  943. }
  944. function expire_lock_files($debug) {
  945. //if ($debug) _debug("Removing old lock files...");
  946. $num_deleted = 0;
  947. if (is_writable(LOCK_DIRECTORY)) {
  948. $files = glob(LOCK_DIRECTORY . "/*.lock");
  949. if ($files) {
  950. foreach ($files as $file) {
  951. if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
  952. unlink($file);
  953. ++$num_deleted;
  954. }
  955. }
  956. }
  957. }
  958. if ($debug) _debug("Removed $num_deleted old lock files.");
  959. }
  960. function expire_cached_files($debug) {
  961. foreach (array("simplepie", "images", "export", "upload") as $dir) {
  962. $cache_dir = CACHE_DIR . "/$dir";
  963. // if ($debug) _debug("Expiring $cache_dir");
  964. $num_deleted = 0;
  965. if (is_writable($cache_dir)) {
  966. $files = glob("$cache_dir/*");
  967. if ($files) {
  968. foreach ($files as $file) {
  969. if (time() - filemtime($file) > 86400*7) {
  970. unlink($file);
  971. ++$num_deleted;
  972. }
  973. }
  974. }
  975. }
  976. if ($debug) _debug("$cache_dir: removed $num_deleted files.");
  977. }
  978. }
  979. /**
  980. * Source: http://www.php.net/manual/en/function.parse-url.php#104527
  981. * Returns the url query as associative array
  982. *
  983. * @param string query
  984. * @return array params
  985. */
  986. function convertUrlQuery($query) {
  987. $queryParts = explode('&', $query);
  988. $params = array();
  989. foreach ($queryParts as $param) {
  990. $item = explode('=', $param);
  991. $params[$item[0]] = $item[1];
  992. }
  993. return $params;
  994. }
  995. function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags, &$matched_rules = false) {
  996. $matches = array();
  997. foreach ($filters as $filter) {
  998. $match_any_rule = $filter["match_any_rule"];
  999. $inverse = $filter["inverse"];
  1000. $filter_match = false;
  1001. foreach ($filter["rules"] as $rule) {
  1002. $match = false;
  1003. $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
  1004. $rule_inverse = $rule["inverse"];
  1005. if (!$reg_exp)
  1006. continue;
  1007. switch ($rule["type"]) {
  1008. case "title":
  1009. $match = @preg_match("/$reg_exp/iu", $title);
  1010. break;
  1011. case "content":
  1012. // we don't need to deal with multiline regexps
  1013. $content = preg_replace("/[\r\n\t]/", "", $content);
  1014. $match = @preg_match("/$reg_exp/iu", $content);
  1015. break;
  1016. case "both":
  1017. // we don't need to deal with multiline regexps
  1018. $content = preg_replace("/[\r\n\t]/", "", $content);
  1019. $match = (@preg_match("/$reg_exp/iu", $title) || @preg_match("/$reg_exp/iu", $content));
  1020. break;
  1021. case "link":
  1022. $match = @preg_match("/$reg_exp/iu", $link);
  1023. break;
  1024. case "author":
  1025. $match = @preg_match("/$reg_exp/iu", $author);
  1026. break;
  1027. case "tag":
  1028. foreach ($tags as $tag) {
  1029. if (@preg_match("/$reg_exp/iu", $tag)) {
  1030. $match = true;
  1031. break;
  1032. }
  1033. }
  1034. break;
  1035. }
  1036. if ($rule_inverse) $match = !$match;
  1037. if ($match_any_rule) {
  1038. if ($match) {
  1039. $filter_match = true;
  1040. break;
  1041. }
  1042. } else {
  1043. $filter_match = $match;
  1044. if (!$match) {
  1045. break;
  1046. }
  1047. }
  1048. }
  1049. if ($inverse) $filter_match = !$filter_match;
  1050. if ($filter_match) {
  1051. if (is_array($matched_rules)) array_push($matched_rules, $rule);
  1052. foreach ($filter["actions"] AS $action) {
  1053. array_push($matches, $action);
  1054. // if Stop action encountered, perform no further processing
  1055. if (isset($action["type"]) && $action["type"] == "stop") return $matches;
  1056. }
  1057. }
  1058. }
  1059. return $matches;
  1060. }
  1061. function find_article_filter($filters, $filter_name) {
  1062. foreach ($filters as $f) {
  1063. if ($f["type"] == $filter_name) {
  1064. return $f;
  1065. };
  1066. }
  1067. return false;
  1068. }
  1069. function find_article_filters($filters, $filter_name) {
  1070. $results = array();
  1071. foreach ($filters as $f) {
  1072. if ($f["type"] == $filter_name) {
  1073. array_push($results, $f);
  1074. };
  1075. }
  1076. return $results;
  1077. }
  1078. function calculate_article_score($filters) {
  1079. $score = 0;
  1080. foreach ($filters as $f) {
  1081. if ($f["type"] == "score") {
  1082. $score += $f["param"];
  1083. };
  1084. }
  1085. return $score;
  1086. }
  1087. function labels_contains_caption($labels, $caption) {
  1088. foreach ($labels as $label) {
  1089. if ($label[1] == $caption) {
  1090. return true;
  1091. }
  1092. }
  1093. return false;
  1094. }
  1095. function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
  1096. foreach ($filters as $f) {
  1097. if ($f["type"] == "label") {
  1098. if (!labels_contains_caption($article_labels, $f["param"])) {
  1099. label_add_article($id, $f["param"], $owner_uid);
  1100. }
  1101. }
  1102. }
  1103. }
  1104. function make_guid_from_title($title) {
  1105. return preg_replace("/[ \"\',.:;]/", "-",
  1106. mb_strtolower(strip_tags($title), 'utf-8'));
  1107. }
  1108. /* function verify_feed_xml($feed_data) {
  1109. libxml_use_internal_errors(true);
  1110. $doc = new DOMDocument();
  1111. $doc->loadXML($feed_data);
  1112. $error = libxml_get_last_error();
  1113. libxml_clear_errors();
  1114. return $error;
  1115. } */
  1116. function cleanup_counters_cache($debug) {
  1117. $result = db_query("DELETE FROM ttrss_counters_cache
  1118. WHERE feed_id > 0 AND
  1119. (SELECT COUNT(id) FROM ttrss_feeds WHERE
  1120. id = feed_id AND
  1121. ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
  1122. $frows = db_affected_rows($result);
  1123. $result = db_query("DELETE FROM ttrss_cat_counters_cache
  1124. WHERE feed_id > 0 AND
  1125. (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
  1126. id = feed_id AND
  1127. ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
  1128. $crows = db_affected_rows($result);
  1129. _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
  1130. }
  1131. function housekeeping_user($owner_uid) {
  1132. $tmph = new PluginHost();
  1133. load_user_plugins($owner_uid, $tmph);
  1134. $tmph->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
  1135. }
  1136. function housekeeping_common($debug) {
  1137. expire_cached_files($debug);
  1138. expire_lock_files($debug);
  1139. expire_error_log($debug);
  1140. $count = update_feedbrowser_cache();
  1141. _debug("Feedbrowser updated, $count feeds processed.");
  1142. purge_orphans( true);
  1143. cleanup_counters_cache($debug);
  1144. //$rc = cleanup_tags( 14, 50000);
  1145. //_debug("Cleaned $rc cached tags.");
  1146. PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
  1147. }
  1148. ?>