rssfuncs.php 44 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534
  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 $count;
  41. }
  42. function update_daemon_common($limit = DAEMON_FEED_LIMIT, $debug = true) {
  43. // Process all other feeds using last_updated and interval parameters
  44. $schema_version = get_schema_version();
  45. if ($schema_version != SCHEMA_VERSION) {
  46. die("Schema version is wrong, please upgrade the database.\n");
  47. }
  48. define('PREFS_NO_CACHE', true);
  49. // Test if the user has loggued in recently. If not, it does not update its feeds.
  50. if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
  51. if (DB_TYPE == "pgsql") {
  52. $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
  53. } else {
  54. $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
  55. }
  56. } else {
  57. $login_thresh_qpart = "";
  58. }
  59. // Test if the feed need a update (update interval exceeded).
  60. if (DB_TYPE == "pgsql") {
  61. $update_limit_qpart = "AND ((
  62. ttrss_feeds.update_interval = 0
  63. AND ttrss_user_prefs.value != '-1'
  64. AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
  65. ) OR (
  66. ttrss_feeds.update_interval > 0
  67. AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
  68. ) OR (ttrss_feeds.last_updated IS NULL
  69. AND ttrss_user_prefs.value != '-1')
  70. OR (last_updated = '1970-01-01 00:00:00'
  71. AND ttrss_user_prefs.value != '-1'))";
  72. } else {
  73. $update_limit_qpart = "AND ((
  74. ttrss_feeds.update_interval = 0
  75. AND ttrss_user_prefs.value != '-1'
  76. AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
  77. ) OR (
  78. ttrss_feeds.update_interval > 0
  79. AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
  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. }
  85. // Test if feed is currently being updated by another process.
  86. if (DB_TYPE == "pgsql") {
  87. $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
  88. } else {
  89. $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
  90. }
  91. // Test if there is a limit to number of updated feeds
  92. $query_limit = "";
  93. if($limit) $query_limit = sprintf("LIMIT %d", $limit);
  94. // Update the least recently updated feeds first
  95. $query_order = "ORDER BY last_updated";
  96. if (DB_TYPE == "pgsql") $query_order .= " NULLS FIRST";
  97. $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
  98. FROM
  99. ttrss_feeds, ttrss_users, ttrss_user_prefs
  100. WHERE
  101. ttrss_feeds.owner_uid = ttrss_users.id
  102. AND ttrss_user_prefs.profile IS NULL
  103. AND ttrss_users.id = ttrss_user_prefs.owner_uid
  104. AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
  105. $login_thresh_qpart $update_limit_qpart
  106. $updstart_thresh_qpart
  107. $query_order $query_limit";
  108. // We search for feed needing update.
  109. $result = db_query($query);
  110. if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
  111. // Here is a little cache magic in order to minimize risk of double feed updates.
  112. $feeds_to_update = array();
  113. while ($line = db_fetch_assoc($result)) {
  114. array_push($feeds_to_update, db_escape_string($line['feed_url']));
  115. }
  116. // We update the feed last update started date before anything else.
  117. // There is no lag due to feed contents downloads
  118. // It prevent an other process to update the same feed.
  119. if(count($feeds_to_update) > 0) {
  120. $feeds_quoted = array();
  121. foreach ($feeds_to_update as $feed) {
  122. array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
  123. }
  124. db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
  125. WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
  126. }
  127. $nf = 0;
  128. $bstarted = microtime(true);
  129. $batch_owners = array();
  130. // For each feed, we call the feed update function.
  131. foreach ($feeds_to_update as $feed) {
  132. if($debug) _debug("Base feed: $feed");
  133. //update_rss_feed($line["id"], true);
  134. // since we have the data cached, we can deal with other feeds with the same url
  135. $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
  136. FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
  137. ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
  138. ttrss_users.id = ttrss_user_prefs.owner_uid AND
  139. ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
  140. ttrss_user_prefs.profile IS NULL AND
  141. feed_url = '".db_escape_string($feed)."'
  142. $update_limit_qpart
  143. $login_thresh_qpart
  144. ORDER BY ttrss_feeds.id $query_limit");
  145. if (db_num_rows($tmp_result) > 0) {
  146. while ($tline = db_fetch_assoc($tmp_result)) {
  147. if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
  148. if (array_search($tline["owner_uid"], $batch_owners) === FALSE)
  149. array_push($batch_owners, $tline["owner_uid"]);
  150. $fstarted = microtime(true);
  151. update_rss_feed($tline["id"], true, false);
  152. _debug_suppress(false);
  153. _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
  154. ++$nf;
  155. }
  156. }
  157. }
  158. if ($nf > 0) {
  159. _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
  160. microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
  161. }
  162. foreach ($batch_owners as $owner_uid) {
  163. _debug("Running housekeeping tasks for user $owner_uid...");
  164. housekeeping_user($owner_uid);
  165. }
  166. require_once "digest.php";
  167. // Send feed digests by email if needed.
  168. send_headlines_digests($debug);
  169. return $nf;
  170. }
  171. // this is used when subscribing
  172. function set_basic_feed_info($feed) {
  173. $feed = db_escape_string($feed);
  174. $result = db_query("SELECT feed_url,auth_pass,auth_login,auth_pass_encrypted
  175. FROM ttrss_feeds WHERE id = '$feed'");
  176. $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
  177. 0, "auth_pass_encrypted"));
  178. $auth_login = db_fetch_result($result, 0, "auth_login");
  179. $auth_pass = db_fetch_result($result, 0, "auth_pass");
  180. if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
  181. require_once "crypt.php";
  182. $auth_pass = decrypt_string($auth_pass);
  183. }
  184. $fetch_url = db_fetch_result($result, 0, "feed_url");
  185. $feed_data = fetch_file_contents($fetch_url, false,
  186. $auth_login, $auth_pass, false,
  187. FEED_FETCH_TIMEOUT,
  188. 0);
  189. global $fetch_curl_used;
  190. if (!$fetch_curl_used) {
  191. $tmp = @gzdecode($feed_data);
  192. if ($tmp) $feed_data = $tmp;
  193. }
  194. $feed_data = trim($feed_data);
  195. $rss = new FeedParser($feed_data);
  196. $rss->init();
  197. if (!$rss->error()) {
  198. $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
  199. $registered_title = db_fetch_result($result, 0, "title");
  200. $orig_site_url = db_fetch_result($result, 0, "site_url");
  201. $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
  202. $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
  203. if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
  204. db_query("UPDATE ttrss_feeds SET
  205. title = '$feed_title' WHERE id = '$feed'");
  206. }
  207. if ($site_url && $orig_site_url != $site_url) {
  208. db_query("UPDATE ttrss_feeds SET
  209. site_url = '$site_url' WHERE id = '$feed'");
  210. }
  211. }
  212. }
  213. /**
  214. * @SuppressWarnings(PHPMD.UnusedFormalParameter)
  215. */
  216. function update_rss_feed($feed, $no_cache = false) {
  217. $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
  218. _debug_suppress(!$debug_enabled);
  219. _debug("start", $debug_enabled);
  220. $result = db_query("SELECT title FROM ttrss_feeds
  221. WHERE id = '$feed'");
  222. if (db_num_rows($result) == 0) {
  223. _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
  224. user_error("Attempt to update unknown/invalid feed $feed", E_USER_WARNING);
  225. return false;
  226. }
  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. $owner_uid = db_fetch_result($result, 0, "owner_uid");
  241. $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
  242. 0, "mark_unread_on_update"));
  243. $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
  244. $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
  245. 0, "auth_pass_encrypted"));
  246. db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
  247. WHERE id = '$feed'");
  248. $auth_login = db_fetch_result($result, 0, "auth_login");
  249. $auth_pass = db_fetch_result($result, 0, "auth_pass");
  250. if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
  251. require_once "crypt.php";
  252. $auth_pass = decrypt_string($auth_pass);
  253. }
  254. $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
  255. $fetch_url = db_fetch_result($result, 0, "feed_url");
  256. $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
  257. if (!$feed_language) $feed_language = 'english';
  258. $feed = db_escape_string($feed);
  259. $date_feed_processed = date('Y-m-d H:i');
  260. $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
  261. $pluginhost = new PluginHost();
  262. $pluginhost->set_debug($debug_enabled);
  263. $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
  264. $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
  265. $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
  266. $pluginhost->load_data();
  267. $rss_hash = false;
  268. $force_refetch = isset($_REQUEST["force_refetch"]);
  269. $feed_data = "";
  270. foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
  271. $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
  272. }
  273. // try cache
  274. if (!$feed_data &&
  275. file_exists($cache_filename) &&
  276. is_readable($cache_filename) &&
  277. !$auth_login && !$auth_pass &&
  278. filemtime($cache_filename) > time() - 30) {
  279. _debug("using local cache [$cache_filename].", $debug_enabled);
  280. @$feed_data = file_get_contents($cache_filename);
  281. if ($feed_data) {
  282. $rss_hash = sha1($feed_data);
  283. }
  284. } else {
  285. _debug("local cache will not be used for this feed", $debug_enabled);
  286. }
  287. // fetch feed from source
  288. if (!$feed_data) {
  289. _debug("fetching [$fetch_url]...", $debug_enabled);
  290. if (ini_get("open_basedir") && function_exists("curl_init")) {
  291. _debug("not using CURL due to open_basedir restrictions");
  292. }
  293. $feed_data = fetch_file_contents($fetch_url, false,
  294. $auth_login, $auth_pass, false,
  295. $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
  296. 0);
  297. global $fetch_curl_used;
  298. if (!$fetch_curl_used) {
  299. $tmp = @gzdecode($feed_data);
  300. if ($tmp) $feed_data = $tmp;
  301. }
  302. $feed_data = trim($feed_data);
  303. _debug("fetch done.", $debug_enabled);
  304. // cache vanilla feed data for re-use
  305. if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
  306. $new_rss_hash = sha1($feed_data);
  307. if ($new_rss_hash != $rss_hash) {
  308. _debug("saving $cache_filename", $debug_enabled);
  309. @file_put_contents($cache_filename, $feed_data);
  310. }
  311. }
  312. }
  313. if (!$feed_data) {
  314. global $fetch_last_error;
  315. global $fetch_last_error_code;
  316. _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
  317. $error_escaped = '';
  318. // If-Modified-Since
  319. if ($fetch_last_error_code != 304) {
  320. $error_escaped = db_escape_string($fetch_last_error);
  321. } else {
  322. _debug("source claims data not modified, nothing to do.", $debug_enabled);
  323. }
  324. db_query(
  325. "UPDATE ttrss_feeds SET last_error = '$error_escaped',
  326. last_updated = NOW() WHERE id = '$feed'");
  327. return;
  328. }
  329. foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
  330. $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
  331. }
  332. // set last update to now so if anything *simplepie* crashes later we won't be
  333. // continuously failing on the same feed
  334. //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
  335. if (!$rss) {
  336. $rss = new FeedParser($feed_data);
  337. $rss->init();
  338. }
  339. // print_r($rss);
  340. $feed = db_escape_string($feed);
  341. if (!$rss->error()) {
  342. // We use local pluginhost here because we need to load different per-user feed plugins
  343. $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
  344. _debug("language: $feed_language", $debug_enabled);
  345. _debug("processing feed data...", $debug_enabled);
  346. // db_query("BEGIN");
  347. if (DB_TYPE == "pgsql") {
  348. $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
  349. } else {
  350. $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
  351. }
  352. $result = db_query("SELECT owner_uid,favicon_avg_color,
  353. (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
  354. favicon_needs_check
  355. FROM ttrss_feeds WHERE id = '$feed'");
  356. $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
  357. "favicon_needs_check"));
  358. $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
  359. $owner_uid = db_fetch_result($result, 0, "owner_uid");
  360. $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
  361. _debug("site_url: $site_url", $debug_enabled);
  362. _debug("feed_title: " . $rss->get_title(), $debug_enabled);
  363. if ($favicon_needs_check || $force_refetch) {
  364. /* terrible hack: if we crash on floicon shit here, we won't check
  365. * the icon avgcolor again (unless the icon got updated) */
  366. $favicon_file = ICONS_DIR . "/$feed.ico";
  367. $favicon_modified = @filemtime($favicon_file);
  368. _debug("checking favicon...", $debug_enabled);
  369. check_feed_favicon($site_url, $feed);
  370. $favicon_modified_new = @filemtime($favicon_file);
  371. if ($favicon_modified_new > $favicon_modified)
  372. $favicon_avg_color = '';
  373. if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
  374. require_once "colors.php";
  375. db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
  376. id = '$feed'");
  377. $favicon_color = db_escape_string(
  378. calculate_avg_color($favicon_file));
  379. $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
  380. } else if ($favicon_avg_color == 'fail') {
  381. _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
  382. }
  383. db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
  384. $favicon_colorstring
  385. WHERE id = '$feed'");
  386. }
  387. _debug("loading filters & labels...", $debug_enabled);
  388. $filters = load_filters($feed, $owner_uid);
  389. _debug("" . count($filters) . " filters loaded.", $debug_enabled);
  390. $items = $rss->get_items();
  391. if (!is_array($items)) {
  392. _debug("no articles found.", $debug_enabled);
  393. db_query("UPDATE ttrss_feeds
  394. SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
  395. return; // no articles
  396. }
  397. if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
  398. _debug("checking for PUSH hub...", $debug_enabled);
  399. $feed_hub_url = false;
  400. $links = $rss->get_links('hub');
  401. if ($links && is_array($links)) {
  402. foreach ($links as $l) {
  403. $feed_hub_url = $l;
  404. break;
  405. }
  406. }
  407. _debug("feed hub url: $feed_hub_url", $debug_enabled);
  408. $feed_self_url = $fetch_url;
  409. $links = $rss->get_links('self');
  410. if ($links && is_array($links)) {
  411. foreach ($links as $l) {
  412. $feed_self_url = $l;
  413. break;
  414. }
  415. }
  416. _debug("feed self url = $feed_self_url");
  417. if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
  418. !ini_get("open_basedir")) {
  419. require_once 'lib/pubsubhubbub/Subscriber.php';
  420. $callback_url = get_self_url_prefix() .
  421. "/public.php?op=pubsub&id=$feed";
  422. $s = new Pubsubhubbub\Subscriber\Subscriber($feed_hub_url, $callback_url);
  423. $rc = $s->subscribe($feed_self_url);
  424. _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
  425. db_query("UPDATE ttrss_feeds SET pubsub_state = 1
  426. WHERE id = '$feed'");
  427. }
  428. }
  429. _debug("processing articles...", $debug_enabled);
  430. $tstart = time();
  431. foreach ($items as $item) {
  432. if ($_REQUEST['xdebug'] == 3) {
  433. print_r($item);
  434. }
  435. if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
  436. _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
  437. break;
  438. }
  439. $entry_guid = $item->get_id();
  440. if (!$entry_guid) $entry_guid = $item->get_link();
  441. if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
  442. if (!$entry_guid) continue;
  443. $entry_guid = "$owner_uid,$entry_guid";
  444. $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
  445. _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
  446. $entry_timestamp = "";
  447. $entry_timestamp = $item->get_date();
  448. _debug("orig date: " . $item->get_date(), $debug_enabled);
  449. if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
  450. $entry_timestamp = time();
  451. }
  452. $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
  453. _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
  454. // $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
  455. // $entry_title = decode_numeric_entities($entry_title);
  456. $entry_title = $item->get_title();
  457. $entry_link = rewrite_relative_url($site_url, $item->get_link());
  458. _debug("title $entry_title", $debug_enabled);
  459. _debug("link $entry_link", $debug_enabled);
  460. if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
  461. $entry_content = $item->get_content();
  462. if (!$entry_content) $entry_content = $item->get_description();
  463. if ($_REQUEST["xdebug"] == 2) {
  464. print "content: ";
  465. print htmlspecialchars($entry_content);
  466. print "\n";
  467. }
  468. $entry_comments = db_escape_string(mb_substr($item->get_comments_url(), 0, 245));
  469. $num_comments = (int) $item->get_comments_count();
  470. $entry_author = $item->get_author(); // escaped later
  471. $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
  472. _debug("author $entry_author", $debug_enabled);
  473. _debug("num_comments: $num_comments", $debug_enabled);
  474. _debug("looking for tags...", $debug_enabled);
  475. // parse <category> entries into tags
  476. $additional_tags = array();
  477. $additional_tags_src = $item->get_categories();
  478. if (is_array($additional_tags_src)) {
  479. foreach ($additional_tags_src as $tobj) {
  480. array_push($additional_tags, $tobj);
  481. }
  482. }
  483. $entry_tags = array_unique($additional_tags);
  484. for ($i = 0; $i < count($entry_tags); $i++)
  485. $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
  486. _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
  487. _debug("done collecting data.", $debug_enabled);
  488. $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
  489. WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
  490. if (db_num_rows($result) != 0) {
  491. $base_entry_id = db_fetch_result($result, 0, "id");
  492. $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
  493. $article_labels = get_article_labels($base_entry_id, $owner_uid);
  494. $entry_language = db_fetch_result($result, 0, "lang");
  495. $existing_tags = get_article_tags($base_entry_id, $owner_uid);
  496. $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
  497. } else {
  498. $base_entry_id = false;
  499. $entry_stored_hash = "";
  500. $article_labels = array();
  501. $entry_language = "";
  502. }
  503. $article = array("owner_uid" => $owner_uid, // read only
  504. "guid" => $entry_guid, // read only
  505. "guid_hashed" => $entry_guid_hashed, // read only
  506. "title" => $entry_title,
  507. "content" => $entry_content,
  508. "link" => $entry_link,
  509. "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
  510. "tags" => $entry_tags,
  511. "author" => $entry_author,
  512. "force_catchup" => false, // ugly hack for the time being
  513. "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
  514. "language" => $entry_language,
  515. "feed" => array("id" => $feed,
  516. "fetch_url" => $fetch_url,
  517. "site_url" => $site_url,
  518. "cache_images" => $cache_images)
  519. );
  520. $entry_plugin_data = "";
  521. $entry_current_hash = calculate_article_hash($article, $pluginhost);
  522. _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
  523. if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
  524. _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
  525. // we keep encountering the entry in feeds, so we need to
  526. // update date_updated column so that we don't get horrible
  527. // dupes when the entry gets purged and reinserted again e.g.
  528. // in the case of SLOW SLOW OMG SLOW updating feeds
  529. $base_entry_id = db_fetch_result($result, 0, "id");
  530. db_query("UPDATE ttrss_entries SET date_updated = NOW()
  531. WHERE id = '$base_entry_id'");
  532. continue;
  533. }
  534. _debug("hash differs, applying plugin filters:", $debug_enabled);
  535. foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
  536. _debug("... " . get_class($plugin), $debug_enabled);
  537. $start = microtime(true);
  538. $article = $plugin->hook_article_filter($article);
  539. _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
  540. $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
  541. }
  542. if ($_REQUEST["xdebug"] == 2) {
  543. print "processed content: ";
  544. print htmlspecialchars($article["content"]);
  545. print "\n";
  546. }
  547. $entry_plugin_data = db_escape_string($entry_plugin_data);
  548. _debug("plugin data: $entry_plugin_data", $debug_enabled);
  549. // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
  550. if (DB_TYPE == "mysql") {
  551. foreach ($article as $k => $v) {
  552. // i guess we'll have to take the risk of 4byte unicode labels & tags here
  553. if (is_string($article[$k])) {
  554. $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
  555. }
  556. }
  557. }
  558. /* Collect article tags here so we could filter by them: */
  559. $matched_rules = array();
  560. $article_filters = get_article_filters($filters, $article["title"],
  561. $article["content"], $article["link"], $article["author"],
  562. $article["tags"], $matched_rules);
  563. if ($debug_enabled) {
  564. _debug("matched filter rules: ", $debug_enabled);
  565. if (count($matched_rules) != 0) {
  566. print_r($matched_rules);
  567. }
  568. _debug("filter actions: ", $debug_enabled);
  569. if (count($article_filters) != 0) {
  570. print_r($article_filters);
  571. }
  572. }
  573. $plugin_filter_names = find_article_filters($article_filters, "plugin");
  574. $plugin_filter_actions = $pluginhost->get_filter_actions();
  575. if (count($plugin_filter_names) > 0) {
  576. _debug("applying plugin filter actions...", $debug_enabled);
  577. foreach ($plugin_filter_names as $pfn) {
  578. list($pfclass,$pfaction) = explode(":", $pfn["param"]);
  579. if (isset($plugin_filter_actions[$pfclass])) {
  580. $plugin = $pluginhost->get_plugin($pfclass);
  581. _debug("... $pfclass: $pfaction", $debug_enabled);
  582. if ($plugin) {
  583. $start = microtime(true);
  584. $article = $plugin->hook_article_filter_action($article, $pfaction);
  585. _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
  586. } else {
  587. _debug("??? $pfclass: plugin object not found.");
  588. }
  589. } else {
  590. _debug("??? $pfclass: filter plugin not registered.");
  591. }
  592. }
  593. }
  594. $entry_tags = $article["tags"];
  595. $entry_guid = db_escape_string($entry_guid);
  596. $entry_title = db_escape_string($article["title"]);
  597. $entry_author = db_escape_string(mb_substr($article["author"], 0, 245));
  598. $entry_link = db_escape_string($article["link"]);
  599. $entry_content = $article["content"]; // escaped below
  600. $entry_force_catchup = $article["force_catchup"];
  601. $article_labels = $article["labels"];
  602. $entry_score_modifier = (int) $article["score_modifier"];
  603. $entry_language = db_escape_string($article["language"]);
  604. if ($debug_enabled) {
  605. _debug("article labels:", $debug_enabled);
  606. if (count($article_labels) != 0) {
  607. print_r($article_labels);
  608. }
  609. }
  610. _debug("force catchup: $entry_force_catchup");
  611. if ($cache_images && is_writable(CACHE_DIR . '/images'))
  612. cache_media($entry_content, $site_url, $debug_enabled);
  613. $entry_content = db_escape_string($entry_content, false);
  614. //db_query("BEGIN");
  615. $result = db_query("SELECT id FROM ttrss_entries
  616. WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
  617. if (db_num_rows($result) == 0) {
  618. _debug("base guid [$entry_guid] not found", $debug_enabled);
  619. // base post entry does not exist, create it
  620. $result = db_query(
  621. "INSERT INTO ttrss_entries
  622. (title,
  623. guid,
  624. link,
  625. updated,
  626. content,
  627. content_hash,
  628. no_orig_date,
  629. date_updated,
  630. date_entered,
  631. comments,
  632. num_comments,
  633. plugin_data,
  634. lang,
  635. author)
  636. VALUES
  637. ('$entry_title',
  638. '$entry_guid_hashed',
  639. '$entry_link',
  640. '$entry_timestamp_fmt',
  641. '$entry_content',
  642. '$entry_current_hash',
  643. false,
  644. NOW(),
  645. '$date_feed_processed',
  646. '$entry_comments',
  647. '$num_comments',
  648. '$entry_plugin_data',
  649. '$entry_language',
  650. '$entry_author')");
  651. } else {
  652. $base_entry_id = db_fetch_result($result, 0, "id");
  653. }
  654. // now it should exist, if not - bad luck then
  655. $result = db_query("SELECT id FROM ttrss_entries
  656. WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
  657. $entry_ref_id = 0;
  658. $entry_int_id = 0;
  659. if (db_num_rows($result) == 1) {
  660. _debug("base guid found, checking for user record", $debug_enabled);
  661. $ref_id = db_fetch_result($result, 0, "id");
  662. $entry_ref_id = $ref_id;
  663. /* $stored_guid = db_fetch_result($result, 0, "guid");
  664. if ($stored_guid != $entry_guid_hashed) {
  665. if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
  666. db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
  667. id = '$ref_id'");
  668. } */
  669. if (find_article_filter($article_filters, "filter")) {
  670. //db_query("COMMIT"); // close transaction in progress
  671. continue;
  672. }
  673. $score = calculate_article_score($article_filters) + $entry_score_modifier;
  674. _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
  675. // check for user post link to main table
  676. $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
  677. ref_id = '$ref_id' AND owner_uid = '$owner_uid'";
  678. // if ($_REQUEST["xdebug"]) print "$query\n";
  679. $result = db_query($query);
  680. // okay it doesn't exist - create user entry
  681. if (db_num_rows($result) == 0) {
  682. _debug("user record not found, creating...", $debug_enabled);
  683. if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
  684. $unread = 'true';
  685. $last_read_qpart = 'NULL';
  686. } else {
  687. $unread = 'false';
  688. $last_read_qpart = 'NOW()';
  689. }
  690. if (find_article_filter($article_filters, 'mark') || $score > 1000) {
  691. $marked = 'true';
  692. } else {
  693. $marked = 'false';
  694. }
  695. if (find_article_filter($article_filters, 'publish')) {
  696. $published = 'true';
  697. } else {
  698. $published = 'false';
  699. }
  700. $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
  701. $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
  702. $result = db_query(
  703. "INSERT INTO ttrss_user_entries
  704. (ref_id, owner_uid, feed_id, unread, last_read, marked,
  705. published, score, tag_cache, label_cache, uuid,
  706. last_marked, last_published)
  707. VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
  708. $last_read_qpart, $marked, $published, '$score', '', '',
  709. '', $last_marked, $last_published)");
  710. if (PUBSUBHUBBUB_HUB && $published == 'true') {
  711. $rss_link = get_self_url_prefix() .
  712. "/public.php?op=rss&id=-2&key=" .
  713. get_feed_access_key(-2, false, $owner_uid);
  714. $p = new pubsubhubbub\publisher\Publisher(PUBSUBHUBBUB_HUB);
  715. /* $pubsub_result = */ $p->publish_update($rss_link);
  716. }
  717. $result = db_query(
  718. "SELECT int_id FROM ttrss_user_entries WHERE
  719. ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
  720. feed_id = '$feed' LIMIT 1");
  721. if (db_num_rows($result) == 1) {
  722. $entry_int_id = db_fetch_result($result, 0, "int_id");
  723. }
  724. } else {
  725. _debug("user record FOUND", $debug_enabled);
  726. $entry_ref_id = db_fetch_result($result, 0, "ref_id");
  727. $entry_int_id = db_fetch_result($result, 0, "int_id");
  728. }
  729. _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
  730. if (DB_TYPE == "pgsql") {
  731. $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
  732. 0, 1000000));
  733. $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
  734. } else {
  735. $tsvector_qpart = "";
  736. }
  737. db_query("UPDATE ttrss_entries
  738. SET title = '$entry_title',
  739. content = '$entry_content',
  740. content_hash = '$entry_current_hash',
  741. updated = '$entry_timestamp_fmt',
  742. $tsvector_qpart
  743. num_comments = '$num_comments',
  744. plugin_data = '$entry_plugin_data',
  745. author = '$entry_author',
  746. lang = '$entry_language'
  747. WHERE id = '$ref_id'");
  748. // update aux data
  749. db_query("UPDATE ttrss_user_entries
  750. SET score = '$score' WHERE ref_id = '$ref_id'");
  751. if ($mark_unread_on_update) {
  752. _debug("article updated, marking unread as requested.", $debug_enabled);
  753. db_query("UPDATE ttrss_user_entries
  754. SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
  755. }
  756. }
  757. //db_query("COMMIT");
  758. _debug("assigning labels [other]...", $debug_enabled);
  759. foreach ($article_labels as $label) {
  760. label_add_article($entry_ref_id, $label[1], $owner_uid);
  761. }
  762. _debug("assigning labels [filters]...", $debug_enabled);
  763. assign_article_to_label_filters($entry_ref_id, $article_filters,
  764. $owner_uid, $article_labels);
  765. _debug("looking for enclosures...", $debug_enabled);
  766. // enclosures
  767. $enclosures = array();
  768. $encs = $item->get_enclosures();
  769. if (is_array($encs)) {
  770. foreach ($encs as $e) {
  771. $e_item = array(
  772. rewrite_relative_url($site_url, $e->link),
  773. $e->type, $e->length, $e->title, $e->width, $e->height);
  774. array_push($enclosures, $e_item);
  775. }
  776. }
  777. if ($cache_images && is_writable(CACHE_DIR . '/images'))
  778. cache_enclosures($enclosures, $site_url, $debug_enabled);
  779. if ($debug_enabled) {
  780. _debug("article enclosures:", $debug_enabled);
  781. print_r($enclosures);
  782. }
  783. //db_query("BEGIN");
  784. // debugging
  785. // db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
  786. foreach ($enclosures as $enc) {
  787. $enc_url = db_escape_string($enc[0]);
  788. $enc_type = db_escape_string($enc[1]);
  789. $enc_dur = db_escape_string($enc[2]);
  790. $enc_title = db_escape_string($enc[3]);
  791. $enc_width = intval($enc[4]);
  792. $enc_height = intval($enc[5]);
  793. $result = db_query("SELECT id FROM ttrss_enclosures
  794. WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
  795. if (db_num_rows($result) == 0) {
  796. db_query("INSERT INTO ttrss_enclosures
  797. (content_url, content_type, title, duration, post_id, width, height) VALUES
  798. ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
  799. }
  800. }
  801. //db_query("COMMIT");
  802. // check for manual tags (we have to do it here since they're loaded from filters)
  803. foreach ($article_filters as $f) {
  804. if ($f["type"] == "tag") {
  805. $manual_tags = trim_array(explode(",", $f["param"]));
  806. foreach ($manual_tags as $tag) {
  807. if (tag_is_valid($tag)) {
  808. array_push($entry_tags, $tag);
  809. }
  810. }
  811. }
  812. }
  813. // Skip boring tags
  814. $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
  815. 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
  816. $filtered_tags = array();
  817. $tags_to_cache = array();
  818. if ($entry_tags && is_array($entry_tags)) {
  819. foreach ($entry_tags as $tag) {
  820. if (array_search($tag, $boring_tags) === false) {
  821. array_push($filtered_tags, $tag);
  822. }
  823. }
  824. }
  825. $filtered_tags = array_unique($filtered_tags);
  826. if ($debug_enabled) {
  827. _debug("filtered article tags:", $debug_enabled);
  828. print_r($filtered_tags);
  829. }
  830. // Save article tags in the database
  831. if (count($filtered_tags) > 0) {
  832. //db_query("BEGIN");
  833. foreach ($filtered_tags as $tag) {
  834. $tag = sanitize_tag($tag);
  835. $tag = db_escape_string($tag);
  836. if (!tag_is_valid($tag)) continue;
  837. $result = db_query("SELECT id FROM ttrss_tags
  838. WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
  839. owner_uid = '$owner_uid' LIMIT 1");
  840. if ($result && db_num_rows($result) == 0) {
  841. db_query("INSERT INTO ttrss_tags
  842. (owner_uid,tag_name,post_int_id)
  843. VALUES ('$owner_uid','$tag', '$entry_int_id')");
  844. }
  845. array_push($tags_to_cache, $tag);
  846. }
  847. /* update the cache */
  848. $tags_to_cache = array_unique($tags_to_cache);
  849. $tags_str = db_escape_string(join(",", $tags_to_cache));
  850. db_query("UPDATE ttrss_user_entries
  851. SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
  852. AND owner_uid = $owner_uid");
  853. //db_query("COMMIT");
  854. }
  855. _debug("article processed", $debug_enabled);
  856. }
  857. _debug("purging feed...", $debug_enabled);
  858. purge_feed($feed, 0, $debug_enabled);
  859. db_query("UPDATE ttrss_feeds
  860. SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
  861. // db_query("COMMIT");
  862. } else {
  863. $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
  864. _debug("fetch error: $error_msg", $debug_enabled);
  865. if (count($rss->errors()) > 1) {
  866. foreach ($rss->errors() as $error) {
  867. _debug("+ $error");
  868. }
  869. }
  870. db_query(
  871. "UPDATE ttrss_feeds SET last_error = '$error_msg',
  872. last_updated = NOW() WHERE id = '$feed'");
  873. unset($rss);
  874. return;
  875. }
  876. _debug("done", $debug_enabled);
  877. return true;
  878. }
  879. function cache_enclosures($enclosures, $site_url, $debug) {
  880. foreach ($enclosures as $enc) {
  881. if (preg_match("/(image|audio|video)/", $enc[1])) {
  882. $src = rewrite_relative_url($site_url, $enc[0]);
  883. $local_filename = CACHE_DIR . "/images/" . sha1($src);
  884. if ($debug) _debug("cache_enclosures: downloading: $src to $local_filename");
  885. if (!file_exists($local_filename)) {
  886. $file_content = fetch_file_contents($src);
  887. if ($file_content && strlen($file_content) > _MIN_CACHE_FILE_SIZE) {
  888. file_put_contents($local_filename, $file_content);
  889. }
  890. } else {
  891. touch($local_filename);
  892. }
  893. }
  894. }
  895. }
  896. function cache_media($html, $site_url, $debug) {
  897. libxml_use_internal_errors(true);
  898. $charset_hack = '<head>
  899. <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
  900. </head>';
  901. $doc = new DOMDocument();
  902. $doc->loadHTML($charset_hack . $html);
  903. $xpath = new DOMXPath($doc);
  904. $entries = $xpath->query('(//img[@src])|(//video/source[@src])|(//audio/source[@src])');
  905. foreach ($entries as $entry) {
  906. if ($entry->hasAttribute('src') && strpos($entry->getAttribute('src'), "data:") !== 0) {
  907. $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
  908. $local_filename = CACHE_DIR . "/images/" . sha1($src);
  909. if ($debug) _debug("cache_media: downloading: $src to $local_filename");
  910. if (!file_exists($local_filename)) {
  911. $file_content = fetch_file_contents($src);
  912. if ($file_content && strlen($file_content) > _MIN_CACHE_FILE_SIZE) {
  913. file_put_contents($local_filename, $file_content);
  914. }
  915. } else {
  916. touch($local_filename);
  917. }
  918. }
  919. }
  920. }
  921. function expire_error_log($debug) {
  922. if ($debug) _debug("Removing old error log entries...");
  923. if (DB_TYPE == "pgsql") {
  924. db_query("DELETE FROM ttrss_error_log
  925. WHERE created_at < NOW() - INTERVAL '7 days'");
  926. } else {
  927. db_query("DELETE FROM ttrss_error_log
  928. WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
  929. }
  930. }
  931. function expire_lock_files($debug) {
  932. //if ($debug) _debug("Removing old lock files...");
  933. $num_deleted = 0;
  934. if (is_writable(LOCK_DIRECTORY)) {
  935. $files = glob(LOCK_DIRECTORY . "/*.lock");
  936. if ($files) {
  937. foreach ($files as $file) {
  938. if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
  939. unlink($file);
  940. ++$num_deleted;
  941. }
  942. }
  943. }
  944. }
  945. if ($debug) _debug("Removed $num_deleted old lock files.");
  946. }
  947. function expire_cached_files($debug) {
  948. foreach (array("simplepie", "images", "export", "upload") as $dir) {
  949. $cache_dir = CACHE_DIR . "/$dir";
  950. // if ($debug) _debug("Expiring $cache_dir");
  951. $num_deleted = 0;
  952. if (is_writable($cache_dir)) {
  953. $files = glob("$cache_dir/*");
  954. if ($files) {
  955. foreach ($files as $file) {
  956. if (time() - filemtime($file) > 86400*7) {
  957. unlink($file);
  958. ++$num_deleted;
  959. }
  960. }
  961. }
  962. }
  963. if ($debug) _debug("$cache_dir: removed $num_deleted files.");
  964. }
  965. }
  966. /**
  967. * Source: http://www.php.net/manual/en/function.parse-url.php#104527
  968. * Returns the url query as associative array
  969. *
  970. * @param string query
  971. * @return array params
  972. */
  973. function convertUrlQuery($query) {
  974. $queryParts = explode('&', $query);
  975. $params = array();
  976. foreach ($queryParts as $param) {
  977. $item = explode('=', $param);
  978. $params[$item[0]] = $item[1];
  979. }
  980. return $params;
  981. }
  982. function get_article_filters($filters, $title, $content, $link, $author, $tags, &$matched_rules = false) {
  983. $matches = array();
  984. foreach ($filters as $filter) {
  985. $match_any_rule = $filter["match_any_rule"];
  986. $inverse = $filter["inverse"];
  987. $filter_match = false;
  988. foreach ($filter["rules"] as $rule) {
  989. $match = false;
  990. $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
  991. $rule_inverse = $rule["inverse"];
  992. if (!$reg_exp)
  993. continue;
  994. switch ($rule["type"]) {
  995. case "title":
  996. $match = @preg_match("/$reg_exp/iu", $title);
  997. break;
  998. case "content":
  999. // we don't need to deal with multiline regexps
  1000. $content = preg_replace("/[\r\n\t]/", "", $content);
  1001. $match = @preg_match("/$reg_exp/iu", $content);
  1002. break;
  1003. case "both":
  1004. // we don't need to deal with multiline regexps
  1005. $content = preg_replace("/[\r\n\t]/", "", $content);
  1006. $match = (@preg_match("/$reg_exp/iu", $title) || @preg_match("/$reg_exp/iu", $content));
  1007. break;
  1008. case "link":
  1009. $match = @preg_match("/$reg_exp/iu", $link);
  1010. break;
  1011. case "author":
  1012. $match = @preg_match("/$reg_exp/iu", $author);
  1013. break;
  1014. case "tag":
  1015. foreach ($tags as $tag) {
  1016. if (@preg_match("/$reg_exp/iu", $tag)) {
  1017. $match = true;
  1018. break;
  1019. }
  1020. }
  1021. break;
  1022. }
  1023. if ($rule_inverse) $match = !$match;
  1024. if ($match_any_rule) {
  1025. if ($match) {
  1026. $filter_match = true;
  1027. break;
  1028. }
  1029. } else {
  1030. $filter_match = $match;
  1031. if (!$match) {
  1032. break;
  1033. }
  1034. }
  1035. }
  1036. if ($inverse) $filter_match = !$filter_match;
  1037. if ($filter_match) {
  1038. if (is_array($matched_rules)) array_push($matched_rules, $rule);
  1039. foreach ($filter["actions"] AS $action) {
  1040. array_push($matches, $action);
  1041. // if Stop action encountered, perform no further processing
  1042. if (isset($action["type"]) && $action["type"] == "stop") return $matches;
  1043. }
  1044. }
  1045. }
  1046. return $matches;
  1047. }
  1048. function find_article_filter($filters, $filter_name) {
  1049. foreach ($filters as $f) {
  1050. if ($f["type"] == $filter_name) {
  1051. return $f;
  1052. };
  1053. }
  1054. return false;
  1055. }
  1056. function find_article_filters($filters, $filter_name) {
  1057. $results = array();
  1058. foreach ($filters as $f) {
  1059. if ($f["type"] == $filter_name) {
  1060. array_push($results, $f);
  1061. };
  1062. }
  1063. return $results;
  1064. }
  1065. function calculate_article_score($filters) {
  1066. $score = 0;
  1067. foreach ($filters as $f) {
  1068. if ($f["type"] == "score") {
  1069. $score += $f["param"];
  1070. };
  1071. }
  1072. return $score;
  1073. }
  1074. function labels_contains_caption($labels, $caption) {
  1075. foreach ($labels as $label) {
  1076. if ($label[1] == $caption) {
  1077. return true;
  1078. }
  1079. }
  1080. return false;
  1081. }
  1082. function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
  1083. foreach ($filters as $f) {
  1084. if ($f["type"] == "label") {
  1085. if (!labels_contains_caption($article_labels, $f["param"])) {
  1086. label_add_article($id, $f["param"], $owner_uid);
  1087. }
  1088. }
  1089. }
  1090. }
  1091. function make_guid_from_title($title) {
  1092. return preg_replace("/[ \"\',.:;]/", "-",
  1093. mb_strtolower(strip_tags($title), 'utf-8'));
  1094. }
  1095. function cleanup_counters_cache($debug) {
  1096. $result = db_query("DELETE FROM ttrss_counters_cache
  1097. WHERE feed_id > 0 AND
  1098. (SELECT COUNT(id) FROM ttrss_feeds WHERE
  1099. id = feed_id AND
  1100. ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
  1101. $frows = db_affected_rows($result);
  1102. $result = db_query("DELETE FROM ttrss_cat_counters_cache
  1103. WHERE feed_id > 0 AND
  1104. (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
  1105. id = feed_id AND
  1106. ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
  1107. $crows = db_affected_rows($result);
  1108. if ($debug) _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
  1109. }
  1110. function housekeeping_user($owner_uid) {
  1111. $tmph = new PluginHost();
  1112. load_user_plugins($owner_uid, $tmph);
  1113. $tmph->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
  1114. }
  1115. function housekeeping_common($debug) {
  1116. expire_cached_files($debug);
  1117. expire_lock_files($debug);
  1118. expire_error_log($debug);
  1119. $count = update_feedbrowser_cache();
  1120. _debug("Feedbrowser updated, $count feeds processed.");
  1121. purge_orphans( true);
  1122. cleanup_counters_cache($debug);
  1123. //$rc = cleanup_tags( 14, 50000);
  1124. //_debug("Cleaned $rc cached tags.");
  1125. PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
  1126. }