Feed.php 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555
  1. <?php
  2. class FreshRSS_Feed extends Minz_Model {
  3. const PRIORITY_MAIN_STREAM = 10;
  4. const PRIORITY_NORMAL = 0;
  5. const PRIORITY_ARCHIVED = -10;
  6. const TTL_DEFAULT = 0;
  7. const KEEP_HISTORY_DEFAULT = -2;
  8. const KEEP_HISTORY_INFINITE = -1;
  9. private $id = 0;
  10. private $url;
  11. private $category = 1;
  12. private $nbEntries = -1;
  13. private $nbNotRead = -1;
  14. private $entries = null;
  15. private $name = '';
  16. private $website = '';
  17. private $description = '';
  18. private $lastUpdate = 0;
  19. private $priority = self::PRIORITY_MAIN_STREAM;
  20. private $pathEntries = '';
  21. private $httpAuth = '';
  22. private $error = false;
  23. private $keep_history = self::KEEP_HISTORY_DEFAULT;
  24. private $ttl = self::TTL_DEFAULT;
  25. private $attributes = array();
  26. private $mute = false;
  27. private $hash = null;
  28. private $lockPath = '';
  29. private $hubUrl = '';
  30. private $selfUrl = '';
  31. public function __construct($url, $validate = true) {
  32. if ($validate) {
  33. $this->_url($url);
  34. } else {
  35. $this->url = $url;
  36. }
  37. }
  38. public static function example() {
  39. $f = new FreshRSS_Feed('http://example.net/', false);
  40. $f->faviconPrepare();
  41. return $f;
  42. }
  43. public function id() {
  44. return $this->id;
  45. }
  46. public function hash() {
  47. if ($this->hash === null) {
  48. $salt = FreshRSS_Context::$system_conf->salt;
  49. $this->hash = hash('crc32b', $salt . $this->url);
  50. }
  51. return $this->hash;
  52. }
  53. public function url() {
  54. return $this->url;
  55. }
  56. public function selfUrl() {
  57. return $this->selfUrl;
  58. }
  59. public function hubUrl() {
  60. return $this->hubUrl;
  61. }
  62. public function category() {
  63. return $this->category;
  64. }
  65. public function entries() {
  66. return $this->entries === null ? array() : $this->entries;
  67. }
  68. public function name() {
  69. return $this->name;
  70. }
  71. public function website() {
  72. return $this->website;
  73. }
  74. public function description() {
  75. return $this->description;
  76. }
  77. public function lastUpdate() {
  78. return $this->lastUpdate;
  79. }
  80. public function priority() {
  81. return $this->priority;
  82. }
  83. public function pathEntries() {
  84. return $this->pathEntries;
  85. }
  86. public function httpAuth($raw = true) {
  87. if ($raw) {
  88. return $this->httpAuth;
  89. } else {
  90. $pos_colon = strpos($this->httpAuth, ':');
  91. $user = substr($this->httpAuth, 0, $pos_colon);
  92. $pass = substr($this->httpAuth, $pos_colon + 1);
  93. return array(
  94. 'username' => $user,
  95. 'password' => $pass
  96. );
  97. }
  98. }
  99. public function inError() {
  100. return $this->error;
  101. }
  102. public function keepHistory() {
  103. return $this->keep_history;
  104. }
  105. public function ttl() {
  106. return $this->ttl;
  107. }
  108. public function attributes($key = '') {
  109. if ($key == '') {
  110. return $this->attributes;
  111. } else {
  112. return isset($this->attributes[$key]) ? $this->attributes[$key] : null;
  113. }
  114. }
  115. public function mute() {
  116. return $this->mute;
  117. }
  118. // public function ttlExpire() {
  119. // $ttl = $this->ttl;
  120. // if ($ttl == self::TTL_DEFAULT) { //Default
  121. // $ttl = FreshRSS_Context::$user_conf->ttl_default;
  122. // }
  123. // if ($ttl == -1) { //Never
  124. // $ttl = 64000000; //~2 years. Good enough for PubSubHubbub logic
  125. // }
  126. // return $this->lastUpdate + $ttl;
  127. // }
  128. public function nbEntries() {
  129. if ($this->nbEntries < 0) {
  130. $feedDAO = FreshRSS_Factory::createFeedDao();
  131. $this->nbEntries = $feedDAO->countEntries($this->id());
  132. }
  133. return $this->nbEntries;
  134. }
  135. public function nbNotRead() {
  136. if ($this->nbNotRead < 0) {
  137. $feedDAO = FreshRSS_Factory::createFeedDao();
  138. $this->nbNotRead = $feedDAO->countNotRead($this->id());
  139. }
  140. return $this->nbNotRead;
  141. }
  142. public function faviconPrepare() {
  143. global $favicons_dir;
  144. require_once(LIB_PATH . '/favicons.php');
  145. $url = $this->website;
  146. if ($url == '') {
  147. $url = $this->url;
  148. }
  149. $txt = $favicons_dir . $this->hash() . '.txt';
  150. if (!file_exists($txt)) {
  151. file_put_contents($txt, $url);
  152. }
  153. if (FreshRSS_Context::$isCli) {
  154. $ico = $favicons_dir . $this->hash() . '.ico';
  155. $ico_mtime = @filemtime($ico);
  156. $txt_mtime = @filemtime($txt);
  157. if ($txt_mtime != false &&
  158. ($ico_mtime == false || $ico_mtime < $txt_mtime || ($ico_mtime < time() - (14 * 86400)))) {
  159. // no ico file or we should download a new one.
  160. $url = file_get_contents($txt);
  161. download_favicon($url, $ico) || touch($ico);
  162. }
  163. }
  164. }
  165. public static function faviconDelete($hash) {
  166. $path = DATA_PATH . '/favicons/' . $hash;
  167. @unlink($path . '.ico');
  168. @unlink($path . '.txt');
  169. }
  170. public function favicon() {
  171. return Minz_Url::display('/f.php?' . $this->hash());
  172. }
  173. public function _id($value) {
  174. $this->id = $value;
  175. }
  176. public function _url($value, $validate = true) {
  177. $this->hash = null;
  178. if ($validate) {
  179. $value = checkUrl($value);
  180. }
  181. if (empty($value)) {
  182. throw new FreshRSS_BadUrl_Exception($value);
  183. }
  184. $this->url = $value;
  185. }
  186. public function _category($value) {
  187. $value = intval($value);
  188. $this->category = $value >= 0 ? $value : 0;
  189. }
  190. public function _name($value) {
  191. $this->name = $value === null ? '' : $value;
  192. }
  193. public function _website($value, $validate = true) {
  194. if ($validate) {
  195. $value = checkUrl($value);
  196. }
  197. if (empty($value)) {
  198. $value = '';
  199. }
  200. $this->website = $value;
  201. }
  202. public function _description($value) {
  203. $this->description = $value === null ? '' : $value;
  204. }
  205. public function _lastUpdate($value) {
  206. $this->lastUpdate = $value;
  207. }
  208. public function _priority($value) {
  209. $this->priority = intval($value);
  210. }
  211. public function _pathEntries($value) {
  212. $this->pathEntries = $value;
  213. }
  214. public function _httpAuth($value) {
  215. $this->httpAuth = $value;
  216. }
  217. public function _error($value) {
  218. $this->error = (bool)$value;
  219. }
  220. public function _keepHistory($value) {
  221. $value = intval($value);
  222. $value = min($value, 1000000);
  223. $value = max($value, self::KEEP_HISTORY_DEFAULT);
  224. $this->keep_history = $value;
  225. }
  226. public function _ttl($value) {
  227. $value = intval($value);
  228. $value = min($value, 100000000);
  229. $this->ttl = abs($value);
  230. $this->mute = $value < self::TTL_DEFAULT;
  231. }
  232. public function _attributes($key, $value) {
  233. if ($key == '') {
  234. if (is_string($value)) {
  235. $value = json_decode($value, true);
  236. }
  237. if (is_array($value)) {
  238. $this->attributes = $value;
  239. }
  240. } elseif ($value === null) {
  241. unset($this->attributes[$key]);
  242. } else {
  243. $this->attributes[$key] = $value;
  244. }
  245. }
  246. public function _nbNotRead($value) {
  247. $this->nbNotRead = intval($value);
  248. }
  249. public function _nbEntries($value) {
  250. $this->nbEntries = intval($value);
  251. }
  252. public function load($loadDetails = false, $noCache = false) {
  253. if ($this->url !== null) {
  254. if (CACHE_PATH === false) {
  255. throw new Minz_FileNotExistException(
  256. 'CACHE_PATH',
  257. Minz_Exception::ERROR
  258. );
  259. } else {
  260. $url = htmlspecialchars_decode($this->url, ENT_QUOTES);
  261. if ($this->httpAuth != '') {
  262. $url = preg_replace('#((.+)://)(.+)#', '${1}' . $this->httpAuth . '@${3}', $url);
  263. }
  264. $feed = customSimplePie($this->attributes());
  265. if (substr($url, -11) === '#force_feed') {
  266. $feed->force_feed(true);
  267. $url = substr($url, 0, -11);
  268. }
  269. $feed->set_feed_url($url);
  270. if (!$loadDetails) { //Only activates auto-discovery when adding a new feed
  271. $feed->set_autodiscovery_level(SIMPLEPIE_LOCATOR_NONE);
  272. }
  273. $mtime = $feed->init();
  274. if ((!$mtime) || $feed->error()) {
  275. $errorMessage = $feed->error();
  276. throw new FreshRSS_Feed_Exception(
  277. ($errorMessage == '' ? 'Unknown error for feed' : $errorMessage) . ' [' . $url . ']'
  278. );
  279. }
  280. $links = $feed->get_links('self');
  281. $this->selfUrl = isset($links[0]) ? $links[0] : null;
  282. $links = $feed->get_links('hub');
  283. $this->hubUrl = isset($links[0]) ? $links[0] : null;
  284. if ($loadDetails) {
  285. // si on a utilisé l'auto-discover, notre url va avoir changé
  286. $subscribe_url = $feed->subscribe_url(false);
  287. $title = strtr(html_only_entity_decode($feed->get_title()), array('<' => '&lt;', '>' => '&gt;', '"' => '&quot;')); //HTML to HTML-PRE //ENT_COMPAT except &
  288. $this->_name($title == '' ? $url : $title);
  289. $this->_website(html_only_entity_decode($feed->get_link()));
  290. $this->_description(html_only_entity_decode($feed->get_description()));
  291. } else {
  292. //The case of HTTP 301 Moved Permanently
  293. $subscribe_url = $feed->subscribe_url(true);
  294. }
  295. $clean_url = SimplePie_Misc::url_remove_credentials($subscribe_url);
  296. if ($subscribe_url !== null && $subscribe_url !== $url) {
  297. $this->_url($clean_url);
  298. }
  299. if (($mtime === true) || ($mtime > $this->lastUpdate) || $noCache) {
  300. //Minz_Log::debug('FreshRSS no cache ' . $mtime . ' > ' . $this->lastUpdate . ' for ' . $clean_url);
  301. $this->loadEntries($feed); // et on charge les articles du flux
  302. } else {
  303. //Minz_Log::debug('FreshRSS use cache for ' . $clean_url);
  304. $this->entries = array();
  305. }
  306. $feed->__destruct(); //http://simplepie.org/wiki/faq/i_m_getting_memory_leaks
  307. unset($feed);
  308. }
  309. }
  310. }
  311. public function loadEntries($feed) {
  312. $entries = array();
  313. foreach ($feed->get_items() as $item) {
  314. $title = html_only_entity_decode(strip_tags($item->get_title()));
  315. $author = $item->get_author();
  316. $link = $item->get_permalink();
  317. $date = @strtotime($item->get_date());
  318. // gestion des tags (catégorie == tag)
  319. $tags_tmp = $item->get_categories();
  320. $tags = array();
  321. if ($tags_tmp !== null) {
  322. foreach ($tags_tmp as $tag) {
  323. $tags[] = html_only_entity_decode($tag->get_label());
  324. }
  325. }
  326. $content = html_only_entity_decode($item->get_content());
  327. $elinks = array();
  328. foreach ($item->get_enclosures() as $enclosure) {
  329. $elink = $enclosure->get_link();
  330. if ($elink != '' && empty($elinks[$elink])) {
  331. $elinks[$elink] = '1';
  332. $mime = strtolower($enclosure->get_type());
  333. if (strpos($mime, 'image/') === 0) {
  334. $content .= '<p class="enclosure"><img src="' . $elink . '" alt="" /></p>';
  335. } elseif (strpos($mime, 'audio/') === 0) {
  336. $content .= '<p class="enclosure"><audio preload="none" src="' . $elink
  337. . '" controls="controls"></audio> <a download="" href="' . $elink . '">💾</a></p>';
  338. } elseif (strpos($mime, 'video/') === 0) {
  339. $content .= '<p class="enclosure"><video preload="none" src="' . $elink
  340. . '" controls="controls"></video> <a download="" href="' . $elink . '">💾</a></p>';
  341. } elseif (strpos($mime, 'application/') === 0 || strpos($mime, 'text/') === 0) {
  342. $content .= '<p class="enclosure"><a download="" href="' . $elink . '">💾</a></p>';
  343. } else {
  344. unset($elinks[$elink]);
  345. }
  346. }
  347. }
  348. $entry = new FreshRSS_Entry(
  349. $this->id(),
  350. $item->get_id(false, false),
  351. $title === null ? '' : $title,
  352. $author === null ? '' : html_only_entity_decode(strip_tags($author->name == null ? $author->email : $author->name)),
  353. $content === null ? '' : $content,
  354. $link === null ? '' : $link,
  355. $date ? $date : time()
  356. );
  357. $entry->_tags($tags);
  358. // permet de récupérer le contenu des flux tronqués
  359. $entry->loadCompleteContent($this->pathEntries());
  360. $entries[] = $entry;
  361. unset($item);
  362. }
  363. $this->entries = $entries;
  364. }
  365. function cacheModifiedTime() {
  366. return @filemtime(CACHE_PATH . '/' . md5($this->url) . '.spc');
  367. }
  368. function lock() {
  369. $this->lockPath = TMP_PATH . '/' . $this->hash() . '.freshrss.lock';
  370. if (file_exists($this->lockPath) && ((time() - @filemtime($this->lockPath)) > 3600)) {
  371. @unlink($this->lockPath);
  372. }
  373. if (($handle = @fopen($this->lockPath, 'x')) === false) {
  374. return false;
  375. }
  376. //register_shutdown_function('unlink', $this->lockPath);
  377. @fclose($handle);
  378. return true;
  379. }
  380. function unlock() {
  381. @unlink($this->lockPath);
  382. }
  383. //<PubSubHubbub>
  384. function pubSubHubbubEnabled() {
  385. $url = $this->selfUrl ? $this->selfUrl : $this->url;
  386. $hubFilename = PSHB_PATH . '/feeds/' . base64url_encode($url) . '/!hub.json';
  387. if ($hubFile = @file_get_contents($hubFilename)) {
  388. $hubJson = json_decode($hubFile, true);
  389. if ($hubJson && empty($hubJson['error']) &&
  390. (empty($hubJson['lease_end']) || $hubJson['lease_end'] > time())) {
  391. return true;
  392. }
  393. }
  394. return false;
  395. }
  396. function pubSubHubbubError($error = true) {
  397. $url = $this->selfUrl ? $this->selfUrl : $this->url;
  398. $hubFilename = PSHB_PATH . '/feeds/' . base64url_encode($url) . '/!hub.json';
  399. $hubFile = @file_get_contents($hubFilename);
  400. $hubJson = $hubFile ? json_decode($hubFile, true) : array();
  401. if (!isset($hubJson['error']) || $hubJson['error'] !== (bool)$error) {
  402. $hubJson['error'] = (bool)$error;
  403. file_put_contents($hubFilename, json_encode($hubJson));
  404. Minz_Log::warning('Set error to ' . ($error ? 1 : 0) . ' for ' . $url, PSHB_LOG);
  405. }
  406. return false;
  407. }
  408. function pubSubHubbubPrepare() {
  409. $key = '';
  410. if (FreshRSS_Context::$system_conf->base_url && $this->hubUrl && $this->selfUrl && @is_dir(PSHB_PATH)) {
  411. $path = PSHB_PATH . '/feeds/' . base64url_encode($this->selfUrl);
  412. $hubFilename = $path . '/!hub.json';
  413. if ($hubFile = @file_get_contents($hubFilename)) {
  414. $hubJson = json_decode($hubFile, true);
  415. if (!$hubJson || empty($hubJson['key']) || !ctype_xdigit($hubJson['key'])) {
  416. $text = 'Invalid JSON for PubSubHubbub: ' . $this->url;
  417. Minz_Log::warning($text);
  418. Minz_Log::warning($text, PSHB_LOG);
  419. return false;
  420. }
  421. if ((!empty($hubJson['lease_end'])) && ($hubJson['lease_end'] < (time() + (3600 * 23)))) { //TODO: Make a better policy
  422. $text = 'PubSubHubbub lease ends at '
  423. . date('c', empty($hubJson['lease_end']) ? time() : $hubJson['lease_end'])
  424. . ' and needs renewal: ' . $this->url;
  425. Minz_Log::warning($text);
  426. Minz_Log::warning($text, PSHB_LOG);
  427. $key = $hubJson['key']; //To renew our lease
  428. } elseif (((!empty($hubJson['error'])) || empty($hubJson['lease_end'])) &&
  429. (empty($hubJson['lease_start']) || $hubJson['lease_start'] < time() - (3600 * 23))) { //Do not renew too often
  430. $key = $hubJson['key']; //To renew our lease
  431. }
  432. } else {
  433. @mkdir($path, 0777, true);
  434. $key = sha1($path . FreshRSS_Context::$system_conf->salt);
  435. $hubJson = array(
  436. 'hub' => $this->hubUrl,
  437. 'key' => $key,
  438. );
  439. file_put_contents($hubFilename, json_encode($hubJson));
  440. @mkdir(PSHB_PATH . '/keys/');
  441. file_put_contents(PSHB_PATH . '/keys/' . $key . '.txt', base64url_encode($this->selfUrl));
  442. $text = 'PubSubHubbub prepared for ' . $this->url;
  443. Minz_Log::debug($text);
  444. Minz_Log::debug($text, PSHB_LOG);
  445. }
  446. $currentUser = Minz_Session::param('currentUser');
  447. if (FreshRSS_user_Controller::checkUsername($currentUser) && !file_exists($path . '/' . $currentUser . '.txt')) {
  448. touch($path . '/' . $currentUser . '.txt');
  449. }
  450. }
  451. return $key;
  452. }
  453. //Parameter true to subscribe, false to unsubscribe.
  454. function pubSubHubbubSubscribe($state) {
  455. $url = $this->selfUrl ? $this->selfUrl : $this->url;
  456. if (FreshRSS_Context::$system_conf->base_url && $url) {
  457. $hubFilename = PSHB_PATH . '/feeds/' . base64url_encode($url) . '/!hub.json';
  458. $hubFile = @file_get_contents($hubFilename);
  459. if ($hubFile === false) {
  460. Minz_Log::warning('JSON not found for PubSubHubbub: ' . $this->url);
  461. return false;
  462. }
  463. $hubJson = json_decode($hubFile, true);
  464. if (!$hubJson || empty($hubJson['key']) || !ctype_xdigit($hubJson['key']) || empty($hubJson['hub'])) {
  465. Minz_Log::warning('Invalid JSON for PubSubHubbub: ' . $this->url);
  466. return false;
  467. }
  468. $callbackUrl = checkUrl(Minz_Request::getBaseUrl() . '/api/pshb.php?k=' . $hubJson['key']);
  469. if ($callbackUrl == '') {
  470. Minz_Log::warning('Invalid callback for PubSubHubbub: ' . $this->url);
  471. return false;
  472. }
  473. if (!$state) { //unsubscribe
  474. $hubJson['lease_end'] = time() - 60;
  475. file_put_contents($hubFilename, json_encode($hubJson));
  476. }
  477. $ch = curl_init();
  478. curl_setopt_array($ch, array(
  479. CURLOPT_URL => $hubJson['hub'],
  480. CURLOPT_RETURNTRANSFER => true,
  481. CURLOPT_POSTFIELDS => http_build_query(array(
  482. 'hub.verify' => 'sync',
  483. 'hub.mode' => $state ? 'subscribe' : 'unsubscribe',
  484. 'hub.topic' => $url,
  485. 'hub.callback' => $callbackUrl,
  486. )),
  487. CURLOPT_USERAGENT => FRESHRSS_USERAGENT,
  488. CURLOPT_MAXREDIRS => 10,
  489. ));
  490. if (version_compare(PHP_VERSION, '5.6.0') >= 0 || ini_get('open_basedir') == '') {
  491. curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true); //Keep option separated for open_basedir PHP bug 65646
  492. }
  493. if (defined('CURLOPT_ENCODING')) {
  494. curl_setopt($ch, CURLOPT_ENCODING, ''); //Enable all encodings
  495. }
  496. $response = curl_exec($ch);
  497. $info = curl_getinfo($ch);
  498. Minz_Log::warning('PubSubHubbub ' . ($state ? 'subscribe' : 'unsubscribe') . ' to ' . $url .
  499. ' with callback ' . $callbackUrl . ': ' . $info['http_code'] . ' ' . $response, PSHB_LOG);
  500. if (substr($info['http_code'], 0, 1) == '2') {
  501. return true;
  502. } else {
  503. $hubJson['lease_start'] = time(); //Prevent trying again too soon
  504. $hubJson['error'] = true;
  505. file_put_contents($hubFilename, json_encode($hubJson));
  506. return false;
  507. }
  508. }
  509. return false;
  510. }
  511. //</PubSubHubbub>
  512. }