Feed.php 33 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088
  1. <?php
  2. class FreshRSS_Feed extends Minz_Model {
  3. /**
  4. * Normal RSS or Atom feed
  5. * @var int
  6. */
  7. const KIND_RSS = 0;
  8. /**
  9. * Invalid RSS or Atom feed
  10. * @var int
  11. */
  12. const KIND_RSS_FORCED = 2;
  13. /**
  14. * Normal HTML with XPath scraping
  15. * @var int
  16. */
  17. const KIND_HTML_XPATH = 10;
  18. /**
  19. * Normal XML with XPath scraping
  20. * @var int
  21. */
  22. const KIND_XML_XPATH = 15;
  23. /**
  24. * Normal JSON with XPath scraping
  25. * @var int
  26. */
  27. const KIND_JSON_XPATH = 20;
  28. const PRIORITY_MAIN_STREAM = 10;
  29. const PRIORITY_NORMAL = 0;
  30. const PRIORITY_ARCHIVED = -10;
  31. const TTL_DEFAULT = 0;
  32. const ARCHIVING_RETENTION_COUNT_LIMIT = 10000;
  33. const ARCHIVING_RETENTION_PERIOD = 'P3M';
  34. /** @var int */
  35. private $id = 0;
  36. /** @var string */
  37. private $url = '';
  38. /** @var int */
  39. private $kind = 0;
  40. /** @var int */
  41. private $categoryId = 1;
  42. /** @var FreshRSS_Category|null */
  43. private $category;
  44. /** @var int */
  45. private $nbEntries = -1;
  46. /** @var int */
  47. private $nbNotRead = -1;
  48. /** @var int */
  49. private $nbPendingNotRead = 0;
  50. /** @var string */
  51. private $name = '';
  52. /** @var string */
  53. private $website = '';
  54. /** @var string */
  55. private $description = '';
  56. /** @var int */
  57. private $lastUpdate = 0;
  58. /** @var int */
  59. private $priority = self::PRIORITY_MAIN_STREAM;
  60. /** @var string */
  61. private $pathEntries = '';
  62. /** @var string */
  63. private $httpAuth = '';
  64. /** @var bool */
  65. private $error = false;
  66. /** @var int */
  67. private $ttl = self::TTL_DEFAULT;
  68. private $attributes = [];
  69. /** @var bool */
  70. private $mute = false;
  71. /** @var string */
  72. private $hash = '';
  73. /** @var string */
  74. private $lockPath = '';
  75. /** @var string */
  76. private $hubUrl = '';
  77. /** @var string */
  78. private $selfUrl = '';
  79. /** @var array<FreshRSS_FilterAction> $filterActions */
  80. private $filterActions = null;
  81. public function __construct(string $url, bool $validate = true) {
  82. if ($validate) {
  83. $this->_url($url);
  84. } else {
  85. $this->url = $url;
  86. }
  87. }
  88. /**
  89. * @return FreshRSS_Feed
  90. */
  91. public static function example() {
  92. $f = new FreshRSS_Feed('http://example.net/', false);
  93. $f->faviconPrepare();
  94. return $f;
  95. }
  96. public function id(): int {
  97. return $this->id;
  98. }
  99. public function hash(): string {
  100. if ($this->hash == '') {
  101. $salt = FreshRSS_Context::$system_conf->salt;
  102. $this->hash = hash('crc32b', $salt . $this->url);
  103. }
  104. return $this->hash;
  105. }
  106. public function url(bool $includeCredentials = true): string {
  107. return $includeCredentials ? $this->url : SimplePie_Misc::url_remove_credentials($this->url);
  108. }
  109. public function selfUrl(): string {
  110. return $this->selfUrl;
  111. }
  112. public function kind(): int {
  113. return $this->kind;
  114. }
  115. public function hubUrl(): string {
  116. return $this->hubUrl;
  117. }
  118. /**
  119. * @return FreshRSS_Category|null|false
  120. */
  121. public function category() {
  122. if ($this->category === null) {
  123. $catDAO = FreshRSS_Factory::createCategoryDao();
  124. $this->category = $catDAO->searchById($this->categoryId);
  125. }
  126. return $this->category;
  127. }
  128. public function categoryId(): int {
  129. return $this->categoryId;
  130. }
  131. public function entries() {
  132. Minz_Log::warning(__method__ . ' is deprecated since FreshRSS 1.16.1!');
  133. $simplePie = $this->load(false, true);
  134. return $simplePie == null ? [] : iterator_to_array($this->loadEntries($simplePie));
  135. }
  136. public function name($raw = false): string {
  137. return $raw || $this->name != '' ? $this->name : preg_replace('%^https?://(www[.])?%i', '', $this->url);
  138. }
  139. /** @return string HTML-encoded URL of the Web site of the feed */
  140. public function website(): string {
  141. return $this->website;
  142. }
  143. public function description(): string {
  144. return $this->description;
  145. }
  146. public function lastUpdate(): int {
  147. return $this->lastUpdate;
  148. }
  149. public function priority(): int {
  150. return $this->priority;
  151. }
  152. /** @return string HTML-encoded CSS selector */
  153. public function pathEntries(): string {
  154. return $this->pathEntries;
  155. }
  156. public function httpAuth($raw = true) {
  157. if ($raw) {
  158. return $this->httpAuth;
  159. } else {
  160. $pos_colon = strpos($this->httpAuth, ':');
  161. $user = substr($this->httpAuth, 0, $pos_colon);
  162. $pass = substr($this->httpAuth, $pos_colon + 1);
  163. return array(
  164. 'username' => $user,
  165. 'password' => $pass
  166. );
  167. }
  168. }
  169. public function inError(): bool {
  170. return $this->error;
  171. }
  172. /**
  173. * @param bool $raw true for database version combined with mute information, false otherwise
  174. */
  175. public function ttl(bool $raw = false): int {
  176. if ($raw) {
  177. $ttl = $this->ttl;
  178. if ($this->mute && FreshRSS_Feed::TTL_DEFAULT === $ttl) {
  179. $ttl = FreshRSS_Context::$user_conf ? FreshRSS_Context::$user_conf->ttl_default : 3600;
  180. }
  181. return $ttl * ($this->mute ? -1 : 1);
  182. }
  183. return $this->ttl;
  184. }
  185. /**
  186. * @phpstan-return ($key is non-empty-string ? mixed : array<string,mixed>)
  187. * @return array<string,mixed>|mixed|null
  188. */
  189. public function attributes(string $key = '') {
  190. if ($key === '') {
  191. return $this->attributes;
  192. } else {
  193. return $this->attributes[$key] ?? null;
  194. }
  195. }
  196. public function mute(): bool {
  197. return $this->mute;
  198. }
  199. public function nbEntries(): int {
  200. if ($this->nbEntries < 0) {
  201. $feedDAO = FreshRSS_Factory::createFeedDao();
  202. $this->nbEntries = $feedDAO->countEntries($this->id());
  203. }
  204. return $this->nbEntries;
  205. }
  206. public function nbNotRead($includePending = false): int {
  207. if ($this->nbNotRead < 0) {
  208. $feedDAO = FreshRSS_Factory::createFeedDao();
  209. $this->nbNotRead = $feedDAO->countNotRead($this->id());
  210. }
  211. return $this->nbNotRead + ($includePending ? $this->nbPendingNotRead : 0);
  212. }
  213. public function faviconPrepare() {
  214. require_once(LIB_PATH . '/favicons.php');
  215. $url = $this->website;
  216. if ($url == '') {
  217. $url = $this->url;
  218. }
  219. $txt = FAVICONS_DIR . $this->hash() . '.txt';
  220. if (@file_get_contents($txt) !== $url) {
  221. file_put_contents($txt, $url);
  222. }
  223. if (FreshRSS_Context::$isCli) {
  224. $ico = FAVICONS_DIR . $this->hash() . '.ico';
  225. $ico_mtime = @filemtime($ico);
  226. $txt_mtime = @filemtime($txt);
  227. if ($txt_mtime != false &&
  228. ($ico_mtime == false || $ico_mtime < $txt_mtime || ($ico_mtime < time() - (14 * 86400)))) {
  229. // no ico file or we should download a new one.
  230. $url = file_get_contents($txt);
  231. download_favicon($url, $ico) || touch($ico);
  232. }
  233. }
  234. }
  235. public static function faviconDelete($hash) {
  236. $path = DATA_PATH . '/favicons/' . $hash;
  237. @unlink($path . '.ico');
  238. @unlink($path . '.txt');
  239. }
  240. public function favicon(): string {
  241. return Minz_Url::display('/f.php?' . $this->hash());
  242. }
  243. public function _id($value) {
  244. $this->id = intval($value);
  245. }
  246. public function _url(string $value, bool $validate = true) {
  247. $this->hash = '';
  248. $url = $value;
  249. if ($validate) {
  250. $url = checkUrl($url);
  251. }
  252. if ($url == '') {
  253. throw new FreshRSS_BadUrl_Exception($value);
  254. }
  255. $this->url = $url;
  256. }
  257. public function _kind(int $value) {
  258. $this->kind = $value;
  259. }
  260. /** @param FreshRSS_Category|null $cat */
  261. public function _category($cat) {
  262. $this->category = $cat;
  263. $this->categoryId = $this->category == null ? 0 : $this->category->id();
  264. }
  265. /** @param int|string $id */
  266. public function _categoryId($id) {
  267. $this->category = null;
  268. $this->categoryId = intval($id);
  269. }
  270. public function _name(string $value) {
  271. $this->name = $value == '' ? '' : trim($value);
  272. }
  273. public function _website(string $value, bool $validate = true) {
  274. if ($validate) {
  275. $value = checkUrl($value);
  276. }
  277. if ($value == '') {
  278. $value = '';
  279. }
  280. $this->website = $value;
  281. }
  282. public function _description(string $value) {
  283. $this->description = $value == '' ? '' : $value;
  284. }
  285. public function _lastUpdate($value) {
  286. $this->lastUpdate = intval($value);
  287. }
  288. public function _priority($value) {
  289. $this->priority = intval($value);
  290. }
  291. /** @param string $value HTML-encoded CSS selector */
  292. public function _pathEntries(string $value) {
  293. $this->pathEntries = $value;
  294. }
  295. public function _httpAuth(string $value) {
  296. $this->httpAuth = $value;
  297. }
  298. public function _error($value) {
  299. $this->error = (bool)$value;
  300. }
  301. public function _mute(bool $value) {
  302. $this->mute = $value;
  303. }
  304. public function _ttl($value) {
  305. $value = intval($value);
  306. $value = min($value, 100000000);
  307. $this->ttl = abs($value);
  308. $this->mute = $value < self::TTL_DEFAULT;
  309. }
  310. /** @param string|array<mixed>|bool|int|null $value Value, not HTML-encoded */
  311. public function _attributes(string $key, $value) {
  312. if ($key == '') {
  313. if (is_string($value)) {
  314. $value = json_decode($value, true);
  315. }
  316. if (is_array($value)) {
  317. $this->attributes = $value;
  318. }
  319. } elseif ($value === null) {
  320. unset($this->attributes[$key]);
  321. } else {
  322. $this->attributes[$key] = $value;
  323. }
  324. }
  325. public function _nbNotRead($value) {
  326. $this->nbNotRead = intval($value);
  327. }
  328. public function _nbEntries($value) {
  329. $this->nbEntries = intval($value);
  330. }
  331. /**
  332. * @return SimplePie|null
  333. */
  334. public function load(bool $loadDetails = false, bool $noCache = false) {
  335. if ($this->url != '') {
  336. // @phpstan-ignore-next-line
  337. if (CACHE_PATH === false) {
  338. throw new Minz_FileNotExistException(
  339. 'CACHE_PATH',
  340. Minz_Exception::ERROR
  341. );
  342. } else {
  343. $url = htmlspecialchars_decode($this->url, ENT_QUOTES);
  344. if ($this->httpAuth != '') {
  345. $url = preg_replace('#((.+)://)(.+)#', '${1}' . $this->httpAuth . '@${3}', $url);
  346. }
  347. $simplePie = customSimplePie($this->attributes());
  348. if (substr($url, -11) === '#force_feed') {
  349. $simplePie->force_feed(true);
  350. $url = substr($url, 0, -11);
  351. }
  352. $simplePie->set_feed_url($url);
  353. if (!$loadDetails) { //Only activates auto-discovery when adding a new feed
  354. $simplePie->set_autodiscovery_level(SIMPLEPIE_LOCATOR_NONE);
  355. }
  356. if ($this->attributes('clear_cache')) {
  357. // Do not use `$simplePie->enable_cache(false);` as it would prevent caching in multiuser context
  358. $this->clearCache();
  359. }
  360. Minz_ExtensionManager::callHook('simplepie_before_init', $simplePie, $this);
  361. $mtime = $simplePie->init();
  362. if ((!$mtime) || $simplePie->error()) {
  363. $errorMessage = $simplePie->error();
  364. throw new FreshRSS_Feed_Exception(
  365. ($errorMessage == '' ? 'Unknown error for feed' : $errorMessage) . ' [' . $this->url . ']',
  366. $simplePie->status_code()
  367. );
  368. }
  369. $links = $simplePie->get_links('self');
  370. $this->selfUrl = empty($links[0]) ? '' : checkUrl($links[0]);
  371. if ($this->selfUrl == false) {
  372. $this->selfUrl = '';
  373. }
  374. $links = $simplePie->get_links('hub');
  375. $this->hubUrl = empty($links[0]) ? '' : checkUrl($links[0]);
  376. if ($this->hubUrl == false) {
  377. $this->hubUrl = '';
  378. }
  379. if ($loadDetails) {
  380. // si on a utilisé l’auto-discover, notre url va avoir changé
  381. $subscribe_url = $simplePie->subscribe_url(false);
  382. //HTML to HTML-PRE //ENT_COMPAT except '&'
  383. $title = strtr(html_only_entity_decode($simplePie->get_title()), array('<' => '&lt;', '>' => '&gt;', '"' => '&quot;'));
  384. $this->_name($title == '' ? $this->url : $title);
  385. $this->_website(html_only_entity_decode($simplePie->get_link()));
  386. $this->_description(html_only_entity_decode($simplePie->get_description()));
  387. } else {
  388. //The case of HTTP 301 Moved Permanently
  389. $subscribe_url = $simplePie->subscribe_url(true);
  390. }
  391. $clean_url = SimplePie_Misc::url_remove_credentials($subscribe_url);
  392. if ($subscribe_url !== null && $subscribe_url !== $url) {
  393. $this->_url($clean_url);
  394. }
  395. if (($mtime === true) || ($mtime > $this->lastUpdate) || $noCache) {
  396. //Minz_Log::debug('FreshRSS no cache ' . $mtime . ' > ' . $this->lastUpdate . ' for ' . $clean_url);
  397. return $simplePie;
  398. }
  399. //Minz_Log::debug('FreshRSS use cache for ' . $clean_url);
  400. }
  401. }
  402. return null;
  403. }
  404. /**
  405. * @return array<string>
  406. */
  407. public function loadGuids(SimplePie $simplePie) {
  408. $hasUniqueGuids = true;
  409. $testGuids = [];
  410. $guids = [];
  411. $hasBadGuids = $this->attributes('hasBadGuids');
  412. $items = $simplePie->get_items();
  413. if (empty($items)) {
  414. return $guids;
  415. }
  416. for ($i = count($items) - 1; $i >= 0; $i--) {
  417. $item = $items[$i];
  418. if ($item == null) {
  419. continue;
  420. }
  421. $guid = safe_ascii($item->get_id(false, false));
  422. $hasUniqueGuids &= empty($testGuids['_' . $guid]);
  423. $testGuids['_' . $guid] = true;
  424. $guids[] = $guid;
  425. }
  426. if ($hasBadGuids != !$hasUniqueGuids) {
  427. $hasBadGuids = !$hasUniqueGuids;
  428. if ($hasBadGuids) {
  429. Minz_Log::warning('Feed has invalid GUIDs: ' . $this->url);
  430. } else {
  431. Minz_Log::warning('Feed has valid GUIDs again: ' . $this->url);
  432. }
  433. $feedDAO = FreshRSS_Factory::createFeedDao();
  434. $feedDAO->updateFeedAttribute($this, 'hasBadGuids', $hasBadGuids);
  435. }
  436. return $guids;
  437. }
  438. public function loadEntries(SimplePie $simplePie) {
  439. $hasBadGuids = $this->attributes('hasBadGuids');
  440. $items = $simplePie->get_items();
  441. if (empty($items)) {
  442. return;
  443. }
  444. // We want chronological order and SimplePie uses reverse order.
  445. for ($i = count($items) - 1; $i >= 0; $i--) {
  446. $item = $items[$i];
  447. if ($item == null) {
  448. continue;
  449. }
  450. $title = html_only_entity_decode(strip_tags($item->get_title() ?? ''));
  451. $authors = $item->get_authors();
  452. $link = $item->get_permalink();
  453. $date = @strtotime($item->get_date() ?? '');
  454. //Tag processing (tag == category)
  455. $categories = $item->get_categories();
  456. $tags = array();
  457. if (is_array($categories)) {
  458. foreach ($categories as $category) {
  459. $text = html_only_entity_decode($category->get_label());
  460. //Some feeds use a single category with comma-separated tags
  461. $labels = explode(',', $text);
  462. if (is_array($labels)) {
  463. foreach ($labels as $label) {
  464. $tags[] = trim($label);
  465. }
  466. }
  467. }
  468. $tags = array_unique($tags);
  469. }
  470. $content = html_only_entity_decode($item->get_content());
  471. $attributeThumbnail = $item->get_thumbnail() ?? [];
  472. if (empty($attributeThumbnail['url'])) {
  473. $attributeThumbnail['url'] = '';
  474. }
  475. $attributeEnclosures = [];
  476. if (!empty($item->get_enclosures())) {
  477. foreach ($item->get_enclosures() as $enclosure) {
  478. $elink = $enclosure->get_link();
  479. if ($elink != '') {
  480. $etitle = $enclosure->get_title() ?? '';
  481. $credit = $enclosure->get_credit() ?? null;
  482. $description = $enclosure->get_description() ?? '';
  483. $mime = strtolower($enclosure->get_type() ?? '');
  484. $medium = strtolower($enclosure->get_medium() ?? '');
  485. $height = $enclosure->get_height();
  486. $width = $enclosure->get_width();
  487. $length = $enclosure->get_length();
  488. $attributeEnclosure = [
  489. 'url' => $elink,
  490. ];
  491. if ($etitle != '') $attributeEnclosure['title'] = $etitle;
  492. if ($credit != null) $attributeEnclosure['credit'] = $credit->get_name();
  493. if ($description != '') $attributeEnclosure['description'] = $description;
  494. if ($mime != '') $attributeEnclosure['type'] = $mime;
  495. if ($medium != '') $attributeEnclosure['medium'] = $medium;
  496. if ($length != '') $attributeEnclosure['length'] = intval($length);
  497. if ($height != '') $attributeEnclosure['height'] = intval($height);
  498. if ($width != '') $attributeEnclosure['width'] = intval($width);
  499. if (!empty($enclosure->get_thumbnails())) {
  500. foreach ($enclosure->get_thumbnails() as $thumbnail) {
  501. if ($thumbnail !== $attributeThumbnail['url']) {
  502. $attributeEnclosure['thumbnails'][] = $thumbnail;
  503. }
  504. }
  505. }
  506. $attributeEnclosures[] = $attributeEnclosure;
  507. }
  508. }
  509. }
  510. $guid = safe_ascii($item->get_id(false, false));
  511. unset($item);
  512. $authorNames = '';
  513. if (is_array($authors)) {
  514. foreach ($authors as $author) {
  515. $authorName = $author->name != '' ? $author->name : $author->email;
  516. if ($authorName != '') {
  517. $authorNames .= escapeToUnicodeAlternative(strip_tags($authorName), true) . '; ';
  518. }
  519. }
  520. }
  521. $authorNames = substr($authorNames, 0, -2);
  522. $entry = new FreshRSS_Entry(
  523. $this->id(),
  524. $hasBadGuids ? '' : $guid,
  525. $title == '' ? '' : $title,
  526. $authorNames,
  527. $content == '' ? '' : $content,
  528. $link == '' ? '' : $link,
  529. $date ? $date : time()
  530. );
  531. $entry->_tags($tags);
  532. $entry->_feed($this);
  533. if (!empty($attributeThumbnail['url'])) {
  534. $entry->_attributes('thumbnail', $attributeThumbnail);
  535. }
  536. $entry->_attributes('enclosures', $attributeEnclosures);
  537. $entry->hash(); //Must be computed before loading full content
  538. $entry->loadCompleteContent(); // Optionally load full content for truncated feeds
  539. yield $entry;
  540. }
  541. }
  542. /**
  543. * @return SimplePie|null
  544. */
  545. public function loadHtmlXpath() {
  546. if ($this->url == '') {
  547. return null;
  548. }
  549. $feedSourceUrl = htmlspecialchars_decode($this->url, ENT_QUOTES);
  550. if ($this->httpAuth != '') {
  551. $feedSourceUrl = preg_replace('#((.+)://)(.+)#', '${1}' . $this->httpAuth . '@${3}', $feedSourceUrl);
  552. }
  553. // Same naming conventions than https://rss-bridge.github.io/rss-bridge/Bridge_API/XPathAbstract.html
  554. // https://rss-bridge.github.io/rss-bridge/Bridge_API/BridgeAbstract.html#collectdata
  555. /** @var array<string,string> */
  556. $xPathSettings = $this->attributes('xpath');
  557. $xPathFeedTitle = $xPathSettings['feedTitle'] ?? '';
  558. $xPathItem = $xPathSettings['item'] ?? '';
  559. $xPathItemTitle = $xPathSettings['itemTitle'] ?? '';
  560. $xPathItemContent = $xPathSettings['itemContent'] ?? '';
  561. $xPathItemUri = $xPathSettings['itemUri'] ?? '';
  562. $xPathItemAuthor = $xPathSettings['itemAuthor'] ?? '';
  563. $xPathItemTimestamp = $xPathSettings['itemTimestamp'] ?? '';
  564. $xPathItemTimeFormat = $xPathSettings['itemTimeFormat'] ?? '';
  565. $xPathItemThumbnail = $xPathSettings['itemThumbnail'] ?? '';
  566. $xPathItemCategories = $xPathSettings['itemCategories'] ?? '';
  567. $xPathItemUid = $xPathSettings['itemUid'] ?? '';
  568. if ($xPathItem == '') {
  569. return null;
  570. }
  571. $cachePath = FreshRSS_Feed::cacheFilename($feedSourceUrl, $this->attributes(), $this->kind());
  572. $html = httpGet($feedSourceUrl, $cachePath,
  573. $this->kind() === FreshRSS_Feed::KIND_XML_XPATH ? 'xml' : 'html', $this->attributes());
  574. if (strlen($html) <= 0) {
  575. return null;
  576. }
  577. $view = new FreshRSS_View();
  578. $view->_path('index/rss.phtml');
  579. $view->internal_rendering = true;
  580. $view->rss_url = $feedSourceUrl;
  581. $view->entries = [];
  582. try {
  583. $doc = new DOMDocument();
  584. $doc->recover = true;
  585. $doc->strictErrorChecking = false;
  586. switch ($this->kind()) {
  587. case FreshRSS_Feed::KIND_HTML_XPATH:
  588. $doc->loadHTML($html, LIBXML_NONET | LIBXML_NOERROR | LIBXML_NOWARNING);
  589. break;
  590. case FreshRSS_Feed::KIND_XML_XPATH:
  591. $doc->loadXML($html, LIBXML_NONET | LIBXML_NOERROR | LIBXML_NOWARNING);
  592. break;
  593. default:
  594. return null;
  595. }
  596. $xpath = new DOMXPath($doc);
  597. $view->rss_title = $xPathFeedTitle == '' ? $this->name() :
  598. htmlspecialchars(@$xpath->evaluate('normalize-space(' . $xPathFeedTitle . ')'), ENT_COMPAT, 'UTF-8');
  599. $view->rss_base = htmlspecialchars(trim($xpath->evaluate('normalize-space(//base/@href)')), ENT_COMPAT, 'UTF-8');
  600. $nodes = $xpath->query($xPathItem);
  601. if (empty($nodes)) {
  602. return null;
  603. }
  604. foreach ($nodes as $node) {
  605. $item = [];
  606. $item['title'] = $xPathItemTitle == '' ? '' : @$xpath->evaluate('normalize-space(' . $xPathItemTitle . ')', $node);
  607. $item['content'] = '';
  608. if ($xPathItemContent != '') {
  609. $result = @$xpath->evaluate($xPathItemContent, $node);
  610. if ($result instanceof DOMNodeList) {
  611. // List of nodes, save as HTML
  612. $content = '';
  613. foreach ($result as $child) {
  614. $content .= $doc->saveHTML($child) . "\n";
  615. }
  616. $item['content'] = $content;
  617. } else {
  618. // Typed expression, save as-is
  619. $item['content'] = strval($result);
  620. }
  621. }
  622. $item['link'] = $xPathItemUri == '' ? '' : @$xpath->evaluate('normalize-space(' . $xPathItemUri . ')', $node);
  623. $item['author'] = $xPathItemAuthor == '' ? '' : @$xpath->evaluate('normalize-space(' . $xPathItemAuthor . ')', $node);
  624. $item['timestamp'] = $xPathItemTimestamp == '' ? '' : @$xpath->evaluate('normalize-space(' . $xPathItemTimestamp . ')', $node);
  625. if ($xPathItemTimeFormat != '') {
  626. $dateTime = DateTime::createFromFormat($xPathItemTimeFormat, $item['timestamp'] ?? '');
  627. if ($dateTime != false) {
  628. $item['timestamp'] = $dateTime->format(DateTime::ATOM);
  629. }
  630. }
  631. $item['thumbnail'] = $xPathItemThumbnail == '' ? '' : @$xpath->evaluate('normalize-space(' . $xPathItemThumbnail . ')', $node);
  632. if ($xPathItemCategories != '') {
  633. $itemCategories = @$xpath->query($xPathItemCategories, $node);
  634. if ($itemCategories) {
  635. foreach ($itemCategories as $itemCategory) {
  636. $item['categories'][] = $itemCategory->textContent;
  637. }
  638. }
  639. }
  640. if ($xPathItemUid != '') {
  641. $item['guid'] = @$xpath->evaluate('normalize-space(' . $xPathItemUid . ')', $node);
  642. }
  643. if (empty($item['guid'])) {
  644. $item['guid'] = 'urn:sha1:' . sha1($item['title'] . $item['content'] . $item['link']);
  645. }
  646. if ($item['title'] != '' || $item['content'] != '' || $item['link'] != '') {
  647. // HTML-encoding/escaping of the relevant fields (all except 'content')
  648. foreach (['author', 'categories', 'guid', 'link', 'thumbnail', 'timestamp', 'title'] as $key) {
  649. if (!empty($item[$key])) {
  650. $item[$key] = Minz_Helper::htmlspecialchars_utf8($item[$key]);
  651. }
  652. }
  653. // CDATA protection
  654. $item['content'] = str_replace(']]>', ']]&gt;', $item['content']);
  655. $view->entries[] = FreshRSS_Entry::fromArray($item);
  656. }
  657. }
  658. } catch (Exception $ex) {
  659. Minz_Log::warning($ex->getMessage());
  660. return null;
  661. }
  662. $simplePie = customSimplePie();
  663. $simplePie->set_raw_data($view->renderToString());
  664. $simplePie->init();
  665. return $simplePie;
  666. }
  667. /**
  668. * To keep track of some new potentially unread articles since last commit+fetch from database
  669. */
  670. public function incPendingUnread(int $n = 1) {
  671. $this->nbPendingNotRead += $n;
  672. }
  673. /**
  674. * Remember to call updateCachedValue($id_feed) or updateCachedValues() just after.
  675. * @return int|false the number of lines affected, or false if not applicable
  676. */
  677. public function keepMaxUnread() {
  678. $keepMaxUnread = $this->attributes('keep_max_n_unread');
  679. if ($keepMaxUnread === null) {
  680. $keepMaxUnread = FreshRSS_Context::$user_conf->mark_when['max_n_unread'];
  681. }
  682. if ($keepMaxUnread > 0 && $this->nbNotRead(false) + $this->nbPendingNotRead > $keepMaxUnread) {
  683. $feedDAO = FreshRSS_Factory::createFeedDao();
  684. return $feedDAO->keepMaxUnread($this->id(), max(0, $keepMaxUnread - $this->nbPendingNotRead));
  685. }
  686. return false;
  687. }
  688. /**
  689. * Applies the *mark as read upon gone* policy, if enabled.
  690. * Remember to call updateCachedValue($id_feed) or updateCachedValues() just after.
  691. * @return int|false the number of lines affected, or false if not applicable
  692. */
  693. public function markAsReadUponGone() {
  694. $readUponGone = $this->attributes('read_upon_gone');
  695. if ($readUponGone === null) {
  696. $readUponGone = FreshRSS_Context::$user_conf->mark_when['gone'];
  697. }
  698. if ($readUponGone) {
  699. $feedDAO = FreshRSS_Factory::createFeedDao();
  700. return $feedDAO->markAsReadUponGone($this->id());
  701. }
  702. return false;
  703. }
  704. /**
  705. * Remember to call updateCachedValue($id_feed) or updateCachedValues() just after
  706. */
  707. public function cleanOldEntries() {
  708. $archiving = $this->attributes('archiving');
  709. if ($archiving == null) {
  710. $catDAO = FreshRSS_Factory::createCategoryDao();
  711. $category = $catDAO->searchById($this->categoryId);
  712. $archiving = $category == null ? null : $category->attributes('archiving');
  713. if ($archiving == null) {
  714. $archiving = FreshRSS_Context::$user_conf->archiving;
  715. }
  716. }
  717. if (is_array($archiving)) {
  718. $entryDAO = FreshRSS_Factory::createEntryDao();
  719. $nb = $entryDAO->cleanOldEntries($this->id(), $archiving);
  720. if ($nb > 0) {
  721. $needFeedCacheRefresh = true;
  722. Minz_Log::debug($nb . ' entries cleaned in feed [' . $this->url(false) . '] with: ' . json_encode($archiving));
  723. }
  724. return $nb;
  725. }
  726. return false;
  727. }
  728. public static function cacheFilename(string $url, array $attributes, int $kind = FreshRSS_Feed::KIND_RSS): string {
  729. $simplePie = customSimplePie($attributes);
  730. $filename = $simplePie->get_cache_filename($url);
  731. if ($kind === FreshRSS_Feed::KIND_HTML_XPATH) {
  732. return CACHE_PATH . '/' . $filename . '.html';
  733. } elseif ($kind === FreshRSS_Feed::KIND_XML_XPATH) {
  734. return CACHE_PATH . '/' . $filename . '.xml';
  735. } else {
  736. return CACHE_PATH . '/' . $filename . '.spc';
  737. }
  738. }
  739. public function clearCache(): bool {
  740. return @unlink(FreshRSS_Feed::cacheFilename($this->url, $this->attributes(), $this->kind));
  741. }
  742. /** @return int|false */
  743. public function cacheModifiedTime() {
  744. return @filemtime(FreshRSS_Feed::cacheFilename($this->url, $this->attributes(), $this->kind));
  745. }
  746. public function lock(): bool {
  747. $this->lockPath = TMP_PATH . '/' . $this->hash() . '.freshrss.lock';
  748. if (file_exists($this->lockPath) && ((time() - @filemtime($this->lockPath)) > 3600)) {
  749. @unlink($this->lockPath);
  750. }
  751. if (($handle = @fopen($this->lockPath, 'x')) === false) {
  752. return false;
  753. }
  754. //register_shutdown_function('unlink', $this->lockPath);
  755. @fclose($handle);
  756. return true;
  757. }
  758. public function unlock(): bool {
  759. return @unlink($this->lockPath);
  760. }
  761. /**
  762. * @return array<FreshRSS_FilterAction>
  763. */
  764. public function filterActions(): array {
  765. if (empty($this->filterActions)) {
  766. $this->filterActions = array();
  767. $filters = $this->attributes('filters');
  768. if (is_array($filters)) {
  769. foreach ($filters as $filter) {
  770. $filterAction = FreshRSS_FilterAction::fromJSON($filter);
  771. if ($filterAction != null) {
  772. $this->filterActions[] = $filterAction;
  773. }
  774. }
  775. }
  776. }
  777. return $this->filterActions;
  778. }
  779. /**
  780. * @param array<FreshRSS_FilterAction> $filterActions
  781. */
  782. private function _filterActions($filterActions) {
  783. $this->filterActions = $filterActions;
  784. if (is_array($this->filterActions) && !empty($this->filterActions)) {
  785. $this->_attributes('filters', array_map(function ($af) {
  786. return $af == null ? null : $af->toJSON();
  787. }, $this->filterActions));
  788. } else {
  789. $this->_attributes('filters', null);
  790. }
  791. }
  792. /** @return array<FreshRSS_BooleanSearch> */
  793. public function filtersAction(string $action): array {
  794. $action = trim($action);
  795. if ($action == '') {
  796. return array();
  797. }
  798. $filters = array();
  799. $filterActions = $this->filterActions();
  800. for ($i = count($filterActions) - 1; $i >= 0; $i--) {
  801. $filterAction = $filterActions[$i];
  802. if ($filterAction != null && $filterAction->booleanSearch() != null &&
  803. $filterAction->actions() != null && in_array($action, $filterAction->actions(), true)) {
  804. $filters[] = $filterAction->booleanSearch();
  805. }
  806. }
  807. return $filters;
  808. }
  809. /**
  810. * @param array<string> $filters
  811. */
  812. public function _filtersAction(string $action, $filters) {
  813. $action = trim($action);
  814. if ($action == '' || !is_array($filters)) {
  815. return false;
  816. }
  817. $filters = array_unique(array_map('trim', $filters));
  818. $filterActions = $this->filterActions();
  819. //Check existing filters
  820. for ($i = count($filterActions) - 1; $i >= 0; $i--) {
  821. $filterAction = $filterActions[$i];
  822. if ($filterAction == null || !is_array($filterAction->actions()) ||
  823. $filterAction->booleanSearch() == null || trim($filterAction->booleanSearch()->getRawInput()) == '') {
  824. array_splice($filterActions, $i, 1);
  825. continue;
  826. }
  827. $actions = $filterAction->actions();
  828. //Remove existing rules with same action
  829. for ($j = count($actions) - 1; $j >= 0; $j--) {
  830. if ($actions[$j] === $action) {
  831. array_splice($actions, $j, 1);
  832. }
  833. }
  834. //Update existing filter with new action
  835. for ($k = count($filters) - 1; $k >= 0; $k --) {
  836. $filter = $filters[$k];
  837. if ($filter === $filterAction->booleanSearch()->getRawInput()) {
  838. $actions[] = $action;
  839. array_splice($filters, $k, 1);
  840. }
  841. }
  842. //Save result
  843. if (empty($actions)) {
  844. array_splice($filterActions, $i, 1);
  845. } else {
  846. $filterAction->_actions($actions);
  847. }
  848. }
  849. //Add new filters
  850. for ($k = count($filters) - 1; $k >= 0; $k --) {
  851. $filter = $filters[$k];
  852. if ($filter != '') {
  853. $filterAction = FreshRSS_FilterAction::fromJSON(array(
  854. 'search' => $filter,
  855. 'actions' => array($action),
  856. ));
  857. if ($filterAction != null) {
  858. $filterActions[] = $filterAction;
  859. }
  860. }
  861. }
  862. if (empty($filterActions)) {
  863. $filterActions = null;
  864. }
  865. $this->_filterActions($filterActions);
  866. }
  867. //<WebSub>
  868. public function pubSubHubbubEnabled(): bool {
  869. $url = $this->selfUrl ? $this->selfUrl : $this->url;
  870. $hubFilename = PSHB_PATH . '/feeds/' . sha1($url) . '/!hub.json';
  871. if ($hubFile = @file_get_contents($hubFilename)) {
  872. $hubJson = json_decode($hubFile, true);
  873. if ($hubJson && empty($hubJson['error']) &&
  874. (empty($hubJson['lease_end']) || $hubJson['lease_end'] > time())) {
  875. return true;
  876. }
  877. }
  878. return false;
  879. }
  880. public function pubSubHubbubError(bool $error = true): bool {
  881. $url = $this->selfUrl ? $this->selfUrl : $this->url;
  882. $hubFilename = PSHB_PATH . '/feeds/' . sha1($url) . '/!hub.json';
  883. $hubFile = @file_get_contents($hubFilename);
  884. $hubJson = $hubFile ? json_decode($hubFile, true) : array();
  885. if (!isset($hubJson['error']) || $hubJson['error'] !== (bool)$error) {
  886. $hubJson['error'] = (bool)$error;
  887. file_put_contents($hubFilename, json_encode($hubJson));
  888. Minz_Log::warning('Set error to ' . ($error ? 1 : 0) . ' for ' . $url, PSHB_LOG);
  889. }
  890. return false;
  891. }
  892. /**
  893. * @return string|false
  894. */
  895. public function pubSubHubbubPrepare() {
  896. $key = '';
  897. if (Minz_Request::serverIsPublic(FreshRSS_Context::$system_conf->base_url) &&
  898. $this->hubUrl && $this->selfUrl && @is_dir(PSHB_PATH)) {
  899. $path = PSHB_PATH . '/feeds/' . sha1($this->selfUrl);
  900. $hubFilename = $path . '/!hub.json';
  901. if ($hubFile = @file_get_contents($hubFilename)) {
  902. $hubJson = json_decode($hubFile, true);
  903. if (!$hubJson || empty($hubJson['key']) || !ctype_xdigit($hubJson['key'])) {
  904. $text = 'Invalid JSON for WebSub: ' . $this->url;
  905. Minz_Log::warning($text);
  906. Minz_Log::warning($text, PSHB_LOG);
  907. return false;
  908. }
  909. if ((!empty($hubJson['lease_end'])) && ($hubJson['lease_end'] < (time() + (3600 * 23)))) { //TODO: Make a better policy
  910. $text = 'WebSub lease ends at '
  911. . date('c', empty($hubJson['lease_end']) ? time() : $hubJson['lease_end'])
  912. . ' and needs renewal: ' . $this->url;
  913. Minz_Log::warning($text);
  914. Minz_Log::warning($text, PSHB_LOG);
  915. $key = $hubJson['key']; //To renew our lease
  916. } elseif (((!empty($hubJson['error'])) || empty($hubJson['lease_end'])) &&
  917. (empty($hubJson['lease_start']) || $hubJson['lease_start'] < time() - (3600 * 23))) { //Do not renew too often
  918. $key = $hubJson['key']; //To renew our lease
  919. }
  920. } else {
  921. @mkdir($path, 0770, true);
  922. $key = sha1($path . FreshRSS_Context::$system_conf->salt);
  923. $hubJson = array(
  924. 'hub' => $this->hubUrl,
  925. 'key' => $key,
  926. );
  927. file_put_contents($hubFilename, json_encode($hubJson));
  928. @mkdir(PSHB_PATH . '/keys/', 0770, true);
  929. file_put_contents(PSHB_PATH . '/keys/' . $key . '.txt', $this->selfUrl);
  930. $text = 'WebSub prepared for ' . $this->url;
  931. Minz_Log::debug($text);
  932. Minz_Log::debug($text, PSHB_LOG);
  933. }
  934. $currentUser = Minz_User::name();
  935. if (FreshRSS_user_Controller::checkUsername($currentUser) && !file_exists($path . '/' . $currentUser . '.txt')) {
  936. touch($path . '/' . $currentUser . '.txt');
  937. }
  938. }
  939. return $key;
  940. }
  941. //Parameter true to subscribe, false to unsubscribe.
  942. public function pubSubHubbubSubscribe(bool $state): bool {
  943. if ($state) {
  944. $url = $this->selfUrl ? $this->selfUrl : $this->url;
  945. } else {
  946. $url = $this->url; //Always use current URL during unsubscribe
  947. }
  948. if ($url && (Minz_Request::serverIsPublic(FreshRSS_Context::$system_conf->base_url) || !$state)) {
  949. $hubFilename = PSHB_PATH . '/feeds/' . sha1($url) . '/!hub.json';
  950. $hubFile = @file_get_contents($hubFilename);
  951. if ($hubFile === false) {
  952. Minz_Log::warning('JSON not found for WebSub: ' . $this->url);
  953. return false;
  954. }
  955. $hubJson = json_decode($hubFile, true);
  956. if (!$hubJson || empty($hubJson['key']) || !ctype_xdigit($hubJson['key']) || empty($hubJson['hub'])) {
  957. Minz_Log::warning('Invalid JSON for WebSub: ' . $this->url);
  958. return false;
  959. }
  960. $callbackUrl = checkUrl(Minz_Request::getBaseUrl() . '/api/pshb.php?k=' . $hubJson['key']);
  961. if ($callbackUrl == '') {
  962. Minz_Log::warning('Invalid callback for WebSub: ' . $this->url);
  963. return false;
  964. }
  965. if (!$state) { //unsubscribe
  966. $hubJson['lease_end'] = time() - 60;
  967. file_put_contents($hubFilename, json_encode($hubJson));
  968. }
  969. $ch = curl_init();
  970. curl_setopt_array($ch, [
  971. CURLOPT_URL => $hubJson['hub'],
  972. CURLOPT_RETURNTRANSFER => true,
  973. CURLOPT_POSTFIELDS => http_build_query(array(
  974. 'hub.verify' => 'sync',
  975. 'hub.mode' => $state ? 'subscribe' : 'unsubscribe',
  976. 'hub.topic' => $url,
  977. 'hub.callback' => $callbackUrl,
  978. )),
  979. CURLOPT_USERAGENT => FRESHRSS_USERAGENT,
  980. CURLOPT_MAXREDIRS => 10,
  981. CURLOPT_FOLLOWLOCATION => true,
  982. CURLOPT_ENCODING => '', //Enable all encodings
  983. ]);
  984. $response = curl_exec($ch);
  985. $info = curl_getinfo($ch);
  986. Minz_Log::warning('WebSub ' . ($state ? 'subscribe' : 'unsubscribe') . ' to ' . $url .
  987. ' via hub ' . $hubJson['hub'] .
  988. ' with callback ' . $callbackUrl . ': ' . $info['http_code'] . ' ' . $response, PSHB_LOG);
  989. if (substr('' . $info['http_code'], 0, 1) == '2') {
  990. return true;
  991. } else {
  992. $hubJson['lease_start'] = time(); //Prevent trying again too soon
  993. $hubJson['error'] = true;
  994. file_put_contents($hubFilename, json_encode($hubJson));
  995. return false;
  996. }
  997. }
  998. return false;
  999. }
  1000. //</WebSub>
  1001. }