Feed.php 47 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439
  1. <?php
  2. declare(strict_types=1);
  3. class FreshRSS_Feed extends Minz_Model {
  4. use FreshRSS_AttributesTrait, FreshRSS_FilterActionsTrait;
  5. /**
  6. * Normal RSS or Atom feed
  7. * @var int
  8. */
  9. public const KIND_RSS = 0;
  10. /**
  11. * Invalid RSS or Atom feed
  12. * @var int
  13. */
  14. public const KIND_RSS_FORCED = 2;
  15. /**
  16. * Normal HTML with XPath scraping
  17. * @var int
  18. */
  19. public const KIND_HTML_XPATH = 10;
  20. /**
  21. * Normal XML with XPath scraping
  22. * @var int
  23. */
  24. public const KIND_XML_XPATH = 15;
  25. /**
  26. * Normal JSON with XPath scraping
  27. * @var int
  28. */
  29. public const KIND_JSON_XPATH = 20;
  30. public const KIND_JSONFEED = 25;
  31. public const KIND_JSON_DOTNOTATION = 30;
  32. /** JSON embedded in HTML */
  33. public const KIND_HTML_XPATH_JSON_DOTNOTATION = 35;
  34. public const PRIORITY_IMPORTANT = 20;
  35. public const PRIORITY_MAIN_STREAM = 10;
  36. public const PRIORITY_CATEGORY = 0;
  37. public const PRIORITY_HIDDEN = -10;
  38. /** @deprecated use PRIORITY_HIDDEN instead */
  39. public const PRIORITY_ARCHIVED = -10;
  40. public const TTL_DEFAULT = 0;
  41. public const ARCHIVING_RETENTION_COUNT_LIMIT = 10000;
  42. public const ARCHIVING_RETENTION_PERIOD = 'P3M';
  43. private int $id = 0;
  44. private string $url = '';
  45. private int $kind = 0;
  46. private int $categoryId = 0;
  47. private ?FreshRSS_Category $category = null;
  48. private int $nbEntries = -1;
  49. private int $nbNotRead = -1;
  50. private string $name = '';
  51. private string $website = '';
  52. private string $description = '';
  53. private int $lastUpdate = 0;
  54. private int $priority = self::PRIORITY_MAIN_STREAM;
  55. private string $pathEntries = '';
  56. private string $httpAuth = '';
  57. private bool $error = false;
  58. private int $ttl = self::TTL_DEFAULT;
  59. private bool $mute = false;
  60. private string $hash = '';
  61. private string $hashFavicon = '';
  62. private string $lockPath = '';
  63. private string $hubUrl = '';
  64. private string $selfUrl = '';
  65. /**
  66. * @throws FreshRSS_BadUrl_Exception
  67. */
  68. public function __construct(string $url, bool $validate = true) {
  69. if ($validate) {
  70. $this->_url($url);
  71. } else {
  72. $this->url = $url;
  73. }
  74. }
  75. public static function default(): FreshRSS_Feed {
  76. $f = new FreshRSS_Feed('http://example.net/', false);
  77. $f->faviconPrepare();
  78. return $f;
  79. }
  80. public function id(): int {
  81. return $this->id;
  82. }
  83. public function hash(): string {
  84. if ($this->hash == '') {
  85. $salt = FreshRSS_Context::systemConf()->salt;
  86. $params = $this->url . $this->proxyParam();
  87. $this->hash = sha1($salt . $params);
  88. }
  89. return $this->hash;
  90. }
  91. public function resetFaviconHash(): void {
  92. $this->hashFavicon(skipCache: true);
  93. }
  94. public function proxyParam(): string {
  95. $curl_params = $this->attributeArray('curl_params');
  96. if (is_array($curl_params)) {
  97. // Content provided through a proxy may be completely different
  98. return is_string($curl_params[CURLOPT_PROXY] ?? null) ? $curl_params[CURLOPT_PROXY] : '';
  99. }
  100. return '';
  101. }
  102. /**
  103. * Resets the custom favicon to the default one. Also deletes the favicon when allowed by extension.
  104. *
  105. * @param array{'url'?:string,'kind'?:int,'category'?:int,'name'?:string,'website'?:string,'description'?:string,'lastUpdate'?:int,'priority'?:int,
  106. * 'pathEntries'?:string,'httpAuth'?:string,'error'?:int,'ttl'?:int,'attributes'?:string|array<string,mixed>} &$values &$values
  107. *
  108. * @param bool $updateFeed Whether `updateFeed()` should be called immediately. If false, it must be handled manually.
  109. *
  110. * @return void
  111. *
  112. * @throws FreshRSS_Feed_Exception
  113. */
  114. public function resetCustomFavicon(?array &$values = null, bool $updateFeed = true) {
  115. if (!$this->customFavicon()) {
  116. return;
  117. }
  118. if (!$this->attributeBoolean('customFaviconDisallowDel')) {
  119. FreshRSS_Feed::faviconDelete($this->hashFavicon());
  120. }
  121. $this->_attribute('customFavicon', false);
  122. $this->_attribute('customFaviconExt', null);
  123. $this->_attribute('customFaviconDisallowDel', false);
  124. if ($values !== null) {
  125. $values['attributes'] = $this->attributes();
  126. $feedDAO = FreshRSS_Factory::createFeedDao();
  127. if ($updateFeed && !$feedDAO->updateFeed($this->id(), $values)) {
  128. throw new FreshRSS_Feed_Exception();
  129. }
  130. }
  131. $this->resetFaviconHash();
  132. }
  133. /**
  134. * Set a custom favicon for the feed.
  135. *
  136. * @param string $contents Contents of the favicon file. Optional if $tmpPath is set.
  137. * @param string $tmpPath Use only when handling file uploads. (value from `tmp_name` goes here)
  138. *
  139. * @param array{'url'?:string,'kind'?:int,'category'?:int,'name'?:string,'website'?:string,'description'?:string,'lastUpdate'?:int,'priority'?:int,
  140. * 'pathEntries'?:string,'httpAuth'?:string,'error'?:int,'ttl'?:int,'attributes'?:string|array<string,mixed>} &$values &$values
  141. *
  142. * @param bool $updateFeed Whether `updateFeed()` should be called immediately. If false, it must be handled manually.
  143. * @param string $extName The extension name of the calling extension.
  144. * @param bool $disallowDelete Whether the icon can be later deleted when it's being reset. Intended for use by extensions.
  145. * @param bool $overrideCustomIcon Whether a custom favicon set by a user can be overridden.
  146. *
  147. * @return string|null Path where the favicon can be found. Useful for checking if the favicon already exists, before downloading it for example.
  148. *
  149. * @throws FreshRSS_UnsupportedImageFormat_Exception
  150. * @throws FreshRSS_Feed_Exception
  151. */
  152. public function setCustomFavicon(
  153. ?string $contents = null,
  154. string $tmpPath = '',
  155. ?array &$values = null,
  156. bool $updateFeed = true,
  157. ?string $extName = null,
  158. bool $disallowDelete = false,
  159. bool $overrideCustomIcon = false
  160. ): ?string {
  161. if ($contents === null && $tmpPath !== '') {
  162. $contents = file_get_contents($tmpPath);
  163. }
  164. $attributesOnly = $contents === null && $tmpPath === '';
  165. require_once LIB_PATH . '/favicons.php';
  166. if (!$attributesOnly && !isImgMime(is_string($contents) ? $contents : '')) {
  167. throw new FreshRSS_UnsupportedImageFormat_Exception();
  168. }
  169. $oldHash = '';
  170. $oldDisallowDelete = false;
  171. if ($this->customFavicon()) {
  172. /* If $overrideCustomFavicon is true, custom favicons set by extensions can be overridden,
  173. * but not ones explicitly set by the user */
  174. if (!$overrideCustomIcon && $this->customFaviconExt() === null) {
  175. return null;
  176. }
  177. $oldHash = $this->hashFavicon(skipCache: true);
  178. $oldDisallowDelete = $this->attributeBoolean('customFaviconDisallowDel');
  179. }
  180. $this->_attribute('customFavicon', true);
  181. $this->_attribute('customFaviconExt', $extName);
  182. $this->_attribute('customFaviconDisallowDel', $disallowDelete);
  183. $newPath = FAVICONS_DIR . $this->hashFavicon(skipCache: true) . '.ico';
  184. if ($attributesOnly && !file_exists($newPath)) {
  185. $updateFeed = false;
  186. }
  187. if ($values !== null) {
  188. $values['attributes'] = $this->attributes();
  189. $feedDAO = FreshRSS_Factory::createFeedDao();
  190. if ($updateFeed && !$feedDAO->updateFeed($this->id(), $values)) {
  191. throw new FreshRSS_Feed_Exception();
  192. }
  193. }
  194. if ($tmpPath !== '') {
  195. move_uploaded_file($tmpPath, $newPath);
  196. } elseif ($contents !== null) {
  197. file_put_contents($newPath, $contents);
  198. }
  199. if ($oldHash !== '' && !$oldDisallowDelete) {
  200. FreshRSS_Feed::faviconDelete($oldHash);
  201. }
  202. return $newPath;
  203. }
  204. /**
  205. * Checks if the feed has a custom favicon set by an extension.
  206. * Additionally, it also checks if the extension that set the icon is still enabled
  207. * And if not, it resets attributes related to custom favicons.
  208. *
  209. * @return string|null The name of the extension that set the icon.
  210. */
  211. public function customFaviconExt(): ?string {
  212. $customFaviconExt = $this->attributeString('customFaviconExt');
  213. if ($customFaviconExt !== null && !Minz_ExtensionManager::isExtensionEnabled($customFaviconExt)) {
  214. $this->_attribute('customFavicon', false);
  215. $this->_attribute('customFaviconExt', null);
  216. $this->_attribute('customFaviconDisallowDel', false);
  217. $customFaviconExt = null;
  218. }
  219. return $customFaviconExt;
  220. }
  221. public function customFavicon(): bool {
  222. $this->customFaviconExt();
  223. return $this->attributeBoolean('customFavicon') ?? false;
  224. }
  225. public function hashFavicon(bool $skipCache = false): string {
  226. if ($this->hashFavicon == '' || $skipCache) {
  227. $salt = FreshRSS_Context::systemConf()->salt;
  228. $params = '';
  229. if ($this->customFavicon()) {
  230. $current = $this->id . Minz_User::name();
  231. $hookParams = Minz_ExtensionManager::callHook('custom_favicon_hash', $this);
  232. $params = $hookParams !== null ? $hookParams : $current;
  233. } else {
  234. $params = $this->website(fallback: true) . $this->proxyParam();
  235. }
  236. $this->hashFavicon = hash('crc32b', $salt . (is_string($params) ? $params : ''));
  237. }
  238. return $this->hashFavicon;
  239. }
  240. public function url(bool $includeCredentials = true): string {
  241. return $includeCredentials ? $this->url : \SimplePie\Misc::url_remove_credentials($this->url);
  242. }
  243. public function selfUrl(): string {
  244. return $this->selfUrl;
  245. }
  246. public function kind(): int {
  247. return $this->kind;
  248. }
  249. public function hubUrl(): string {
  250. return $this->hubUrl;
  251. }
  252. public function category(): ?FreshRSS_Category {
  253. if ($this->category === null && $this->categoryId > 0) {
  254. $catDAO = FreshRSS_Factory::createCategoryDao();
  255. $this->category = $catDAO->searchById($this->categoryId);
  256. }
  257. return $this->category;
  258. }
  259. public function categoryId(): int {
  260. return $this->category?->id() ?: $this->categoryId;
  261. }
  262. /**
  263. * @return list<FreshRSS_Entry>|null
  264. * @deprecated
  265. */
  266. public function entries(): ?array {
  267. Minz_Log::warning(__METHOD__ . ' is deprecated since FreshRSS 1.16.1!');
  268. $simplePie = $this->load(false, true);
  269. return $simplePie == null ? [] : array_values(iterator_to_array($this->loadEntries($simplePie)));
  270. }
  271. public function name(bool $raw = false): string {
  272. return $raw || $this->name != '' ? $this->name : (preg_replace('%^https?://(www[.])?%i', '', $this->url) ?? '');
  273. }
  274. /**
  275. * @param bool $fallback true to return the URL of the feed if the Web site is blank
  276. * @return string HTML-encoded URL of the Web site of the feed
  277. */
  278. public function website(bool $fallback = false): string {
  279. $url = $this->website;
  280. if ($fallback && !preg_match('%^https?://.%i', $url)) {
  281. $url = $this->url;
  282. }
  283. return $url;
  284. }
  285. public function description(): string {
  286. return $this->description;
  287. }
  288. public function lastUpdate(): int {
  289. return $this->lastUpdate;
  290. }
  291. public function priority(): int {
  292. return $this->priority;
  293. }
  294. /** @return string HTML-encoded CSS selector */
  295. public function pathEntries(): string {
  296. return $this->pathEntries;
  297. }
  298. /**
  299. * @phpstan-return ($raw is true ? string : array{'username':string,'password':string})
  300. * @return array{'username':string,'password':string}|string
  301. */
  302. public function httpAuth(bool $raw = true): array|string {
  303. if ($raw) {
  304. return $this->httpAuth;
  305. } else {
  306. $pos_colon = strpos($this->httpAuth, ':');
  307. if ($pos_colon !== false) {
  308. $user = substr($this->httpAuth, 0, $pos_colon);
  309. $pass = substr($this->httpAuth, $pos_colon + 1);
  310. } else {
  311. $user = '';
  312. $pass = '';
  313. }
  314. return [
  315. 'username' => $user,
  316. 'password' => $pass,
  317. ];
  318. }
  319. }
  320. /** @return array<int,mixed> */
  321. public function curlOptions(): array {
  322. $curl_options = [];
  323. if ($this->httpAuth !== '') {
  324. $curl_options[CURLOPT_USERPWD] = htmlspecialchars_decode($this->httpAuth, ENT_QUOTES);
  325. }
  326. return $curl_options;
  327. }
  328. public function inError(): bool {
  329. return $this->error;
  330. }
  331. /**
  332. * @param bool $raw true for database version combined with mute information, false otherwise
  333. */
  334. public function ttl(bool $raw = false): int {
  335. if ($raw) {
  336. $ttl = $this->ttl;
  337. if ($this->mute && FreshRSS_Feed::TTL_DEFAULT === $ttl) {
  338. $ttl = FreshRSS_Context::userConf()->ttl_default;
  339. }
  340. return $ttl * ($this->mute ? -1 : 1);
  341. }
  342. if ($this->mute && $this->ttl === FreshRSS_Context::userConf()->ttl_default) {
  343. return FreshRSS_Feed::TTL_DEFAULT;
  344. }
  345. return $this->ttl;
  346. }
  347. public function mute(): bool {
  348. return $this->mute;
  349. }
  350. public function nbEntries(): int {
  351. if ($this->nbEntries < 0) {
  352. $feedDAO = FreshRSS_Factory::createFeedDao();
  353. $this->nbEntries = $feedDAO->countEntries($this->id());
  354. }
  355. return $this->nbEntries;
  356. }
  357. public function nbNotRead(): int {
  358. if ($this->nbNotRead < 0) {
  359. $feedDAO = FreshRSS_Factory::createFeedDao();
  360. $this->nbNotRead = $feedDAO->countNotRead($this->id());
  361. }
  362. return $this->nbNotRead;
  363. }
  364. public function faviconPrepare(bool $force = false): void {
  365. require_once LIB_PATH . '/favicons.php';
  366. if ($this->customFavicon()) {
  367. return;
  368. }
  369. $url = $this->website(fallback: false);
  370. if ($url === '' || $url === $this->url) {
  371. // Get root URL from the feed URL
  372. $url = preg_replace('%^(https?://[^/]+).*$%i', '$1/', $this->url) ?? $this->url;
  373. }
  374. $txt = FAVICONS_DIR . $this->hashFavicon() . '.txt';
  375. if (@file_get_contents($txt) !== $url) {
  376. file_put_contents($txt, $url);
  377. }
  378. if (FreshRSS_Context::$isCli || $force) {
  379. $ico = FAVICONS_DIR . $this->hashFavicon() . '.ico';
  380. $ico_mtime = @filemtime($ico);
  381. $txt_mtime = @filemtime($txt);
  382. if ($txt_mtime != false &&
  383. ($ico_mtime == false || $ico_mtime < $txt_mtime || ($ico_mtime < time() - (14 * 86400)))) {
  384. // no ico file or we should download a new one.
  385. $url = file_get_contents($txt);
  386. if ($url == false || !download_favicon($url, $ico)) {
  387. touch($ico);
  388. }
  389. }
  390. }
  391. }
  392. public static function faviconDelete(string $hash): void {
  393. if (!ctype_xdigit($hash)) {
  394. return;
  395. }
  396. $path = DATA_PATH . '/favicons/' . $hash;
  397. @unlink($path . '.ico');
  398. @unlink($path . '.txt');
  399. }
  400. public function favicon(bool $absolute = false): string {
  401. $hash = $this->hashFavicon();
  402. $url = '/f.php?h=' . $hash;
  403. if ($this->customFavicon()
  404. // when the below attribute is set, icon won't be changing frequently so cache buster is not needed
  405. && !$this->attributeBoolean('customFaviconDisallowDel')) {
  406. $url .= '&t=' . @filemtime(DATA_PATH . '/favicons/' . $hash . '.ico');
  407. }
  408. return Minz_Url::display($url, absolute: $absolute);
  409. }
  410. public function _id(int $value): void {
  411. $this->id = $value;
  412. }
  413. /**
  414. * @throws FreshRSS_BadUrl_Exception
  415. */
  416. public function _url(string $value, bool $validate = true): void {
  417. $this->hash = '';
  418. $this->hashFavicon = '';
  419. $url = $value;
  420. if ($validate) {
  421. $url = checkUrl($url);
  422. }
  423. if ($url == false) {
  424. throw new FreshRSS_BadUrl_Exception($value);
  425. }
  426. $this->url = $url;
  427. }
  428. public function _selfUrl(string $value): void {
  429. $this->selfUrl = $value;
  430. }
  431. public function _kind(int $value): void {
  432. $this->kind = $value;
  433. }
  434. public function _category(?FreshRSS_Category $cat): void {
  435. $this->category = $cat;
  436. $this->categoryId = $this->category == null ? 0 : $this->category->id();
  437. }
  438. /** @param int|numeric-string $id */
  439. public function _categoryId(int|string $id): void {
  440. $this->category = null;
  441. $this->categoryId = (int)$id;
  442. }
  443. public function _name(string $value): void {
  444. $this->name = $value == '' ? '' : trim($value);
  445. }
  446. public function _website(string $value, bool $validate = true): void {
  447. $this->hashFavicon = '';
  448. if ($validate) {
  449. $value = checkUrl($value);
  450. }
  451. if ($value == false) {
  452. $value = '';
  453. }
  454. $this->website = $value;
  455. }
  456. public function _description(string $value): void {
  457. $this->description = $value == '' ? '' : $value;
  458. }
  459. /**
  460. * @param int|numeric-string $value
  461. * 32-bit systems provide a string and will fail in year 2038
  462. */
  463. public function _lastUpdate(int|string $value): void {
  464. $this->lastUpdate = (int)$value;
  465. }
  466. public function _priority(int $value): void {
  467. $this->priority = $value;
  468. }
  469. /** @param string $value HTML-encoded CSS selector */
  470. public function _pathEntries(string $value): void {
  471. $this->pathEntries = $value;
  472. }
  473. public function _httpAuth(string $value): void {
  474. $this->httpAuth = $value;
  475. }
  476. public function _error(bool|int $value): void {
  477. $this->error = (bool)$value;
  478. }
  479. public function _mute(bool $value): void {
  480. $this->mute = $value;
  481. }
  482. public function _ttl(int $value): void {
  483. $value = min($value, 100_000_000);
  484. $this->ttl = abs($value);
  485. $this->mute = $value < self::TTL_DEFAULT;
  486. }
  487. public function _nbNotRead(int $value): void {
  488. $this->nbNotRead = $value;
  489. }
  490. public function _nbEntries(int $value): void {
  491. $this->nbEntries = $value;
  492. }
  493. /**
  494. * @throws Minz_FileNotExistException
  495. * @throws FreshRSS_Feed_Exception
  496. */
  497. public function load(bool $loadDetails = false, bool $noCache = false): ?\SimplePie\SimplePie {
  498. if ($this->url != '') {
  499. /**
  500. * @throws Minz_FileNotExistException
  501. */
  502. if (trim(CACHE_PATH) === '') {
  503. throw new Minz_FileNotExistException(
  504. 'CACHE_PATH',
  505. Minz_Exception::ERROR
  506. );
  507. } else {
  508. if (($retryAfter = FreshRSS_http_Util::getRetryAfter($this->url)) > 0) {
  509. throw new FreshRSS_Feed_Exception('For that domain, will first retry after ' . date('c', $retryAfter) .
  510. '. ' . $this->url(includeCredentials: false), code: 503);
  511. }
  512. $simplePie = customSimplePie($this->attributes(), $this->curlOptions());
  513. $url = htmlspecialchars_decode($this->url, ENT_QUOTES);
  514. if (str_ends_with($url, '#force_feed')) {
  515. $simplePie->force_feed(true);
  516. $url = substr($url, 0, -11);
  517. }
  518. $simplePie->set_feed_url($url);
  519. if (!$loadDetails) { //Only activates auto-discovery when adding a new feed
  520. $simplePie->set_autodiscovery_level(\SimplePie\SimplePie::LOCATOR_NONE);
  521. }
  522. if ($this->attributeBoolean('clear_cache')) {
  523. // Do not use `$simplePie->enable_cache(false);` as it would prevent caching in multiuser context
  524. $this->clearCache();
  525. }
  526. Minz_ExtensionManager::callHook('simplepie_before_init', $simplePie, $this);
  527. $simplePieResult = $simplePie->init();
  528. Minz_ExtensionManager::callHook('simplepie_after_init', $simplePie, $this, $simplePieResult);
  529. if ($simplePieResult === false || $simplePie->get_hash() === '' || !empty($simplePie->error())) {
  530. if ($simplePie->status_code() === 429) {
  531. $errorMessage = 'HTTP 429 Too Many Requests!';
  532. } elseif ($simplePie->status_code() === 503) {
  533. $errorMessage = 'HTTP 503 Service Unavailable!';
  534. } else {
  535. $errorMessage = $simplePie->error();
  536. if (empty($errorMessage)) {
  537. $errorMessage = '';
  538. } elseif (is_array($errorMessage)) {
  539. $errorMessage = json_encode($errorMessage, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_LINE_TERMINATORS) ?: '';
  540. }
  541. }
  542. throw new FreshRSS_Feed_Exception(
  543. ($errorMessage == '' ? 'Unknown error for feed' : $errorMessage) .
  544. ' [' . $this->url(includeCredentials: false) . ']',
  545. $simplePie->status_code()
  546. );
  547. }
  548. $links = $simplePie->get_links('self');
  549. $this->selfUrl = empty($links[0]) ? '' : (checkUrl($links[0]) ?: '');
  550. $links = $simplePie->get_links('hub');
  551. $this->hubUrl = empty($links[0]) ? '' : (checkUrl($links[0]) ?: '');
  552. if ($loadDetails) {
  553. // si on a utilisé l’auto-discover, notre url va avoir changé
  554. $subscribe_url = $simplePie->subscribe_url(false) ?? '';
  555. if ($this->name(true) === '') {
  556. //HTML to HTML-PRE //ENT_COMPAT except '&'
  557. $title = strtr(html_only_entity_decode($simplePie->get_title()), ['<' => '&lt;', '>' => '&gt;', '"' => '&quot;']);
  558. $this->_name($title == '' ? $this->url : $title);
  559. }
  560. if ($this->website() === '') {
  561. $this->_website(html_only_entity_decode($simplePie->get_link()));
  562. }
  563. if ($this->description() === '') {
  564. $this->_description(html_only_entity_decode($simplePie->get_description()));
  565. }
  566. } else {
  567. //The case of HTTP 301 Moved Permanently
  568. $subscribe_url = $simplePie->subscribe_url(true) ?? '';
  569. }
  570. $clean_url = \SimplePie\Misc::url_remove_credentials($subscribe_url);
  571. if ($subscribe_url !== '' && $subscribe_url !== $url) {
  572. $this->_url($clean_url);
  573. }
  574. if ($noCache || $simplePie->get_hash() !== $this->attributeString('SimplePieHash')) {
  575. // syslog(LOG_DEBUG, 'FreshRSS no cache ' . $simplePie->get_hash() . ' !== ' . $this->attributeString('SimplePieHash') . ' for ' . $clean_url);
  576. $this->_attribute('SimplePieHash', $simplePie->get_hash());
  577. return $simplePie;
  578. }
  579. syslog(LOG_DEBUG, 'FreshRSS SimplePie uses cache for ' . $clean_url);
  580. }
  581. }
  582. return null;
  583. }
  584. /**
  585. * Decide the GUID of an entry based on the feed’s policy.
  586. * @param \SimplePie\Item $item The item to decide the GUID for.
  587. * @param bool $fallback Whether to automatically switch to the next policy in case of blank GUID.
  588. * @return string The decided GUID for the entry.
  589. */
  590. protected function decideEntryGuid(\SimplePie\Item $item, bool $fallback = false): string {
  591. $unicityCriteria = $this->attributeString('unicityCriteria');
  592. if ($this->attributeBoolean('hasBadGuids')) { // Legacy
  593. $unicityCriteria = 'link';
  594. }
  595. $entryId = safe_ascii($item->get_id(false, false));
  596. $guid = match ($unicityCriteria) {
  597. null => $entryId,
  598. 'link' => $item->get_permalink() ?? '',
  599. 'sha1:link_published' => sha1($item->get_permalink() . $item->get_date('U')),
  600. 'sha1:link_published_title' => sha1($item->get_permalink() . $item->get_date('U') . $item->get_title()),
  601. 'sha1:link_published_title_content' => sha1($item->get_permalink() . $item->get_date('U') . $item->get_title() . $item->get_content()),
  602. 'sha1:title' => sha1($item->get_title() ?? ''),
  603. 'sha1:title_published' => sha1($item->get_title() . $item->get_date('U')),
  604. 'sha1:title_published_content' => sha1($item->get_title() . $item->get_date('U') . $item->get_content()),
  605. 'sha1:content' => sha1($item->get_content() ?? ''),
  606. 'sha1:content_published' => sha1($item->get_content() . $item->get_date('U')),
  607. 'sha1:published' => sha1((string)($item->get_date('U') ?? '')),
  608. default => $entryId,
  609. };
  610. $blankHash = 'da39a3ee5e6b4b0d3255bfef95601890afd80709'; // sha1('')
  611. if ($guid === $blankHash) {
  612. $guid = '';
  613. }
  614. if ($fallback && $guid === '') {
  615. if ($entryId !== '') {
  616. $guid = $entryId;
  617. } elseif (($item->get_permalink() ?? '') !== '') {
  618. $guid = sha1($item->get_permalink() . $item->get_date('U'));
  619. } elseif (($item->get_title() ?? '') !== '') {
  620. $guid = sha1($item->get_permalink() . $item->get_date('U') . $item->get_title());
  621. } else {
  622. $guid = sha1($item->get_permalink() . $item->get_date('U') . $item->get_title() . $item->get_content());
  623. }
  624. if ($guid === $blankHash) {
  625. $guid = '';
  626. }
  627. }
  628. return $guid;
  629. }
  630. /**
  631. * @param float $invalidGuidsTolerance (default 0.05) The maximum ratio (rounded) of invalid GUIDs to tolerate before degrading the unicity criteria.
  632. * Example for 0.05 (5% rounded): tolerate 0 invalid GUIDs for up to 9 articles, 1 for 10, 2 for 30, 3 for 50, 4 for 70, 5 for 90, 6 for 110, etc.
  633. * The default value of 5% rounded was chosen to allow 1 invalid GUID for feeds of 10 articles, which is a frequently observed amount of articles.
  634. * @return list<string>
  635. */
  636. public function loadGuids(\SimplePie\SimplePie $simplePie, float $invalidGuidsTolerance = 0.05): array {
  637. $invalidGuids = 0;
  638. $testGuids = [];
  639. $guids = [];
  640. $items = $simplePie->get_items();
  641. if (empty($items)) {
  642. return $guids;
  643. }
  644. for ($i = count($items) - 1; $i >= 0; $i--) {
  645. $item = $items[$i];
  646. if ($item == null) {
  647. continue;
  648. }
  649. $guid = $this->decideEntryGuid($item, fallback: true);
  650. if ($guid === '' || !empty($testGuids['_' . $guid])) {
  651. $invalidGuids++;
  652. Minz_Log::debug('Invalid GUID [' . $guid . '] for feed ' . $this->url);
  653. }
  654. $testGuids['_' . $guid] = true;
  655. $guids[] = $guid;
  656. }
  657. if ($invalidGuids > 0) {
  658. Minz_Log::warning("Feed has {$invalidGuids} invalid GUIDs: " . $this->url(includeCredentials: false));
  659. if (!$this->attributeBoolean('unicityCriteriaForced') && $invalidGuids > round($invalidGuidsTolerance * count($items))) {
  660. $unicityCriteria = $this->attributeString('unicityCriteria');
  661. if ($this->attributeBoolean('hasBadGuids')) { // Legacy
  662. $unicityCriteria = 'link';
  663. }
  664. // Automatic fallback to next (degraded) unicity criteria
  665. $newUnicityCriteria = match ($unicityCriteria) {
  666. null => 'sha1:link_published',
  667. 'link' => 'sha1:link_published',
  668. 'sha1:link_published' => 'sha1:link_published_title',
  669. default => $unicityCriteria,
  670. };
  671. if ($newUnicityCriteria !== $unicityCriteria) {
  672. $this->_attribute('hasBadGuids', null); // Remove legacy
  673. $this->_attribute('unicityCriteria', $newUnicityCriteria);
  674. Minz_Log::warning('Feed unicity policy degraded (' . ($unicityCriteria ?: 'id') . ' → ' . $newUnicityCriteria . '): ' .
  675. $this->url(includeCredentials: false));
  676. return $this->loadGuids($simplePie, $invalidGuidsTolerance);
  677. }
  678. }
  679. $this->_error(true);
  680. }
  681. return $guids;
  682. }
  683. /** @return Traversable<FreshRSS_Entry> */
  684. public function loadEntries(\SimplePie\SimplePie $simplePie): Traversable {
  685. $items = $simplePie->get_items();
  686. if (empty($items)) {
  687. return;
  688. }
  689. // We want chronological order and SimplePie uses reverse order.
  690. for ($i = count($items) - 1; $i >= 0; $i--) {
  691. $item = $items[$i];
  692. if ($item == null) {
  693. continue;
  694. }
  695. $title = html_only_entity_decode(strip_tags($item->get_title() ?? ''));
  696. $authors = $item->get_authors();
  697. $link = $item->get_permalink();
  698. $date = $item->get_date('U');
  699. if (!is_numeric($date)) {
  700. $date = 0;
  701. }
  702. //Tag processing (tag == category)
  703. $categories = $item->get_categories();
  704. $tags = [];
  705. if (is_array($categories)) {
  706. foreach ($categories as $category) {
  707. $text = html_only_entity_decode($category->get_label());
  708. //Some feeds use a single category with comma-separated tags
  709. $labels = explode(',', $text);
  710. if (!empty($labels)) {
  711. foreach ($labels as $label) {
  712. $tags[] = trim($label);
  713. }
  714. }
  715. }
  716. $tags = array_unique($tags);
  717. }
  718. $content = html_only_entity_decode($item->get_content());
  719. $attributeThumbnail = $item->get_thumbnail() ?? [];
  720. if (empty($attributeThumbnail['url'])) {
  721. $attributeThumbnail['url'] = '';
  722. }
  723. $attributeEnclosures = [];
  724. if (!empty($item->get_enclosures())) {
  725. foreach ($item->get_enclosures() as $enclosure) {
  726. $elink = $enclosure->get_link();
  727. if ($elink != '') {
  728. $etitle = $enclosure->get_title() ?? '';
  729. $credits = $enclosure->get_credits() ?? null;
  730. $description = $enclosure->get_description() ?? '';
  731. $mime = strtolower($enclosure->get_type() ?? '');
  732. $medium = strtolower($enclosure->get_medium() ?? '');
  733. $height = $enclosure->get_height();
  734. $width = $enclosure->get_width();
  735. $length = $enclosure->get_length();
  736. $attributeEnclosure = [
  737. 'url' => $elink,
  738. ];
  739. if ($etitle != '') {
  740. $attributeEnclosure['title'] = $etitle;
  741. }
  742. if (is_array($credits)) {
  743. $attributeEnclosure['credit'] = [];
  744. foreach ($credits as $credit) {
  745. $attributeEnclosure['credit'][] = $credit->get_name();
  746. }
  747. }
  748. if ($description != '') {
  749. $attributeEnclosure['description'] = $description;
  750. }
  751. if ($mime != '') {
  752. $attributeEnclosure['type'] = $mime;
  753. }
  754. if ($medium != '') {
  755. $attributeEnclosure['medium'] = $medium;
  756. }
  757. if ($length != '') {
  758. $attributeEnclosure['length'] = (int)$length;
  759. }
  760. if ($height != '') {
  761. $attributeEnclosure['height'] = (int)$height;
  762. }
  763. if ($width != '') {
  764. $attributeEnclosure['width'] = (int)$width;
  765. }
  766. if (!empty($enclosure->get_thumbnails())) {
  767. foreach ($enclosure->get_thumbnails() as $thumbnail) {
  768. if ($thumbnail !== $attributeThumbnail['url']) {
  769. $attributeEnclosure['thumbnails'][] = $thumbnail;
  770. }
  771. }
  772. }
  773. $attributeEnclosures[] = $attributeEnclosure;
  774. }
  775. }
  776. }
  777. $guid = $this->decideEntryGuid($item, fallback: true);
  778. unset($item);
  779. $authorNames = '';
  780. if (is_array($authors)) {
  781. foreach ($authors as $author) {
  782. $authorName = $author->name != '' ? $author->name : $author->email;
  783. if (is_string($authorName) && $authorName !== '') {
  784. $authorNames .= html_only_entity_decode(strip_tags($authorName)) . '; ';
  785. }
  786. }
  787. }
  788. $authorNames = substr($authorNames, 0, -2) ?: '';
  789. $entry = new FreshRSS_Entry(
  790. $this->id(),
  791. $guid,
  792. $title == '' ? '' : $title,
  793. $authorNames,
  794. $content == '' ? '' : $content,
  795. $link == null ? '' : $link,
  796. $date ?: time()
  797. );
  798. $entry->_tags($tags);
  799. $entry->_feed($this);
  800. if (!empty($attributeThumbnail['url'])) {
  801. $entry->_attribute('thumbnail', $attributeThumbnail);
  802. }
  803. $entry->_attribute('enclosures', $attributeEnclosures);
  804. $entry->hash(); //Must be computed before loading full content
  805. $entry->loadCompleteContent(); // Optionally load full content for truncated feeds
  806. yield $entry;
  807. }
  808. }
  809. /**
  810. * Given a feed content generated from a FreshRSS_View
  811. * returns a SimplePie initialized already with that content
  812. * @param string $feedContent the content of the feed, typically generated via FreshRSS_View::renderToString()
  813. */
  814. private function simplePieFromContent(string $feedContent): \SimplePie\SimplePie {
  815. $simplePie = customSimplePie();
  816. $simplePie->enable_cache(false);
  817. $simplePie->set_raw_data($feedContent);
  818. $simplePie->init();
  819. return $simplePie;
  820. }
  821. /** @return array<string,string> */
  822. private function dotNotationForStandardJsonFeed(): array {
  823. return [
  824. 'feedTitle' => 'title',
  825. 'item' => 'items',
  826. 'itemTitle' => 'title',
  827. 'itemContent' => 'content_text',
  828. 'itemContentHTML' => 'content_html',
  829. 'itemUri' => 'url',
  830. 'itemTimestamp' => 'date_published',
  831. 'itemTimeFormat' => DateTimeInterface::RFC3339_EXTENDED,
  832. 'itemThumbnail' => 'image',
  833. 'itemCategories' => 'tags',
  834. 'itemUid' => 'id',
  835. 'itemAttachment' => 'attachments',
  836. 'itemAttachmentUrl' => 'url',
  837. 'itemAttachmentType' => 'mime_type',
  838. 'itemAttachmentLength' => 'size_in_bytes',
  839. ];
  840. }
  841. private function extractJsonFromHtml(string $html): ?string {
  842. $xPathToJson = $this->attributeString('xPathToJson') ?? '';
  843. if ($xPathToJson === '') {
  844. return null;
  845. }
  846. $doc = new DOMDocument();
  847. $doc->recover = true;
  848. $doc->strictErrorChecking = false;
  849. if (!$doc->loadHTML($html, LIBXML_NONET | LIBXML_NOERROR | LIBXML_NOWARNING)) {
  850. return null;
  851. }
  852. $xpath = new DOMXPath($doc);
  853. $jsonFragments = @$xpath->evaluate($xPathToJson);
  854. if ($jsonFragments === false) {
  855. return null;
  856. }
  857. if (is_string($jsonFragments)) {
  858. return $jsonFragments;
  859. }
  860. if ($jsonFragments instanceof DOMNodeList && $jsonFragments->length > 0) {
  861. // If the result is a list, then aggregate as a JSON array
  862. $result = [];
  863. foreach ($jsonFragments as $node) {
  864. $json = json_decode($node->textContent, true);
  865. if (json_last_error() === JSON_ERROR_NONE && is_array($json)) {
  866. $result[] = $json;
  867. }
  868. }
  869. return json_encode($result, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE) ?: null;
  870. }
  871. return null;
  872. }
  873. public function loadJson(): ?\SimplePie\SimplePie {
  874. if ($this->url == '') {
  875. return null;
  876. }
  877. $feedSourceUrl = htmlspecialchars_decode($this->url, ENT_QUOTES);
  878. if ($feedSourceUrl == null) {
  879. return null;
  880. }
  881. $httpAccept = $this->kind() === FreshRSS_Feed::KIND_HTML_XPATH_JSON_DOTNOTATION ? 'html' : 'json';
  882. $content = httpGet($feedSourceUrl, $this->cacheFilename(), $httpAccept, $this->attributes(), $this->curlOptions())['body'];
  883. if (strlen($content) <= 0) {
  884. return null;
  885. }
  886. if ($this->kind() === FreshRSS_Feed::KIND_HTML_XPATH_JSON_DOTNOTATION) {
  887. $content = $this->extractJsonFromHtml($content);
  888. if ($content == null) {
  889. return null;
  890. }
  891. }
  892. //check if the content is actual JSON
  893. $jf = json_decode($content, true);
  894. if (json_last_error() !== JSON_ERROR_NONE || !is_array($jf)) {
  895. return null;
  896. }
  897. /** @var array<string,string> $json_dotnotation */
  898. $json_dotnotation = $this->attributeArray('json_dotnotation') ?? [];
  899. $dotnotations = $this->kind() === FreshRSS_Feed::KIND_JSONFEED ? $this->dotNotationForStandardJsonFeed() : $json_dotnotation;
  900. $feedContent = FreshRSS_dotNotation_Util::convertJsonToRss($jf, $feedSourceUrl, $dotnotations, $this->name());
  901. if ($feedContent == null) {
  902. return null;
  903. }
  904. return $this->simplePieFromContent($feedContent);
  905. }
  906. public function loadHtmlXpath(): ?\SimplePie\SimplePie {
  907. if ($this->url == '') {
  908. return null;
  909. }
  910. $feedSourceUrl = htmlspecialchars_decode($this->url, ENT_QUOTES);
  911. if ($feedSourceUrl == null) {
  912. return null;
  913. }
  914. // Same naming conventions than https://rss-bridge.github.io/rss-bridge/Bridge_API/XPathAbstract.html
  915. // https://rss-bridge.github.io/rss-bridge/Bridge_API/BridgeAbstract.html#collectdata
  916. /** @var array<string,string> $xPathSettings */
  917. $xPathSettings = $this->attributeArray('xpath');
  918. $xPathFeedTitle = $xPathSettings['feedTitle'] ?? '';
  919. $xPathItem = $xPathSettings['item'] ?? '';
  920. $xPathItemTitle = $xPathSettings['itemTitle'] ?? '';
  921. $xPathItemContent = $xPathSettings['itemContent'] ?? '';
  922. $xPathItemUri = $xPathSettings['itemUri'] ?? '';
  923. $xPathItemAuthor = $xPathSettings['itemAuthor'] ?? '';
  924. $xPathItemTimestamp = $xPathSettings['itemTimestamp'] ?? '';
  925. $xPathItemTimeFormat = $xPathSettings['itemTimeFormat'] ?? '';
  926. $xPathItemThumbnail = $xPathSettings['itemThumbnail'] ?? '';
  927. $xPathItemCategories = $xPathSettings['itemCategories'] ?? '';
  928. $xPathItemUid = $xPathSettings['itemUid'] ?? '';
  929. if ($xPathItem == '') {
  930. return null;
  931. }
  932. $httpAccept = $this->kind() === FreshRSS_Feed::KIND_XML_XPATH ? 'xml' : 'html';
  933. $html = httpGet($feedSourceUrl, $this->cacheFilename(), $httpAccept, $this->attributes(), $this->curlOptions())['body'];
  934. if (strlen($html) <= 0) {
  935. return null;
  936. }
  937. $view = new FreshRSS_View();
  938. $view->_path('index/rss.phtml');
  939. $view->internal_rendering = true;
  940. $view->rss_url = htmlspecialchars($feedSourceUrl, ENT_COMPAT, 'UTF-8');
  941. $view->html_url = $view->rss_url;
  942. $view->entries = [];
  943. try {
  944. $doc = new DOMDocument();
  945. $doc->recover = true;
  946. $doc->strictErrorChecking = false;
  947. $ok = false;
  948. switch ($this->kind()) {
  949. case FreshRSS_Feed::KIND_HTML_XPATH:
  950. $ok = $doc->loadHTML($html, LIBXML_NONET | LIBXML_NOERROR | LIBXML_NOWARNING) !== false;
  951. break;
  952. case FreshRSS_Feed::KIND_XML_XPATH:
  953. $ok = $doc->loadXML($html, LIBXML_NONET | LIBXML_NOERROR | LIBXML_NOWARNING) !== false;
  954. break;
  955. }
  956. if (!$ok) {
  957. return null;
  958. }
  959. $xpath = new DOMXPath($doc);
  960. $xpathEvaluateString = function (string $expression, ?DOMNode $contextNode = null) use ($xpath): string {
  961. $result = @$xpath->evaluate('normalize-space(' . $expression . ')', $contextNode);
  962. return is_string($result) ? $result : '';
  963. };
  964. $view->rss_title = $xPathFeedTitle == '' ? $this->name() :
  965. htmlspecialchars($xpathEvaluateString($xPathFeedTitle), ENT_COMPAT, 'UTF-8');
  966. $view->rss_base = htmlspecialchars(trim($xpathEvaluateString('//base/@href')), ENT_COMPAT, 'UTF-8');
  967. $nodes = $xpath->query($xPathItem);
  968. if ($nodes === false || $nodes->length === 0) {
  969. return null;
  970. }
  971. foreach ($nodes as $node) {
  972. $item = [];
  973. $item['title'] = $xPathItemTitle == '' ? '' : $xpathEvaluateString($xPathItemTitle, $node);
  974. $item['content'] = '';
  975. if ($xPathItemContent != '') {
  976. $result = @$xpath->evaluate($xPathItemContent, $node);
  977. if ($result instanceof DOMNodeList) {
  978. // List of nodes, save as HTML
  979. $content = '';
  980. foreach ($result as $child) {
  981. $content .= $doc->saveHTML($child) . "\n";
  982. }
  983. $item['content'] = $content;
  984. } elseif (is_string($result) || is_int($result) || is_bool($result)) {
  985. // Typed expression, save as-is
  986. $item['content'] = (string)$result;
  987. }
  988. }
  989. $item['link'] = $xPathItemUri == '' ? '' : $xpathEvaluateString($xPathItemUri, $node);
  990. $item['author'] = $xPathItemAuthor == '' ? '' : $xpathEvaluateString($xPathItemAuthor, $node);
  991. $item['timestamp'] = $xPathItemTimestamp == '' ? '' : $xpathEvaluateString($xPathItemTimestamp, $node);
  992. if ($xPathItemTimeFormat != '') {
  993. $dateTime = DateTime::createFromFormat($xPathItemTimeFormat, $item['timestamp']);
  994. if ($dateTime != false) {
  995. $item['timestamp'] = $dateTime->format(DateTime::ATOM);
  996. }
  997. }
  998. $item['thumbnail'] = $xPathItemThumbnail == '' ? '' : $xpathEvaluateString($xPathItemThumbnail, $node);
  999. if ($xPathItemCategories != '') {
  1000. $itemCategories = @$xpath->evaluate($xPathItemCategories, $node);
  1001. if (is_string($itemCategories) && $itemCategories !== '') {
  1002. $item['tags'] = [$itemCategories];
  1003. } elseif ($itemCategories instanceof DOMNodeList && $itemCategories->length > 0) {
  1004. $item['tags'] = [];
  1005. foreach ($itemCategories as $itemCategory) {
  1006. $item['tags'][] = $itemCategory->textContent;
  1007. }
  1008. }
  1009. }
  1010. if ($xPathItemUid != '') {
  1011. $item['guid'] = $xpathEvaluateString($xPathItemUid, $node);
  1012. }
  1013. if (empty($item['guid'])) {
  1014. $item['guid'] = 'urn:sha1:' . sha1($item['title'] . $item['content'] . $item['link']);
  1015. }
  1016. if ($item['title'] != '' || $item['content'] != '' || $item['link'] != '') {
  1017. // HTML-encoding/escaping of the relevant fields (all except 'content')
  1018. foreach (['author', 'guid', 'link', 'thumbnail', 'timestamp', 'title'] as $key) {
  1019. if (isset($item[$key])) {
  1020. $item[$key] = htmlspecialchars($item[$key], ENT_COMPAT, 'UTF-8');
  1021. }
  1022. }
  1023. if (isset($item['tags'])) {
  1024. $item['tags'] = Minz_Helper::htmlspecialchars_utf8($item['tags']);
  1025. }
  1026. // CDATA protection
  1027. $item['content'] = str_replace(']]>', ']]&gt;', $item['content']);
  1028. $view->entries[] = FreshRSS_Entry::fromArray($item);
  1029. }
  1030. }
  1031. } catch (Exception $ex) {
  1032. Minz_Log::warning($ex->getMessage());
  1033. return null;
  1034. }
  1035. return $this->simplePieFromContent($view->renderToString());
  1036. }
  1037. /**
  1038. * @return int|null The max number of unread articles to keep, or null if disabled.
  1039. */
  1040. public function keepMaxUnread(): ?int {
  1041. $keepMaxUnread = $this->attributeInt('keep_max_n_unread');
  1042. if ($keepMaxUnread === null) {
  1043. $keepMaxUnread = FreshRSS_Context::userConf()->mark_when['max_n_unread'];
  1044. }
  1045. return is_int($keepMaxUnread) && $keepMaxUnread >= 0 ? $keepMaxUnread : null;
  1046. }
  1047. /**
  1048. * @return int|false The number of articles marked as read, of false if error
  1049. */
  1050. public function markAsReadMaxUnread(): int|false {
  1051. $keepMaxUnread = $this->keepMaxUnread();
  1052. if ($keepMaxUnread === null) {
  1053. return false;
  1054. }
  1055. $feedDAO = FreshRSS_Factory::createFeedDao();
  1056. $affected = $feedDAO->markAsReadMaxUnread($this->id(), $keepMaxUnread);
  1057. return $affected;
  1058. }
  1059. /**
  1060. * Applies the *mark as read upon gone* policy, if enabled.
  1061. * Remember to call `updateCachedValues($id_feed)` or `updateCachedValues()` just after.
  1062. * @return int|false the number of lines affected, or false if not applicable
  1063. */
  1064. public function markAsReadUponGone(bool $upstreamIsEmpty, int $minLastSeen = 0): int|false {
  1065. $readUponGone = $this->attributeBoolean('read_upon_gone');
  1066. if ($readUponGone === null) {
  1067. $readUponGone = FreshRSS_Context::userConf()->mark_when['gone'];
  1068. }
  1069. if (!$readUponGone) {
  1070. return false;
  1071. }
  1072. if ($upstreamIsEmpty) {
  1073. if ($minLastSeen <= 0) {
  1074. $minLastSeen = time();
  1075. }
  1076. $entryDAO = FreshRSS_Factory::createEntryDao();
  1077. $affected = $entryDAO->markReadFeed($this->id(), $minLastSeen . '000000');
  1078. } else {
  1079. $feedDAO = FreshRSS_Factory::createFeedDao();
  1080. $affected = $feedDAO->markAsReadNotSeen($this->id(), $minLastSeen);
  1081. }
  1082. if ($affected > 0) {
  1083. Minz_Log::debug(__METHOD__ . " $affected items" . ($upstreamIsEmpty ? ' (all)' : '') . ' [' . $this->url(includeCredentials: false) . ']');
  1084. }
  1085. return $affected;
  1086. }
  1087. /**
  1088. * Remember to call `updateCachedValues($id_feed)` or `updateCachedValues()` just after
  1089. */
  1090. public function cleanOldEntries(): int|false {
  1091. /** @var array<string,bool|int|string>|null $archiving */
  1092. $archiving = $this->attributeArray('archiving');
  1093. if ($archiving === null) {
  1094. $catDAO = FreshRSS_Factory::createCategoryDao();
  1095. $category = $catDAO->searchById($this->categoryId);
  1096. $archiving = $category === null ? null : $category->attributeArray('archiving');
  1097. /** @var array<string,bool|int|string>|null $archiving */
  1098. if ($archiving === null) {
  1099. $archiving = FreshRSS_Context::userConf()->archiving;
  1100. }
  1101. }
  1102. if (is_array($archiving)) {
  1103. $entryDAO = FreshRSS_Factory::createEntryDao();
  1104. $nb = $entryDAO->cleanOldEntries($this->id(), $archiving);
  1105. if ($nb > 0) {
  1106. Minz_Log::debug($nb . ' entries cleaned in feed [' . $this->url(false) . '] with: ' . json_encode($archiving));
  1107. }
  1108. return $nb;
  1109. }
  1110. return false;
  1111. }
  1112. /**
  1113. * @param string $url Overridden URL. Will default to the feed URL.
  1114. * @throws FreshRSS_Context_Exception
  1115. */
  1116. public function cacheFilename(string $url = ''): string {
  1117. $simplePie = customSimplePie($this->attributes(), $this->curlOptions());
  1118. if ($url !== '') {
  1119. $filename = $simplePie->get_cache_filename($url);
  1120. return CACHE_PATH . '/' . $filename . '.html';
  1121. }
  1122. $url = htmlspecialchars_decode($this->url);
  1123. $filename = $simplePie->get_cache_filename($url);
  1124. switch ($this->kind) {
  1125. case FreshRSS_Feed::KIND_HTML_XPATH:
  1126. return CACHE_PATH . '/' . $filename . '.html';
  1127. case FreshRSS_Feed::KIND_XML_XPATH:
  1128. return CACHE_PATH . '/' . $filename . '.xml';
  1129. case FreshRSS_Feed::KIND_JSON_DOTNOTATION:
  1130. case FreshRSS_Feed::KIND_JSON_XPATH:
  1131. case FreshRSS_Feed::KIND_JSONFEED:
  1132. return CACHE_PATH . '/' . $filename . '.json';
  1133. case FreshRSS_Feed::KIND_RSS:
  1134. case FreshRSS_Feed::KIND_RSS_FORCED:
  1135. default:
  1136. return CACHE_PATH . '/' . $filename . '.spc';
  1137. }
  1138. }
  1139. private function faviconRebuild(): void {
  1140. if ($this->customFavicon()) {
  1141. return;
  1142. }
  1143. FreshRSS_Feed::faviconDelete($this->hashFavicon());
  1144. $this->faviconPrepare(true);
  1145. }
  1146. public function clearCache(): bool {
  1147. $this->faviconRebuild();
  1148. return @unlink($this->cacheFilename());
  1149. }
  1150. /** @return int|false */
  1151. public function cacheModifiedTime(): int|false {
  1152. $filename = $this->cacheFilename();
  1153. clearstatcache(true, $filename);
  1154. return @filemtime($filename);
  1155. }
  1156. public function lock(): bool {
  1157. $this->lockPath = TMP_PATH . '/' . $this->hash() . '.freshrss.lock';
  1158. if (file_exists($this->lockPath) && ((time() - (@filemtime($this->lockPath) ?: 0)) > 3600)) {
  1159. @unlink($this->lockPath);
  1160. }
  1161. if (($handle = @fopen($this->lockPath, 'x')) === false) {
  1162. return false;
  1163. }
  1164. //register_shutdown_function('unlink', $this->lockPath);
  1165. @fclose($handle);
  1166. return true;
  1167. }
  1168. public function unlock(): bool {
  1169. return @unlink($this->lockPath);
  1170. }
  1171. //<WebSub>
  1172. public function pubSubHubbubEnabled(): bool {
  1173. $url = $this->selfUrl ?: $this->url;
  1174. $hubFilename = PSHB_PATH . '/feeds/' . sha1($url) . '/!hub.json';
  1175. if (($hubFile = @file_get_contents($hubFilename)) != false) {
  1176. $hubJson = json_decode($hubFile, true);
  1177. if (is_array($hubJson) && empty($hubJson['error']) &&
  1178. (empty($hubJson['lease_end']) || $hubJson['lease_end'] > time())) {
  1179. return true;
  1180. }
  1181. }
  1182. return false;
  1183. }
  1184. public function pubSubHubbubError(bool $error = true): bool {
  1185. $url = $this->selfUrl ?: $this->url;
  1186. $hubFilename = PSHB_PATH . '/feeds/' . sha1($url) . '/!hub.json';
  1187. $hubFile = @file_get_contents($hubFilename);
  1188. $hubJson = is_string($hubFile) ? json_decode($hubFile, true) : null;
  1189. if (is_array($hubJson) && (!isset($hubJson['error']) || $hubJson['error'] !== $error)) {
  1190. $hubJson['error'] = $error;
  1191. file_put_contents($hubFilename, json_encode($hubJson));
  1192. Minz_Log::warning('Set error to ' . ($error ? 1 : 0) . ' for ' . $url, PSHB_LOG);
  1193. }
  1194. return false;
  1195. }
  1196. public function pubSubHubbubPrepare(): string|false {
  1197. $key = '';
  1198. if (Minz_Request::serverIsPublic(FreshRSS_Context::systemConf()->base_url) &&
  1199. $this->hubUrl !== '' && $this->selfUrl !== '' && @is_dir(PSHB_PATH)) {
  1200. $path = PSHB_PATH . '/feeds/' . sha1($this->selfUrl);
  1201. $hubFilename = $path . '/!hub.json';
  1202. if (($hubFile = @file_get_contents($hubFilename)) != false) {
  1203. $hubJson = json_decode($hubFile, true);
  1204. if (!is_array($hubJson) || empty($hubJson['key']) || !is_string($hubJson['key']) || !ctype_xdigit($hubJson['key'])) {
  1205. $text = 'Invalid JSON for WebSub: ' . $this->url;
  1206. Minz_Log::warning($text);
  1207. Minz_Log::warning($text, PSHB_LOG);
  1208. return false;
  1209. }
  1210. if (!empty($hubJson['lease_end']) && is_int($hubJson['lease_end']) && $hubJson['lease_end'] < (time() + (3600 * 23))) { //TODO: Make a better policy
  1211. $text = 'WebSub lease ends at '
  1212. . date('c', empty($hubJson['lease_end']) ? time() : $hubJson['lease_end'])
  1213. . ' and needs renewal: ' . $this->url;
  1214. Minz_Log::warning($text);
  1215. Minz_Log::warning($text, PSHB_LOG);
  1216. $key = $hubJson['key']; //To renew our lease
  1217. } elseif (((!empty($hubJson['error'])) || empty($hubJson['lease_end'])) &&
  1218. (empty($hubJson['lease_start']) || $hubJson['lease_start'] < time() - (3600 * 23))) { //Do not renew too often
  1219. $key = $hubJson['key']; //To renew our lease
  1220. }
  1221. } else {
  1222. @mkdir($path, 0770, true);
  1223. $key = sha1($path . FreshRSS_Context::systemConf()->salt);
  1224. $hubJson = [
  1225. 'hub' => $this->hubUrl,
  1226. 'key' => $key,
  1227. ];
  1228. file_put_contents($hubFilename, json_encode($hubJson));
  1229. @mkdir(PSHB_PATH . '/keys/', 0770, true);
  1230. file_put_contents(PSHB_PATH . '/keys/' . $key . '.txt', $this->selfUrl);
  1231. $text = 'WebSub prepared for ' . $this->url;
  1232. Minz_Log::debug($text);
  1233. Minz_Log::debug($text, PSHB_LOG);
  1234. }
  1235. $currentUser = Minz_User::name() ?? '';
  1236. if (FreshRSS_user_Controller::checkUsername($currentUser) && !file_exists($path . '/' . $currentUser . '.txt')) {
  1237. touch($path . '/' . $currentUser . '.txt');
  1238. }
  1239. }
  1240. return $key;
  1241. }
  1242. //Parameter true to subscribe, false to unsubscribe.
  1243. public function pubSubHubbubSubscribe(bool $state): bool {
  1244. if ($state) {
  1245. $url = $this->selfUrl ?: $this->url;
  1246. } else {
  1247. $url = $this->url; //Always use current URL during unsubscribe
  1248. }
  1249. if ($url !== '' && (Minz_Request::serverIsPublic(FreshRSS_Context::systemConf()->base_url) || !$state)) {
  1250. $hubFilename = PSHB_PATH . '/feeds/' . sha1($url) . '/!hub.json';
  1251. $hubFile = @file_get_contents($hubFilename);
  1252. if ($hubFile === false) {
  1253. Minz_Log::warning('JSON not found for WebSub: ' . $this->url);
  1254. return false;
  1255. }
  1256. $hubJson = json_decode($hubFile, true);
  1257. if (!is_array($hubJson) || empty($hubJson['key']) || !is_string($hubJson['key']) || !ctype_xdigit($hubJson['key']) ||
  1258. empty($hubJson['hub']) || !is_string($hubJson['hub'])) {
  1259. Minz_Log::warning('Invalid JSON for WebSub: ' . $this->url);
  1260. return false;
  1261. }
  1262. $callbackUrl = checkUrl(Minz_Request::getBaseUrl() . '/api/pshb.php?k=' . $hubJson['key']);
  1263. if ($callbackUrl == '') {
  1264. Minz_Log::warning('Invalid callback for WebSub: ' . $this->url);
  1265. return false;
  1266. }
  1267. if (!$state) { //unsubscribe
  1268. $hubJson['lease_end'] = time() - 60;
  1269. file_put_contents($hubFilename, json_encode($hubJson));
  1270. }
  1271. $ch = curl_init();
  1272. if ($ch === false) {
  1273. Minz_Log::warning('curl_init() failed in ' . __METHOD__);
  1274. return false;
  1275. }
  1276. curl_setopt_array($ch, [
  1277. CURLOPT_URL => $hubJson['hub'],
  1278. CURLOPT_RETURNTRANSFER => true,
  1279. CURLOPT_POSTFIELDS => http_build_query([
  1280. 'hub.verify' => 'sync',
  1281. 'hub.mode' => $state ? 'subscribe' : 'unsubscribe',
  1282. 'hub.topic' => $url,
  1283. 'hub.callback' => $callbackUrl,
  1284. ]),
  1285. CURLOPT_USERAGENT => FRESHRSS_USERAGENT,
  1286. CURLOPT_MAXREDIRS => 10,
  1287. CURLOPT_FOLLOWLOCATION => true,
  1288. CURLOPT_ENCODING => '', //Enable all encodings
  1289. //CURLOPT_VERBOSE => 1, // To debug sent HTTP headers
  1290. ]);
  1291. $response = curl_exec($ch);
  1292. $info = curl_getinfo($ch);
  1293. if (!is_array($info)) {
  1294. Minz_Log::warning('curl_getinfo() failed in ' . __METHOD__);
  1295. return false;
  1296. }
  1297. Minz_Log::warning('WebSub ' . ($state ? 'subscribe' : 'unsubscribe') . ' to ' . $url .
  1298. ' via hub ' . $hubJson['hub'] .
  1299. ' with callback ' . $callbackUrl . ': ' . $info['http_code'] . ' ' . $response, PSHB_LOG);
  1300. if (str_starts_with('' . $info['http_code'], '2')) {
  1301. return true;
  1302. } else {
  1303. $hubJson['lease_start'] = time(); //Prevent trying again too soon
  1304. $hubJson['error'] = true;
  1305. file_put_contents($hubFilename, json_encode($hubJson));
  1306. return false;
  1307. }
  1308. }
  1309. return false;
  1310. }
  1311. //</WebSub>
  1312. }