lib_rss.php 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448
  1. <?php
  2. if (!function_exists('json_decode')) {
  3. require_once('JSON.php');
  4. function json_decode($var) {
  5. $JSON = new Services_JSON;
  6. return (array)($JSON->decode($var));
  7. }
  8. }
  9. if (!function_exists('json_encode')) {
  10. require_once('JSON.php');
  11. function json_encode($var) {
  12. $JSON = new Services_JSON;
  13. return $JSON->encodeUnsafe($var);
  14. }
  15. }
  16. /**
  17. * Build a directory path by concatenating a list of directory names.
  18. *
  19. * @param $path_parts a list of directory names
  20. * @return a string corresponding to the final pathname
  21. */
  22. function join_path() {
  23. $path_parts = func_get_args();
  24. return join(DIRECTORY_SEPARATOR, $path_parts);
  25. }
  26. //<Auto-loading>
  27. function classAutoloader($class) {
  28. if (strpos($class, 'FreshRSS') === 0) {
  29. $components = explode('_', $class);
  30. switch (count($components)) {
  31. case 1:
  32. include(APP_PATH . '/' . $components[0] . '.php');
  33. return;
  34. case 2:
  35. include(APP_PATH . '/Models/' . $components[1] . '.php');
  36. return;
  37. case 3: //Controllers, Exceptions
  38. include(APP_PATH . '/' . $components[2] . 's/' . $components[1] . $components[2] . '.php');
  39. return;
  40. }
  41. } elseif (strpos($class, 'Minz') === 0) {
  42. include(LIB_PATH . '/' . str_replace('_', '/', $class) . '.php');
  43. } elseif (strpos($class, 'SimplePie') === 0) {
  44. include(LIB_PATH . '/SimplePie/' . str_replace('_', '/', $class) . '.php');
  45. }
  46. }
  47. spl_autoload_register('classAutoloader');
  48. //</Auto-loading>
  49. function idn_to_punny($url) {
  50. if (function_exists('idn_to_ascii')) {
  51. $parts = parse_url($url);
  52. if (!empty($parts['host'])) {
  53. $idn = $parts['host'];
  54. $punny = idn_to_ascii($idn);
  55. $pos = strpos($url, $idn);
  56. if ($pos !== false) {
  57. return substr_replace($url, $punny, $pos, strlen($idn));
  58. }
  59. }
  60. }
  61. return $url;
  62. }
  63. function checkUrl($url) {
  64. if (empty ($url)) {
  65. return '';
  66. }
  67. if (!preg_match ('#^https?://#i', $url)) {
  68. $url = 'http://' . $url;
  69. }
  70. $url = idn_to_punny($url); //PHP bug #53474 IDN
  71. if (filter_var($url, FILTER_VALIDATE_URL) ||
  72. (version_compare(PHP_VERSION, '5.3.3', '<') && (strpos($url, '-') > 0) && //PHP bug #51192
  73. ($url === filter_var($url, FILTER_SANITIZE_URL)))) {
  74. return $url;
  75. } else {
  76. return false;
  77. }
  78. }
  79. function format_number($n, $precision = 0) {
  80. // number_format does not seem to be Unicode-compatible
  81. return str_replace(' ', ' ', //Espace fine insécable
  82. number_format($n, $precision, '.', ' ')
  83. );
  84. }
  85. function format_bytes($bytes, $precision = 2, $system = 'IEC') {
  86. if ($system === 'IEC') {
  87. $base = 1024;
  88. $units = array('B', 'KiB', 'MiB', 'GiB', 'TiB');
  89. } elseif ($system === 'SI') {
  90. $base = 1000;
  91. $units = array('B', 'KB', 'MB', 'GB', 'TB');
  92. }
  93. $bytes = max(intval($bytes), 0);
  94. $pow = $bytes === 0 ? 0 : floor(log($bytes) / log($base));
  95. $pow = min($pow, count($units) - 1);
  96. $bytes /= pow($base, $pow);
  97. return format_number($bytes, $precision) . ' ' . $units[$pow];
  98. }
  99. function timestamptodate ($t, $hour = true) {
  100. $month = _t('gen.date.' . date('M', $t));
  101. if ($hour) {
  102. $date = _t('gen.date.format_date_hour', $month);
  103. } else {
  104. $date = _t('gen.date.format_date', $month);
  105. }
  106. return @date ($date, $t);
  107. }
  108. function html_only_entity_decode($text) {
  109. static $htmlEntitiesOnly = null;
  110. if ($htmlEntitiesOnly === null) {
  111. if (version_compare(PHP_VERSION, '5.3.4') >= 0) {
  112. $htmlEntitiesOnly = array_flip(array_diff(
  113. get_html_translation_table(HTML_ENTITIES, ENT_NOQUOTES, 'UTF-8'), //Decode HTML entities
  114. get_html_translation_table(HTML_SPECIALCHARS, ENT_NOQUOTES, 'UTF-8') //Preserve XML entities
  115. ));
  116. } else {
  117. $htmlEntitiesOnly = array_map('utf8_encode', array_flip(array_diff(
  118. get_html_translation_table(HTML_ENTITIES, ENT_NOQUOTES), //Decode HTML entities
  119. get_html_translation_table(HTML_SPECIALCHARS, ENT_NOQUOTES) //Preserve XML entities
  120. )));
  121. }
  122. }
  123. return strtr($text, $htmlEntitiesOnly);
  124. }
  125. function customSimplePie() {
  126. $system_conf = Minz_Configuration::get('system');
  127. $limits = $system_conf->limits;
  128. $simplePie = new SimplePie();
  129. $simplePie->set_useragent(_t('gen.freshrss') . '/' . FRESHRSS_VERSION . ' (' . PHP_OS . '; ' . FRESHRSS_WEBSITE . ') ' . SIMPLEPIE_NAME . '/' . SIMPLEPIE_VERSION);
  130. $simplePie->set_syslog($system_conf->simplepie_syslog_enabled);
  131. $simplePie->set_cache_location(CACHE_PATH);
  132. $simplePie->set_cache_duration($limits['cache_duration']);
  133. $simplePie->set_timeout($limits['timeout']);
  134. $simplePie->strip_htmltags(array(
  135. 'base', 'blink', 'body', 'doctype', 'embed',
  136. 'font', 'form', 'frame', 'frameset', 'html',
  137. 'link', 'input', 'marquee', 'meta', 'noscript',
  138. 'object', 'param', 'plaintext', 'script', 'style',
  139. ));
  140. $simplePie->strip_attributes(array_merge($simplePie->strip_attributes, array(
  141. 'autoplay', 'onload', 'onunload', 'onclick', 'ondblclick', 'onmousedown', 'onmouseup',
  142. 'onmouseover', 'onmousemove', 'onmouseout', 'onfocus', 'onblur',
  143. 'onkeypress', 'onkeydown', 'onkeyup', 'onselect', 'onchange', 'seamless')));
  144. $simplePie->add_attributes(array(
  145. 'img' => array('lazyload' => '', 'postpone' => ''), //http://www.w3.org/TR/resource-priorities/
  146. 'audio' => array('lazyload' => '', 'postpone' => '', 'preload' => 'none'),
  147. 'iframe' => array('lazyload' => '', 'postpone' => '', 'sandbox' => 'allow-scripts allow-same-origin'),
  148. 'video' => array('lazyload' => '', 'postpone' => '', 'preload' => 'none'),
  149. ));
  150. $simplePie->set_url_replacements(array(
  151. 'a' => 'href',
  152. 'area' => 'href',
  153. 'audio' => 'src',
  154. 'blockquote' => 'cite',
  155. 'del' => 'cite',
  156. 'form' => 'action',
  157. 'iframe' => 'src',
  158. 'img' => array(
  159. 'longdesc',
  160. 'src'
  161. ),
  162. 'input' => 'src',
  163. 'ins' => 'cite',
  164. 'q' => 'cite',
  165. 'source' => 'src',
  166. 'track' => 'src',
  167. 'video' => array(
  168. 'poster',
  169. 'src',
  170. ),
  171. ));
  172. return $simplePie;
  173. }
  174. function sanitizeHTML($data, $base = '') {
  175. static $simplePie = null;
  176. if ($simplePie == null) {
  177. $simplePie = customSimplePie();
  178. $simplePie->init();
  179. }
  180. return html_only_entity_decode($simplePie->sanitize->sanitize($data, SIMPLEPIE_CONSTRUCT_HTML, $base));
  181. }
  182. /* permet de récupérer le contenu d'un article pour un flux qui n'est pas complet */
  183. function get_content_by_parsing ($url, $path) {
  184. require_once (LIB_PATH . '/lib_phpQuery.php');
  185. Minz_Log::notice('FreshRSS GET ' . SimplePie_Misc::url_remove_credentials($url));
  186. $html = file_get_contents ($url);
  187. if ($html) {
  188. $doc = phpQuery::newDocument ($html);
  189. $content = $doc->find ($path);
  190. return sanitizeHTML($content->__toString(), $url);
  191. } else {
  192. throw new Exception ();
  193. }
  194. }
  195. /**
  196. * Add support of image lazy loading
  197. * Move content from src attribute to data-original
  198. * @param content is the text we want to parse
  199. */
  200. function lazyimg($content) {
  201. return preg_replace(
  202. '/<((?:img|iframe)[^>]+?)src=[\'"]([^"\']+)[\'"]([^>]*)>/i',
  203. '<$1src="' . Minz_Url::display('/themes/icons/grey.gif') . '" data-original="$2"$3>',
  204. $content
  205. );
  206. }
  207. function uTimeString() {
  208. $t = @gettimeofday();
  209. return $t['sec'] . str_pad($t['usec'], 6, '0');
  210. }
  211. function uSecString() {
  212. $t = @gettimeofday();
  213. return str_pad($t['usec'], 6, '0');
  214. }
  215. function invalidateHttpCache() {
  216. Minz_Session::_param('touch', uTimeString());
  217. return touch(join_path(DATA_PATH, 'users', Minz_Session::param('currentUser', '_'), 'log.txt'));
  218. }
  219. function listUsers() {
  220. $final_list = array();
  221. $base_path = join_path(DATA_PATH, 'users');
  222. $dir_list = array_values(array_diff(
  223. scandir($base_path),
  224. array('..', '.', '_')
  225. ));
  226. foreach ($dir_list as $file) {
  227. if (is_dir(join_path($base_path, $file))) {
  228. $final_list[] = $file;
  229. }
  230. }
  231. return $final_list;
  232. }
  233. /**
  234. * Register and return the configuration for a given user.
  235. *
  236. * Note this function has been created to generate temporary configuration
  237. * objects. If you need a long-time configuration, please don't use this function.
  238. *
  239. * @param $username the name of the user of which we want the configuration.
  240. * @return a Minz_Configuration object, null if the configuration cannot be loaded.
  241. */
  242. function get_user_configuration($username) {
  243. $namespace = 'user_' . $username;
  244. try {
  245. Minz_Configuration::register($namespace,
  246. join_path(USERS_PATH, $username, 'config.php'),
  247. join_path(USERS_PATH, '_', 'config.default.php'));
  248. } catch (Minz_ConfigurationNamespaceException $e) {
  249. // namespace already exists, do nothing.
  250. } catch (Minz_FileNotExistException $e) {
  251. Minz_Log::warning($e->getMessage());
  252. return null;
  253. }
  254. return Minz_Configuration::get($namespace);
  255. }
  256. function httpAuthUser() {
  257. return isset($_SERVER['REMOTE_USER']) ? $_SERVER['REMOTE_USER'] : '';
  258. }
  259. function cryptAvailable() {
  260. if (version_compare(PHP_VERSION, '5.3.3', '>=')) {
  261. try {
  262. $hash = '$2y$04$usesomesillystringfore7hnbRJHxXVLeakoG8K30oukPsA.ztMG';
  263. return $hash === @crypt('password', $hash);
  264. } catch (Exception $e) {
  265. }
  266. }
  267. return false;
  268. }
  269. function is_referer_from_same_domain() {
  270. if (empty($_SERVER['HTTP_REFERER'])) {
  271. return false;
  272. }
  273. $host = parse_url(((isset($_SERVER['HTTPS']) && $_SERVER['HTTPS'] === 'on') ? 'https://' : 'http://') .
  274. (empty($_SERVER['HTTP_HOST']) ? $_SERVER['SERVER_NAME'] : $_SERVER['HTTP_HOST']));
  275. $referer = parse_url($_SERVER['HTTP_REFERER']);
  276. if (empty($host['host']) || empty($referer['host']) || $host['host'] !== $referer['host']) {
  277. return false;
  278. }
  279. //TODO: check 'scheme', taking into account the case of a proxy
  280. if ((isset($host['port']) ? $host['port'] : 0) !== (isset($referer['port']) ? $referer['port'] : 0)) {
  281. return false;
  282. }
  283. return true;
  284. }
  285. /**
  286. * Check PHP and its extensions are well-installed.
  287. *
  288. * @return array of tested values.
  289. */
  290. function check_install_php() {
  291. $pdo_mysql = extension_loaded('pdo_mysql');
  292. $pdo_sqlite = extension_loaded('pdo_sqlite');
  293. return array(
  294. 'php' => version_compare(PHP_VERSION, '5.2.1') >= 0,
  295. 'minz' => file_exists(LIB_PATH . '/Minz'),
  296. 'curl' => extension_loaded('curl'),
  297. 'pdo' => $pdo_mysql || $pdo_sqlite,
  298. 'pcre' => extension_loaded('pcre'),
  299. 'ctype' => extension_loaded('ctype'),
  300. 'dom' => class_exists('DOMDocument'),
  301. 'json' => extension_loaded('json'),
  302. 'zip' => extension_loaded('zip'),
  303. );
  304. }
  305. /**
  306. * Check different data files and directories exist.
  307. *
  308. * @return array of tested values.
  309. */
  310. function check_install_files() {
  311. return array(
  312. 'data' => DATA_PATH && is_writable(DATA_PATH),
  313. 'cache' => CACHE_PATH && is_writable(CACHE_PATH),
  314. 'users' => USERS_PATH && is_writable(USERS_PATH),
  315. 'favicons' => is_writable(DATA_PATH . '/favicons'),
  316. 'persona' => is_writable(DATA_PATH . '/persona'),
  317. 'tokens' => is_writable(DATA_PATH . '/tokens'),
  318. );
  319. }
  320. /**
  321. * Check database is well-installed.
  322. *
  323. * @return array of tested values.
  324. */
  325. function check_install_database() {
  326. $status = array(
  327. 'connection' => true,
  328. 'tables' => false,
  329. 'categories' => false,
  330. 'feeds' => false,
  331. 'entries' => false,
  332. );
  333. try {
  334. $dbDAO = FreshRSS_Factory::createDatabaseDAO();
  335. $status['tables'] = $dbDAO->tablesAreCorrect();
  336. $status['categories'] = $dbDAO->categoryIsCorrect();
  337. $status['feeds'] = $dbDAO->feedIsCorrect();
  338. $status['entries'] = $dbDAO->entryIsCorrect();
  339. } catch(Minz_PDOConnectionException $e) {
  340. $status['connection'] = false;
  341. }
  342. return $status;
  343. }
  344. /**
  345. * Remove a directory recursively.
  346. *
  347. * From http://php.net/rmdir#110489
  348. *
  349. * @param $dir the directory to remove
  350. */
  351. function recursive_unlink($dir) {
  352. if (!is_dir($dir)) {
  353. return true;
  354. }
  355. $files = array_diff(scandir($dir), array('.', '..'));
  356. foreach ($files as $filename) {
  357. $filename = $dir . '/' . $filename;
  358. if (is_dir($filename)) {
  359. @chmod($filename, 0777);
  360. recursive_unlink($filename);
  361. } else {
  362. unlink($filename);
  363. }
  364. }
  365. return rmdir($dir);
  366. }
  367. /**
  368. * Remove queries where $get is appearing.
  369. * @param $get the get attribute which should be removed.
  370. * @param $queries an array of queries.
  371. * @return the same array whithout those where $get is appearing.
  372. */
  373. function remove_query_by_get($get, $queries) {
  374. $final_queries = array();
  375. foreach ($queries as $key => $query) {
  376. if (empty($query['get']) || $query['get'] !== $get) {
  377. $final_queries[$key] = $query;
  378. }
  379. }
  380. return $final_queries;
  381. }
  382. /**
  383. * Add a value in an array and take care it is unique.
  384. * @param $array the array in which we add the value.
  385. * @param $value the value to add.
  386. */
  387. function array_push_unique(&$array, $value) {
  388. $found = array_search($value, $array) !== false;
  389. if (!$found) {
  390. $array[] = $value;
  391. }
  392. }
  393. /**
  394. * Remove a value from an array.
  395. * @param $array the array from wich value is removed.
  396. * @param $value the value to remove.
  397. */
  398. function array_remove(&$array, $value) {
  399. $array = array_diff($array, array($value));
  400. }