lib_rss.php 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468
  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('_', '/', $class) . '.php');
  45. }
  46. }
  47. spl_autoload_register('classAutoloader');
  48. //</Auto-loading>
  49. function idn_to_puny($url) {
  50. if (function_exists('idn_to_ascii')) {
  51. $parts = parse_url($url);
  52. if (!empty($parts['host'])) {
  53. $idn = $parts['host'];
  54. $puny = idn_to_ascii($idn);
  55. $pos = strpos($url, $idn);
  56. if ($pos !== false) {
  57. return substr_replace($url, $puny, $pos, strlen($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. $url = idn_to_puny($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. $simplePie->set_curl_options($system_conf->curl_options);
  135. $simplePie->strip_htmltags(array(
  136. 'base', 'blink', 'body', 'doctype', 'embed',
  137. 'font', 'form', 'frame', 'frameset', 'html',
  138. 'link', 'input', 'marquee', 'meta', 'noscript',
  139. 'object', 'param', 'plaintext', 'script', 'style',
  140. ));
  141. $simplePie->strip_attributes(array_merge($simplePie->strip_attributes, array(
  142. 'autoplay', 'onload', 'onunload', 'onclick', 'ondblclick', 'onmousedown', 'onmouseup',
  143. 'onmouseover', 'onmousemove', 'onmouseout', 'onfocus', 'onblur',
  144. 'onkeypress', 'onkeydown', 'onkeyup', 'onselect', 'onchange', 'seamless')));
  145. $simplePie->add_attributes(array(
  146. 'img' => array('lazyload' => '', 'postpone' => ''), //http://www.w3.org/TR/resource-priorities/
  147. 'audio' => array('lazyload' => '', 'postpone' => '', 'preload' => 'none'),
  148. 'iframe' => array('lazyload' => '', 'postpone' => '', 'sandbox' => 'allow-scripts allow-same-origin'),
  149. 'video' => array('lazyload' => '', 'postpone' => '', 'preload' => 'none'),
  150. ));
  151. $simplePie->set_url_replacements(array(
  152. 'a' => 'href',
  153. 'area' => 'href',
  154. 'audio' => 'src',
  155. 'blockquote' => 'cite',
  156. 'del' => 'cite',
  157. 'form' => 'action',
  158. 'iframe' => 'src',
  159. 'img' => array(
  160. 'longdesc',
  161. 'src'
  162. ),
  163. 'input' => 'src',
  164. 'ins' => 'cite',
  165. 'q' => 'cite',
  166. 'source' => 'src',
  167. 'track' => 'src',
  168. 'video' => array(
  169. 'poster',
  170. 'src',
  171. ),
  172. ));
  173. return $simplePie;
  174. }
  175. function sanitizeHTML($data, $base = '') {
  176. static $simplePie = null;
  177. if ($simplePie == null) {
  178. $simplePie = customSimplePie();
  179. $simplePie->init();
  180. }
  181. return html_only_entity_decode($simplePie->sanitize->sanitize($data, SIMPLEPIE_CONSTRUCT_HTML, $base));
  182. }
  183. /* permet de récupérer le contenu d'un article pour un flux qui n'est pas complet */
  184. function get_content_by_parsing ($url, $path) {
  185. require_once(LIB_PATH . '/lib_phpQuery.php');
  186. Minz_Log::notice('FreshRSS GET ' . SimplePie_Misc::url_remove_credentials($url));
  187. $html = file_get_contents($url);
  188. if ($html) {
  189. $doc = phpQuery::newDocument($html);
  190. $content = $doc->find($path);
  191. foreach (pq('img[data-src]') as $img) {
  192. $imgP = pq($img);
  193. $dataSrc = $imgP->attr('data-src');
  194. if (strlen($dataSrc) > 4) {
  195. $imgP->attr('src', $dataSrc);
  196. $imgP->removeAttr('data-src');
  197. }
  198. }
  199. return sanitizeHTML($content->__toString(), $url);
  200. } else {
  201. throw new Exception();
  202. }
  203. }
  204. /**
  205. * Add support of image lazy loading
  206. * Move content from src attribute to data-original
  207. * @param content is the text we want to parse
  208. */
  209. function lazyimg($content) {
  210. return preg_replace(
  211. '/<((?:img|iframe)[^>]+?)src=[\'"]([^"\']+)[\'"]([^>]*)>/i',
  212. '<$1src="' . Minz_Url::display('/themes/icons/grey.gif') . '" data-original="$2"$3>',
  213. $content
  214. );
  215. }
  216. function uTimeString() {
  217. $t = @gettimeofday();
  218. return $t['sec'] . str_pad($t['usec'], 6, '0');
  219. }
  220. function uSecString() {
  221. $t = @gettimeofday();
  222. return str_pad($t['usec'], 6, '0');
  223. }
  224. function invalidateHttpCache() {
  225. Minz_Session::_param('touch', uTimeString());
  226. return touch(join_path(DATA_PATH, 'users', Minz_Session::param('currentUser', '_'), 'log.txt'));
  227. }
  228. function listUsers() {
  229. $final_list = array();
  230. $base_path = join_path(DATA_PATH, 'users');
  231. $dir_list = array_values(array_diff(
  232. scandir($base_path),
  233. array('..', '.', '_')
  234. ));
  235. foreach ($dir_list as $file) {
  236. if (is_dir(join_path($base_path, $file))) {
  237. $final_list[] = $file;
  238. }
  239. }
  240. return $final_list;
  241. }
  242. /**
  243. * Register and return the configuration for a given user.
  244. *
  245. * Note this function has been created to generate temporary configuration
  246. * objects. If you need a long-time configuration, please don't use this function.
  247. *
  248. * @param $username the name of the user of which we want the configuration.
  249. * @return a Minz_Configuration object, null if the configuration cannot be loaded.
  250. */
  251. function get_user_configuration($username) {
  252. $namespace = 'user_' . $username;
  253. try {
  254. Minz_Configuration::register($namespace,
  255. join_path(USERS_PATH, $username, 'config.php'),
  256. join_path(USERS_PATH, '_', 'config.default.php'));
  257. } catch (Minz_ConfigurationNamespaceException $e) {
  258. // namespace already exists, do nothing.
  259. } catch (Minz_FileNotExistException $e) {
  260. Minz_Log::warning($e->getMessage());
  261. return null;
  262. }
  263. return Minz_Configuration::get($namespace);
  264. }
  265. function httpAuthUser() {
  266. return isset($_SERVER['REMOTE_USER']) ? $_SERVER['REMOTE_USER'] : '';
  267. }
  268. function cryptAvailable() {
  269. if (version_compare(PHP_VERSION, '5.3.3', '>=')) {
  270. try {
  271. $hash = '$2y$04$usesomesillystringfore7hnbRJHxXVLeakoG8K30oukPsA.ztMG';
  272. return $hash === @crypt('password', $hash);
  273. } catch (Exception $e) {
  274. }
  275. }
  276. return false;
  277. }
  278. function is_referer_from_same_domain() {
  279. if (empty($_SERVER['HTTP_REFERER'])) {
  280. return false;
  281. }
  282. $host = parse_url(((isset($_SERVER['HTTPS']) && $_SERVER['HTTPS'] === 'on') ? 'https://' : 'http://') .
  283. (empty($_SERVER['HTTP_HOST']) ? $_SERVER['SERVER_NAME'] : $_SERVER['HTTP_HOST']));
  284. $referer = parse_url($_SERVER['HTTP_REFERER']);
  285. if (empty($host['host']) || empty($referer['host']) || $host['host'] !== $referer['host']) {
  286. return false;
  287. }
  288. //TODO: check 'scheme', taking into account the case of a proxy
  289. if ((isset($host['port']) ? $host['port'] : 0) !== (isset($referer['port']) ? $referer['port'] : 0)) {
  290. return false;
  291. }
  292. return true;
  293. }
  294. /**
  295. * Check PHP and its extensions are well-installed.
  296. *
  297. * @return array of tested values.
  298. */
  299. function check_install_php() {
  300. $pdo_mysql = extension_loaded('pdo_mysql');
  301. $pdo_sqlite = extension_loaded('pdo_sqlite');
  302. return array(
  303. 'php' => version_compare(PHP_VERSION, '5.2.1') >= 0,
  304. 'minz' => file_exists(LIB_PATH . '/Minz'),
  305. 'curl' => extension_loaded('curl'),
  306. 'pdo' => $pdo_mysql || $pdo_sqlite,
  307. 'pcre' => extension_loaded('pcre'),
  308. 'ctype' => extension_loaded('ctype'),
  309. 'dom' => class_exists('DOMDocument'),
  310. 'json' => extension_loaded('json'),
  311. 'zip' => extension_loaded('zip'),
  312. );
  313. }
  314. /**
  315. * Check different data files and directories exist.
  316. *
  317. * @return array of tested values.
  318. */
  319. function check_install_files() {
  320. return array(
  321. 'data' => DATA_PATH && is_writable(DATA_PATH),
  322. 'cache' => CACHE_PATH && is_writable(CACHE_PATH),
  323. 'users' => USERS_PATH && is_writable(USERS_PATH),
  324. 'favicons' => is_writable(DATA_PATH . '/favicons'),
  325. 'persona' => is_writable(DATA_PATH . '/persona'),
  326. 'tokens' => is_writable(DATA_PATH . '/tokens'),
  327. );
  328. }
  329. /**
  330. * Check database is well-installed.
  331. *
  332. * @return array of tested values.
  333. */
  334. function check_install_database() {
  335. $status = array(
  336. 'connection' => true,
  337. 'tables' => false,
  338. 'categories' => false,
  339. 'feeds' => false,
  340. 'entries' => false,
  341. );
  342. try {
  343. $dbDAO = FreshRSS_Factory::createDatabaseDAO();
  344. $status['tables'] = $dbDAO->tablesAreCorrect();
  345. $status['categories'] = $dbDAO->categoryIsCorrect();
  346. $status['feeds'] = $dbDAO->feedIsCorrect();
  347. $status['entries'] = $dbDAO->entryIsCorrect();
  348. } catch(Minz_PDOConnectionException $e) {
  349. $status['connection'] = false;
  350. }
  351. return $status;
  352. }
  353. /**
  354. * Remove a directory recursively.
  355. *
  356. * From http://php.net/rmdir#110489
  357. *
  358. * @param $dir the directory to remove
  359. */
  360. function recursive_unlink($dir) {
  361. if (!is_dir($dir)) {
  362. return true;
  363. }
  364. $files = array_diff(scandir($dir), array('.', '..'));
  365. foreach ($files as $filename) {
  366. $filename = $dir . '/' . $filename;
  367. if (is_dir($filename)) {
  368. @chmod($filename, 0777);
  369. recursive_unlink($filename);
  370. } else {
  371. unlink($filename);
  372. }
  373. }
  374. return rmdir($dir);
  375. }
  376. /**
  377. * Remove queries where $get is appearing.
  378. * @param $get the get attribute which should be removed.
  379. * @param $queries an array of queries.
  380. * @return the same array whithout those where $get is appearing.
  381. */
  382. function remove_query_by_get($get, $queries) {
  383. $final_queries = array();
  384. foreach ($queries as $key => $query) {
  385. if (empty($query['get']) || $query['get'] !== $get) {
  386. $final_queries[$key] = $query;
  387. }
  388. }
  389. return $final_queries;
  390. }
  391. /**
  392. * Add a value in an array and take care it is unique.
  393. * @param $array the array in which we add the value.
  394. * @param $value the value to add.
  395. */
  396. function array_push_unique(&$array, $value) {
  397. $found = array_search($value, $array) !== false;
  398. if (!$found) {
  399. $array[] = $value;
  400. }
  401. }
  402. /**
  403. * Remove a value from an array.
  404. * @param $array the array from wich value is removed.
  405. * @param $value the value to remove.
  406. */
  407. function array_remove(&$array, $value) {
  408. $array = array_diff($array, array($value));
  409. }
  410. //RFC 4648
  411. function base64url_encode($data) {
  412. return strtr(rtrim(base64_encode($data), '='), '+/', '-_');
  413. }
  414. //RFC 4648
  415. function base64url_decode($data) {
  416. return base64_decode(strtr($data, '-_', '+/'));
  417. }