lib_rss.php 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537
  1. <?php
  2. if (version_compare(PHP_VERSION, '5.3.8', '<')) {
  3. die('FreshRSS error: FreshRSS requires PHP 5.3.8+!');
  4. }
  5. if (!function_exists('json_decode')) {
  6. require_once('JSON.php');
  7. function json_decode($var, $assoc = false) {
  8. $JSON = new Services_JSON($assoc ? SERVICES_JSON_LOOSE_TYPE : 0);
  9. return $JSON->decode($var);
  10. }
  11. }
  12. if (!function_exists('json_encode')) {
  13. require_once('JSON.php');
  14. function json_encode($var) {
  15. $JSON = new Services_JSON();
  16. return $JSON->encodeUnsafe($var);
  17. }
  18. }
  19. defined('JSON_UNESCAPED_UNICODE') or define('JSON_UNESCAPED_UNICODE', 256); //PHP 5.3
  20. /**
  21. * Build a directory path by concatenating a list of directory names.
  22. *
  23. * @param $path_parts a list of directory names
  24. * @return a string corresponding to the final pathname
  25. */
  26. function join_path() {
  27. $path_parts = func_get_args();
  28. return join(DIRECTORY_SEPARATOR, $path_parts);
  29. }
  30. //<Auto-loading>
  31. function classAutoloader($class) {
  32. if (strpos($class, 'FreshRSS') === 0) {
  33. $components = explode('_', $class);
  34. switch (count($components)) {
  35. case 1:
  36. include(APP_PATH . '/' . $components[0] . '.php');
  37. return;
  38. case 2:
  39. include(APP_PATH . '/Models/' . $components[1] . '.php');
  40. return;
  41. case 3: //Controllers, Exceptions
  42. include(APP_PATH . '/' . $components[2] . 's/' . $components[1] . $components[2] . '.php');
  43. return;
  44. }
  45. } elseif (strpos($class, 'Minz') === 0) {
  46. include(LIB_PATH . '/' . str_replace('_', '/', $class) . '.php');
  47. } elseif (strpos($class, 'SimplePie') === 0) {
  48. include(LIB_PATH . '/SimplePie/' . str_replace('_', '/', $class) . '.php');
  49. }
  50. }
  51. spl_autoload_register('classAutoloader');
  52. //</Auto-loading>
  53. function idn_to_puny($url) {
  54. if (function_exists('idn_to_ascii')) {
  55. $parts = parse_url($url);
  56. if (!empty($parts['host'])) {
  57. $idn = $parts['host'];
  58. $puny = idn_to_ascii($idn);
  59. $pos = strpos($url, $idn);
  60. if ($pos !== false) {
  61. return substr_replace($url, $puny, $pos, strlen($idn));
  62. }
  63. }
  64. }
  65. return $url;
  66. }
  67. function checkUrl($url) {
  68. if ($url == '') {
  69. return '';
  70. }
  71. if (!preg_match('#^https?://#i', $url)) {
  72. $url = 'http://' . $url;
  73. }
  74. $url = idn_to_puny($url); //PHP bug #53474 IDN
  75. if (filter_var($url, FILTER_VALIDATE_URL)) {
  76. return $url;
  77. } else {
  78. return false;
  79. }
  80. }
  81. function safe_ascii($text) {
  82. return filter_var($text, FILTER_DEFAULT, FILTER_FLAG_STRIP_LOW | FILTER_FLAG_STRIP_HIGH);
  83. }
  84. /**
  85. * Test if a given server address is publicly accessible.
  86. *
  87. * Note: for the moment it tests only if address is corresponding to a
  88. * localhost address.
  89. *
  90. * @param $address the address to test, can be an IP or a URL.
  91. * @return true if server is accessible, false else.
  92. * @todo improve test with a more valid technique (e.g. test with an external server?)
  93. */
  94. function server_is_public($address) {
  95. $host = parse_url($address, PHP_URL_HOST);
  96. $is_public = !in_array($host, array(
  97. '127.0.0.1',
  98. 'localhost',
  99. 'localhost.localdomain',
  100. '[::1]',
  101. 'localhost6',
  102. 'localhost6.localdomain6',
  103. ));
  104. return $is_public;
  105. }
  106. function format_number($n, $precision = 0) {
  107. // number_format does not seem to be Unicode-compatible
  108. return str_replace(' ', ' ', //Espace fine insécable
  109. number_format($n, $precision, '.', ' ')
  110. );
  111. }
  112. function format_bytes($bytes, $precision = 2, $system = 'IEC') {
  113. if ($system === 'IEC') {
  114. $base = 1024;
  115. $units = array('B', 'KiB', 'MiB', 'GiB', 'TiB');
  116. } elseif ($system === 'SI') {
  117. $base = 1000;
  118. $units = array('B', 'KB', 'MB', 'GB', 'TB');
  119. } else {
  120. return format_number($bytes, $precision);
  121. }
  122. $bytes = max(intval($bytes), 0);
  123. $pow = $bytes === 0 ? 0 : floor(log($bytes) / log($base));
  124. $pow = min($pow, count($units) - 1);
  125. $bytes /= pow($base, $pow);
  126. return format_number($bytes, $precision) . ' ' . $units[$pow];
  127. }
  128. function timestamptodate ($t, $hour = true) {
  129. $month = _t('gen.date.' . date('M', $t));
  130. if ($hour) {
  131. $date = _t('gen.date.format_date_hour', $month);
  132. } else {
  133. $date = _t('gen.date.format_date', $month);
  134. }
  135. return @date ($date, $t);
  136. }
  137. function html_only_entity_decode($text) {
  138. static $htmlEntitiesOnly = null;
  139. if ($htmlEntitiesOnly === null) {
  140. if (version_compare(PHP_VERSION, '5.3.4') >= 0) {
  141. $htmlEntitiesOnly = array_flip(array_diff(
  142. get_html_translation_table(HTML_ENTITIES, ENT_NOQUOTES, 'UTF-8'), //Decode HTML entities
  143. get_html_translation_table(HTML_SPECIALCHARS, ENT_NOQUOTES, 'UTF-8') //Preserve XML entities
  144. ));
  145. } else {
  146. $htmlEntitiesOnly = array_map('utf8_encode', array_flip(array_diff(
  147. get_html_translation_table(HTML_ENTITIES, ENT_NOQUOTES), //Decode HTML entities
  148. get_html_translation_table(HTML_SPECIALCHARS, ENT_NOQUOTES) //Preserve XML entities
  149. )));
  150. }
  151. }
  152. return strtr($text, $htmlEntitiesOnly);
  153. }
  154. function customSimplePie() {
  155. $system_conf = Minz_Configuration::get('system');
  156. $limits = $system_conf->limits;
  157. $simplePie = new SimplePie();
  158. $simplePie->set_useragent(FRESHRSS_USERAGENT);
  159. $simplePie->set_syslog($system_conf->simplepie_syslog_enabled);
  160. $simplePie->set_cache_location(CACHE_PATH);
  161. $simplePie->set_cache_duration($limits['cache_duration']);
  162. $simplePie->set_timeout($limits['timeout']);
  163. $simplePie->set_curl_options($system_conf->curl_options);
  164. $simplePie->strip_htmltags(array(
  165. 'base', 'blink', 'body', 'doctype', 'embed',
  166. 'font', 'form', 'frame', 'frameset', 'html',
  167. 'link', 'input', 'marquee', 'meta', 'noscript',
  168. 'object', 'param', 'plaintext', 'script', 'style',
  169. ));
  170. $simplePie->strip_attributes(array_merge($simplePie->strip_attributes, array(
  171. 'autoplay', 'class', 'onload', 'onunload', 'onclick', 'ondblclick', 'onmousedown', 'onmouseup',
  172. 'onmouseover', 'onmousemove', 'onmouseout', 'onfocus', 'onblur',
  173. 'onkeypress', 'onkeydown', 'onkeyup', 'onselect', 'onchange', 'seamless', 'sizes', 'srcset')));
  174. $simplePie->add_attributes(array(
  175. 'audio' => array('controls' => 'controls', 'preload' => 'none'),
  176. 'iframe' => array('sandbox' => 'allow-scripts allow-same-origin'),
  177. 'video' => array('controls' => 'controls', 'preload' => 'none'),
  178. ));
  179. $simplePie->set_url_replacements(array(
  180. 'a' => 'href',
  181. 'area' => 'href',
  182. 'audio' => 'src',
  183. 'blockquote' => 'cite',
  184. 'del' => 'cite',
  185. 'form' => 'action',
  186. 'iframe' => 'src',
  187. 'img' => array(
  188. 'longdesc',
  189. 'src'
  190. ),
  191. 'input' => 'src',
  192. 'ins' => 'cite',
  193. 'q' => 'cite',
  194. 'source' => 'src',
  195. 'track' => 'src',
  196. 'video' => array(
  197. 'poster',
  198. 'src',
  199. ),
  200. ));
  201. $https_domains = array();
  202. $force = @file(FRESHRSS_PATH . '/force-https.default.txt', FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
  203. if (is_array($force)) {
  204. $https_domains = array_merge($https_domains, $force);
  205. }
  206. $force = @file(DATA_PATH . '/force-https.txt', FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
  207. if (is_array($force)) {
  208. $https_domains = array_merge($https_domains, $force);
  209. }
  210. $simplePie->set_https_domains($https_domains);
  211. return $simplePie;
  212. }
  213. function sanitizeHTML($data, $base = '') {
  214. static $simplePie = null;
  215. if ($simplePie == null) {
  216. $simplePie = customSimplePie();
  217. $simplePie->init();
  218. }
  219. return html_only_entity_decode($simplePie->sanitize->sanitize($data, SIMPLEPIE_CONSTRUCT_HTML, $base));
  220. }
  221. /* permet de récupérer le contenu d'un article pour un flux qui n'est pas complet */
  222. function get_content_by_parsing ($url, $path) {
  223. require_once(LIB_PATH . '/lib_phpQuery.php');
  224. Minz_Log::notice('FreshRSS GET ' . SimplePie_Misc::url_remove_credentials($url));
  225. $html = file_get_contents($url);
  226. if ($html) {
  227. $doc = phpQuery::newDocument($html);
  228. $content = $doc->find($path);
  229. foreach (pq('img[data-src]') as $img) {
  230. $imgP = pq($img);
  231. $dataSrc = $imgP->attr('data-src');
  232. if (strlen($dataSrc) > 4) {
  233. $imgP->attr('src', $dataSrc);
  234. $imgP->removeAttr('data-src');
  235. }
  236. }
  237. return sanitizeHTML($content->__toString(), $url);
  238. } else {
  239. throw new Exception();
  240. }
  241. }
  242. /**
  243. * Add support of image lazy loading
  244. * Move content from src attribute to data-original
  245. * @param content is the text we want to parse
  246. */
  247. function lazyimg($content) {
  248. return preg_replace(
  249. '/<((?:img|iframe)[^>]+?)src=[\'"]([^"\']+)[\'"]([^>]*)>/i',
  250. '<$1src="' . Minz_Url::display('/themes/icons/grey.gif') . '" data-original="$2"$3>',
  251. $content
  252. );
  253. }
  254. function uTimeString() {
  255. $t = @gettimeofday();
  256. return $t['sec'] . str_pad($t['usec'], 6, '0');
  257. }
  258. function uSecString() {
  259. $t = @gettimeofday();
  260. return str_pad($t['usec'], 6, '0');
  261. }
  262. function invalidateHttpCache($username = '') {
  263. if (!FreshRSS_user_Controller::checkUsername($username)) {
  264. Minz_Session::_param('touch', uTimeString());
  265. $username = Minz_Session::param('currentUser', '_');
  266. }
  267. return touch(join_path(DATA_PATH, 'users', $username, 'log.txt'));
  268. }
  269. function listUsers() {
  270. $final_list = array();
  271. $base_path = join_path(DATA_PATH, 'users');
  272. $dir_list = array_values(array_diff(
  273. scandir($base_path),
  274. array('..', '.', '_')
  275. ));
  276. foreach ($dir_list as $file) {
  277. if ($file[0] !== '.' && is_dir(join_path($base_path, $file)) && file_exists(join_path($base_path, $file, 'config.php'))) {
  278. $final_list[] = $file;
  279. }
  280. }
  281. return $final_list;
  282. }
  283. /**
  284. * Return if the maximum number of registrations has been reached.
  285. *
  286. * Note a max_regstrations of 0 means there is no limit.
  287. *
  288. * @return true if number of users >= max registrations, false else.
  289. */
  290. function max_registrations_reached() {
  291. $system_conf = Minz_Configuration::get('system');
  292. $limit_registrations = $system_conf->limits['max_registrations'];
  293. $number_accounts = count(listUsers());
  294. return $limit_registrations > 0 && $number_accounts >= $limit_registrations;
  295. }
  296. /**
  297. * Register and return the configuration for a given user.
  298. *
  299. * Note this function has been created to generate temporary configuration
  300. * objects. If you need a long-time configuration, please don't use this function.
  301. *
  302. * @param $username the name of the user of which we want the configuration.
  303. * @return a Minz_Configuration object, null if the configuration cannot be loaded.
  304. */
  305. function get_user_configuration($username) {
  306. if (!FreshRSS_user_Controller::checkUsername($username)) {
  307. return null;
  308. }
  309. $namespace = 'user_' . $username;
  310. try {
  311. Minz_Configuration::register($namespace,
  312. join_path(USERS_PATH, $username, 'config.php'),
  313. join_path(FRESHRSS_PATH, 'config-user.default.php'));
  314. } catch (Minz_ConfigurationNamespaceException $e) {
  315. // namespace already exists, do nothing.
  316. Minz_Log::warning($e->getMessage());
  317. } catch (Minz_FileNotExistException $e) {
  318. Minz_Log::warning($e->getMessage());
  319. return null;
  320. }
  321. return Minz_Configuration::get($namespace);
  322. }
  323. function httpAuthUser() {
  324. return isset($_SERVER['REMOTE_USER']) ? $_SERVER['REMOTE_USER'] : '';
  325. }
  326. function cryptAvailable() {
  327. try {
  328. $hash = '$2y$04$usesomesillystringfore7hnbRJHxXVLeakoG8K30oukPsA.ztMG';
  329. return $hash === @crypt('password', $hash);
  330. } catch (Exception $e) {
  331. Minz_Log::warning($e->getMessage());
  332. }
  333. return false;
  334. }
  335. function is_referer_from_same_domain() {
  336. if (empty($_SERVER['HTTP_REFERER'])) {
  337. return true; //Accept empty referer while waiting for good support of meta referrer same-origin policy in browsers
  338. }
  339. $host = parse_url(((isset($_SERVER['HTTPS']) && $_SERVER['HTTPS'] === 'on') ? 'https://' : 'http://') .
  340. (empty($_SERVER['HTTP_HOST']) ? $_SERVER['SERVER_NAME'] : $_SERVER['HTTP_HOST']));
  341. $referer = parse_url($_SERVER['HTTP_REFERER']);
  342. if (empty($host['host']) || empty($referer['host']) || $host['host'] !== $referer['host']) {
  343. return false;
  344. }
  345. //TODO: check 'scheme', taking into account the case of a proxy
  346. if ((isset($host['port']) ? $host['port'] : 0) !== (isset($referer['port']) ? $referer['port'] : 0)) {
  347. return false;
  348. }
  349. return true;
  350. }
  351. /**
  352. * Check PHP and its extensions are well-installed.
  353. *
  354. * @return array of tested values.
  355. */
  356. function check_install_php() {
  357. $pdo_mysql = extension_loaded('pdo_mysql');
  358. $pdo_sqlite = extension_loaded('pdo_sqlite');
  359. return array(
  360. 'php' => version_compare(PHP_VERSION, '5.3.8') >= 0,
  361. 'minz' => file_exists(LIB_PATH . '/Minz'),
  362. 'curl' => extension_loaded('curl'),
  363. 'pdo' => $pdo_mysql || $pdo_sqlite,
  364. 'pcre' => extension_loaded('pcre'),
  365. 'ctype' => extension_loaded('ctype'),
  366. 'fileinfo' => extension_loaded('fileinfo'),
  367. 'dom' => class_exists('DOMDocument'),
  368. 'json' => extension_loaded('json'),
  369. 'zip' => extension_loaded('zip'),
  370. );
  371. }
  372. /**
  373. * Check different data files and directories exist.
  374. *
  375. * @return array of tested values.
  376. */
  377. function check_install_files() {
  378. return array(
  379. 'data' => DATA_PATH && is_writable(DATA_PATH),
  380. 'cache' => CACHE_PATH && is_writable(CACHE_PATH),
  381. 'users' => USERS_PATH && is_writable(USERS_PATH),
  382. 'favicons' => is_writable(DATA_PATH . '/favicons'),
  383. 'tokens' => is_writable(DATA_PATH . '/tokens'),
  384. );
  385. }
  386. /**
  387. * Check database is well-installed.
  388. *
  389. * @return array of tested values.
  390. */
  391. function check_install_database() {
  392. $status = array(
  393. 'connection' => true,
  394. 'tables' => false,
  395. 'categories' => false,
  396. 'feeds' => false,
  397. 'entries' => false,
  398. );
  399. try {
  400. $dbDAO = FreshRSS_Factory::createDatabaseDAO();
  401. $status['tables'] = $dbDAO->tablesAreCorrect();
  402. $status['categories'] = $dbDAO->categoryIsCorrect();
  403. $status['feeds'] = $dbDAO->feedIsCorrect();
  404. $status['entries'] = $dbDAO->entryIsCorrect();
  405. } catch(Minz_PDOConnectionException $e) {
  406. $status['connection'] = false;
  407. }
  408. return $status;
  409. }
  410. /**
  411. * Remove a directory recursively.
  412. *
  413. * From http://php.net/rmdir#110489
  414. *
  415. * @param $dir the directory to remove
  416. */
  417. function recursive_unlink($dir) {
  418. if (!is_dir($dir)) {
  419. return true;
  420. }
  421. $files = array_diff(scandir($dir), array('.', '..'));
  422. foreach ($files as $filename) {
  423. $filename = $dir . '/' . $filename;
  424. if (is_dir($filename)) {
  425. @chmod($filename, 0777);
  426. recursive_unlink($filename);
  427. } else {
  428. unlink($filename);
  429. }
  430. }
  431. return rmdir($dir);
  432. }
  433. /**
  434. * Remove queries where $get is appearing.
  435. * @param $get the get attribute which should be removed.
  436. * @param $queries an array of queries.
  437. * @return the same array whithout those where $get is appearing.
  438. */
  439. function remove_query_by_get($get, $queries) {
  440. $final_queries = array();
  441. foreach ($queries as $key => $query) {
  442. if (empty($query['get']) || $query['get'] !== $get) {
  443. $final_queries[$key] = $query;
  444. }
  445. }
  446. return $final_queries;
  447. }
  448. /**
  449. * Add a value in an array and take care it is unique.
  450. * @param $array the array in which we add the value.
  451. * @param $value the value to add.
  452. */
  453. function array_push_unique(&$array, $value) {
  454. $found = array_search($value, $array) !== false;
  455. if (!$found) {
  456. $array[] = $value;
  457. }
  458. }
  459. /**
  460. * Remove a value from an array.
  461. * @param $array the array from wich value is removed.
  462. * @param $value the value to remove.
  463. */
  464. function array_remove(&$array, $value) {
  465. $array = array_diff($array, array($value));
  466. }
  467. //RFC 4648
  468. function base64url_encode($data) {
  469. return strtr(rtrim(base64_encode($data), '='), '+/', '-_');
  470. }
  471. //RFC 4648
  472. function base64url_decode($data) {
  473. return base64_decode(strtr($data, '-_', '+/'));
  474. }
  475. function _i($icon, $url_only = false) {
  476. return FreshRSS_Themes::icon($icon, $url_only);
  477. }