lib_rss.php 16 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551
  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(__DIR__ . '/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(__DIR__ . '/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. if (!function_exists('mb_strcut')) {
  21. function mb_strcut($str, $start, $length = null, $encoding = 'UTF-8') {
  22. return substr($str, $start, $length);
  23. }
  24. }
  25. /**
  26. * Build a directory path by concatenating a list of directory names.
  27. *
  28. * @param $path_parts a list of directory names
  29. * @return a string corresponding to the final pathname
  30. */
  31. function join_path() {
  32. $path_parts = func_get_args();
  33. return join(DIRECTORY_SEPARATOR, $path_parts);
  34. }
  35. //<Auto-loading>
  36. function classAutoloader($class) {
  37. if (strpos($class, 'FreshRSS') === 0) {
  38. $components = explode('_', $class);
  39. switch (count($components)) {
  40. case 1:
  41. include(APP_PATH . '/' . $components[0] . '.php');
  42. return;
  43. case 2:
  44. include(APP_PATH . '/Models/' . $components[1] . '.php');
  45. return;
  46. case 3: //Controllers, Exceptions
  47. include(APP_PATH . '/' . $components[2] . 's/' . $components[1] . $components[2] . '.php');
  48. return;
  49. }
  50. } elseif (strpos($class, 'Minz') === 0) {
  51. include(LIB_PATH . '/' . str_replace('_', '/', $class) . '.php');
  52. } elseif (strpos($class, 'SimplePie') === 0) {
  53. include(LIB_PATH . '/SimplePie/' . str_replace('_', '/', $class) . '.php');
  54. }
  55. }
  56. spl_autoload_register('classAutoloader');
  57. //</Auto-loading>
  58. function idn_to_puny($url) {
  59. if (function_exists('idn_to_ascii')) {
  60. $parts = parse_url($url);
  61. if (!empty($parts['host'])) {
  62. $idn = $parts['host'];
  63. // INTL_IDNA_VARIANT_UTS46 is defined starting in PHP 5.4
  64. if (defined('INTL_IDNA_VARIANT_UTS46')) {
  65. $puny = idn_to_ascii($idn, 0, INTL_IDNA_VARIANT_UTS46);
  66. } else {
  67. $puny = idn_to_ascii($idn);
  68. }
  69. $pos = strpos($url, $idn);
  70. if ($pos !== false) {
  71. return substr_replace($url, $puny, $pos, strlen($idn));
  72. }
  73. }
  74. }
  75. return $url;
  76. }
  77. function checkUrl($url) {
  78. if ($url == '') {
  79. return '';
  80. }
  81. if (!preg_match('#^https?://#i', $url)) {
  82. $url = 'http://' . $url;
  83. }
  84. $url = idn_to_puny($url); //PHP bug #53474 IDN
  85. if (filter_var($url, FILTER_VALIDATE_URL)) {
  86. return $url;
  87. } else {
  88. return false;
  89. }
  90. }
  91. function safe_ascii($text) {
  92. return filter_var($text, FILTER_DEFAULT, FILTER_FLAG_STRIP_LOW | FILTER_FLAG_STRIP_HIGH);
  93. }
  94. function escapeToUnicodeAlternative($text, $extended = true) {
  95. $text = htmlspecialchars_decode($text, ENT_QUOTES);
  96. //Problematic characters
  97. $problem = array('&', '<', '>');
  98. //Use their fullwidth Unicode form instead:
  99. $replace = array('&', '<', '>');
  100. // https://raw.githubusercontent.com/mihaip/google-reader-api/master/wiki/StreamId.wiki
  101. if ($extended) {
  102. $problem += array("'", '"', '^', '?', '\\', '/', ',', ';');
  103. $replace += array("’", '"', '^', '?', '\', '/', ',', ';');
  104. }
  105. return trim(str_replace($problem, $replace, $text));
  106. }
  107. /**
  108. * Test if a given server address is publicly accessible.
  109. *
  110. * Note: for the moment it tests only if address is corresponding to a
  111. * localhost address.
  112. *
  113. * @param $address the address to test, can be an IP or a URL.
  114. * @return true if server is accessible, false otherwise.
  115. * @todo improve test with a more valid technique (e.g. test with an external server?)
  116. */
  117. function server_is_public($address) {
  118. $host = parse_url($address, PHP_URL_HOST);
  119. $is_public = !in_array($host, array(
  120. 'localhost',
  121. 'localhost.localdomain',
  122. '[::1]',
  123. 'ip6-localhost',
  124. 'localhost6',
  125. 'localhost6.localdomain6',
  126. ));
  127. if ($is_public) {
  128. $is_public &= !preg_match('/^(10|127|172[.]16|192[.]168)[.]/', $host);
  129. $is_public &= !preg_match('/^(\[)?(::1$|fc00::|fe80::)/i', $host);
  130. }
  131. return (bool)$is_public;
  132. }
  133. function format_number($n, $precision = 0) {
  134. // number_format does not seem to be Unicode-compatible
  135. return str_replace(' ', ' ', //Espace fine insécable
  136. number_format($n, $precision, '.', ' ')
  137. );
  138. }
  139. function format_bytes($bytes, $precision = 2, $system = 'IEC') {
  140. if ($system === 'IEC') {
  141. $base = 1024;
  142. $units = array('B', 'KiB', 'MiB', 'GiB', 'TiB');
  143. } elseif ($system === 'SI') {
  144. $base = 1000;
  145. $units = array('B', 'KB', 'MB', 'GB', 'TB');
  146. } else {
  147. return format_number($bytes, $precision);
  148. }
  149. $bytes = max(intval($bytes), 0);
  150. $pow = $bytes === 0 ? 0 : floor(log($bytes) / log($base));
  151. $pow = min($pow, count($units) - 1);
  152. $bytes /= pow($base, $pow);
  153. return format_number($bytes, $precision) . ' ' . $units[$pow];
  154. }
  155. function timestamptodate ($t, $hour = true) {
  156. $month = _t('gen.date.' . date('M', $t));
  157. if ($hour) {
  158. $date = _t('gen.date.format_date_hour', $month);
  159. } else {
  160. $date = _t('gen.date.format_date', $month);
  161. }
  162. return @date ($date, $t);
  163. }
  164. function html_only_entity_decode($text) {
  165. static $htmlEntitiesOnly = null;
  166. if ($htmlEntitiesOnly === null) {
  167. if (version_compare(PHP_VERSION, '5.3.4') >= 0) {
  168. $htmlEntitiesOnly = array_flip(array_diff(
  169. get_html_translation_table(HTML_ENTITIES, ENT_NOQUOTES, 'UTF-8'), //Decode HTML entities
  170. get_html_translation_table(HTML_SPECIALCHARS, ENT_NOQUOTES, 'UTF-8') //Preserve XML entities
  171. ));
  172. } else {
  173. $htmlEntitiesOnly = array_map('utf8_encode', array_flip(array_diff(
  174. get_html_translation_table(HTML_ENTITIES, ENT_NOQUOTES), //Decode HTML entities
  175. get_html_translation_table(HTML_SPECIALCHARS, ENT_NOQUOTES) //Preserve XML entities
  176. )));
  177. }
  178. }
  179. return strtr($text, $htmlEntitiesOnly);
  180. }
  181. function prepareSyslog() {
  182. return openlog("FreshRSS", LOG_PERROR | LOG_PID, LOG_USER);
  183. }
  184. function customSimplePie($attributes = array()) {
  185. $system_conf = Minz_Configuration::get('system');
  186. $limits = $system_conf->limits;
  187. $simplePie = new SimplePie();
  188. $simplePie->set_useragent(FRESHRSS_USERAGENT);
  189. $simplePie->set_syslog($system_conf->simplepie_syslog_enabled);
  190. if ($system_conf->simplepie_syslog_enabled) {
  191. prepareSyslog();
  192. }
  193. $simplePie->set_cache_location(CACHE_PATH);
  194. $simplePie->set_cache_duration($limits['cache_duration']);
  195. $feed_timeout = empty($attributes['timeout']) ? 0 : intval($attributes['timeout']);
  196. $simplePie->set_timeout($feed_timeout > 0 ? $feed_timeout : $limits['timeout']);
  197. $curl_options = $system_conf->curl_options;
  198. if (isset($attributes['ssl_verify'])) {
  199. $curl_options[CURLOPT_SSL_VERIFYHOST] = $attributes['ssl_verify'] ? 2 : 0;
  200. $curl_options[CURLOPT_SSL_VERIFYPEER] = $attributes['ssl_verify'] ? true : false;
  201. }
  202. $simplePie->set_curl_options($curl_options);
  203. $simplePie->strip_htmltags(array(
  204. 'base', 'blink', 'body', 'doctype', 'embed',
  205. 'font', 'form', 'frame', 'frameset', 'html',
  206. 'link', 'input', 'marquee', 'meta', 'noscript',
  207. 'object', 'param', 'plaintext', 'script', 'style',
  208. 'svg', //TODO: Support SVG after sanitizing and URL rewriting of xlink:href
  209. ));
  210. $simplePie->strip_attributes(array_merge($simplePie->strip_attributes, array(
  211. 'autoplay', 'class', 'onload', 'onunload', 'onclick', 'ondblclick', 'onmousedown', 'onmouseup',
  212. 'onmouseover', 'onmousemove', 'onmouseout', 'onfocus', 'onblur',
  213. 'onkeypress', 'onkeydown', 'onkeyup', 'onselect', 'onchange', 'seamless', 'sizes', 'srcset')));
  214. $simplePie->add_attributes(array(
  215. 'audio' => array('controls' => 'controls', 'preload' => 'none'),
  216. 'iframe' => array('sandbox' => 'allow-scripts allow-same-origin'),
  217. 'video' => array('controls' => 'controls', 'preload' => 'none'),
  218. ));
  219. $simplePie->set_url_replacements(array(
  220. 'a' => 'href',
  221. 'area' => 'href',
  222. 'audio' => 'src',
  223. 'blockquote' => 'cite',
  224. 'del' => 'cite',
  225. 'form' => 'action',
  226. 'iframe' => 'src',
  227. 'img' => array(
  228. 'longdesc',
  229. 'src'
  230. ),
  231. 'input' => 'src',
  232. 'ins' => 'cite',
  233. 'q' => 'cite',
  234. 'source' => 'src',
  235. 'track' => 'src',
  236. 'video' => array(
  237. 'poster',
  238. 'src',
  239. ),
  240. ));
  241. $https_domains = array();
  242. $force = @file(FRESHRSS_PATH . '/force-https.default.txt', FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
  243. if (is_array($force)) {
  244. $https_domains = array_merge($https_domains, $force);
  245. }
  246. $force = @file(DATA_PATH . '/force-https.txt', FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
  247. if (is_array($force)) {
  248. $https_domains = array_merge($https_domains, $force);
  249. }
  250. $simplePie->set_https_domains($https_domains);
  251. return $simplePie;
  252. }
  253. function sanitizeHTML($data, $base = '') {
  254. static $simplePie = null;
  255. if ($simplePie == null) {
  256. $simplePie = customSimplePie();
  257. $simplePie->init();
  258. }
  259. return html_only_entity_decode($simplePie->sanitize->sanitize($data, SIMPLEPIE_CONSTRUCT_HTML, $base));
  260. }
  261. /**
  262. * Add support of image lazy loading
  263. * Move content from src attribute to data-original
  264. * @param content is the text we want to parse
  265. */
  266. function lazyimg($content) {
  267. return preg_replace(
  268. '/<((?:img|iframe)[^>]+?)src=[\'"]([^"\']+)[\'"]([^>]*)>/i',
  269. '<$1src="' . Minz_Url::display('/themes/icons/grey.gif') . '" data-original="$2"$3>',
  270. $content
  271. );
  272. }
  273. function uTimeString() {
  274. $t = @gettimeofday();
  275. return $t['sec'] . str_pad($t['usec'], 6, '0');
  276. }
  277. function uSecString() {
  278. $t = @gettimeofday();
  279. return str_pad($t['usec'], 6, '0');
  280. }
  281. function invalidateHttpCache($username = '') {
  282. if (!FreshRSS_user_Controller::checkUsername($username)) {
  283. Minz_Session::_param('touch', uTimeString());
  284. $username = Minz_Session::param('currentUser', '_');
  285. }
  286. return touch(join_path(DATA_PATH, 'users', $username, 'log.txt'));
  287. }
  288. function listUsers() {
  289. $final_list = array();
  290. $base_path = join_path(DATA_PATH, 'users');
  291. $dir_list = array_values(array_diff(
  292. scandir($base_path),
  293. array('..', '.', '_')
  294. ));
  295. foreach ($dir_list as $file) {
  296. if ($file[0] !== '.' && is_dir(join_path($base_path, $file)) && file_exists(join_path($base_path, $file, 'config.php'))) {
  297. $final_list[] = $file;
  298. }
  299. }
  300. return $final_list;
  301. }
  302. /**
  303. * Return if the maximum number of registrations has been reached.
  304. *
  305. * Note a max_regstrations of 0 means there is no limit.
  306. *
  307. * @return true if number of users >= max registrations, false else.
  308. */
  309. function max_registrations_reached() {
  310. $system_conf = Minz_Configuration::get('system');
  311. $limit_registrations = $system_conf->limits['max_registrations'];
  312. $number_accounts = count(listUsers());
  313. return $limit_registrations > 0 && $number_accounts >= $limit_registrations;
  314. }
  315. /**
  316. * Register and return the configuration for a given user.
  317. *
  318. * Note this function has been created to generate temporary configuration
  319. * objects. If you need a long-time configuration, please don't use this function.
  320. *
  321. * @param $username the name of the user of which we want the configuration.
  322. * @return a Minz_Configuration object, null if the configuration cannot be loaded.
  323. */
  324. function get_user_configuration($username) {
  325. if (!FreshRSS_user_Controller::checkUsername($username)) {
  326. return null;
  327. }
  328. $namespace = 'user_' . $username;
  329. try {
  330. Minz_Configuration::register($namespace,
  331. join_path(USERS_PATH, $username, 'config.php'),
  332. join_path(FRESHRSS_PATH, 'config-user.default.php'));
  333. } catch (Minz_ConfigurationNamespaceException $e) {
  334. // namespace already exists, do nothing.
  335. Minz_Log::warning($e->getMessage(), USERS_PATH . '/_/log.txt');
  336. } catch (Minz_FileNotExistException $e) {
  337. Minz_Log::warning($e->getMessage(), USERS_PATH . '/_/log.txt');
  338. return null;
  339. }
  340. return Minz_Configuration::get($namespace);
  341. }
  342. function httpAuthUser() {
  343. if (!empty($_SERVER['REMOTE_USER'])) {
  344. return $_SERVER['REMOTE_USER'];
  345. } elseif (!empty($_SERVER['REDIRECT_REMOTE_USER'])) {
  346. return $_SERVER['REDIRECT_REMOTE_USER'];
  347. } elseif (!empty($_SERVER['HTTP_X_WEBAUTH_USER'])) {
  348. return $_SERVER['HTTP_X_WEBAUTH_USER'];
  349. }
  350. return '';
  351. }
  352. function cryptAvailable() {
  353. try {
  354. $hash = '$2y$04$usesomesillystringfore7hnbRJHxXVLeakoG8K30oukPsA.ztMG';
  355. return $hash === @crypt('password', $hash);
  356. } catch (Exception $e) {
  357. Minz_Log::warning($e->getMessage());
  358. }
  359. return false;
  360. }
  361. function is_referer_from_same_domain() {
  362. if (empty($_SERVER['HTTP_REFERER'])) {
  363. return true; //Accept empty referer while waiting for good support of meta referrer same-origin policy in browsers
  364. }
  365. $host = parse_url(((isset($_SERVER['HTTPS']) && $_SERVER['HTTPS'] === 'on') ? 'https://' : 'http://') .
  366. (empty($_SERVER['HTTP_HOST']) ? $_SERVER['SERVER_NAME'] : $_SERVER['HTTP_HOST']));
  367. $referer = parse_url($_SERVER['HTTP_REFERER']);
  368. if (empty($host['host']) || empty($referer['host']) || $host['host'] !== $referer['host']) {
  369. return false;
  370. }
  371. //TODO: check 'scheme', taking into account the case of a proxy
  372. if ((isset($host['port']) ? $host['port'] : 0) !== (isset($referer['port']) ? $referer['port'] : 0)) {
  373. return false;
  374. }
  375. return true;
  376. }
  377. /**
  378. * Check PHP and its extensions are well-installed.
  379. *
  380. * @return array of tested values.
  381. */
  382. function check_install_php() {
  383. $pdo_mysql = extension_loaded('pdo_mysql');
  384. $pdo_pgsql = extension_loaded('pdo_pgsql');
  385. $pdo_sqlite = extension_loaded('pdo_sqlite');
  386. return array(
  387. 'php' => version_compare(PHP_VERSION, '5.3.8') >= 0,
  388. 'minz' => file_exists(LIB_PATH . '/Minz'),
  389. 'curl' => extension_loaded('curl'),
  390. 'pdo' => $pdo_mysql || $pdo_sqlite || $pdo_pgsql,
  391. 'pcre' => extension_loaded('pcre'),
  392. 'ctype' => extension_loaded('ctype'),
  393. 'fileinfo' => extension_loaded('fileinfo'),
  394. 'dom' => class_exists('DOMDocument'),
  395. 'json' => extension_loaded('json'),
  396. 'mbstring' => extension_loaded('mbstring'),
  397. 'zip' => extension_loaded('zip'),
  398. );
  399. }
  400. /**
  401. * Check different data files and directories exist.
  402. *
  403. * @return array of tested values.
  404. */
  405. function check_install_files() {
  406. return array(
  407. 'data' => DATA_PATH && is_writable(DATA_PATH),
  408. 'cache' => CACHE_PATH && is_writable(CACHE_PATH),
  409. 'users' => USERS_PATH && is_writable(USERS_PATH),
  410. 'favicons' => is_writable(DATA_PATH . '/favicons'),
  411. 'tokens' => is_writable(DATA_PATH . '/tokens'),
  412. );
  413. }
  414. /**
  415. * Check database is well-installed.
  416. *
  417. * @return array of tested values.
  418. */
  419. function check_install_database() {
  420. $status = array(
  421. 'connection' => true,
  422. 'tables' => false,
  423. 'categories' => false,
  424. 'feeds' => false,
  425. 'entries' => false,
  426. 'entrytmp' => false,
  427. 'tag' => false,
  428. 'entrytag' => false,
  429. );
  430. try {
  431. $dbDAO = FreshRSS_Factory::createDatabaseDAO();
  432. $status['tables'] = $dbDAO->tablesAreCorrect();
  433. $status['categories'] = $dbDAO->categoryIsCorrect();
  434. $status['feeds'] = $dbDAO->feedIsCorrect();
  435. $status['entries'] = $dbDAO->entryIsCorrect();
  436. $status['entrytmp'] = $dbDAO->entrytmpIsCorrect();
  437. $status['tag'] = $dbDAO->tagIsCorrect();
  438. $status['entrytag'] = $dbDAO->entrytagIsCorrect();
  439. } catch(Minz_PDOConnectionException $e) {
  440. $status['connection'] = false;
  441. }
  442. return $status;
  443. }
  444. /**
  445. * Remove a directory recursively.
  446. *
  447. * From http://php.net/rmdir#110489
  448. *
  449. * @param $dir the directory to remove
  450. */
  451. function recursive_unlink($dir) {
  452. if (!is_dir($dir)) {
  453. return true;
  454. }
  455. $files = array_diff(scandir($dir), array('.', '..'));
  456. foreach ($files as $filename) {
  457. $filename = $dir . '/' . $filename;
  458. if (is_dir($filename)) {
  459. @chmod($filename, 0777);
  460. recursive_unlink($filename);
  461. } else {
  462. unlink($filename);
  463. }
  464. }
  465. return rmdir($dir);
  466. }
  467. /**
  468. * Remove queries where $get is appearing.
  469. * @param $get the get attribute which should be removed.
  470. * @param $queries an array of queries.
  471. * @return the same array whithout those where $get is appearing.
  472. */
  473. function remove_query_by_get($get, $queries) {
  474. $final_queries = array();
  475. foreach ($queries as $key => $query) {
  476. if (empty($query['get']) || $query['get'] !== $get) {
  477. $final_queries[$key] = $query;
  478. }
  479. }
  480. return $final_queries;
  481. }
  482. //RFC 4648
  483. function base64url_encode($data) {
  484. return strtr(rtrim(base64_encode($data), '='), '+/', '-_');
  485. }
  486. //RFC 4648
  487. function base64url_decode($data) {
  488. return base64_decode(strtr($data, '-_', '+/'));
  489. }
  490. function _i($icon, $url_only = false) {
  491. return FreshRSS_Themes::icon($icon, $url_only);
  492. }