X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/4336e714ebb2cf1b07c59a4534b9d9acd244c6f9..1e7a09621f810d54f9badc8bcf3f01a6c6ae2347:/src/Tweeper.php diff --git a/src/Tweeper.php b/src/Tweeper.php index 7318900..e98623b 100644 --- a/src/Tweeper.php +++ b/src/Tweeper.php @@ -36,7 +36,9 @@ date_default_timezone_set('UTC'); */ class Tweeper { - private static $userAgent = "Mozilla/5.0 (Windows NT 6.1; WOW64; rv:60.0) Gecko/20100101 Firefox/60.0"; + private static $userAgent = "Mozilla/5.0"; + private static $maxConnectionTimeout = 5; + private static $maxConnectionRetries = 5; /** * Create a new Tweeper object controlling optional settings. @@ -47,10 +49,18 @@ class Tweeper { * Enables showing the username in front of the content for multi-user * sites (enabled by default). Only some stylesheets supports this * functionality (twitter, instagram, pump.io). + * @param bool $show_multimedia + * Enables showing multimedia content (images, videos) directly in the + * item description (enabled by default). Only some stylesheets supports + * this functionality (twitter, instagram, dilbert). + * @param bool $verbose_output + * Enables showing non-fatal errors like XML parsing errors. */ - public function __construct($generate_enclosure = FALSE, $show_usernames = TRUE) { + public function __construct($generate_enclosure = FALSE, $show_usernames = TRUE, $show_multimedia = TRUE, $verbose_output = TRUE) { $this->generate_enclosure = $generate_enclosure; $this->show_usernames = $show_usernames; + $this->show_multimedia = $show_multimedia; + $this->verbose_output = $verbose_output; } /** @@ -86,24 +96,40 @@ class Tweeper { } /** + * Perform a cURL session multiple times when it fails with a timeout. + * + * @param resource $ch + * a cURL session handle. + */ + private static function curlExec($ch) { + $ret = FALSE; + $attempt = 0; + do { + $ret = curl_exec($ch); + if (FALSE === $ret) { + trigger_error(curl_error($ch), E_USER_WARNING); + } + } while (curl_errno($ch) == CURLE_OPERATION_TIMEDOUT && ++$attempt < Tweeper::$maxConnectionRetries); + + return $ret; + } + + /** * Get the contents from a URL. */ private static function getUrlContents($url) { $ch = curl_init($url); curl_setopt_array($ch, array( CURLOPT_HEADER => FALSE, + CURLOPT_CONNECTTIMEOUT => Tweeper::$maxConnectionTimeout, // Follow http redirects to get the real URL. CURLOPT_FOLLOWLOCATION => TRUE, + CURLOPT_COOKIEFILE => "", CURLOPT_RETURNTRANSFER => TRUE, - CURLOPT_SSL_VERIFYHOST => FALSE, - CURLOPT_SSL_VERIFYPEER => FALSE, CURLOPT_HTTPHEADER => array('Accept-language: en'), CURLOPT_USERAGENT => Tweeper::$userAgent, )); - $contents = curl_exec($ch); - if (FALSE === $contents) { - trigger_error(curl_error($ch), E_USER_WARNING); - } + $contents = Tweeper::curlExec($ch); curl_close($ch); return $contents; @@ -117,17 +143,15 @@ class Tweeper { curl_setopt_array($ch, array( CURLOPT_HEADER => TRUE, CURLOPT_NOBODY => TRUE, + CURLOPT_CONNECTTIMEOUT => Tweeper::$maxConnectionTimeout, // Follow http redirects to get the real URL. CURLOPT_FOLLOWLOCATION => TRUE, CURLOPT_RETURNTRANSFER => TRUE, - CURLOPT_SSL_VERIFYHOST => FALSE, - CURLOPT_SSL_VERIFYPEER => FALSE, CURLOPT_USERAGENT => Tweeper::$userAgent, )); - $ret = curl_exec($ch); + $ret = Tweeper::curlExec($ch); if (FALSE === $ret) { - trigger_error(curl_error($ch), E_USER_WARNING); curl_close($ch); return FALSE; } @@ -182,10 +206,17 @@ class Tweeper { // See http://sourceforge.net/p/feedvalidator/bugs/72/ $http_url = preg_replace("/^https/", "http", $url_info['url']); + // When the server does not provide a Content-Length header, + // curl_getinfo() would return a negative value for + // "download_content_length", however RSS recommends to use 0 when the + // enclosure's size cannot be determined. + // See: https://www.feedvalidator.org/docs/error/UseZeroForUnknown.html + $length = max($url_info['download_content_length'], 0); + $dom = new DOMDocument(); $enc = $dom->createElement('enclosure'); $enc->setAttribute('url', $http_url); - $enc->setAttribute('length', $url_info['download_content_length']); + $enc->setAttribute('length', $length); $enc->setAttribute('type', $url_info['content_type']); return $enc; @@ -311,8 +342,10 @@ class Tweeper { $xmlDoc->loadHTML($html); } - foreach (libxml_get_errors() as $xml_error) { - Tweeper::logXmlError($xml_error); + if ($this->verbose_output) { + foreach (libxml_get_errors() as $xml_error) { + Tweeper::logXmlError($xml_error); + } } libxml_clear_errors(); libxml_use_internal_errors($xml_errors_value); @@ -332,6 +365,7 @@ class Tweeper { $stylesheet_contents = Tweeper::getUrlContents($stylesheet); if (FALSE === $stylesheet_contents) { + trigger_error("Cannot open $stylesheet", E_USER_WARNING); return NULL; } @@ -342,6 +376,7 @@ class Tweeper { $xsltProcessor->registerPHPFunctions(); $xsltProcessor->setParameter('', 'generate-enclosure', $this->generate_enclosure); $xsltProcessor->setParameter('', 'show-usernames', $this->show_usernames); + $xsltProcessor->setParameter('', 'show-multimedia', $this->show_multimedia); $xsltProcessor->importStylesheet($xslDoc); return $xsltProcessor; @@ -350,7 +385,7 @@ class Tweeper { /** * Convert the site content to RSS. */ - public function tweep($src_url, $host=NULL, $validate_scheme=TRUE) { + public function tweep($src_url, $host = NULL, $validate_scheme = TRUE) { $url = parse_url($src_url); if (FALSE === $url) { trigger_error("Invalid URL: $src_url", E_USER_WARNING); @@ -365,7 +400,7 @@ class Tweeper { } } - // if the host is not given derive it from the URL + // If the host is not given derive it from the URL. if (NULL === $host) { if (empty($url["host"])) { trigger_error("Invalid host in URL: $src_url", E_USER_WARNING); @@ -382,6 +417,7 @@ class Tweeper { $html = Tweeper::getUrlContents($src_url); if (FALSE === $html) { + trigger_error("Failed to retrieve $src_url", E_USER_WARNING); return NULL; }