X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/4c4771f61019b68bdf3d6896995fe9fb24b5fd45..727664e9cb005bb788b455ec637bc3aa0857d382:/tweeper.php?ds=sidebyside diff --git a/tweeper.php b/tweeper.php index 37b73f5..d534a10 100644 --- a/tweeper.php +++ b/tweeper.php @@ -18,7 +18,11 @@ * along with this program. If not, see . */ -require_once 'XML/Serializer.php'; +require_once 'Symfony/Component/Serializer/autoload.php'; + +use Symfony\Component\Serializer\Serializer; +use Symfony\Component\Serializer\Encoder\XmlEncoder; +use Symfony\Component\Serializer\Normalizer\ObjectNormalizer; date_default_timezone_set('UTC'); @@ -30,19 +34,24 @@ class Tweeper { $this->generate_enclosure = $generate_enclosure; } - public static function epoch_to_gmdate($timestamp) - { + public static function epoch_to_gmdate($timestamp) { + if (!is_numeric($timestamp) || is_nan($timestamp)) { + $timestamp = 0; + } + return gmdate('D, d M Y H:i:s', $timestamp) . ' GMT'; } - public static function str_to_gmdate($date) - { + public static function str_to_gmdate($date) { $timestamp = strtotime($date); + if (FALSE === $timestamp) { + $timestamp = 0; + } + return Tweeper::epoch_to_gmdate($timestamp); } - private static function get_contents($url) - { + private static function get_contents($url) { $ch = curl_init($url); curl_setopt_array($ch, array( CURLOPT_HEADER => FALSE, @@ -59,8 +68,7 @@ class Tweeper { return $contents; } - private static function get_info($url) - { + private static function get_info($url) { $ch = curl_init($url); curl_setopt_array($ch, array( CURLOPT_HEADER => TRUE, @@ -78,8 +86,7 @@ class Tweeper { return $url_info; } - public static function generate_enclosure($url) - { + public static function generate_enclosure($url) { $supported_content_types = array( "application/ogg", "audio/aac", @@ -153,7 +160,7 @@ class Tweeper { private function load_stylesheet($host) { $stylesheet = "file://" . __DIR__ . "/rss_converter_" . $host . ".xsl"; if (FALSE === file_exists($stylesheet)) { - trigger_error("Conversion to RSS not supported: $host", E_USER_ERROR); + trigger_error("Conversion to RSS not supported for $host ($stylesheet not found)", E_USER_ERROR); return NULL; } @@ -170,39 +177,50 @@ class Tweeper { return $xsltProcessor; } - private function json_to_xml($html, $json_match_expr, $rootName) { - // pre-process, convert json to XML - $ret = preg_match($json_match_expr, $html, $matches); - if ($ret !== 1) { - trigger_error("Cannot match expression: $json_match_expr\n", E_USER_ERROR); - return NULL; - } - - $data = json_decode($matches[1]); + private function json_to_xml($json, $root_node_name) { + // Apparenty the ObjectNormalizer used afterwards is not able to handle + // the stdClass object created by json_decode() with the default setting + // $assoc = false; so use $assoc = true + $data = json_decode($json, $assoc = true); if (!$data) { return NULL; } + $encoder = new XmlEncoder(); + $normalizer = new ObjectNormalizer(); + $serializer = new Serializer(array($normalizer), array($encoder)); + $serializer_options = array ( - 'addDecl' => TRUE, - 'encoding' => "UTF-8", - 'indent' => ' ', - 'rootName' => $rootName, + 'xml_encoding' => "UTF-8", + 'xml_format_output' => TRUE, + 'xml_root_node_name' => $root_node_name, ); - $serializer = new XML_Serializer($serializer_options); - - $status = $serializer->serialize($data); - if (PEAR::isError($status)) { - trigger_error($status->getMessage(), E_USER_ERROR); + $xml_data = $serializer->serialize($data, 'xml', $serializer_options); + if (!$xml_data) { + trigger_error("Cannot serialize data", E_USER_ERROR); return NULL; } - return $serializer->getSerializedData(); + return $xml_data; } private function get_xml_instagram_com($html) { - return $this->json_to_xml($html, '/window._sharedData = (.*);/', 'instagram'); + // extract the json data from the html code + $json_match_expr = '/window._sharedData = (.*);/'; + $ret = preg_match($json_match_expr, $html, $matches); + if ($ret !== 1) { + trigger_error("Cannot match expression: $json_match_expr\n", E_USER_ERROR); + return NULL; + } + + return $this->json_to_xml($matches[1], 'instagram'); + } + + private function preprocess_html_facebook_com($html) { + $html = str_replace('', '', $html); + return $html; } private function html_to_xml($html, $host) { @@ -236,7 +254,10 @@ class Tweeper { return NULL; } - $xsltProcessor = $this->load_stylesheet($url["host"]); + // Strip the leading www. to be more forgiving on input URLs + $host = preg_replace('/^www\./', '', $url["host"]); + + $xsltProcessor = $this->load_stylesheet($host); if (NULL === $xsltProcessor) { return NULL; } @@ -246,7 +267,12 @@ class Tweeper { return NULL; } - $xmlDoc = $this->html_to_xml($html, $url["host"]); + $preprocess_html_host_method = 'preprocess_html_' . str_replace(".", "_", $host); + if (method_exists($this, $preprocess_html_host_method)) { + $html = call_user_func_array(array($this, $preprocess_html_host_method), array($html)); + } + + $xmlDoc = $this->html_to_xml($html, $host); if (NULL === $xmlDoc) { return NULL; } @@ -261,13 +287,11 @@ class Tweeper { } } -function is_cli() -{ - return (substr(php_sapi_name(), 0, 3) == 'cli'); +function is_cli() { + return (php_sapi_name() === "cli"); } -function usage($argv) -{ +function usage($argv) { if (is_cli()) { $usage = "{$argv[0]} [-e|-h|--help] \n"; } else { @@ -277,8 +301,7 @@ function usage($argv) return "usage: $usage"; } -function parse_options_cli($argv, $argc) -{ +function parse_options_cli($argv, $argc) { $options = array( 'generate_enclosure' => FALSE ); @@ -307,8 +330,7 @@ function parse_options_cli($argv, $argc) return $options; } -function parse_options_query_string() -{ +function parse_options_query_string() { $options = array( 'generate_enclosure' => FALSE ); @@ -332,7 +354,7 @@ if (is_cli()) { } if (!isset($options['src_url'])) { - fwrite($ERROR_STREAM, usage($argv)); + fwrite($ERROR_STREAM, usage(is_cli() ? $argv : NULL)); exit(1); }