X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/84dc8897a796b91632cee1e0d736f811978022f6..b25986590ebff9cfbf6b1a5b55c1ba9dc748d210:/tweeper.php diff --git a/tweeper.php b/tweeper.php index bed6350..293595f 100644 --- a/tweeper.php +++ b/tweeper.php @@ -182,18 +182,11 @@ 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: $json_match_expr\n", E_USER_ERROR); - return NULL; - } - + private function json_to_xml($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 - $data = json_decode($matches[1], $assoc = true); + $data = json_decode($json, $assoc = true); if (!$data) { return NULL; } @@ -205,7 +198,7 @@ class Tweeper { $serializer_options = array ( 'xml_encoding' => "UTF-8", 'xml_format_output' => TRUE, - 'xml_root_node_name' => $rootName, + 'xml_root_node_name' => $root_node_name, ); $xml_data = $serializer->serialize($data, 'xml', $serializer_options); @@ -218,7 +211,15 @@ class Tweeper { } private function get_xml_instagram_com($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) {