+ return($complete);
+}
+
+function parseurl_getsiteinfo_cached($url, $no_guessing = false, $do_oembed = true) {
+
+ $data = Cache::get("parse_url:".$no_guessing.":".$do_oembed.":".$url);
+ if (!is_null($data)) {
+ $data = unserialize($data);
+ return $data;
+ }
+
+ $data = parseurl_getsiteinfo($url, $no_guessing, $do_oembed);
+
+ Cache::set("parse_url:".$no_guessing.":".$do_oembed.":".$url,serialize($data), CACHE_DAY);
+
+ return $data;
}
function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1) {
}
function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1) {
$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_HEADER, 1);
$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_HEADER, 1);
if ((($curl_info['http_code'] == "301") OR ($curl_info['http_code'] == "302") OR ($curl_info['http_code'] == "303") OR ($curl_info['http_code'] == "307"))
AND (($curl_info['redirect_url'] != "") OR ($curl_info['location'] != ""))) {
if ($curl_info['redirect_url'] != "")
if ((($curl_info['http_code'] == "301") OR ($curl_info['http_code'] == "302") OR ($curl_info['http_code'] == "303") OR ($curl_info['http_code'] == "307"))
AND (($curl_info['redirect_url'] != "") OR ($curl_info['location'] != ""))) {
if ($curl_info['redirect_url'] != "")
+ // if the file is too large then exit
+ if ($curl_info["download_content_length"] > 1000000)
+ return($siteinfo);
+
+ // if it isn't a HTML file then exit
+ if (($curl_info["content_type"] != "") AND !strstr(strtolower($curl_info["content_type"]),"html"))
+ return($siteinfo);
+
- // if the file is too large then exit
- if ($curl_info["download_content_length"] > 1000000)
- return($siteinfo);
+ if (($oembed_data->type == "link") AND ($siteinfo["type"] != "photo")) {
+ if (isset($oembed_data->title))
+ $siteinfo["title"] = $oembed_data->title;
+ if (isset($oembed_data->description))
+ $siteinfo["text"] = trim($oembed_data->description);
+ if (isset($oembed_data->thumbnail_url))
+ $siteinfo["image"] = $oembed_data->thumbnail_url;
+ }
+ }
// Fetch the first mentioned charset. Can be in body or header
$charset = "";
if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
// Fetch the first mentioned charset. Can be in body or header
$charset = "";
if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
- foreach ($list as $node) {
- $attr = array();
- if ($node->attributes->length)
- foreach ($node->attributes as $attribute)
- $attr[$attribute->name] = $attribute->value;
-
- if (@$attr["http-equiv"] == 'refresh') {
- $path = $attr["content"];
- $pathinfo = explode(";", $path);
- $content = "";
- foreach ($pathinfo AS $value) {
- if (substr(strtolower($value), 0, 4) == "url=")
- $content = substr($value, 4);
- }
- if ($content != "") {
- $siteinfo = parseurl_getsiteinfo($content, $no_guessing, $do_oembed, ++$count);
- return($siteinfo);
- }
- }
+ foreach ($list as $node) {
+ $attr = array();
+ if ($node->attributes->length)
+ foreach ($node->attributes as $attribute)
+ $attr[$attribute->name] = $attribute->value;
+
+ if (@$attr["http-equiv"] == 'refresh') {
+ $path = $attr["content"];
+ $pathinfo = explode(";", $path);
+ $content = "";
+ foreach ($pathinfo AS $value) {
+ if (substr(strtolower($value), 0, 4) == "url=")
+ $content = substr($value, 4);
+ }
+ if ($content != "") {
+ $siteinfo = parseurl_getsiteinfo($content, $no_guessing, $do_oembed, ++$count);
+ return($siteinfo);
+ }
+ }
- if (isset($oembed_data) AND ($oembed_data->type == "link") AND ($siteinfo["type"] != "photo")) {
- if (isset($oembed_data->title) AND (trim($oembed_data->title) != ""))
- $siteinfo["title"] = $oembed_data->title;
- if (isset($oembed_data->description) AND (trim($oembed_data->description) != ""))
- $siteinfo["text"] = trim($oembed_data->description);
- if (isset($oembed_data->thumbnail_url) AND (trim($oembed_data->thumbnail_url) != ""))
- $siteinfo["image"] = $oembed_data->thumbnail_url;
- }
-
- $list = $xpath->query("//img[@src]");
- foreach ($list as $node) {
- $attr = array();
- if ($node->attributes->length)
- foreach ($node->attributes as $attribute)
- $attr[$attribute->name] = $attribute->value;
+ $list = $xpath->query("//img[@src]");
+ foreach ($list as $node) {
+ $attr = array();
+ if ($node->attributes->length)
+ foreach ($node->attributes as $attribute)
+ $attr[$attribute->name] = $attribute->value;
$src = completeurl($attr["src"], $url);
$photodata = @getimagesize($src);
$src = completeurl($attr["src"], $url);
$photodata = @getimagesize($src);
+ // add url scheme if missing
+ $arrurl = parse_url($url);
+ if (!x($arrurl, 'scheme')) {
+ if (x($arrurl, 'host'))
+ $url = "http:".$url;
+ else
+ $url = "http://".$url;
+ }
+
- if($siteinfo["title"] == "") {
- $sitedata .= sprintf($template,$url,$url,'') . $str_tags;
- killme();
- } else {
+ if($siteinfo["title"] != "") {