}
function completeurl($url, $scheme) {
- $urlarr = parse_url($url);
+ $urlarr = parse_url($url);
- if (isset($urlarr["scheme"]))
- return($url);
+ if (isset($urlarr["scheme"]))
+ return($url);
- $schemearr = parse_url($scheme);
+ $schemearr = parse_url($scheme);
- $complete = $schemearr["scheme"]."://".$schemearr["host"];
+ $complete = $schemearr["scheme"]."://".$schemearr["host"];
- if (@$schemearr["port"] != "")
- $complete .= ":".$schemearr["port"];
+ if (@$schemearr["port"] != "")
+ $complete .= ":".$schemearr["port"];
if(strpos($urlarr['path'],'/') !== 0)
$complete .= '/';
- $complete .= $urlarr["path"];
+ $complete .= $urlarr["path"];
- if (@$urlarr["query"] != "")
- $complete .= "?".$urlarr["query"];
+ if (@$urlarr["query"] != "")
+ $complete .= "?".$urlarr["query"];
- if (@$urlarr["fragment"] != "")
- $complete .= "#".$urlarr["fragment"];
+ if (@$urlarr["fragment"] != "")
+ $complete .= "#".$urlarr["fragment"];
- return($complete);
+ return($complete);
}
function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1) {
$siteinfo["url"] = $url;
$siteinfo["type"] = "link";
+ $stamp1 = microtime(true);
+
$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_HEADER, 1);
$header = curl_exec($ch);
$curl_info = @curl_getinfo($ch);
- $http_code = $curl_info['http_code'];
+ $http_code = $curl_info['http_code'];
curl_close($ch);
+ $a->save_timestamp($stamp1, "network");
+
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["content_type"] != "") AND !strstr(strtolower($curl_info["content_type"]),"html"))
return($siteinfo);
+ $stamp1 = microtime(true);
+
// Now fetch the body as well
$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $url);
$header = curl_exec($ch);
$curl_info = @curl_getinfo($ch);
- $http_code = $curl_info['http_code'];
+ $http_code = $curl_info['http_code'];
curl_close($ch);
+ $a->save_timestamp($stamp1, "network");
+
// Fetch the first mentioned charset. Can be in body or header
$charset = "";
if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
$xpath = new DomXPath($doc);
$list = $xpath->query("//meta[@content]");
- foreach ($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($siteinfo);
+ }
+ }
}
//$list = $xpath->query("head/title");
foreach ($list as $node) {
$attr = array();
if ($node->attributes->length)
- foreach ($node->attributes as $attribute)
- $attr[$attribute->name] = $attribute->value;
+ foreach ($node->attributes as $attribute)
+ $attr[$attribute->name] = $attribute->value;
$attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
foreach ($list as $node) {
$attr = array();
if ($node->attributes->length)
- foreach ($node->attributes as $attribute)
- $attr[$attribute->name] = $attribute->value;
+ foreach ($node->attributes as $attribute)
+ $attr[$attribute->name] = $attribute->value;
$attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
}
if ((@$siteinfo["image"] == "") AND !$no_guessing) {
- $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);
"height"=>$photodata[1]);
}
- }
+ }
} else {
$src = completeurl($siteinfo["image"], $url);