photos-albms widget: add some classes

[friendica.git] / mod / parse_url.php

diff --git a/mod/parse_url.php b/mod/parse_url.php
index 9df35c76125c652e8bfa9f34b6ca8f85da93a590..ef051d9f6b433c0e8152c39168c0debd8cfbee0a 100644 (file)
--- a/mod/parse_url.php
+++ b/mod/parse_url.php
@@ -24,30 +24,45 @@ if(!function_exists('deletenode')) {
 }
 
 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_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, $count = 1) {
@@ -70,6 +85,8 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
        $siteinfo["url"] = $url;
        $siteinfo["type"] = "link";
 
+       $stamp1 = microtime(true);
+
        $ch = curl_init();
        curl_setopt($ch, CURLOPT_URL, $url);
        curl_setopt($ch, CURLOPT_HEADER, 1);
@@ -81,9 +98,11 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
 
        $header = curl_exec($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 ($curl_info['redirect_url'] != "")
@@ -101,6 +120,26 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
        if (($curl_info["content_type"] != "") AND !strstr(strtolower($curl_info["content_type"]),"html"))
                return($siteinfo);
 
+       if ($do_oembed) {
+               require_once("include/oembed.php");
+
+               $oembed_data = oembed_fetch_url($url);
+
+               if ($oembed_data->type != "error")
+                       $siteinfo["type"] = $oembed_data->type;
+
+               if (($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;
+               }
+       }
+
+       $stamp1 = microtime(true);
+
        // Now fetch the body as well
        $ch = curl_init();
        curl_setopt($ch, CURLOPT_URL, $url);
@@ -112,17 +151,10 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
 
        $header = curl_exec($ch);
        $curl_info = @curl_getinfo($ch);
-        $http_code = $curl_info['http_code'];
+       $http_code = $curl_info['http_code'];
        curl_close($ch);
 
-       if ($do_oembed) {
-               require_once("include/oembed.php");
-
-               $oembed_data = oembed_fetch_url($url);
-
-               if ($oembed_data->type != "error")
-                       $siteinfo["type"] = $oembed_data->type;
-       }
+       $a->save_timestamp($stamp1, "network");
 
        // Fetch the first mentioned charset. Can be in body or header
        $charset = "";
@@ -165,25 +197,25 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
        $xpath = new DomXPath($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 = $xpath->query("head/title");
@@ -196,8 +228,8 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
        foreach ($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"));
 
@@ -248,7 +280,8 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
        if (isset($keywords)) {
                $siteinfo["keywords"] = array();
                foreach ($keywords as $keyword)
-                       $siteinfo["keywords"][] = trim($keyword);
+                       if (!in_array(trim($keyword), $siteinfo["keywords"]))
+                               $siteinfo["keywords"][] = trim($keyword);
        }
 
        //$list = $xpath->query("head/meta[@property]");
@@ -256,8 +289,8 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
        foreach ($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"));
 
@@ -275,22 +308,13 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
                        }
        }
 
-       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"] = $oembed_data->thumbnail_url;
-       }
-
        if ((@$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);
@@ -309,7 +333,7 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
                                                                "height"=>$photodata[1]);
                        }
 
-               }
+               }
     } else {
                $src = completeurl($siteinfo["image"], $url);
 
@@ -399,6 +423,15 @@ function parse_url_content(&$a) {
                }
        }
 
+       // add url scheme if missing
+       $arrurl = parse_url($url);
+       if (!x($arrurl, 'scheme')) {
+               if (x($arrurl, 'host'))
+                       $url = "http:".$url;
+               else
+                       $url = "http://".$url;
+       }
+
        logger('parse_url: ' . $url);
 
        if($textmode)
@@ -454,10 +487,7 @@ function parse_url_content(&$a) {
 
        $sitedata = "";
 
-       if($siteinfo["title"] == "") {
-               $sitedata .= sprintf($template,$url,$url,'') . $str_tags;
-               killme();
-       } else {
+       if($siteinfo["title"] != "") {
                $text = $siteinfo["text"];
                $title = $siteinfo["title"];
        }