X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=mod%2Fparse_url.php;h=ef051d9f6b433c0e8152c39168c0debd8cfbee0a;hb=89c5989cfb679449fb776df7380e972d0f676d8d;hp=cd2263dbe2622d8ae532ab87dcce935ac037b916;hpb=35f623ae4f82ac8167ad98b38fab39e449ffa1ea;p=friendica.git

diff --git a/mod/parse_url.php b/mod/parse_url.php
index cd2263dbe2..ef051d9f6b 100644
--- a/mod/parse_url.php
+++ b/mod/parse_url.php
@@ -50,6 +50,21 @@ function completeurl($url, $scheme) {
 	return($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) {
 	require_once("include/network.php");
 
@@ -97,6 +112,14 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
 		return($siteinfo);
 	}
 
+	// 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($siteinfo);
+
 	if ($do_oembed) {
 		require_once("include/oembed.php");
 
@@ -104,15 +127,16 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
 
 		if ($oembed_data->type != "error")
 			$siteinfo["type"] = $oembed_data->type;
-	}
 
-	// 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($siteinfo);
+		if (($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);
 
@@ -256,7 +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 = $xpath->query("head/meta[@property]");
@@ -283,15 +308,6 @@ function parseurl_getsiteinfo($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 $node) {
@@ -407,6 +423,15 @@ function parse_url_content(&$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($textmode)
@@ -462,10 +487,7 @@ function parse_url_content(&$a) {
 
 	$sitedata = "";
 
-	if($siteinfo["title"] == "") {
-		$sitedata .= sprintf($template,$url,$url,'') . $str_tags;
-		killme();
-	} else {
+	if($siteinfo["title"] != "") {
 		$text = $siteinfo["text"];
 		$title = $siteinfo["title"];
 	}