Merge pull request #2906 from annando/1611-parseurl

Improvements to page_info (bad ssl, bad formatted url)

include/items.php

@@ -147,19 +147,23 @@ function add_page_info_data($data) {
 	// It maybe is a rich content, but if it does have everything that a link has,
 	// then treat it that way
 	if (($data["type"] == "rich") AND is_string($data["title"]) AND
-		is_string($data["text"]) AND (sizeof($data["images"]) > 0))
+		is_string($data["text"]) AND (sizeof($data["images"]) > 0)) {
 		$data["type"] = "link";
+	}
 
-	if ((($data["type"] != "link") AND ($data["type"] != "video") AND ($data["type"] != "photo")) OR ($data["title"] == $url))
+	if ((($data["type"] != "link") AND ($data["type"] != "video") AND ($data["type"] != "photo")) OR ($data["title"] == $data["url"])) {
-		return("");
+		return "";
+	}
 
-	if ($no_photos AND ($data["type"] == "photo"))
+	if ($no_photos AND ($data["type"] == "photo")) {
-		return("");
+		return "";
+	}
 
-	if (sizeof($data["images"]) > 0)
+	if (sizeof($data["images"]) > 0) {
 		$preview = $data["images"][0];
-	else
+	} else {
 		$preview = "";
+	}
 
 	// Escape some bad characters
 	$data["url"] = str_replace(array("[", "]"), array("[", "]"), htmlentities($data["url"], ENT_QUOTES, 'UTF-8', false));
@@ -167,19 +171,33 @@ function add_page_info_data($data) {
 
 	$text = "[attachment type='".$data["type"]."'";
 
-	if ($data["url"] != "")
+	if ($data["text"] == "") {
+		$data["text"] = $data["title"];
+	}
+
+	if ($data["text"] == "") {
+		$data["text"] = $data["url"];
+	}
+
+	if ($data["url"] != "") {
 		$text .= " url='".$data["url"]."'";
-	if ($data["title"] != "")
+	}
+
+	if ($data["title"] != "") {
 		$text .= " title='".$data["title"]."'";
+	}
+
 	if (sizeof($data["images"]) > 0) {
 		$preview = str_replace(array("[", "]"), array("[", "]"), htmlentities($data["images"][0]["src"], ENT_QUOTES, 'UTF-8', false));
 		// if the preview picture is larger than 500 pixels then show it in a larger mode
 		// But only, if the picture isn't higher than large (To prevent huge posts)
-		if (($data["images"][0]["width"] >= 500) AND ($data["images"][0]["width"] >= $data["images"][0]["height"]))
+		if (($data["images"][0]["width"] >= 500) AND ($data["images"][0]["width"] >= $data["images"][0]["height"])) {
 			$text .= " image='".$preview."'";
-		else
+		} else {
 			$text .= " preview='".$preview."'";
 		}
+	}
+
 	$text .= "]".$data["text"]."[/attachment]";
 
 	$hashtags = "";

mod/parse_url.php

@@ -89,6 +89,13 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
 
 	$siteinfo = array();
 
+	// Check if the URL does contain a scheme
+	$scheme = parse_url($url, PHP_URL_SCHEME);
+
+	if ($scheme == "") {
+		$url = "http://".trim($url, "/");
+	}
+
 	if ($count > 10) {
 		logger("parseurl_getsiteinfo: Endless loop detected for ".$url, LOGGER_DEBUG);
 		return($siteinfo);
@@ -102,6 +109,8 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
 	$siteinfo["url"] = $url;
 	$siteinfo["type"] = "link";
 
+	$check_cert = get_config('system','verifyssl');
+
 	$stamp1 = microtime(true);
 
 	$ch = curl_init();
@@ -110,8 +119,9 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
 	curl_setopt($ch, CURLOPT_NOBODY, 1);
 	curl_setopt($ch, CURLOPT_TIMEOUT, 3);
 	curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
-	//curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
 	curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
+	curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
+	curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, (($check_cert) ? 2 : false));
 
 	$header = curl_exec($ch);
 	$curl_info = @curl_getinfo($ch);
@@ -142,8 +152,9 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
 
 		$oembed_data = oembed_fetch_url($url);
 
-		if ($oembed_data->type != "error")
+		if (!in_array($oembed_data->type, array("error", "rich"))) {
 			$siteinfo["type"] = $oembed_data->type;
+		}
 
 		if (($oembed_data->type == "link") AND ($siteinfo["type"] != "photo")) {
 			if (isset($oembed_data->title))
@@ -165,6 +176,8 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co
 	curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 	curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 	curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
+	curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
+	curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, (($check_cert) ? 2 : false));
 
 	$header = curl_exec($ch);
 	$curl_info = @curl_getinfo($ch);