Merge pull request #3795 from annando/complete-url

Complete incomplete URL during feed import

include/feed.php

@@ -27,6 +27,14 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
 		return;
 	}
 
+	if (!empty($contact['poll'])) {
+		$basepath = $contact['poll'];
+	} elseif (!empty($contact['url'])) {
+		$basepath = $contact['url'];
+	} else {
+		$basepath = '';
+	}
+
 	$doc = new DOMDocument();
 	@$doc->loadXML(trim($xml));
 	$xpath = new DomXPath($doc);
@@ -344,7 +352,7 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
 			if (title_is_body($item["title"], $body)) {
 				$item["title"] = "";
 			}
-			$item["body"] = html2bbcode($body);
+			$item["body"] = html2bbcode($body, $basepath);
 
 			if (($item["body"] == '') && ($item["title"] != '')) {
 				$item["body"] = $item["title"];

include/html2bbcode.php

@@ -79,7 +79,7 @@ function node2bbcodesub(&$doc, $oldnode, $attributes, $startbb, $endbb)
 	return($replace);
 }
 
-function html2bbcode($message)
+function html2bbcode($message, $basepath = '')
 {
 
 	$message = str_replace("\r", "", $message);
@@ -90,10 +90,10 @@ function html2bbcode($message)
 		function ($matches) use (&$codeblocks) {
 			$return = '[codeblock-' . count($codeblocks) . ']';
 
-            $prefix = '[code]';
-            if ($matches[1] != '') {
-                $prefix = '[code=' . $matches[1] . ']';
-            }
+			$prefix = '[code]';
+			if ($matches[1] != '') {
+				$prefix = '[code=' . $matches[1] . ']';
+			}
 			$codeblocks[] = $prefix . $matches[2] . '[/code]';
 			return $return;
 		}
@@ -313,15 +313,69 @@ function html2bbcode($message)
 	// Restore code blocks
 	$message = preg_replace_callback('#\[codeblock-([0-9]+)\]#iU',
 		function ($matches) use ($codeblocks) {
-            $return = '';
-            if (isset($codeblocks[intval($matches[1])])) {
-                $return = $codeblocks[$matches[1]];
-            }
+			$return = '';
+			if (isset($codeblocks[intval($matches[1])])) {
+				$return = $codeblocks[$matches[1]];
+			}
 			return $return;
 		}
 	, $message);
 
 	$message = trim($message);
 
+	if ($basepath != '') {
+		$message = addHostname($message, $basepath);
+	}
+
 	return $message;
 }
+
+/**
+ * @brief Sub function to complete incomplete URL
+ *
+ * @param array $matches Result of preg_replace_callback
+ * @param string $basepath Basepath that is used to complete the URL
+ *
+ * @return string The expanded URL
+ */
+function addHostnameSub($matches, $basepath) {
+	$base = parse_url($basepath);
+
+	$link = $matches[0];
+	$url = $matches[1];
+
+	$parts = array_merge($base, parse_url($url));
+	$url2 = unParseUrl($parts);
+
+	return str_replace($url, $url2, $link);
+}
+
+/**
+ * @brief Complete incomplete URLs in BBCode
+ *
+ * @param string $body Body with URLs
+ * @param string $basepath Basepath that is used to complete the URL
+ *
+ * @return string Body with expanded URLs
+ */
+function addHostname($body, $basepath) {
+	$URLSearchString = "^\[\]";
+
+	$matches = array("/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
+			"/\[url\]([$URLSearchString]*)\[\/url\]/ism",
+			"/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
+			"/\[img\](.*?)\[\/img\]/ism",
+			"/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
+			"/\[zmg\](.*?)\[\/zmg\]/ism",
+			"/\[video\](.*?)\[\/video\]/ism",
+			"/\[audio\](.*?)\[\/audio\]/ism",
+			);
+
+	foreach ($matches AS $match) {
+		$body = preg_replace_callback($match,
+						function ($match) use ($basepath) {
+							return addHostnameSub($match, $basepath);
+						}, $body);
+	}
+	return $body;
+}

include/network.php

@@ -1003,3 +1003,34 @@ function matching_url($url1, $url2) {
 
 	return normalise_link($match);
 }
+
+/**
+ * @brief Glue url parts together
+ *
+ * @param array $parsed URL parts
+ *
+ * @return string The glued URL
+ */
+function unParseUrl($parsed) {
+	$get = function ($key) use ($parsed) {
+		return isset($parsed[$key]) ? $parsed[$key] : null;
+	};
+
+	$pass      = $get('pass');
+	$user      = $get('user');
+	$userinfo  = $pass !== null ? "$user:$pass" : $user;
+	$port      = $get('port');
+	$scheme    = $get('scheme');
+	$query     = $get('query');
+	$fragment  = $get('fragment');
+	$authority =
+		($userinfo !== null ? $userinfo."@" : '') .
+		$get('host') .
+		($port ? ":$port" : '');
+
+	return	(strlen($scheme) ? $scheme.":" : '') .
+		(strlen($authority) ? "//".$authority : '') .
+		$get('path') .
+		(strlen($query) ? "?".$query : '') .
+		(strlen($fragment) ? "#".$fragment : '');
+}