fix unsafe urls stripping

1 files changed, 9 insertions, 5 deletions

diff --git a/juick-common/src/main/java/com/juick/util/MessageUtils.java b/juick-common/src/main/java/com/juick/util/MessageUtils.java
index 5a4b05ca..c9796809 100644
--- a/juick-common/src/main/java/com/juick/util/MessageUtils.java
+++ b/juick-common/src/main/java/com/juick/util/MessageUtils.java
@@ -57,8 +57,12 @@ public class MessageUtils {
         return result;
     }
 
-    private final static String regexUrl =
-            "((?<=\\s)|(?<=\\A))((?:ht|f)tps?://(?:www\\.)?([^\\/\\s\\n\\\"]+)/?[^\\s\\n\\\"]*)";
+    private final static String urlWhiteSpacePrefix = "((?<=\\s)|(?<=\\A))";
+
+    private final static String urlRegex = "((?:ht|f)tps?://(?:www\\.)?([^\\/\\s\\n\\\"]+)/?[^\\s\\n\\\"\\>]*)";
+
+    private final static String urlWithWhitespacesRegex =
+            urlWhiteSpacePrefix + urlRegex;
 
     private final static Pattern regexLinks2 = Pattern.compile("((?<=\\s)|(?<=\\A))([\\[\\{]|&lt;)((?:ht|f)tps?://(?:www\\.)?([^\\/\\s\\\"\\)\\!]+)/?(?:[^\\]\\}](?<!&gt;))*)([\\]\\}]|&gt;)");
 
@@ -69,7 +73,7 @@ public class MessageUtils {
 
         // http://juick.com/last?page=2
         // <a href="http://juick.com/last?page=2" rel="nofollow">http://juick.com/last?page=2</a>
-        msg = msg.replaceAll(regexUrl, "$1<a href=\"$2\" rel=\"nofollow\">$2</a>");
+        msg = msg.replaceAll(urlWithWhitespacesRegex, "$1<a href=\"$2\" rel=\"nofollow\">$2</a>");
 
         // (http://juick.com/last?page=2)
         // (<a href="http://juick.com/last?page=2" rel="nofollow">http://juick.com/last?page=2</a>)
@@ -96,7 +100,7 @@ public class MessageUtils {
 
         // http://juick.com/last?page=2
         // <a href="http://juick.com/last?page=2" rel="nofollow">juick.com</a>
-        msg = msg.replaceAll(regexUrl, "$1<a href=\"$2\" rel=\"nofollow\">$3</a>");
+        msg = msg.replaceAll(urlWithWhitespacesRegex, "$1<a href=\"$2\" rel=\"nofollow\">$3</a>");
 
         // [link text][http://juick.com/last?page=2]
         // <a href="http://juick.com/last?page=2" rel="nofollow">link text</a>
@@ -276,7 +280,7 @@ public class MessageUtils {
 
     public static String stripNonSafeUrls(String input) {
         // strip login urls
-        Matcher urlMatcher = Pattern.compile(MessageUtils.regexUrl).matcher(input);
+        Matcher urlMatcher = Pattern.compile(MessageUtils.urlRegex).matcher(input);
         while (urlMatcher.find()) {
             URI uri = URI.create(urlMatcher.group(0));
             if (uri.getHost().equals("juick.com")) {