Fix URL regex

2 files changed, 11 insertions, 3 deletions

diff --git a/src/main/java/com/juick/util/MessageUtils.java b/src/main/java/com/juick/util/MessageUtils.java
index 42c97d84..316adb90 100644
--- a/src/main/java/com/juick/util/MessageUtils.java
+++ b/src/main/java/com/juick/util/MessageUtils.java
@@ -57,7 +57,7 @@ public class MessageUtils {
 
     private final static String urlWhiteSpacePrefix = "((?<=\\s)|(?<=\\A))";
 
-    private final static String urlRegex = "((?:ht|f)tps?://(?:www\\.)?([^\\/\\s\\n\\\"]+)/?[^\\]\\s\\n\\\"\\>]*)";
+    private final static String urlRegex = "((?:(?:ht|f)tps?://(?:www\\.)?([^\\s()<>/?#]+)([^\\s()<>]*)?))";
 
     private final static String urlWithWhitespacesRegex =
             urlWhiteSpacePrefix + urlRegex;
diff --git a/src/test/java/com/juick/MessageTest.java b/src/test/java/com/juick/MessageTest.java
index 14d6175a..6ce01f4a 100644
--- a/src/test/java/com/juick/MessageTest.java
+++ b/src/test/java/com/juick/MessageTest.java
@@ -213,8 +213,8 @@ public class MessageTest {
         assertThat(MessageUtils.formatMessage(msg), is("<blockquote>quote</blockquote>message"));
         String brokenComment = "<!-- read next";
         assertThat(MessageUtils.formatMessage(brokenComment), is("&lt;!-- read next"));
-        String url = "[ya](http://ya.ru)";
-        assertThat(MessageUtils.formatMessage(url), is("<a href=\"http://ya.ru\" rel=\"nofollow\">ya</a>"));
+        String url = "test [ya](https://juick.com/a/1)? hh";
+        assertThat(MessageUtils.formatMessage(url), is("test <a href=\"https://juick.com/a/1\" rel=\"nofollow\">ya</a>? hh"));
         String complexMessage = "У футболистов нет мозгов. Что в России, что в Беларуси:\n" +
                 "\n" +
                 ">Отец футболиста Лухвича, объехавшего пробку по тротуару: «Сына задержали, Infiniti арестовали» https://auto.onliner.by/2019/01/23/probka-9\n" +
@@ -238,4 +238,12 @@ public class MessageTest {
         testMessage.setTags(MessageUtils.parseTags("NSFW test"));
         assertThat(MessageUtils.isSensitive(testMessage), is(true));
     }
+    @Test
+    public void stripHashesTest() {
+        var safeMessage = "And that is a [odd](https://juick.com/a/1)? aaapaa";
+        var nonsafeMessage = "[Here is my link](https://juick.com?hash=12345)";
+        var filteredMessage = "[Here is my link](https://juick.com)";
+        assertThat(MessageUtils.stripNonSafeUrls(safeMessage), is(safeMessage));
+        assertThat(MessageUtils.stripNonSafeUrls(nonsafeMessage), is(filteredMessage));
+    }
 }