fix for linkedin#13

pgalbraith · Sep 21, 2018 · 0bd1932 · 0bd1932
1 parent 83c0c36
commit 0bd1932
Show file tree

Hide file tree

Showing 4 changed files with 27 additions and 2 deletions.
diff --git a/url-detector/src/main/java/com/linkedin/urls/detection/DomainNameReader.java b/url-detector/src/main/java/com/linkedin/urls/detection/DomainNameReader.java
@@ -102,7 +102,11 @@ public enum ReaderNextState {
     /**
      * Finished reading, next step should be to read the query string.
      */
-    ReadQueryString
+    ReadQueryString,
+    /**
+     * This was actually not a domain at all.
+     */
+    ReadUserPass
   }
 
   /**
@@ -332,6 +336,10 @@ public ReaderNextState readDomainName() {
       } else if (curr == '#') {
         //continue by reading the fragment
         return checkDomainNameValid(ReaderNextState.ReadFragment, curr);
+      } else if (curr == '@') {
+        //this may not have been a domain after all, but rather a username/password instead
+        _reader.goBack();
+        return ReaderNextState.ReadUserPass;
       } else if (CharUtils.isDot(curr)
           || (curr == '%' && _reader.canReadChars(2) && _reader.peek(2).equalsIgnoreCase(HEX_ENCODED_DOT))) {
         //if the current character is a dot or a urlEncodedDot

diff --git a/url-detector/src/main/java/com/linkedin/urls/detection/UrlDetector.java b/url-detector/src/main/java/com/linkedin/urls/detection/UrlDetector.java
@@ -553,6 +553,10 @@ public void addCharacter(char character) {
         return readPort();
       case ReadQueryString:
         return readQueryString();
+      case ReadUserPass:
+        int host = _currentUrlMarker.indexOf(UrlPart.HOST);
+        _currentUrlMarker.unsetIndex(UrlPart.HOST);
+        return readUserPass(host);
       default:
         return readEnd(ReadEndState.InvalidUrl);
     }

diff --git a/url-detector/src/test/java/com/linkedin/urls/TestUrl.java b/url-detector/src/test/java/com/linkedin/urls/TestUrl.java
@@ -28,7 +28,8 @@ private Object[][] getUsernamePasswordUrls() {
         {"@www.google.com", "www.google.com", "/", "", ""},
         {"lalal:@www.gogo.com", "www.gogo.com", "/", "lalal", ""},
         {"nono:boo@[::1]", "[::1]", "/", "nono", "boo"},
-        {"nono:[email protected]/@1234", "yahoo.com", "/@1234", "nono", "boo"}
+        {"nono:[email protected]/@1234", "yahoo.com", "/@1234", "nono", "boo"},
+        {"[email protected]", "google.com", "/", "big.big.boss", ""}
     };
   }
 

diff --git a/url-detector/src/test/java/com/linkedin/urls/detection/TestUriDetection.java b/url-detector/src/test/java/com/linkedin/urls/detection/TestUriDetection.java
@@ -657,6 +657,18 @@ public void testIssue12() {
     runTest("http://user:[email protected] host.com", UrlDetectorOptions.Default, "http://user:[email protected]", "host.com");
   }
 
+  /*
+   * https://github.com/linkedin/URL-Detector/issues/13
+   */
+  @Test
+  public void testIssue13() {
+    runTest("[email protected]/page", UrlDetectorOptions.Default, "[email protected]/page");
+    runTest("[email protected]", UrlDetectorOptions.Default, "[email protected]");
+    runTest("[email protected]", UrlDetectorOptions.Default, "[email protected]");
+    runTest("[email protected]", UrlDetectorOptions.Default, "[email protected]");
+    runTest("first.middle.reallyreallyreallyreallyreallyreallyreallyreallyreallyreallylonglastname@gmail.com", UrlDetectorOptions.Default, "first.middle.reallyreallyreallyreallyreallyreallyreallyreallyreallyreallylonglastname@gmail.com");
+  }
+
   /*
    * https://github.com/linkedin/URL-Detector/issues/16
    */