Make speculative link extractor ignore urls with two consecutive slashes (not preceded by a colon as in http://). We've seen the double slashes lead to crawler traps.

* UriUtils.java
    isLikelyFalsePositive() - consider likely false positive if it contains '//' (but not '://')
* UriUtilsTest.java
    testDoubleSlashFalsePositives()

commons/src/main/java/org/archive/util/UriUtils.java

@@ -280,6 +280,13 @@ public class UriUtils {
             return true;
         }
         
+        if (TextUtils.matches("^.*[^:]//.*$", candidate)) {
+            if (LOGGER.isLoggable(Level.FINE)) {
+                LOGGER.fine("rejected: contains '//' (but not '://'): " + candidate);
+            }
+            return true;
+        }
+        
         // look for things that look like hostnames and not filenames?
         // look for too many dots but make sure we take into account that url may have hostname?
 

commons/src/test/java/org/archive/util/UriUtilsTest.java

@@ -210,6 +210,14 @@ public class UriUtilsTest extends TestCase {
         tryAll(startsOrEndsWithPlusFalsePositives, false);
     }
     
+    protected static final String[] doubleSlashFalsePositives = new String[] {
+        ".//*",
+        "http://example.com/monkey//foo/whatever"
+    };
+    public void testDoubleSlashFalsePositives() {
+        tryAll(startsOrEndsWithPlusFalsePositives, false);
+    }
+
     /**
      * Test that all supplied candidates give the expected result, for each of 
      * the 'legacy' (H1) likely-URI-tests