JavaScript: Fix known-broken diffs from objective-c

Author: dmsnell

javascript/diff_match_patch_uncompressed.js

@@ -1512,7 +1512,13 @@ diff_match_patch.prototype.decodeURI = function(text) {
       throw new URIError('URI malformed');
     }
 
-    return decoded;
+    // some objective-c versions of the library produced patches with
+    // (null) in the place where surrogates were split across diff
+    // boundaries. if we leave those in we'll be stuck with a
+    // high-surrogate (null) low-surrogate pattern that will break
+    // deeper in the library or consumping application. we'll "fix"
+    // these by dropping the (null) and re-joining the surrogate halves
+    return decoded.replace(/([\uD800-\uDBFF])\(null\)([\uDC00-\uDFFF])/g, "$1$2");
   }
 };
 

javascript/tests/diff_match_patch_test.js

@@ -605,6 +605,15 @@ function testDiffDelta() {
     assertEquals('Swap surrogate pair', 'crashed');
   }
 
+  try {
+    assertEquivalent(
+      dmp.diff_fromDelta('', '+%ED%A0%BC%28null%29%ED%B5%B0'),
+      [[DIFF_INSERT, '\ud83c\udd70']]
+    );
+  } catch ( e ) {
+    assertEquals('Invalid diff from objective-c with (null) string' );
+  }
+
   // Empty diff groups
   assertEquivalent(
     dmp.diff_toDelta([[DIFF_EQUAL, 'abcdef'], [DIFF_DELETE, ''], [DIFF_INSERT, 'ghijk']]),