JS: Handle surrogate pairs correctly

javascript/diff_match_patch_uncompressed.js

@@ -22,6 +22,16 @@
  * @author [email protected] (Neil Fraser)
  */
 
+/**
+ * Determine if the index is inside a surrogate pair.
+ * @param {string} str The string
+ * @param {numer} idx The index
+ */
+function insideSurrogate(str, idx) {
+  var code = str.charCodeAt(idx);
+  return code >= 0xDC00 && code <= 0xDFFF;
+}
+
 /**
  * Class containing the diff, match and patch methods.
  * @constructor
@@ -361,6 +371,11 @@ diff_match_patch.prototype.diff_bisect_ = function(text1, text2, deadline) {
         x1++;
         y1++;
       }
+      if (insideSurrogate(text1, x1)) {
+        x1--;
+        y1--;
+      }
+
       v1[k1_offset] = x1;
       if (x1 > text1_length) {
         // Ran off the right of the graph.
@@ -569,6 +584,9 @@ diff_match_patch.prototype.diff_commonPrefix = function(text1, text2) {
     }
     pointermid = Math.floor((pointermax - pointermin) / 2 + pointermin);
   }
+  if (insideSurrogate(text1, pointermid)) {
+    pointermid--;
+  }
   return pointermid;
 };
 
@@ -601,6 +619,9 @@ diff_match_patch.prototype.diff_commonSuffix = function(text1, text2) {
     }
     pointermid = Math.floor((pointermax - pointermin) / 2 + pointermin);
   }
+  if (insideSurrogate(text1, text1.length - pointermid)) {
+    pointermid--;
+  }
   return pointermid;
 };
 

javascript/tests/diff_match_patch_test.html

@@ -113,6 +113,7 @@
           'testPatchObj',
           'testPatchFromText',
           'testPatchToText',
+          'testPatchSurrogates',
           'testPatchAddContext',
           'testPatchMake',
           'testPatchSplitMax',

javascript/tests/diff_match_patch_test.js

@@ -767,6 +767,57 @@ function testPatchToText() {
   strp = '@@ -1,9 +1,9 @@\n-f\n+F\n oo+fooba\n@@ -7,9 +7,9 @@\n obar\n-,\n+.\n  tes\n';
   p = dmp.patch_fromText(strp);
   assertEquals(strp, dmp.patch_toText(p));
+
+}
+
+function testPatchSurrogates() {
+  var p, p2, strp;
+
+  // These share the same high surrogate prefix
+  p = dmp.patch_make('\u{1F30D}', '\u{1F308}');
+  strp = dmp.patch_toText(p);
+  p2 = dmp.patch_fromText(strp);
+  assertEquivalent(p, p2);
+
+  // These share the same low surrogate suffix
+  p = dmp.patch_make('\u{10120}', '\u{10520}');
+  strp = dmp.patch_toText(p);
+  p2 = dmp.patch_fromText(strp);
+  assertEquivalent(p, p2);
+
+  // No common prefix, but later there's the same high surrogate char
+  p = dmp.patch_make('abbb\u{1F30D}', 'cbbb\u{1F308}');
+  strp = dmp.patch_toText(p);
+  p2 = dmp.patch_fromText(strp);
+  assertEquivalent(p, p2);
+
+  // No common suffix, but earlier there's the same low surrogate char
+  p = dmp.patch_make('\u{10120}aaac', '\u{10520}aaab');
+  strp = dmp.patch_toText(p);
+  p2 = dmp.patch_fromText(strp);
+  assertEquivalent(p, p2);
+
+  // No common suffix, but earlier there's the same low surrogate char
+  p = dmp.patch_make('abbb\u{10120}aaac', '\u{10520}aaab');
+  strp = dmp.patch_toText(p);
+  p2 = dmp.patch_fromText(strp);
+  assertEquivalent(p, p2);
+
+  var padding1 = "";
+  while (padding1.length < 100) {
+    padding1 += String.fromCharCode(50 + padding1.length);
+  }
+
+  var padding2 = "";
+  while (padding2.length < 100) {
+    padding2 += String.fromCharCode(200 + padding2.length);
+  }
+
+  // Add some random padding
+  p = dmp.patch_make(padding1+'\u{10120}'+padding2, padding2+'\u{10520}'+padding1);
+  strp = dmp.patch_toText(p);
+  p2 = dmp.patch_fromText(strp);
+  assertEquivalent(p, p2);
 }
 
 function testPatchAddContext() {