From a65f65fa7b577911084bbe8b3bbe0375b40ef24e Mon Sep 17 00:00:00 2001
From: keita <keita@twitter.com>
Date: Tue, 14 Feb 2012 16:37:24 -0800
Subject: [PATCH] Fix a bug in
 modifyIndicesFromUTF16ToUnicode/modifyIndicesFromUnicodeToUTF16.

---
 twitter-text.js | 47 +++++++++++++++++++++++++++++++++--------------
 1 file changed, 33 insertions(+), 14 deletions(-)

diff --git a/twitter-text.js b/twitter-text.js
index 926a243..7780186 100644
--- a/twitter-text.js
+++ b/twitter-text.js
@@ -767,25 +767,44 @@ if (typeof twttr === "undefined" || twttr === null) {
   };
 
   twttr.txt.modifyIndicesFromUnicodeToUTF16 = function(text, entities) {
-    twttr.txt.shiftIndices(text, entities, 1);
+    twttr.txt.shiftIndices(text, entities, false);
   };
 
   twttr.txt.modifyIndicesFromUTF16ToUnicode = function(text, entities) {
-    twttr.txt.shiftIndices(text, entities, -1);
+    twttr.txt.shiftIndices(text, entities, true);
   };
 
-  twttr.txt.shiftIndices = function(text, entities, diff) {
-    for (var i = 0; i < text.length - 1; i++) {
-      var c1 = text.charCodeAt(i);
-      var c2 = text.charCodeAt(i + 1);
-      if (0xD800 <= c1 && c1 <= 0xDBFF && 0xDC00 <= c2 && c2 <= 0xDFFF) {
-        // supplementary character
-        i++; // skip surrogate pair character
-        for (var j = 0; j < entities.length; j++) {
-          if (entities[j].indices[0] >= i) {
-            entities[j].indices[0] += diff;
-            entities[j].indices[1] += diff;
-          }
+  twttr.txt.shiftIndices = function(text, entities, indicesInUTF16) {
+    if (entities.length == 0) {
+      return;
+    }
+
+    var charIndex = text.length - 1;
+    // replace surrogate pairs with whitespace, and then count length
+    var codePointIndex = text.replace(/[\uD800-\uDBFF][\uDC00-\uDFFF]/g, " ").length - 1;
+    var entityIndex = entities.length - 1;
+    var entity = entities[entityIndex];
+
+    while (charIndex != codePointIndex) {
+      if (entity.indices[0] == (indicesInUTF16 ? charIndex : codePointIndex)) {
+        var len = entity.indices[1] - entity.indices[0];
+        entity.indices[0] = indicesInUTF16 ? codePointIndex : charIndex;
+        entity.indices[1] = entity.indices[0] + len;
+
+        if (entityIndex == 0) {
+          // no more entity
+          break;
+        }
+        entity = entities[--entityIndex];
+      }
+
+      codePointIndex--;
+      charIndex--;
+      var c = text.charCodeAt(charIndex);
+      if (0xD800 <= c && c <= 0xDBFF && charIndex < text.length - 1) {
+        c = text.charCodeAt(charIndex + 1);
+        if (0xDC00 <= c && c <= 0xDFFF) {
+          charIndex--;
         }
       }
     }