From 1628647e055fcd9e4cb2b8ade80a806f8debbe28 Mon Sep 17 00:00:00 2001
From: Robert Krimen <robertkrimen@gmail.com>
Date: Thu, 2 May 2013 20:23:06 +0200
Subject: [PATCH] For RegExp.exec, return the rune index in the string, not the
 byte index

---
 regexp_test.go |  7 ++++++-
 type_regexp.go | 13 ++++++++++++-
 2 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/regexp_test.go b/regexp_test.go
index 536aa15..26c11b1 100644
--- a/regexp_test.go
+++ b/regexp_test.go
@@ -94,13 +94,18 @@ func TestRegExp_exec(t *testing.T) {
 
 	test(`
 		var abc = /[abc](\d)?/.exec("a0 b c1 d3");
-        [ abc.length, abc.input, abc.index, abc ]
+        [ abc.length, abc.input, abc.index, abc ];
     `, "2,a0 b c1 d3,0,a0,0")
 
 	test(`raise:
 		var exec = RegExp.prototype.exec;
 		exec("Xyzzy");
 	`, "TypeError: Calling RegExp.exec on a non-RegExp object")
+
+	test(`
+        var abc = /\w{3}\d?/.exec("CE\uFFFFL\uFFDDbox127");
+        [ abc.input.length, abc.length, abc.input, abc.index, abc ];
+    `, "11,1,CE\uFFFFL\uFFDDbox127,5,box1")
 }
 
 func TestRegExp_zaacbbbcac(t *testing.T) {
diff --git a/type_regexp.go b/type_regexp.go
index 8aa2d5f..4b7706a 100644
--- a/type_regexp.go
+++ b/type_regexp.go
@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"regexp"
 	"strings"
+	"unicode/utf8"
 )
 
 type _regExpObject struct {
@@ -106,9 +107,19 @@ func execResultToArray(runtime *_runtime, target string, result []int) *_object
 			valueArray[index] = UndefinedValue()
 		}
 	}
+	matchIndex := result[0]
+	if matchIndex != 0 {
+		matchIndex = 0
+		// Find the rune index in the string, not the byte index
+		for index := 0; index < result[0]; {
+			_, size := utf8.DecodeRuneInString(target[index:])
+			matchIndex += 1
+			index += size
+		}
+	}
 	match := runtime.newArray(valueArray)
 	match.set("input", toValue(target), false)
-	match.set("index", toValue(result[0]), false)
+	match.set("index", toValue(matchIndex), false)
 	return match
 }