From 7807497cd83bdbef0acc2ac96da17d59cb0e4e57 Mon Sep 17 00:00:00 2001
From: Steven Hartland <steven.hartland@multiplay.co.uk>
Date: Fri, 25 Nov 2022 20:08:29 +0000
Subject: [PATCH] fix: string lengths

Fix string lengths for strings containing code points which require a
Surrogate pair encoding in utf16.

Fixes #388
---
 builtin_string.go |  9 +++++++--
 regexp_test.go    |  5 +++++
 string_test.go    |  1 +
 type_regexp.go    | 12 ++++--------
 type_string.go    | 18 ++++++++++--------
 5 files changed, 27 insertions(+), 18 deletions(-)

diff --git a/builtin_string.go b/builtin_string.go
index e45b71cd..1823ce7f 100644
--- a/builtin_string.go
+++ b/builtin_string.go
@@ -5,6 +5,7 @@ import (
 	"regexp"
 	"strconv"
 	"strings"
+	"unicode/utf16"
 	"unicode/utf8"
 )
 
@@ -72,18 +73,22 @@ func builtinString_concat(call FunctionCall) Value {
 
 func lastIndexRune(s, substr string) int {
 	if i := strings.LastIndex(s, substr); i >= 0 {
-		return utf8.RuneCountInString(s[:i])
+		return utf16Length(s[:i])
 	}
 	return -1
 }
 
 func indexRune(s, substr string) int {
 	if i := strings.Index(s, substr); i >= 0 {
-		return utf8.RuneCountInString(s[:i])
+		return utf16Length(s[:i])
 	}
 	return -1
 }
 
+func utf16Length(s string) int {
+	return len(utf16.Encode([]rune(s)))
+}
+
 func builtinString_indexOf(call FunctionCall) Value {
 	checkObjectCoercible(call.runtime, call.This)
 	value := call.This.string()
diff --git a/regexp_test.go b/regexp_test.go
index 41b60cc5..d98f3ce4 100644
--- a/regexp_test.go
+++ b/regexp_test.go
@@ -151,6 +151,11 @@ func TestRegExp_exec(t *testing.T) {
             [ abc.input.length, abc.length, abc.input, abc.index, abc ];
         `, "11,1,CE\uFFFFL\uFFDDbox127,5,box1")
 
+		test(`
+        var abc = /\w{3}\d?/.exec("CE😋box127");
+        [ abc.input.length, abc.length, abc.input, abc.index, abc ];
+    `, "10,1,CE😋box127,4,box1")
+
 		test(`RegExp.prototype.exec.length`, 1)
 		test(`RegExp.prototype.exec.prototype`, "undefined")
 	})
diff --git a/string_test.go b/string_test.go
index 509e61ae..c7d31808 100644
--- a/string_test.go
+++ b/string_test.go
@@ -316,6 +316,7 @@ func TestString_length(t *testing.T) {
 
 		test(`"abc".length`, 3)
 		test(`"uñiçode".length`, 7)
+		test(`"😋".length`, 2)
 	})
 }
 
diff --git a/type_regexp.go b/type_regexp.go
index b4c07168..51801788 100644
--- a/type_regexp.go
+++ b/type_regexp.go
@@ -3,7 +3,7 @@ package otto
 import (
 	"fmt"
 	"regexp"
-	"unicode/utf8"
+	"unicode/utf16"
 
 	"github.com/robertkrimen/otto/parser"
 )
@@ -131,13 +131,9 @@ func execResultToArray(runtime *_runtime, target string, result []int) *_object
 	}
 	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
-		}
+		// Find the ut16 rune index in the string, not the byte index.
+		matchPrefix := target[:matchIndex]
+		matchIndex = len(utf16.Encode([]rune(matchPrefix)))
 	}
 	match := runtime.newArrayOf(valueArray)
 	match.defineProperty("input", toValue_string(target), 0111, false)
diff --git a/type_string.go b/type_string.go
index 2aa9ee93..f08a6183 100644
--- a/type_string.go
+++ b/type_string.go
@@ -2,6 +2,7 @@ package otto
 
 import (
 	"strconv"
+	"unicode/utf16"
 	"unicode/utf8"
 )
 
@@ -26,20 +27,22 @@ func (str _stringASCII) String() string {
 }
 
 type _stringWide struct {
-	string string
-	length int
-	runes  []rune
+	string  string
+	value16 []uint16
 }
 
 func (str _stringWide) Length() int {
-	return str.length
+	if str.value16 == nil {
+		str.value16 = utf16.Encode([]rune(str.string))
+	}
+	return len(str.value16)
 }
 
 func (str _stringWide) At(at int) rune {
-	if str.runes == nil {
-		str.runes = []rune(str.string)
+	if str.value16 == nil {
+		str.value16 = utf16.Encode([]rune(str.string))
 	}
-	return str.runes[at]
+	return rune(str.value16[at])
 }
 
 func (str _stringWide) String() string {
@@ -58,7 +61,6 @@ func _newStringObject(str string) _stringObject {
 wide:
 	return &_stringWide{
 		string: str,
-		length: utf8.RuneCountInString(str),
 	}
 }