Files
swift-mirror/stdlib/core/StringUnicodeScalarView.swift
Dmitri Hrybenko 4671211ce3 stdlib/String: fix two bugs in UnicodeScalarView
If underlying NSString contained isolated surrogates, then we were crashing in
following ways:

- subscripting by index could crash;

- index pointing to the second code unit sequence was not moved backwards
  correctly.  Instead of moving it to pointing to the beginning of the view it
  could be moved to point to the code unit before the beginning of the view.



Swift SVN r19230
2014-06-26 13:47:00 +00:00

214 lines
5.9 KiB
Swift

//===----------------------------------------------------------------------===//
//
// This source file is part of the Swift.org open source project
//
// Copyright (c) 2014 - 2015 Apple Inc. and the Swift project authors
// Licensed under Apache License v2.0 with Runtime Library Exception
//
// See http://swift.org/LICENSE.txt for license information
// See http://swift.org/CONTRIBUTORS.txt for the list of Swift project authors
//
//===----------------------------------------------------------------------===//
@public func ==(
lhs: String.UnicodeScalarView.IndexType,
rhs: String.UnicodeScalarView.IndexType
) -> Bool {
return lhs._position == rhs._position
}
extension String {
@public struct UnicodeScalarView : Sliceable, Sequence {
init(_ _base: _StringCore) {
self._base = _base
}
struct _ScratchGenerator : Generator {
var base: _StringCore
var idx: Int
init(_ core: _StringCore, _ pos: Int) {
idx = pos
base = core
}
mutating func next() -> UTF16.CodeUnit? {
return self.base[idx++]
}
}
@public struct IndexType : BidirectionalIndex {
init(_ _position: Int, _ _base: _StringCore) {
self._position = _position
self._base = _base
}
@public func successor() -> IndexType {
var scratch = _ScratchGenerator(_base, _position)
var decoder = UTF16()
let (result, length) = decoder._decodeOne(&scratch)
return IndexType(_position + length, _base)
}
@public func predecessor() -> IndexType {
var i = _position
let codeUnit = _base[--i]
if _slowPath((codeUnit >> 10) == 0b1101_11) {
if i != 0 && (_base[i - 1] >> 10) == 0b1101_10 {
--i
}
}
return IndexType(i, _base)
}
var _position: Int
var _base: _StringCore
}
@public var startIndex: IndexType {
return IndexType(_base.startIndex, _base)
}
@public var endIndex: IndexType {
return IndexType(_base.endIndex, _base)
}
@public subscript(i: IndexType) -> UnicodeScalar {
var scratch = _ScratchGenerator(_base, i._position)
var decoder = UTF16()
switch decoder.decode(&scratch) {
case .Result(let us):
return us
case .EmptyInput:
_fatalError("can not subscript using an endIndex")
case .Error:
return UnicodeScalar(0xfffd)
}
}
@public subscript(r: Range<IndexType>) -> UnicodeScalarView {
return UnicodeScalarView(
_base[r.startIndex._position..<r.endIndex._position])
}
@public struct GeneratorType : Generator {
init(_ _base: _StringCore.GeneratorType) {
self._base = _base
}
@public mutating func next() -> UnicodeScalar? {
switch _decoder.decode(&self._base) {
case .Result(let us):
return us
case .EmptyInput:
return .None
case .Error:
return UnicodeScalar(0xfffd)
}
}
var _decoder: UTF16 = UTF16()
var _base: _StringCore.GeneratorType
}
@public func generate() -> GeneratorType {
return GeneratorType(_base.generate())
}
@conversion @public
func __conversion() -> String {
return String(_base)
}
@public func compare(other : UnicodeScalarView) -> Int {
// Try to compare the string without decoding
// the UTF16 string.
var aIdx = self._base.startIndex
var bIdx = other._base.startIndex
var aEnd = self._base.endIndex
var bEnd = other._base.endIndex
// If this is not a contiguous UTF-16 then use the slow path.
// TODO: when we fix rdar://16740011 we'll need to optimize the ascii
// path.
if ((self._base.elementShift != 1) |
(other._base.elementShift != 1) |
(!self._base.hasContiguousStorage) |
(!other._base.hasContiguousStorage)) {
// This is a cold path and inlining _compareUnicode may
// prevent the inlining of the compare function.
return _compareUnicode(other)
}
while true {
if aIdx < aEnd {
if bIdx < bEnd {
let e1 = self._base._nthContiguous(aIdx)
let e2 = other._base._nthContiguous(bIdx)
// The range 0xD800 .. 0xDFFF is reserved for lead and trail
// surrogates. In this code we are only comparing against the
// lower bound because most interesting characters are in that
// range. This is conservatively correct since the slow path is
// handling the surrogates correctly.
if _slowPath((e1 >= 0xD800) | (e2 >= 0xD800)) {
// Use slow unicode comparator if
// we found multi-byte scalar.
return _compareUnicode(other)
}
if e1 < e2 {
return -1
}
if e2 < e1 {
return 1
}
aIdx++
bIdx++
continue // equivalent
}
return 1
}
if bIdx < bEnd {
return -1
}
return 0
}
}
@noinline
func _compareUnicode(other : UnicodeScalarView) -> Int {
var g1 = self.generate()
var g2 = other.generate()
while true {
let e1_ = g1.next()
let e2_ = g2.next()
if let e1 = e1_ {
if let e2 = e2_ {
if e1 < e2 {
return -1
}
if e2 < e1 {
return 1
}
continue // equivalent
}
return 1
}
if e2_ {
return -1
}
return 0
}
}
var _base: _StringCore
}
}
extension String {
@public func compare(other : String) -> Int {
return(UnicodeScalarView(core).compare(UnicodeScalarView(other.core)))
}
@public var unicodeScalars : UnicodeScalarView {
return UnicodeScalarView(core)
}
}