blob: b14410c3e19138a34cd9092363ff3a25501a5bee [file] [log] [blame]
// Copyright (c) 2018 The Chromium Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#include "base/strings/utf_string_conversions.h"
#include "base/strings/string_piece.h"
#include "base/strings/string_util.h"
#include "base/strings/utf_string_conversion_utils.h"
#include "base/third_party/icu/icu_utf.h"
#include "build/build_config.h"
#include "starboard/types.h"
namespace base {
namespace {
constexpr int32_t kErrorCodePoint = 0xFFFD;
// Size coefficient ----------------------------------------------------------
// The maximum number of codeunits in the destination encoding corresponding to
// one codeunit in the source encoding.
template <typename SrcChar, typename DestChar>
struct SizeCoefficient {
static_assert(sizeof(SrcChar) < sizeof(DestChar),
"Default case: from a smaller encoding to the bigger one");
// ASCII symbols are encoded by one codeunit in all encodings.
static constexpr int value = 1;
};
template <>
struct SizeCoefficient<char16, char> {
// One UTF-16 codeunit corresponds to at most 3 codeunits in UTF-8.
static constexpr int value = 3;
};
#if defined(WCHAR_T_IS_UTF32)
template <>
struct SizeCoefficient<wchar_t, char> {
// UTF-8 uses at most 4 codeunits per character.
static constexpr int value = 4;
};
template <>
struct SizeCoefficient<wchar_t, char16> {
// UTF-16 uses at most 2 codeunits per character.
static constexpr int value = 2;
};
#endif // defined(WCHAR_T_IS_UTF32)
#if __cplusplus >= 201402L
template <typename SrcChar, typename DestChar>
constexpr int size_coefficient_v =
SizeCoefficient<std::decay_t<SrcChar>, std::decay_t<DestChar>>::value;
#endif
// UnicodeAppendUnsafe --------------------------------------------------------
// Function overloads that write code_point to the output string. Output string
// has to have enough space for the codepoint.
void UnicodeAppendUnsafe(char* out, int32_t* size, uint32_t code_point) {
CBU8_APPEND_UNSAFE(out, *size, code_point);
}
void UnicodeAppendUnsafe(char16* out, int32_t* size, uint32_t code_point) {
CBU16_APPEND_UNSAFE(out, *size, code_point);
}
#if defined(WCHAR_T_IS_UTF32)
void UnicodeAppendUnsafe(wchar_t* out, int32_t* size, uint32_t code_point) {
out[(*size)++] = code_point;
}
#endif // defined(WCHAR_T_IS_UTF32)
// DoUTFConversion ------------------------------------------------------------
// Main driver of UTFConversion specialized for different Src encodings.
// dest has to have enough room for the converted text.
template <typename DestChar>
bool DoUTFConversion(const char* src,
int32_t src_len,
DestChar* dest,
int32_t* dest_len) {
bool success = true;
for (int32_t i = 0; i < src_len;) {
int32_t code_point;
CBU8_NEXT(src, i, src_len, code_point);
if (!IsValidCodepoint(code_point)) {
success = false;
code_point = kErrorCodePoint;
}
UnicodeAppendUnsafe(dest, dest_len, code_point);
}
return success;
}
template <typename DestChar>
bool DoUTFConversion(const char16* src,
int32_t src_len,
DestChar* dest,
int32_t* dest_len) {
bool success = true;
auto ConvertSingleChar = [&success](char16 in) -> int32_t {
if (!CBU16_IS_SINGLE(in) || !IsValidCodepoint(in)) {
success = false;
return kErrorCodePoint;
}
return in;
};
int32_t i = 0;
// Always have another symbol in order to avoid checking boundaries in the
// middle of the surrogate pair.
while (i < src_len - 1) {
int32_t code_point;
if (CBU16_IS_LEAD(src[i]) && CBU16_IS_TRAIL(src[i + 1])) {
code_point = CBU16_GET_SUPPLEMENTARY(src[i], src[i + 1]);
if (!IsValidCodepoint(code_point)) {
code_point = kErrorCodePoint;
success = false;
}
i += 2;
} else {
code_point = ConvertSingleChar(src[i]);
++i;
}
UnicodeAppendUnsafe(dest, dest_len, code_point);
}
if (i < src_len)
UnicodeAppendUnsafe(dest, dest_len, ConvertSingleChar(src[i]));
return success;
}
#if defined(WCHAR_T_IS_UTF32)
template <typename DestChar>
bool DoUTFConversion(const wchar_t* src,
int32_t src_len,
DestChar* dest,
int32_t* dest_len) {
bool success = true;
for (int32_t i = 0; i < src_len; ++i) {
int32_t code_point = src[i];
if (!IsValidCodepoint(code_point)) {
success = false;
code_point = kErrorCodePoint;
}
UnicodeAppendUnsafe(dest, dest_len, code_point);
}
return success;
}
#endif // defined(WCHAR_T_IS_UTF32)
// UTFConversion --------------------------------------------------------------
// Function template for generating all UTF conversions.
template <typename InputString, typename DestString>
bool UTFConversion(const InputString& src_str, DestString* dest_str) {
if (IsStringASCII(src_str)) {
dest_str->assign(src_str.begin(), src_str.end());
return true;
}
dest_str->resize(
src_str.length() *
#if __cplusplus < 201402L
SizeCoefficient<std::decay_t<typename InputString::value_type>,
std::decay_t<typename DestString::value_type>>::value);
#else
size_coefficient_v<typename InputString::value_type,
typename DestString::value_type>);
#endif
// Empty string is ASCII => it OK to call operator[].
auto* dest = &(*dest_str)[0];
// ICU requires 32 bit numbers.
int32_t src_len32 = static_cast<int32_t>(src_str.length());
int32_t dest_len32 = 0;
bool res = DoUTFConversion(src_str.data(), src_len32, dest, &dest_len32);
dest_str->resize(dest_len32);
// dest_str->shrink_to_fit();
// Workaround for missing symbols linking issue for Xcode 13.2.1
// Mac builds.
dest_str->reserve(0);
return res;
}
} // namespace
// UTF16 <-> UTF8 --------------------------------------------------------------
bool UTF8ToUTF16(const char* src, size_t src_len, string16* output) {
return UTFConversion(StringPiece(src, src_len), output);
}
string16 UTF8ToUTF16(StringPiece utf8) {
string16 ret;
// Ignore the success flag of this call, it will do the best it can for
// invalid input, which is what we want here.
UTF8ToUTF16(utf8.data(), utf8.size(), &ret);
return ret;
}
bool UTF16ToUTF8(const char16* src, size_t src_len, std::string* output) {
return UTFConversion(StringPiece16(src, src_len), output);
}
std::string UTF16ToUTF8(StringPiece16 utf16) {
std::string ret;
// Ignore the success flag of this call, it will do the best it can for
// invalid input, which is what we want here.
UTF16ToUTF8(utf16.data(), utf16.length(), &ret);
return ret;
}
// UTF-16 <-> Wide -------------------------------------------------------------
#if defined(WCHAR_T_IS_UTF16)
// When wide == UTF-16 the conversions are a NOP.
bool WideToUTF16(const wchar_t* src, size_t src_len, string16* output) {
output->assign(src, src_len);
return true;
}
string16 WideToUTF16(WStringPiece wide) {
return wide.as_string();
}
bool UTF16ToWide(const char16* src, size_t src_len, std::wstring* output) {
output->assign(src, src_len);
return true;
}
std::wstring UTF16ToWide(StringPiece16 utf16) {
return utf16.as_string();
}
#elif defined(WCHAR_T_IS_UTF32)
bool WideToUTF16(const wchar_t* src, size_t src_len, string16* output) {
return UTFConversion(base::WStringPiece(src, src_len), output);
}
string16 WideToUTF16(WStringPiece wide) {
string16 ret;
// Ignore the success flag of this call, it will do the best it can for
// invalid input, which is what we want here.
WideToUTF16(wide.data(), wide.length(), &ret);
return ret;
}
bool UTF16ToWide(const char16* src, size_t src_len, std::wstring* output) {
return UTFConversion(StringPiece16(src, src_len), output);
}
std::wstring UTF16ToWide(StringPiece16 utf16) {
std::wstring ret;
// Ignore the success flag of this call, it will do the best it can for
// invalid input, which is what we want here.
UTF16ToWide(utf16.data(), utf16.length(), &ret);
return ret;
}
#endif // defined(WCHAR_T_IS_UTF32)
// UTF-8 <-> Wide --------------------------------------------------------------
// UTF8ToWide is the same code, regardless of whether wide is 16 or 32 bits
bool UTF8ToWide(const char* src, size_t src_len, std::wstring* output) {
return UTFConversion(StringPiece(src, src_len), output);
}
std::wstring UTF8ToWide(StringPiece utf8) {
std::wstring ret;
// Ignore the success flag of this call, it will do the best it can for
// invalid input, which is what we want here.
UTF8ToWide(utf8.data(), utf8.length(), &ret);
return ret;
}
#if defined(WCHAR_T_IS_UTF16)
// Easy case since we can use the "utf" versions we already wrote above.
bool WideToUTF8(const wchar_t* src, size_t src_len, std::string* output) {
return UTF16ToUTF8(src, src_len, output);
}
std::string WideToUTF8(WStringPiece wide) {
return UTF16ToUTF8(wide);
}
#elif defined(WCHAR_T_IS_UTF32)
bool WideToUTF8(const wchar_t* src, size_t src_len, std::string* output) {
return UTFConversion(WStringPiece(src, src_len), output);
}
std::string WideToUTF8(WStringPiece wide) {
std::string ret;
// Ignore the success flag of this call, it will do the best it can for
// invalid input, which is what we want here.
WideToUTF8(wide.data(), wide.length(), &ret);
return ret;
}
#endif // defined(WCHAR_T_IS_UTF32)
string16 ASCIIToUTF16(StringPiece ascii) {
DCHECK(IsStringASCII(ascii)) << ascii;
return string16(ascii.begin(), ascii.end());
}
std::string UTF16ToASCII(StringPiece16 utf16) {
DCHECK(IsStringASCII(utf16)) << UTF16ToUTF8(utf16);
return std::string(utf16.begin(), utf16.end());
}
} // namespace base