diff options
author | bunnei <bunneidev@gmail.com> | 2018-10-04 00:44:51 +0200 |
---|---|---|
committer | GitHub <noreply@github.com> | 2018-10-04 00:44:51 +0200 |
commit | 137d43fa2f8060ee9052f7541a34b900ecbc7ff0 (patch) | |
tree | 3927c204eda84d63db274f615b7c69137f12dc9c /src | |
parent | Merge pull request #1435 from lioncash/xci (diff) | |
parent | string_util: unify UTF8<->UTF16 conversion to codecvt (diff) | |
download | yuzu-137d43fa2f8060ee9052f7541a34b900ecbc7ff0.tar yuzu-137d43fa2f8060ee9052f7541a34b900ecbc7ff0.tar.gz yuzu-137d43fa2f8060ee9052f7541a34b900ecbc7ff0.tar.bz2 yuzu-137d43fa2f8060ee9052f7541a34b900ecbc7ff0.tar.lz yuzu-137d43fa2f8060ee9052f7541a34b900ecbc7ff0.tar.xz yuzu-137d43fa2f8060ee9052f7541a34b900ecbc7ff0.tar.zst yuzu-137d43fa2f8060ee9052f7541a34b900ecbc7ff0.zip |
Diffstat (limited to 'src')
-rw-r--r-- | src/common/file_util.h | 2 | ||||
-rw-r--r-- | src/common/string_util.cpp | 134 | ||||
-rw-r--r-- | src/common/string_util.h | 21 |
3 files changed, 7 insertions, 150 deletions
diff --git a/src/common/file_util.h b/src/common/file_util.h index 3d8fe6264..571503d2a 100644 --- a/src/common/file_util.h +++ b/src/common/file_util.h @@ -285,7 +285,7 @@ private: template <typename T> void OpenFStream(T& fstream, const std::string& filename, std::ios_base::openmode openmode) { #ifdef _MSC_VER - fstream.open(Common::UTF8ToTStr(filename).c_str(), openmode); + fstream.open(Common::UTF8ToUTF16W(filename).c_str(), openmode); #else fstream.open(filename.c_str(), openmode); #endif diff --git a/src/common/string_util.cpp b/src/common/string_util.cpp index c9a5425a7..731d1db34 100644 --- a/src/common/string_util.cpp +++ b/src/common/string_util.cpp @@ -5,6 +5,7 @@ #include <algorithm> #include <cctype> #include <cerrno> +#include <codecvt> #include <cstdio> #include <cstdlib> #include <cstring> @@ -13,11 +14,7 @@ #include "common/string_util.h" #ifdef _WIN32 -#include <codecvt> #include <windows.h> -#include "common/common_funcs.h" -#else -#include <iconv.h> #endif namespace Common { @@ -195,11 +192,9 @@ std::string ReplaceAll(std::string result, const std::string& src, const std::st return result; } -#ifdef _WIN32 - std::string UTF16ToUTF8(const std::u16string& input) { -#if _MSC_VER >= 1900 - // Workaround for missing char16_t/char32_t instantiations in MSVC2015 +#ifdef _MSC_VER + // Workaround for missing char16_t/char32_t instantiations in MSVC2017 std::wstring_convert<std::codecvt_utf8_utf16<__int16>, __int16> convert; std::basic_string<__int16> tmp_buffer(input.cbegin(), input.cend()); return convert.to_bytes(tmp_buffer); @@ -210,8 +205,8 @@ std::string UTF16ToUTF8(const std::u16string& input) { } std::u16string UTF8ToUTF16(const std::string& input) { -#if _MSC_VER >= 1900 - // Workaround for missing char16_t/char32_t instantiations in MSVC2015 +#ifdef _MSC_VER + // Workaround for missing char16_t/char32_t instantiations in MSVC2017 std::wstring_convert<std::codecvt_utf8_utf16<__int16>, __int16> convert; auto tmp_buffer = convert.from_bytes(input); return std::u16string(tmp_buffer.cbegin(), tmp_buffer.cend()); @@ -221,6 +216,7 @@ std::u16string UTF8ToUTF16(const std::string& input) { #endif } +#ifdef _WIN32 static std::wstring CPToUTF16(u32 code_page, const std::string& input) { const auto size = MultiByteToWideChar(code_page, 0, input.data(), static_cast<int>(input.size()), nullptr, 0); @@ -261,124 +257,6 @@ std::wstring UTF8ToUTF16W(const std::string& input) { return CPToUTF16(CP_UTF8, input); } -std::string SHIFTJISToUTF8(const std::string& input) { - return UTF16ToUTF8(CPToUTF16(932, input)); -} - -std::string CP1252ToUTF8(const std::string& input) { - return UTF16ToUTF8(CPToUTF16(1252, input)); -} - -#else - -template <typename T> -static std::string CodeToUTF8(const char* fromcode, const std::basic_string<T>& input) { - iconv_t const conv_desc = iconv_open("UTF-8", fromcode); - if ((iconv_t)(-1) == conv_desc) { - LOG_ERROR(Common, "Iconv initialization failure [{}]: {}", fromcode, strerror(errno)); - iconv_close(conv_desc); - return {}; - } - - const std::size_t in_bytes = sizeof(T) * input.size(); - // Multiply by 4, which is the max number of bytes to encode a codepoint - const std::size_t out_buffer_size = 4 * in_bytes; - - std::string out_buffer(out_buffer_size, '\0'); - - auto src_buffer = &input[0]; - std::size_t src_bytes = in_bytes; - auto dst_buffer = &out_buffer[0]; - std::size_t dst_bytes = out_buffer.size(); - - while (0 != src_bytes) { - std::size_t const iconv_result = - iconv(conv_desc, (char**)(&src_buffer), &src_bytes, &dst_buffer, &dst_bytes); - - if (static_cast<std::size_t>(-1) == iconv_result) { - if (EILSEQ == errno || EINVAL == errno) { - // Try to skip the bad character - if (0 != src_bytes) { - --src_bytes; - ++src_buffer; - } - } else { - LOG_ERROR(Common, "iconv failure [{}]: {}", fromcode, strerror(errno)); - break; - } - } - } - - std::string result; - out_buffer.resize(out_buffer_size - dst_bytes); - out_buffer.swap(result); - - iconv_close(conv_desc); - - return result; -} - -std::u16string UTF8ToUTF16(const std::string& input) { - iconv_t const conv_desc = iconv_open("UTF-16LE", "UTF-8"); - if ((iconv_t)(-1) == conv_desc) { - LOG_ERROR(Common, "Iconv initialization failure [UTF-8]: {}", strerror(errno)); - iconv_close(conv_desc); - return {}; - } - - const std::size_t in_bytes = sizeof(char) * input.size(); - // Multiply by 4, which is the max number of bytes to encode a codepoint - const std::size_t out_buffer_size = 4 * sizeof(char16_t) * in_bytes; - - std::u16string out_buffer(out_buffer_size, char16_t{}); - - char* src_buffer = const_cast<char*>(&input[0]); - std::size_t src_bytes = in_bytes; - char* dst_buffer = (char*)(&out_buffer[0]); - std::size_t dst_bytes = out_buffer.size(); - - while (0 != src_bytes) { - std::size_t const iconv_result = - iconv(conv_desc, &src_buffer, &src_bytes, &dst_buffer, &dst_bytes); - - if (static_cast<std::size_t>(-1) == iconv_result) { - if (EILSEQ == errno || EINVAL == errno) { - // Try to skip the bad character - if (0 != src_bytes) { - --src_bytes; - ++src_buffer; - } - } else { - LOG_ERROR(Common, "iconv failure [UTF-8]: {}", strerror(errno)); - break; - } - } - } - - std::u16string result; - out_buffer.resize(out_buffer_size - dst_bytes); - out_buffer.swap(result); - - iconv_close(conv_desc); - - return result; -} - -std::string UTF16ToUTF8(const std::u16string& input) { - return CodeToUTF8("UTF-16LE", input); -} - -std::string CP1252ToUTF8(const std::string& input) { - // return CodeToUTF8("CP1252//TRANSLIT", input); - // return CodeToUTF8("CP1252//IGNORE", input); - return CodeToUTF8("CP1252", input); -} - -std::string SHIFTJISToUTF8(const std::string& input) { - // return CodeToUTF8("CP932", input); - return CodeToUTF8("SJIS", input); -} - #endif std::string StringFromFixedZeroTerminatedBuffer(const char* buffer, std::size_t max_len) { diff --git a/src/common/string_util.h b/src/common/string_util.h index dcca6bc38..32bf6a19c 100644 --- a/src/common/string_util.h +++ b/src/common/string_util.h @@ -72,31 +72,10 @@ std::string ReplaceAll(std::string result, const std::string& src, const std::st std::string UTF16ToUTF8(const std::u16string& input); std::u16string UTF8ToUTF16(const std::string& input); -std::string CP1252ToUTF8(const std::string& str); -std::string SHIFTJISToUTF8(const std::string& str); - #ifdef _WIN32 std::string UTF16ToUTF8(const std::wstring& input); std::wstring UTF8ToUTF16W(const std::string& str); -#ifdef _UNICODE -inline std::string TStrToUTF8(const std::wstring& str) { - return UTF16ToUTF8(str); -} - -inline std::wstring UTF8ToTStr(const std::string& str) { - return UTF8ToUTF16W(str); -} -#else -inline std::string TStrToUTF8(const std::string& str) { - return str; -} - -inline std::string UTF8ToTStr(const std::string& str) { - return str; -} -#endif - #endif /** |