unicode.hpp

/*
    SPDX-License-Identifier: MIT
    Copyright © 2023 Amebis
*/
 
#pragma once
 
#include "sal.hpp"
#include <assert.h>
#ifdef _WIN32
#include <windows.h>
#endif
#include <memory>
#include <string>
 
namespace stdex
{
    enum class charset_id {
        default = 0,
#ifdef _WIN32
        utf8 = CP_UTF8,
#endif
    };
 
    inline void str2wstr(
        _Inout_ std::wstring& dst,
        _In_reads_or_z_opt_(count_src) const char* src, _In_ size_t count_src,
        _In_ charset_id charset = charset_id::default)
    {
        assert(src || !count_src);
#ifdef _WIN32
        assert(count_src < INT_MAX || count_src == SIZE_MAX);
        constexpr DWORD dwFlags = MB_PRECOMPOSED;
 
        // Try to convert to stack buffer first.
        WCHAR szStackBuffer[1024/sizeof(WCHAR)];
        int cch = MultiByteToWideChar(static_cast<UINT>(charset), dwFlags, src, static_cast<int>(count_src), szStackBuffer, _countof(szStackBuffer));
        if (cch) {
            // Append from stack.
            dst.append(szStackBuffer, count_src != SIZE_MAX ? wcsnlen(szStackBuffer, cch) : (size_t)cch - 1);
        } else if (::GetLastError() == ERROR_INSUFFICIENT_BUFFER) {
            // Query the required output size. Allocate buffer. Then convert again.
            cch = MultiByteToWideChar(static_cast<UINT>(charset), dwFlags, src, static_cast<int>(count_src), NULL, 0);
            std::unique_ptr<WCHAR[]> szBuffer(new WCHAR[cch]);
            cch = MultiByteToWideChar(static_cast<UINT>(charset), dwFlags, src, static_cast<int>(count_src), szBuffer.get(), cch);
            dst.append(szBuffer.get(), count_src != SIZE_MAX ? wcsnlen(szBuffer.get(), cch) : (size_t)cch - 1);
        }
#else
        throw std::exception("not implemented");
#endif
    }
 
    inline void str2wstr(
        _Inout_ std::wstring& dst,
        _In_ const std::string& src,
        _In_ charset_id charset = charset_id::default)
    {
        str2wstr(dst, src.data(), src.size(), charset);
    }
 
    inline std::wstring str2wstr(
        _In_reads_or_z_opt_(count_src) const char* src, _In_ size_t count_src,
        _In_ charset_id charset = charset_id::default)
    {
        std::wstring dst;
        str2wstr(dst, src, count_src, charset);
        return dst;
    }
 
    inline std::wstring str2wstr(
        _In_ const std::string& src,
        _In_ charset_id charset = charset_id::default)
    {
        return str2wstr(src.c_str(), src.size(), charset);
    }
 
    inline void wstr2str(
        _Inout_ std::string& dst,
        _In_reads_or_z_opt_(count_src) const wchar_t* src,
        _In_ size_t count_src,
        _In_ charset_id charset = charset_id::default)
    {
        assert(src || !count_src);
#ifdef _WIN32
        assert(count_src < INT_MAX || count_src == SIZE_MAX);
        constexpr DWORD dwFlags = 0;
        constexpr LPCCH lpDefaultChar = NULL;
 
        // Try to convert to stack buffer first.
        CHAR szStackBuffer[1024/sizeof(CHAR)];
        int cch = WideCharToMultiByte(static_cast<UINT>(charset), dwFlags, src, static_cast<int>(count_src), szStackBuffer, _countof(szStackBuffer), lpDefaultChar, NULL);
        if (cch) {
            // Copy from stack. Be careful not to include zero terminator.
            dst.append(szStackBuffer, count_src != SIZE_MAX ? strnlen(szStackBuffer, cch) : (size_t)cch - 1);
        } else if (::GetLastError() == ERROR_INSUFFICIENT_BUFFER) {
            // Query the required output size. Allocate buffer. Then convert again.
            cch = WideCharToMultiByte(static_cast<UINT>(charset), dwFlags, src, static_cast<int>(count_src), NULL, 0, lpDefaultChar, NULL);
            std::unique_ptr<CHAR[]> szBuffer(new CHAR[cch]);
            cch = WideCharToMultiByte(static_cast<UINT>(charset), dwFlags, src, static_cast<int>(count_src), szBuffer.get(), cch, lpDefaultChar, NULL);
            dst.append(szBuffer.get(), count_src != SIZE_MAX ? strnlen(szBuffer.get(), cch) : (size_t)cch - 1);
        }
#else
        throw std::exception("not implemented");
#endif
    }
 
    inline void wstr2str(
        _Inout_ std::string& dst,
        _In_ const std::wstring& src,
        _In_ charset_id charset = charset_id::default)
    {
        wstr2str(dst, src.c_str(), src.size(), charset);
    }
 
    inline std::string wstr2str(
        _In_reads_or_z_opt_(count_src) const wchar_t* src,
        _In_ size_t count_src,
        _In_ charset_id charset = charset_id::default)
    {
        std::string dst;
        wstr2str(dst, src, count_src, charset);
        return dst;
    }
 
    inline std::string wstr2str(
        _In_ const std::wstring& src,
        _In_ charset_id charset = charset_id::default)
    {
        return wstr2str(src.c_str(), src.size(), charset);
    }
}