bool convertToUnicode(int srcCodepage, const char * src, size_t * srcbytes, wchar_t * dest, size_t *destchars)
{
UINT uisrcbytes = static_cast<UINT>(*srcbytes), uidestchars = static_cast<UINT>(*destchars);
HRESULT hr = m_pmlang->ConvertStringToUnicode(&m_mlangcookie, srcCodepage, (char *)src, &uisrcbytes, dest, &uidestchars);
*srcbytes = uisrcbytes;
*destchars = uidestchars;
return SUCCEEDED(hr) ? true : false;
}
std::wstring load_text_file(const std::wstring &path, uint32_t codepage)
{
struct F {
static void release(IUnknown *x) { x->Release(); }
};
IStream *stream;
HRESULT hr = SHCreateStreamOnFileW(path.c_str(),
STGM_READ | STGM_SHARE_DENY_WRITE,
&stream);
if (FAILED(hr)) win32::throw_error(path, hr);
std::shared_ptr<IStream> streamPtr(stream, F::release);
LARGE_INTEGER li = { 0 };
ULARGE_INTEGER ui;
HR(stream->Seek(li, STREAM_SEEK_END, &ui));
if (ui.QuadPart > 0x100000) {
throw std::runtime_error(strutil::w2us(path + L": file too big"));
}
size_t fileSize = ui.LowPart;
HR(stream->Seek(li, STREAM_SEEK_SET, &ui));
IMultiLanguage2 *mlang;
HR(CoCreateInstance(CLSID_CMultiLanguage, 0, CLSCTX_INPROC_SERVER,
IID_IMultiLanguage2, (void**)(&mlang)));
std::shared_ptr<IMultiLanguage2> mlangPtr(mlang, F::release);
if (!codepage) {
DetectEncodingInfo encoding[5];
INT nscores = 5;
HR(mlang->DetectCodepageInIStream(0, GetACP(),
stream, encoding, &nscores));
/*
* Usually DetectCodepageInIStream() puts the most appropriate choice
* in the first place.
* However, it tends to pick 8bit locale charset for the first place,
* even if it is really an UTF-8 encoded file.
*/
codepage = encoding[0].nCodePage;
for (size_t i = 0; i < nscores; ++i)
if (encoding[i].nCodePage == 65001) {
codepage = 65001;
break;
}
HR(stream->Seek(li, STREAM_SEEK_SET, &ui));
}
std::vector<char> ibuf(fileSize);
ULONG nread;
HR(stream->Read(&ibuf[0], ibuf.size(), &nread));
DWORD ctx = 0;
UINT size = ibuf.size(), cnt;
HR(mlang->ConvertStringToUnicode(&ctx, codepage,
&ibuf[0], &size, 0, &cnt));
std::vector<wchar_t> obuf(cnt);
size = ibuf.size();
HR(mlang->ConvertStringToUnicode(&ctx, codepage,
&ibuf[0], &size, &obuf[0], &cnt));
obuf.push_back(0);
// chop off BOM
size_t bom = obuf.size() && obuf[0] == 0xfeff;
return strutil::normalize_crlf(&obuf[bom], L"\n");
}
