unsigned char* web_page::load_utf8_file( LPCWSTR path, bool is_html, LPCWSTR defEncoding )
{
unsigned char* ret = NULL;
HANDLE fl = CreateFile(path, GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL);
if(fl != INVALID_HANDLE_VALUE)
{
DWORD size = GetFileSize(fl, NULL);
ret = new unsigned char[size + 1];
DWORD cbRead = 0;
if(size >= 3)
{
ReadFile(fl, ret, 3, &cbRead, NULL);
if(ret[0] == '\xEF' && ret[1] == '\xBB' && ret[2] == '\xBF')
{
ReadFile(fl, ret, size - 3, &cbRead, NULL);
ret[cbRead] = 0;
} else
{
ReadFile(fl, ret + 3, size - 3, &cbRead, NULL);
ret[cbRead + 3] = 0;
}
}
CloseHandle(fl);
}
// try to convert encoding
if(is_html)
{
std::wstring encoding;
char* begin = StrStrIA((LPSTR) ret, "<meta");
while(begin && encoding.empty())
{
char* end = StrStrIA(begin, ">");
char* s1 = StrStrIA(begin, "Content-Type");
if(s1 && s1 < end)
{
s1 = StrStrIA(begin, "charset");
if(s1)
{
s1 += strlen("charset");
while(!isalnum(s1[0]) && s1 < end)
{
s1++;
}
while((isalnum(s1[0]) || s1[0] == '-') && s1 < end)
{
encoding += s1[0];
s1++;
}
}
}
if(encoding.empty())
{
begin = StrStrIA(begin + strlen("<meta"), "<meta");
}
}
if(encoding.empty() && defEncoding)
{
encoding = defEncoding;
}
if(!encoding.empty())
{
if(!StrCmpI(encoding.c_str(), L"UTF-8"))
{
encoding.clear();
}
}
if(!encoding.empty())
{
CoInitialize(NULL);
IMultiLanguage* ml = NULL;
HRESULT hr = CoCreateInstance(CLSID_CMultiLanguage, NULL, CLSCTX_INPROC_SERVER, IID_IMultiLanguage, (LPVOID*) &ml);
MIMECSETINFO charset_src = {0};
MIMECSETINFO charset_dst = {0};
BSTR bstrCharSet = SysAllocString(encoding.c_str());
ml->GetCharsetInfo(bstrCharSet, &charset_src);
SysFreeString(bstrCharSet);
bstrCharSet = SysAllocString(L"utf-8");
ml->GetCharsetInfo(bstrCharSet, &charset_dst);
SysFreeString(bstrCharSet);
DWORD dwMode = 0;
UINT szDst = (UINT) strlen((LPSTR) ret) * 4;
LPSTR dst = new char[szDst];
if(ml->ConvertString(&dwMode, charset_src.uiInternetEncoding, charset_dst.uiInternetEncoding, (LPBYTE) ret, NULL, (LPBYTE) dst, &szDst) == S_OK)
{
dst[szDst] = 0;
delete ret;
ret = (unsigned char*) dst;
} else
{
delete dst;
}
CoUninitialize();
}
}
return ret;
}
