[Openexr-devel] RE: UNICODE support in openexr file I/O

[Nick Porcino]

Hello list, I am stumped.

I am trying to create a file named Kyoto-to in Kanji. name below is the UTF-8 
encoding of Kyoto-to. I've tried a number of experiments, but all yield a 
goobledy-gook filename. I have appended my tests below.

If someone can provide the correct answer that would be swell!



const char name[] =
{0xe4, 0xba, 0xac, 0xe9, 0x83, 0xbd, 0xe5, 0xb8, 0x82, 0x23, 0};





TEST ONE - fopen


fopen(name, "wb") yields a file called:

京都市#

(random junk)






TEST TWO - CreateFile (windows)

    HANDLE hFile = CreateFile(TEXT(name),    // file to open
        GENERIC_WRITE,          // open for writing
        FILE_SHARE_WRITE,       // share for writing
        NULL,                  // default security
        CREATE_ALWAYS,         // create file only
        FILE_ATTRIBUTE_NORMAL, // normal file
        NULL);   

äºéƒ½å¸‚#

(random junk)



TEST THREE - _wfopen


    _wfopen((const wchar_t*)name, (const wchar_t*)"wb");

äºéƒ½å¸‚#

(random junk)






TEST FOUR - _wfopen + UTF-8 -> 16 conversion

Finally:


using std::wstring;

wstring 
toWideString( const char* pStr , int len )
{
    // figure out how many wide characters we are going to get 
    int nChars = MultiByteToWideChar( CP_ACP , 0 , pStr , len , NULL , 0 ) ; 
    if ( len == -1 )
        -- nChars ; 
    if ( nChars == 0 )
        return L"" ;

    wstring buf ;
    buf.resize( nChars ) ; 
    MultiByteToWideChar( CP_ACP , 0 , pStr , len , 
        const_cast<wchar_t*>(buf.c_str()) , nChars ) ; 

    return buf ;
}

int _tmain(int argc, _TCHAR* argv[])
{
    wstring wname = toWideString(name, strlen(name));
    FILE* x = _wfopen(wname.c_str(), (const wchar_t*)"wb");
        return 0;
}

äºéƒ½å¸‚#

(random junk)



-----Original Message-----
From: Florian Kainz [mailto:address@hidden 
Sent: Wednesday, January 25, 2006 3:17 PM
To: Nick Porcino
Cc: Drew Hess
Subject: Re: [Openexr-devel] RE: UNICODE support in openexr file I/O

According to a hex dump of a UTF-8-encoded e-mail message that I sent
to myself, the byte sequence for 京都市 is:

     const char name[] =
         {0xe4, 0xba, 0xac, 0xe9, 0x83, 0xbd, 0xe5, 0xb8, 0x82, 0x23};


Nick Porcino wrote:
> I'm kind of pressed for time... I can try something tho'
> 
> Florian, could you give me something like this -
> 
> char name[] = { 0xa, 0x21, 0x23, ...., 0 };
> 
> that UTF-8 encodes one of those strings like the name of Bangkok or
> whatever and I'll write a little test app for you
> 
> -----Original Message-----
> From: Drew Hess [mailto:address@hidden 
> Sent: Wednesday, January 25, 2006 10:25 AM
> To: Nick Porcino
> Cc: Florian Kainz
> Subject: Re: [Openexr-devel] RE: UNICODE support in openexr file I/O
> 
> 
> "Nick Porcino" <address@hidden> writes:
> 
>>The code I posted earlier converts between UTF-8 and 16. So if you do
>>the conversion of the strings 8<>16 outside of OpenEXR you are covered
>>for OpenEXR's internal strings.
>>
>>Santiago, are you saying that if you convert UTF-16 to UTF-8, then use
>>the converted string as a filename, that the OS itself displays the
>>filename as garbage?
>