Using Unicode in INI files

Introduction

This article shows how to use Unicode in an INI file. Even though we call WritePrivateProfileStringW API function for using Unicode in an INI file, an INI file is not saved as Unicode but saved as ANSI. However, the small code that I show here will let you use that.

What does WritePrivateProfileStringW do?

From my experience, I explain that WritePrivateProfileStringW function copies a string encoded to Unicode (UTF18-little Endian) into an INI file which can be written with UTF18-little Endian encoded character. If the INI file for an application does not exist, the API function will create a plain text file and write a section name and a key name and so on. The plain text file is not a Unicode format text file but an ANSI format text file. So a Unicode format file has to have the "BOM" in the beginning of a text file. When the Unicode format file already exists as INI file, this function works well. After all, the function does not prepare it, and somebody has to prepare it.

How is the file created?

• Manually creating the INI file
• Preparing the INI file by coding

Manually creating the INI file

This is the simplest way. We save the ANSI INI file already existing as UTF16-little Endian using a text editor such as Notepad. This method is suitable for trials.

Preparing the INI file by coding

The difference between a Unicode and an ANSI format file is whether a file has the BOM in the beginning of the file or not. Besides, the BOM has to be UTF16-little Endian's BOM. So, we have to create a plain file and then add the BOM of UTF16-little Endian into the beginning of the file before using initially WritePrivateProfileW.

In the following code, when the INI file is not found, the new INI file is created. Do notice the code uses Generic Text Mappings.

LPTSTR pszINIFilePath;
::GetModuleFileName(NULL, pszINIFilePath, MAX_PATH);
::PathRemoveFileSpec(pszINIFilePath);
::StrCat(pszINIFilePath, _T("\\App.ini"));

LPTSTR pszSectionB = _T("[StringTable]"); // section name with bracket 
LPTSTR pszSection = _T("StringTable"); // section name without bracket
LPTSTR pszKey = _T("String");

if(!::PathFileExists(pszINIFilePath))
{

    // UTF16-LE BOM(FFFE)
    WORD wBOM = 0xFEFF;
    DWORD NumberOfBytesWritten;

    HANDLE hFile = ::CreateFile(pszINIFilePath, GENERIC_WRITE, 0, 
                   NULL, CREATE_NEW, FILE_ATTRIBUTE_NORMAL, NULL);
    ::WriteFile(hFile, &wBOM, sizeof(WORD), &NumberOfBytesWritten, NULL);
    ::WriteFile(hFile, pszSectionB, (_tcslen(pszSectionB)+1)*(sizeof(TCHAR)), 
                &NumberOfBytesWritten, NULL);
    ::CloseHandle(hFile);
}
 
WritePrivateProfileString(pszSection, pszKey, _T(""), pszINIFilePath);

Points of Interest

In the code, I added not only the BOM but also the first section name into the file by using WriteFile() function. This coding can avoid making an empty line in the first line of the file.

Why is it only UTF16-little Endian?

In Notepad included with Windows, we can choose 3 encoding formats in Unicode. These are "Unicode" (UTF16-little Endian), "Unicode big Endian" (UTF16-big Endian), and "UTF-8". We can use only UTF16-little endian of these formats as an INI file format. The other encodings do not work correctly (you examine it once). Probably, the reason is that Windows NT, 2000 or XP uses the encoding internally. This is why Windows particularly names UTF16-little Endian "Unicode".

Reference

• Unicode Enabled - about Unicode around Windows