UTF-8 problem with IniWriteSection?

[Bresacon]

It looks like IniWriteSection doesn't observe the unicode encoding for the .ini file it processes. Please, run this:

Local $sINIfile = "file.ini"
Local $sINIfile_settings = "settings"
Local $sINIfile_settings_data[5][2] = [[4, ""], ["data1", "1±1"], ["data2", "2±1"], ["data3", "3²"], ["data4", "4³"]]

FileWrite($sINIfile, "[" & $sINIfile_settings & "]" & @CRLF)
For $c = 1 To $sINIfile_settings_data[0][0]
    FileWrite($sINIfile, $sINIfile_settings_data[$c][0] & "=" & $sINIfile_settings_data[$c][1] & @CRLF)
Next

This creates a file called file.ini with some data in ini-like file format. The file gets created by default with a UTF-8 no BOM encoding as it uses ASCII extended characters. Everything's perfect.

Now, if you run this:

Local $sINIfile = "file.ini"
Local $sINIfile_settings = "settings"
Local $sINIfile_settings_data[5][2] = [[4, ""], ["data1", "1±1"], ["data2", "2±1"], ["data3", "3²"], ["data4", "4³"]]

IniWriteSection($sINIfile, $sINIfile_settings, $sINIfile_settings_data)

It creates the same file but with ANSI encoding.

Furthermore, if that .ini file was already existing with UTF-8 no BOM encoding and it gets processed by IniWriteSection it will be rewritten as ANSI.

Am I missing something? Thanks in advance.

[FrancescoDiMuro]

@Bresacon
In the Help file, about IniWriteSection function, it is stated:

If you want to use an ini file with unicode encoding, first create an .ini file by using the FileOpen() function with the mode parameter set to "Unicode UTF16 Little Endian".

Edited January 7, 2019 by FrancescoDiMuro

[Bresacon]

I see. But it looks like doesn't have to be UTF-16 Little Endian. If I use the $FO_UNICODE or $FO_UTF16_LE modes, IniWriteSection keeps the encoding. Ok. But if I create the file with $FO_UTF16_LE_NOBOM or even $FO_UTF8 stil keeps it.

It only appears to be overriding the encoding if it's UTF-8 no BOM.

Edited January 7, 2019 by Bresacon

[BrewManNH]

Try using a Unicode language, such as Hebrew, you'll find that it's not going to work with just UTF-8 with or w/o BOM.

Your example with a bit of translated Hebrew in it, try it without the FileOpen and it won't work correctly.

#include <FileConstants.au3>
; Create a constant variable in Local scope of the filepath that will be read/written to.
Local $sINIfile = "file.ini"
Local $sINIfile_settings = "settings"
FileOpen($sINIfile, BitOR($FO_UTF16_LE, $FO_OVERWRITE))
Local $sINIfile_settings_data[5][2] = [[4, ""], ["data1", "1±1"], ["data2", "2±1"], ["data3", "3²"], ["data4", "לפני 87 שנה אבותינו."]]

IniWriteSection($sINIfile, $sINIfile_settings, $sINIfile_settings_data)
ShellExecute($sINIfile, "", @ScriptDir, "Open")

 

[Bresacon]

I still don't get why IniWriteSection should modify the file encoding from UTF-8 to ANSI.

[BrewManNH]

Because INI files by design are ASCII only files, they aren't Unicode compatible out of the box, and the Windows API that is used to create/modify them aren't Unicode aware.

[Bresacon]

But if I use UTF-16 -as you put in your example- the encoding will stay untouched after IniWriteSection.

Ok, I'll let it be, this is going nowhere. Thanks everybody.