Fix Windows encoding convertion

[mizunashi-mana]

I got an error on run phing.

> phing prepare
Notice: iconv(): Wrong charset, conversion from `UTF-8' to `Windows-932//IGNORE' is not allowed in vendor\phing\phing\classes\phing\system\io\Win32FileSystem.php on line 635
Buildfile:  does not exist!

Now, phing use iconv or mb_convert_encoding for fixing windows encoding.
However, Windows-${codepage} is not usual codepage encoding format.

My cmd.exe use codepage 932.
CP932 exists in supported encoding for iconv and mbstring, but Windows-932 is not.
So, fixEncoding function failed and returned not string...

I request to use CP for encoding prefix.
Thanks!

Refs

• https://www.gnu.org/software/libiconv/
• http://php.net/manual/ja/mbstring.supported-encodings.php

[mrook]

Changing "Windows-" to "CP-" will probably cause a regression in 84e4050 / https://www.phing.info/trac/ticket/1198

Is there some other way to determine available codepages? I don't do much Windows development, and I don't have any special codepages set on my Windows machine.

[mizunashi-mana]

@mrook Thank you for feedback!

I try on PHP 7.4 (on psysh v0.7.2):

iconv('UTF-8', 'Windows-1251//IGNORE', 'something'); // success
iconv('UTF-8', 'CP1251//IGNORE', 'something'); // success

If iconv receive not support encoding, return false by ignore.

mb_convert_encoding('something', 'Windows-1251', 'UTF-8'); // success
mb_convert_encoding('something', 'CP1251', 'UTF-8'); // success

If mb_convert_encoding receive not support encoding, throw error.

It seems to be no problem.
On https://gist.github.com/hakre/4188459 (iconv encodings) and http://php.net/manual/ja/mbstring.supported-encodings.php (mbstring encodings), CP- is more supported than Windows- and there are CP- aliases for all of Windows- encoding.
If we trust documents on the net, this request will be fine on Windows ;)

My Japanese Windows is using CP932. This is usual code page in Japan. However, CP932 doesn't have Windows- alias unfortunately...

[siad007]

As shown in the iconv list of encodings https://gist.github.com/hakre/4188459 Windows- is the alias for CP as mentioned by @mizunashi-mana, there will be no regression to this. But i would suggest changeing setlocale(LC_CTYPE, 0) to setlocale(LC_CTYPE, '') to get allways the default locale under windows and not an C if not defined in settings properly.

[mrook]

Thanks for the updates! @mizunashi-mana, can you update your pull request with @siad007 's comments? I'll then merge it.

[mizunashi-mana]

@mrook @siad007 I updated this request!