Encoding problems with uploading media files

[pavelevap]

Testing media filename: "Žluťoučký kůň.png"

1) WP 3.4.1 (standard Linux webhosting, using Chrome):

• File successfully uploaded as "Žluťoučký-kůň.png" (space replaced).
• But filename is displayed as "luťoučký-kůň.png".
• URL is ​http://domain/wp-content/uploads/2012/07/Žluťoučký-kůň.png (and file is also accessible on this URL without problem).

So, why is filename displayed without first character?

2) WP 3.4.1 (xampp, Windows computer, Chrome):

• File uploaded as "Žluťoučký-kůň.png" (space replaced).
• Filename is displayed as "Žluťoučký-kůň.png".
• URL is ​http://localhost/wp-content/uploads/Žluťoučký-kůň.png
• But file is not visible and in uploads directory can be seen as "Ĺ˝luĹĄouÄŤkĂ˝-kůň.png".

In this case I can see this media file only encoded by Javascript (image editing tool), for example on URL ​http://localhost/wp-admin/admin-ajax.php?action=imgedit-preview&_ajax_nonce=8f7fa9ba05&postid=21249&rand=49045

[SergeyBiryukov]

Point 2 is specific to Windows builds of PHP (see ticket:15955:11, #18634, #19842).

Point 1 is caused by basename() usage to get the file name in get_media_item():
​http://core.trac.wordpress.org/browser/tags/3.4.1/wp-admin/includes/media.php#L1063

From ​php.net:

basename() is locale aware, so for it to see the correct basename with multibyte character paths, the matching locale must be set using the setlocale() function.

21217.patch​ fixes point 1 using one of the workarounds suggested there.

[nacin]

We also have wp_basename().

[SergeyBiryukov]

Replying to nacin:

We also have wp_basename().

Thanks for the hint, this works too. wp_basename() is used for similar tasks in image_downsize(), image_resize() and a few other places.

[pavelevap]

SergeyBiryukov: Thank you, Sergey. I did not notice this ticket related to second point.

[bpetty]

21217.2.patch​ looks good, and is a quick win for fixing point (1), and point (2) is a duplicate of #15955, so if @SergeyBiryukov's patch is applied, we can close out this ticket.

[bpetty]

While I didn't add unit tests for get_media_item() patched here (it's way too large of a method to unit test), I did however add ​unit tests for wp_basename for this situation.

[ryan]

In 22367:

Use wp_basename() instead of basename() so that multibyte characters are not stomped. Props SergeyBiryukov. fixes #21217