Image dumps fail when an invalid character is present

[TrashBandatcoot]

I've tried to download a pretty messy wiki in the hopes to transfer it to Miraheze, but during the dumping process, the very first image failed because characters are present that cannot be used to create a file or folder in Windows, such as question marks or quotation marks.

IOError: [Errno 22] invalid mode ('wb') or filename: u'C:\Users\XXXXX\Desktop\PTW-2/images/3LameStudio what did you do to my boys?.png'

Removing the file from the wiki altogether fixed it.

[nemobis]

I'm sorry but Windows isn't an officially supported system on which to run wikiteam software. Things are going to be easier in GNU/Linux (e.g. Debian 11), although some are using WSL as well. If you plan to reimport the wiki elsewhere, it's advisable to use a filesystem similar to the source and destination servers (unless you plan to fix all files manually later).

[ohei2]

I recently ran into the same problem, but you don't need to remove the file from the wiki.
Instead use the wiki's option to "move" the file to a new name and skip the offending characters, in your case above the question mark, which has been ever since a placeholder in the DOS/Windows world and thus cannot be used in file names.

[ohei2]

You can use something like this to sanitize filenames for Windows:

def mwSanitizeFname4Windows(fname):
    """ Returns a sanitized filename for Windows """
    # Windows invalid chars in filenames
    # < (less than)  # > (greater than)  # : (colon)
    # " (double quote)  # / (forward slash)  # \ (backslash)
    # | (vertical bar or pipe)  # ? (question mark)  # * (asterisk)
    
    sanName = fname.translate({ord(i): None for i in '<>;"/\|?*'})
    return sanName

In generateImageDump I added the following:

# At the start
MyOS = platform.system()

[...]

       # In the for loop:
        if MyOS == 'Windows':
            if any (i in '<>:\"/\\|\?*' for i in filename2):
                fnametemp = filename2
                filename2 = mwSanitizeFname4Windows(fnametemp)
                print '(Windows) Illegal character(s) in file name. Sanitizing ',fnametemp
                logerror(
                    config=config,
                    text=u'(Windows) Illegal character(s) found. Changing %s to %s' % (fnametemp, filename2)
                )

The logging is done so you will be able to easily access which filenames have been changed.
This worked for me. I got a lot of errors before.
I don't use the XML functionality, but you could sanitize the filenames there too I suppose.

P.S.: I usually don't do Python, so please forgive if I did something unconventional.