File name convention, reserved file names incomplete?

[madig]

According to https://docs.microsoft.com/en-gb/windows/win32/fileio/naming-a-file?redirectedfrom=MSDN#naming-conventions, the reserved file names are:

Do not use the following reserved names for the name of a file:

CON, PRN, AUX, NUL, COM1, COM2, COM3, COM4, COM5, COM6, COM7, COM8, COM9, LPT1, LPT2, LPT3, LPT4, LPT5, LPT6, LPT7, LPT8, and LPT9. Also avoid these names followed immediately by an extension; for example, NUL.txt is not recommended. For more information, see Namespaces.

Should the spec be upated to contain only those? Note how "CLOCK$" and drive letters aren't mentioned (colons are replaced already anyway).

There's also https://docs.microsoft.com/en-us/openspecs/windows_protocols/ms-fscc/b04c3bd0-79dc-4e58-b8ed-74f19fc2ea0a. Not sure they should be reserved names as well?

[madig]

Also, I can create a file named "hello.com1.txt" on my Windows 10 21H2 (and some Windows XP) PC. Where does the

Use period as a delimiter to split name into parts.

part in the spec come from? Is it maybe enough to just look at the string as a whole?

Edit: I can't create a file named "com1 " (note the space) but "com1 .txt" is fine. userNameToFileName won't prevent the former.

[madig]

Also, the spec says that e.g. "C:" is a reserved file name, but I don't think this needs to be checked, because the colon is already unconditionally replaced in the first step.

[madig]

The following links say somewhat inconsistent things about what is illegal and reserved:

• https://docs.microsoft.com/en-us/windows/win32/fileio/naming-a-file#naming-conventions
• https://en.wikipedia.org/wiki/Filename#Reserved_characters_and_words
• https://en.wikipedia.org/wiki/Filename#Comparison_of_filename_limitations

I also found that I can't name things e.g. "com0.txt" on my Windows machine in Explorer, but can in PowerShell. Hm.

So, maybe the algorithm needs to be changed a bit. If we say we support (V|ex)FAT, NTFS, HFS(+) and any POSIXy filesystem, as the 255 character limit suggests, and none of the old DOS crap, it could instead ensure:

• Maximum file name length as before, 255.
• Illegal characters to be replaced by an underscore: [':', '?', '"', '(', ')', '[', ']', '*', '/', '\\', '+', '<', '>', '|'] (Not sure why "()" are disallowed by the spec?)
• Reserved (full!) file names to be prefixed by an underscore: ["con", "prn", "aux", "nul", "com1", "com2", "com3", "com4", "com5", "com6", "com7", "com8", "com9", "lpt1", "lpt2", "lpt3", "lpt4", "lpt5", "lpt6", "lpt7", "lpt8", "lpt9"] (Note that this means "com1.txt" and "com1" but not "hello.com1.txt" and "hello.com1")
• Trailing periods and spaces of the full name are replaced by underscores.
• Uniqueness as before.

How'd that sound?

[benkiel]

That sounds/seems fine. I would venture to guess that the last item (Trailing periods and spaces) may be a breaking change. For the ( & ) rule; I'm not sure why they aren't allowed, but I would not remove them from that list; not allowing them hasn't been an issue so far, no need to touch want isn't broken.

[benkiel]

(@madig I'm assuming you may PR this change?)

[madig]

I started something at https://github.com/fonttools/fonttools/pull/2506 -- if that's ok to merge, I can expand the spec.

[benkiel]

That PR looks good to merge, but I'll let others chime in there. Thank you for doing the spec update!

[madig]

Leftover todo after merged #204:

1. Reserved filename escaping can be less aggressive that it is currently because only the part before the first period is relevant (i.e. "con" and "con.txt" but not "hello.con" and "hello.con.txt")
2. Trailing periods and spaces of the full name must be escaped for Windows filesystems.