AttributeError: module 'unicodedata' has no attribute 'name'

[JeroenRoodIHS]

Hi, apologies for the dumb question, but I am a complete novice with fontTools or Python in general. I am trying to discover all codepoints in a TTF font file, and I discovered through discussions on StackOverflow that fontTools' TTX might be able to help through extracting font information in XML.

I do have an issue: when I try to run the command python ttx.py -d "path/to/font/file.ttf", I run into the issue AttributeError: module 'unicodedata' has no attribute 'name' .

I don't really know what to do or how to troubleshoot. Would anyone be able to help me out?

[behdad]

What Python version?

[JeroenRoodIHS]

Hi Behdad! I am on Python 3.10.4.

[behdad]

I have no idea.

[JeroenRoodIHS]

No worries, thank you for considering. 👍

[justvanrossum]

It may help if you post a full trace back.

You don’t happen to have a script named unicodedata.py in the current directory?

[JeroenRoodIHS]

Hi Justvanrossum! Ah, of course. I should have known to post it along. :)

Traceback (most recent call last):
  File "C:\Python310\Lib\site-packages\fontTools\ttx.py", line 454, in main
    process(jobs, options)
  File "C:\Python310\Lib\site-packages\fontTools\ttx.py", line 436, in process
    action(input, output, options)
  File "C:\Python310\lib\site-packages\fontTools\misc\loggingTools.py", line 375, in wrapper
    return func(*args, **kwds)
  File "C:\Python310\Lib\site-packages\fontTools\ttx.py", line 297, in ttDump
    ttf.saveXML(
  File "C:\Python310\lib\site-packages\fontTools\ttLib\ttFont.py", line 275, in saveXML
    self._saveXML(writer, **kwargs)
  File "C:\Python310\lib\site-packages\fontTools\ttLib\ttFont.py", line 341, in _saveXML
    self._tableToXML(tableWriter, tag, splitGlyphs=splitGlyphs)
  File "C:\Python310\lib\site-packages\fontTools\ttLib\ttFont.py", line 373, in _tableToXML
    table.toXML(writer, self)
  File "C:\Python310\lib\site-packages\fontTools\ttLib\tables\_c_m_a_p.py", line 221, in toXML
    table.toXML(writer, ttFont)
  File "C:\Python310\lib\site-packages\fontTools\ttLib\tables\_c_m_a_p.py", line 318, in toXML
    self._writeCodes(codes, writer)
  File "C:\Python310\lib\site-packages\fontTools\ttLib\tables\_c_m_a_p.py", line 349, in _writeCodes
    writer.comment(Unicode[code])
  File "C:\Python310\lib\site-packages\fontTools\unicode.py", line 40, in __getitem__
    return unicodedata.name(chr(charCode))
AttributeError: module 'unicodedata' has no attribute 'name'

I executed the command in C:\Python310\Lib\site-packages\fontTools, and in that directory, unicode.py is present.

[justvanrossum]

Unfortunately I also don't have a clue either at this point. I would not execute the command literally in C:\Python310\Lib\site-packages\fontTools, though (meaning it being the current directory that you cd'ed into).

[ea]

It seems like, by default, the code tries to import an existing "unicodedata2" package as unicodedata instead of relying on the one that comes with repository. pip3 install unicodedata2 solves this issues , see this bit of code:

fonttools/Lib/fontTools/unicode.py

Line 39 in bd58e66

return unicodedata.name(chr(charCode))

If unicodedata2 isn't available, it defaults back to unicodedata package from fonttoolsLib, which is likely outdated ?

[ea]

It seems to me that fontTools.unicodedata itself is creating a circular import when trying to do:

from unicodedata import *

[anthrotype]

python3 /Users/user/Library/Python/3.9/lib/python/site-packages/fontTools/ttx.py

that's not how you should invoke ttx script... If you installed fonttools (e.g. via pip install fonttools from PyPI releases or pip install -e . from local git checkout), then you simply run the ttx command, or using the fonttools CLI entry points with fonttools ttx, or if you really want to be explicit about which python interpreter to use python3 -m fontTools.ttx

Running the ttx.py module as a script like you're doing is not supported, and might be the cause of the issue somehow

[ea]

yep , i guess that's it. I didn't realize the import semantics would be different. And I guess that might be the OP's original issue as well.

[anthrotype]

I'll investigate more why this is happening, will get back to you. thanks for bringing this up

[anthrotype]

oh yes I see what's going on now. Basically when you invoke ttx.py module as a script, eg. python3 Lib/fontTools/ttx.py, then the directory where the ttx.py file is located will be at the top of the sys.path list, which makes it so that when later on we attempt to import unicodedata, then it's not the built-in unicodedata that gets imported by actually the unicodedata.py module file that is located inside the Lib/fontTools directory... since it has the same name and the Lib/fontTools directory was added to the sys.path. That's how python works, when a script is run the directory where the script is located gets inserted in the sys.path as the first item. Nothing we can do, besides encouraging to run ttx and other scripts after having installed the package via the CLI entry points, or if the latter aren't available or are not in the shell's $PATH for some reasons, you can always run /path/to/your/python3 -m fontTools.ttx like any other runnable modules (not fonttools specific).