Improve explanation of how to ignore per file if the file name contains spaces

[1kastner]

This discussion started at PyCQA/flake8#1519.

I have a file called 01 Einfuehrung in Python.ipynb and I would like to ignore specific issues here. In the file, I harm some design principles purposefully and show the user how this creates an issue lateron. However, my per-file-ignores are not considered. I am not sure how exactly the glob pattern is working. There are a few examples, e.g. at https://flake8.pycqa.org/en/latest/user/options.html?highlight=glob%20#cmdoption-flake8-filename but they are not exhausting enough for me.

Extract of my config:

per-file-ignores =
    "01 Einfuehrung in Python.ipynb": F403, F405, E116, E731

With or without hyphens around the file name, in both cases the extract of the console output looks like this:

01-einfuehrung-in-toolbox\01 einfuehrung in python.ipynb#In[14]:1:19: E226 missing whitespace around arithmetic operator
01-einfuehrung-in-toolbox\01 einfuehrung in python.ipynb#In[14]:3:1: E731 do not assign a lambda expression, use a def
01-einfuehrung-in-toolbox\01 einfuehrung in python.ipynb#In[23]:2:23: E226 missing whitespace around arithmetic operator
01-einfuehrung-in-toolbox\01 einfuehrung in python.ipynb#In[23]:4:1: F403 'from math import *' used; unable to detect undefined names
01-einfuehrung-in-toolbox\01 einfuehrung in python.ipynb#In[23]:8:7: F405 'sin' may be undefined, or defined from star imports: math
01-einfuehrung-in-toolbox\01 einfuehrung in python.ipynb#In[23]:8:11: F405 'pi' may be undefined, or defined from star imports: math
01-einfuehrung-in-toolbox\01 einfuehrung in python.ipynb#In[23]:8:13: E226 missing whitespace around arithmetic operator

So I would wish to be able how exactly the glob patterns work but in the documentation I could not find sufficient information how to deal with spaces in file names. Thanks a lot for your assistance!

[s-weigand]

Sorry to "send you back" (I will file the issue for that) but this is an upstream issue with flake8😅

Cause

I did a quick test and the problem is with the spaces in the path, since flake8 allows you to define a list of files as well as a list of errors codes for per-file-ignores which might be separated with commas or white spaces or a mix of them.

[flake8_nb]
per-file-ignores =
    ; Single file and single error code
    file1.py:E231
    ; Single file and multiple error codes
    file1.py:E231,F401
    ; Multiple files and single error code
    file1.py file2.py:E231
    ; The following are all equivalent, for multiple files and multiple error code
    file1.py file2.py:E231,E401
    file1.py file2.py:E231 E401
    file1.py,file2.py:E231,E401

This is why 01 Einfuehrung in Python.ipynb gets split up and interpreted as 4 files (01, Einfuehrung, in and Python.ipynb) and the pattern doesn't match.

Since adding support for file paths with white spaces in it would most likely be a breaking change (e.g. file1.py file2.py would be a interpreted as a single file) and files with spaces would only be usable as a script, I'm not sure if flake8 will add support for it.

Workarounds

1. Just replace the with -! Having spaces in your file path leads to problems all over the place (one prime example is LaTeX),
2. (Only if you really insist on the spaces, since it can lead to miss matches) Replace with * in the pattern.

[flake8_nb]
per-file-ignores =
    01*Einfuehrung*in*Python.ipynb: F403, F405, E116, E731

[1kastner]

Thank you very much @s-weigand for your comment and support!