Flake8 extensions/plugin (black, rst-docstrings, and other) entries not reported by vscode

[jraygauthier]

Environment data

• VS Code version: 1.26.1
• Extension version (available under the Extensions sidebar): 2018.7.0
• OS and version: nixos 18.03.git.ccea532 (Impala)
• Python version (& distribution if applicable, e.g. Anaconda): Python 3.6.6
• Type of virtual environment used (N/A | venv | virtualenv | conda | ...): nix
• Relevant/affected Python packages and their versions: None

Expected behaviour

Given the following mymodule.py file:

from collections import \
    namedtuple, \
    deque

class ControlAlgoCoreSimpleSlots:
    """My non pydocstring compliant
    summary which should make `flake8-docstrings` bark.

    Here's some markdown code block (non valid sphinx syntax
    which should make `flake8-rst-docstrings` bark.

    ```
    def my_blocking_closure_fn():
        return long_blocking_call_on(my_argument_data)
    return self.make_blocking_job(my_blocking_closure_fn)
    ```
    """
    pass

which when linted with flake8 and its plugins (see below for details) gives:

$ flake8 '--format=%(row)d,%(col)d,%(code).1s,%(code)s:%(text)s' ./mymodule.py
1,1,F,F401:'collections.namedtuple' imported but unused
1,1,F,F401:'collections.deque' imported but unused
1,1,D,D100:Missing docstring in public module
1,25,B,BLK100:Black would make changes.
5,1,E,E302:expected 2 blank lines, found 1
6,1,D,D204:1 blank line required after class docstring
6,1,D,D205:1 blank line required between summary line and description
6,1,D,D400:First line should end with a period
12,1,R,RST299:Inline literal start-string without end-string.
14,1,R,RST301:Unexpected indentation.
15,1,R,RST201:Block quote ends without a blank line; unexpected unindent.
15,1,R,RST299:Inline literal start-string without end-string.
15,1,R,RST299:Inline interpreted text or phrase reference start-string without end-string.

I would expect to get the same list of issue in vscode.

Note that the format string used is the same as the one used by vscode-python/src/client/linters/flake8.ts to run flake8.

Actual behaviour

Instead, I get:

See that none of the BLK entries from the flake8-black plugin are present, nor any RST entries from the flake8-rst-docstrings plugin. However, the D entries from the flake8-docstrings are properly reported.

The problem is with code prefix of more than a single alphabetical character.

Steps to reproduce:

Make sure you launch vscode in an environement with flake8 installed and the following flake8 plugins:

• flake8-black
• flake8-docstrings
• flake8-rst-docstrings

Make sure vscode has flake8 linter active and global linting:

"python.linting.enabled": true,
"python.linting.lintOnSave": true,
"python.linting.flake8Enabled": true,
"python.linting.flake8Path": "flake8",

Launch Python: Run Linting on the provided mymodule.py module. Look at the Problems listing in vscode.

Logs

Output for Python in the Output panel (View→Output, change the drop-down the upper-right of the Output panel to Python)

##########Linting Output - flake8##########
1,1,F,F401:'collections.namedtuple' imported but unused
1,1,F,F401:'collections.deque' imported but unused
1,1,D,D100:Missing docstring in public module
1,25,B,BLK100:Black would make changes.
5,1,E,E302:expected 2 blank lines, found 1
6,1,D,D204:1 blank line required after class docstring
6,1,D,D205:1 blank line required between summary line and description
6,1,D,D400:First line should end with a period
12,1,R,RST299:Inline literal start-string without end-string.
14,1,R,RST301:Unexpected indentation.
15,1,R,RST201:Block quote ends without a blank line; unexpected unindent.
15,1,R,RST299:Inline literal start-string without end-string.
15,1,R,RST299:Inline interpreted text or phrase reference start-string without end-string.

Output from Console under the Developer Tools panel (toggle Developer Tools on under Help; turn on source maps to make any tracebacks be useful by running Enable source map support for extension debugging)

Console is empty after performing the linting action. I'm not sure about this source map thing however:

Suggested fix

In vscode-python/src/client/linters/baseLinter.ts, change REGEX from:

const REGEX = '(?<line>\\d+),(?<column>-?\\d+),(?<type>\\w+),(?<code>\\w\\d+):(?<message>.*)\\r?(\\n|$)';

which matches only codes with a single alphabetic character

to

const REGEX = '(?<line>\\d+),(?<column>-?\\d+),(?<type>\\w+),(?<code>\\w+\\d+):(?<message>.*)\\r?(\\n|$)';

which should now match codes with more that a single alphabetic character.

[jraygauthier]

Note that as this regexp is not specific to the flake8.ts module but in the baseLinter.ts module, other linters might be affected by the same issue.

[d3r3kk]

You can configure the arguments you wish to use in the "python.linting.flake8Args" setting in your workspace, please see our documentation on linting.

[jraygauthier]

@d3r3kk: Please provide an example of what you suggest because the possibility of customizing flake argument won't allow me to get my extension issues listing in the vscode PROBLEMS tab. The issue is with an hardcoded regexpr in the vscode extension's codebase which couldn't be customized via a flake argument. I specifically listed the problematic module and gave a possible solution.

[peterjc]

This reflects a change in flake8 v3, http://flake8.pycqa.org/en/latest/plugin-development/registering-plugins.html

Please Note: Your entry point does not need to be exactly 4 characters as of Flake8 3.0. Consider using an entry point with 3 letters followed by 3 numbers (i.e. ABC123 ).

The original convention of a single letter and three numbers had lead to a number of flake8 plugin codes clashing.

(Commenting as the author of two of the flake8 plugins mentioned - nice to see them being used)

[peterjc]

Since the triage seems to want this as a pull request, I turned @jraygauthier's sensible suggestion into PR #4443.

Update: I closed the PR, the process was more involved than I expected.

[jraygauthier]

@peterjc thanks for the PR. Can I help in any way to re-open the PR? I yield any copyright on the suggested change (if any).

[peterjc]

@jraygauthier probably simplest to make a new pull request yourself, and if possible do this at home outside of work hours (to avoid the awkward question about if you have your employer's permission to give Microsoft the legal permissions they want).

On the technical side, you need to commit the fix itself in src/client/linters/baseLinter.ts, and they want you to include a short description of the work for their news summary in a file named news/2 Fixes/4443.md (because you are fixing issue 4443).

[jraygauthier]

Alright, will try to take some time to do this.

[FraBle]

FYI: This also applies to wemake-python-styleguide.
The custom warnings for code complexity are not displayed in VS Code, but are visible on the Output for Python.

[adam-grant-hendry]

Has this issue been resolved? I am experiencing this problem as well:

VSCode: 1.67.0
OS: Windows x64 10
Python: 3.8.10
flake8: 4.0.1

[adam-grant-hendry]

Has this issue been resolved? I am experiencing this problem as well:

VSCode: 1.67.0
OS: Windows x64 10
Python: 3.8.10
flake8: 4.0.1

The REGEX fix still does not appear to fix the problem.

[adam-grant-hendry]

The `REGEX` fix still does not appear to fix the problem.

False alarm, apologies. I improperly typed my workspace folder variable as ${workspacefolder} instead of ${workspaceFolder} (capital F). Silly mistake. I didn't get an error that vscode-python couldn't find flake8.exe, so I assumed the path was fine (until I checked the Output tab).

I can confirm the REGEX fix solves the problem for me.

[adam-grant-hendry]

I didn't get an error that vscode-python couldn't find flake8.exe, so I assumed the path was fine (until I checked the Output tab).

FYI, in my case, the OUTPUT tab showed

.\.venv\Scripts\python.exe ~\.vscode\extensions\ms-python.python-2022.6.0\pythonFiles\linter.py -p flake8 ${workspacefolder}/Scripts/flake8.exe the_file.py

and flake8 failed silently.

[karrtikr]

@adam-grant-hendry Can you open a separate issue for "and flake8 failed silently.", we would like to know more details about it.

Also seems like the original bug in the issue was fixed by #6925 but we forgot to close it?

[jraygauthier]

I confirm this is now fixed:

Closing the issue.

[adam-grant-hendry]

@karrtikr Yes, of course. Please see #19086 .