While having an option to specify these files using configuration file is cool, in my use case, it is not the best.

I am creating a tool for code quality checks, which uses pylint and mypy and I want to be able to integrate bellybutton to it. Unfortunately bellybutton doesn't allow to specify, on which files it should run using command line. This turns out to be very limiting, because my tool could be launched on specified files only.

I think, it would be better, if user could specify the files using command line, not only using the configuration file.

I am willing to implement this myself, if you don't {want|have time} to do it yourself, if you like it.

Version for pyyaml has been restricted because it has started to require an explicit loader for loading yaml files (see #20). By applying this commit, we can now remove the restriction, since we explicitly specify the loader.

We had to bump the lower bound for pyyaml to >=4.0 because 3.x versions did not have a FullLoader, which is required for this to work properly.

Bellybutton doesn't like this:

  filter_by:
    description: Use SQLAlchemy's filter_by for concision
    # Find calls to "filter" that:
    # 
    # - are chained after a call to "query" with one argument,
    # - have one argument,
    # - which is an equality comparison
    #   * which has a LHS which is an attribute access on a name
    expr: >
      //Call[ ./func/Attribute[ @attr='filter'
                                and ./value/Call[./func/Attribute[@attr='query']
                                and ./args[count(child::*) = 1]]
                              ]
             and ./args[count(child::*) = 1]
             and ./args/Compare[ ./ops/Eq
                                 and ./left/Attribute[ @attr != 'attributes'
                                                       and ./value/Name
                                                     ]
                               ]
            ]
    example: |
      session.query(Model).filter(Model.attr == baz)
    instead: |
      session.query(Model).filter_by(attr = baz)
    settings: *all_files

it says:

ERROR: /x/frontend/.bellybutton.yml, rule `filter_by`: 'str' object has no attribute 'match'

It works fine if you inline expr, but this is a pain because expr is huge:

    expr: //Call[./func/Attribute[ @attr='filter' and ./value/Call[./func/Attribute[@attr='query'] and ./args[count(child::*) = 1]]] and ./args[count(child::*) = 1] and ./args/Compare[ ./ops/Eq and ./left/Attribute[ @attr != 'attributes' and ./value/Name]]]

This is version 0.3.0.

Noticed that when specifying the --modified-only flag, bellybutton no longer filters for only '.py' files, which causes it to try and parse unwanted files.

I created a simple rule from which I expected to fail on every single file in my project:

settings:
  all_files: &all_files !settings
    included:
      - ~+/*
    excluded:
      - ~+/.tox/*
    allow_ignore: yes

default_settings: *all_files

rules:
  SimpleRule:
    description: "Simple rule"
    expr: !regex .*

And then I launched it:

bellybutton lint

Which results in a crash:

Traceback (most recent call last):
  File "/home/dev_intern/.virtualenvs/bellybutton-test/bin/bellybutton", line 11, in <module>
    sys.exit(main())
  File "/home/dev_intern/.virtualenvs/bellybutton-test/lib/python3.5/site-packages/bellybutton/cli.py", line 203, in main
    for arg in args.func.__code__.co_varnames[:args.func.__code__.co_argcount]
  File "/home/dev_intern/.virtualenvs/bellybutton-test/lib/python3.5/site-packages/bellybutton/cli.py", line 167, in lint
    linting_results = list(lint_file(filepath, file_contents, rules))
  File "/home/dev_intern/.virtualenvs/bellybutton-test/lib/python3.5/site-packages/bellybutton/linting.py", line 65, in lint_file
    for match in re.finditer(rule.expr)
TypeError: finditer() missing 1 required positional argument: 'string'

Can this be fixed soon?

First of all, great work with bellybutton, thanks for that!

We are using pre-commit and I was wondering if we could add bellybutton as a hook as well. We can use our own fork, but other people might want to use it like this too, so here it is. The addition of the .pre-commit-hooks.yaml file allows this repo to be used with pre-commit. It will simply run bellybutton lint for you in your repository and you can add args if you like into your own pre-commit config.

It can be used like this:

# .pre-commit-config.yaml
repos:
  - repo: https://github.com/hchasestevens/bellybutton
    rev: '9653582' # git rev (commit hash in this case)
    hooks:
      - id: bellybutton

More and more projects use a default branch other than 'master'. This PR guesses the default branch based on HEAD of the origin remote.

This PR also removes the 'fetch' call, which shaves off two seconds. When you're working on a relatively up-to-date clone, this shouldn't change the behavior too much.

In this PR, we introduce a simple FileManager class that allows you to open a file as-needed, and caches the contents for future use.

Reasoning: opening all Python files indiscriminately slows us down and is an unnecessary risk.

Longer explanation:

Big repositories have a lot of Python files, many of which might not be covered by the include patterns. It makes no sense to open all these files up front. In the company I work for, bellybutton would actually crash on some test files that contain invalid UTF-8 sequences (by design). These files are not covered by any bellybutton rule, but still crashed it. With this PR, we only open the files when the linter has decided to apply a rule to it.

Hi,

I would love to be able to explicitly tell bellybutton which configuration file to use. Furthermore, I believe it might be sensible to allow for multiple configuration files to be loaded, which might help to apply the same rules over multiple projects w/o config duplication.

bellybutton lint --config /some/path/shared_bellybutton.yml .bellybutton.yml

This could simply mean to have the rules from the files concatenated or result with rules in .bellybutton.yml overwriting rules with the same names in /some/path/shared_bellybutton.yml.

What do you think?

I have a project where I put lots of virtual environments under .mam. I've generated a config using bellybutton init and then excluded that directory with the following settings file:

settings:
  all_files: &all_files !settings
    included:
      - ~+/*
    excluded:
      - ~+/.tox/*
      - ~+/.mam/*
    allow_ignore: yes

default_settings: *all_files

rules:
  ExampleRule:
    description: "Empty module."
    expr: /Module/body[not(./*)]
    example: ""
    instead: |
      """This module has a docstring."""

I get the error in #22 from these files. I know this because I changed cli#109 to use a try except and raise the file name as a new error. The problem file is:

.\.mam\bandit\Lib\functools.py

I'm guessing I'm doing something wrong. But it's not excluding the file.

Traceback:

Traceback (most recent call last):
  File "/usr/local/bin/bellybutton", line 10, in <module>
    sys.exit(main())
  File "/usr/local/lib/python3.6/dist-packages/bellybutton/cli.py", line 218, in main
    for arg in args.func.__code__.co_varnames[:args.func.__code__.co_argcount]
  File "/usr/local/lib/python3.6/dist-packages/bellybutton/cli.py", line 191, in lint
    for failure in linting_failures(filepaths, rules):
  File "/usr/local/lib/python3.6/dist-packages/bellybutton/cli.py", line 134, in linting_failures
    for filepath, file_contents in files:
  File "/usr/local/lib/python3.6/dist-packages/bellybutton/cli.py", line 109, in open_python_files
    contents = f.read()
  File "/usr/lib/python3.6/encodings/ascii.py", line 26, in decode
    return codecs.ascii_decode(input, self.errors)[0]
UnicodeDecodeError: 'ascii' codec can't decode byte 0xe3 in position 3487: ordinal not in range(128)