pylint-errors.py

import re
import subprocess
from collections import defaultdict
from dataclasses import dataclass, field
from pprint import pprint

TEMPLATE = """## {code} ({text_code})

### :x: Problematic code:

```python
# to be addded
```

### :heavy_check_mark: Correct code:

```python
# to be added
```

### Rationale:

{description}

### Related resources:

- [Testcases](#)
- [Issue Tracker](https://github.com/PyCQA/pylint/issues?q=is%3Aissue+%22{text_code}%22+OR+%22{code}%22)
"""

CHECKS_TO_DIRS = {
    'Async Checker Messages': 'async',
    'Basic Checker Messages': 'basic',
    'Broad Try Clause Checker Messages': 'broad-try-clause',
    'Classes Checker Messages': 'classes',
    'Compare-To-Empty-String Checker Messages': 'compare-to-empty-string',
    'Compare-To-Zero Checker Messages': 'compare-to-zero',
    'Deprecated Builtins Checker Messages': 'deprecated-builtins',
    'Design Checker Messages': 'design',
    'Docstyle Checker Messages': 'docstyle',
    'Else If Used Checker Messages': 'else-if-used',
    'Exceptions Checker Messages': 'exceptions',
    'Format Checker Messages': 'format',
    'Imports Checker Messages': 'imports',
    'Logging Checker Messages': 'logging',
    'Miscellaneous Checker Messages': 'miscellaneous',
    'Multiple Types Checker Messages': 'multiple-types',
    'Newstyle Checker Messages': 'newstyle',
    'Overlap-Except Checker Messages': 'overlap-except',
    'Parameter Documentation Checker Messages': 'parameter-documentation',
    'Python3 Checker Messages': 'python3',
    'Refactoring Checker Messages': 'refactoring',
    'Similarities Checker Messages': 'similarities',
    'Spelling Checker Messages': 'spelling',
    'Stdlib Checker Messages': 'stdlib',
    'String Checker Messages': 'string',
    'Typecheck Checker Messages': 'typecheck',
    'Variables Checker Messages': 'variables'
}

CHECK = re.compile(r'^\#\#\# .* checker Messages$')
ERROR = re.compile(r'^[a-z-]+ \([\w]+\)$')


@dataclass
class PylintError:
    code: str
    text_code: str
    description: list = field(repr=False, default=None)


def capitalise_key(words):
    """Capitalises a key and preserves dashes."""

    capitalised_key = []
    for word in words:
        sub_words = []
        for sub_word in word.split('-'):
            sub_words.append(sub_word.capitalize())
        capitalised_key.append('-'.join(sub_words))

    return ' '.join(capitalised_key)


def extract_errors():
    check = None
    checks_to_errors = defaultdict(list)
    with open('pylint.md', 'r') as fh:
        for line in fh.readlines():
            if CHECK.search(line):
                # [1:] to skip ###
                check = capitalise_key(line.split()[1:])

            # Add an error to check's array
            if ERROR.search(line):
                checks_to_errors[check].append(line[:-1])

    return checks_to_errors


def parse_errors(mapping):
    errors = defaultdict(list)
    for key, values in mapping.items():
        for error in values:
            text_error = error.split(' ')
            error_code = text_error[-1].split('(')
            error_code = error_code[-1].split(')')

            errors[key].append(PylintError(error_code[0], text_error[0]))

    return errors


def sort_errors(mapping):
    sorted_errors = {}
    for key, errors in mapping.items():
        sorted_errors[key] = sorted(errors, key=lambda e: e.code)

    return sorted_errors


def set_errors_description(mapping):
    for key, errors in mapping.items():
        for error in errors:
            p = subprocess.Popen(
                ['pylint', '--help-msg', error.code],
                stdout=subprocess.PIPE,
                stderr=subprocess.PIPE
            )
            stdout, stderr = p.communicate()

            stdout = stdout.decode().split(':')[2:]
            error.description = ''.join(stdout)


def write_errors(mapping):
    for key, errors in mapping.items():
        for error in errors:
            path = '{}/{}.md'.format(CHECKS_TO_DIRS[key], error.code)
            with open(path, 'w') as fh:
                fh.write(
                    TEMPLATE.format(
                        code=error.code,
                        text_code=error.text_code,
                        description=error.description
                    )
                )


def render_tree(mapping):
    for key, errors in mapping.items():
        print(f'### {key}')
        print()
        for error in errors:
            print(
                '- [{} ({})](errors/{}/{}.md)'.format(
                    error.code, error.text_code, CHECKS_TO_DIRS[key], error.code
                )
            )
        print()


if __name__ == '__main__':
    # Get errors from a .md doc
    raw_errors = extract_errors()

    # Parse errors
    parsed_errors = parse_errors(raw_errors)

    # Sort errors
    sorted_errors = sort_errors(parsed_errors)
    render_tree(sorted_errors)
    #set_errors_description(sorted_errors)

    # Write errors
    #write_errors(sorted_errors)