snake_cases.py

#!/usr/bin/python3

"""Transform all function names to snake case."""

import os
import re


def extract_function_name(line):
    """Get the test function name written in that line or None."""
    match = re.findall(r'def (test.+?)\(', line)
    return match[0] if len(match) > 0 else None


def is_camel_case(function_name):
    """Check if the function name is in camelCase."""
    uppercase = len(re.findall('[A-Z]', function_name)) > 0
    underscores = '_' in function_name
    return uppercase and not underscores


def transform_snake_case(function_name):
    """Transform a function name to snake_case."""
    function_name = re.sub(r'([A-Z])', r'_\g<1>', function_name).lower()
    function_name = function_name.replace('l_d_a', 'lda')
    function_name = function_name.replace('word2_vec', 'word2vec')
    function_name = function_name.replace('token2_id', 'token2id')
    return function_name


def map_function_names(file):
    """Find all camelCase function names and their replacements."""
    mapping = {}
    for line in file.readlines():
        function_name = extract_function_name(line)
        if function_name is None:
            continue

        if not is_camel_case(function_name):
            continue

        snake_cased = transform_snake_case(function_name)
        mapping[function_name] = snake_cased

    return mapping


def main():
    for parent, subdirs, filenames in os.walk('gensim/test'):
        # for each test file
        for filename in filenames:
            if not filename.startswith('test_'):
                continue

            if not filename.endswith('.py'):
                continue

            path = os.path.join(parent, filename)
            with open(path, 'r') as file:
                mapping = map_function_names(file)
                file.seek(0, 0)
                contents = file.read()

            if len(mapping) == 0:
                continue

            with open(path, 'w') as file:
                for search, replace in mapping.items():
                    print(search, '->', replace)
                    # testPersistence exists, but also testPersistenceIgnore,
                    # make sure only complete matches are replaced
                    contents = contents.replace(
                        f'def {search}(',
                        f'def {replace}('
                    )

                file.write(contents)


if __name__ == '__main__':
    main()