Search function for searching or replacing files that have the search keyword recursively. Made this to help myself change asyncio project into anyio. Pretty sure this is reinvention of wheel but whatever!

string_search_in_files.py

"""
Test use-case

>>> len(list(file_search_gen("./", "asyncio", file_extensions=(".py",))))
6
>>> for f in file_search_gen("./", "asyncio", "REPLACE_THIS", file_extensions=(".py",)):
...     print(f.as_posix())
...

telethon_examples/interactive_telegram_client.py
telethon_examples/payment.py
telethon_examples/quart_login.py
tests/telethon/client/test_messages.py
tests/telethon/events/test_chataction.py
tests/telethon/test_helpers.py

"""


def file_search_gen(
    path,
    str_to_find: str,
    replace="",
    whitelist_mode=True,
    file_extensions=(),
    fallback_encodings=(),
    print_error=True,
    exclude_directory=(".venv", "venv"),
):
    """
    generator I use to find texts in nested python libraries.

    Args:
        path: string or pathlib.Path, folder to start search.
        str_to_find: string to find.
        replace: if not empty, all found strings will be replaced.
        whitelist_mode: if True, file_extensions will act as whitelist
        file_extensions: File extension to search/ignore for. Will act as blacklist if whitelist_mode is False.
        fallback_encodings: encodings to use when utf8 and system default fails.
        print_error: If True, prints decode error.
        exclude_directory: folders to exclude
    """

    import pathlib
    from typing import Generator
    from sys import getdefaultencoding

    path = pathlib.Path(path)
    file_extensions = set(file_extensions)
    fallback_encodings = ["utf8", getdefaultencoding()] + list(fallback_encodings)
    encodings = set(fallback_encodings)
    excluding_paths = tuple(map(pathlib.Path, exclude_directory))

    print_func = print if print_error else lambda x: None

    def _replace(file_: pathlib.Path, src: str, encoding_: str):
        replaced = src.replace(str_to_find, replace)
        file_.write_text(replaced, encoding=encoding_)

    replace_func = _replace if replace else lambda x, y, z: None

    def filtered_path_gen(
        directory: pathlib.Path,
    ) -> Generator[pathlib.Path, None, None]:
        for path_ in directory.iterdir():
            if path_.is_dir():
                if path_ in excluding_paths:
                    continue

                yield from filtered_path_gen(path_)
            elif whitelist_mode * (path_.suffix in file_extensions):
                yield path_

    for file in filtered_path_gen(path):
        for encoding in encodings:
            try:
                data = file.read_text(encoding=encoding)
            except UnicodeDecodeError:
                continue
            success_encoding = encoding
            break
        else:
            # runs when no break is called
            print_func(f"No matching encoding for file '{file.as_posix()}'")
            continue

        if str_to_find in data:
            replace_func(file, data, success_encoding)
            yield file