Find unnecessary spellcheck exclusions in uutils; run this from the top-level directory

check-spellcheck.py

#!/usr/bin/env python3

import re
import subprocess

RE_SPELLCHECK_DISABLE = re.compile("spell-checker:disable-\w+")
CACHED_FILES_STRING = None


def check_right_directory():
    try:
        with open("src/bin/coreutils.rs", "r") as fp:
            pass
    except FileNotFoundError as e:
        print("Started in the wrong directory? cd to $(git root) before invoking.")
        raise e


def does_it_spellcheck():
    global CACHED_FILES_STRING
    if CACHED_FILES_STRING is None:
        result = subprocess.run(
            ["git", "ls-files", ":!:tests/fixtures"],
            check=True,
            capture_output=True,
            text=True,
        )
        CACHED_FILES_STRING = result.stdout
    result = subprocess.run(
        [
            "cspell",
            "--config",
            ".vscode/cSpell.json",
            "--cache",
            "--no-progress",
            "--file-list",
            "stdin",
        ],
        input=CACHED_FILES_STRING,
        capture_output=True,
        text=True,
    )
    return result.returncode == 0


def list_locations():
    result = subprocess.run(
        ["git", "grep", "-Frn", "spell-checker:disable-"],
        check=True,
        capture_output=True,
        text=True,
    )
    locations = []
    for match_line in result.stdout.split("\n"):
        if not match_line:
            continue
        # 'match_line' could be something like:
        # src/uu/tail/src/parse.rs:35:/// tail -\[NUM\]\[bcl\]\[f\] and tail +\[NUM\]\[bcl\]\[f\] // spell-checker:disable-line
        parts = match_line.split(":", 2)
        filepath = parts[0]
        linenumber = int(parts[1])
        locations.append((filepath, linenumber))
    return locations


def try_replacing(location):
    filepath, linenumber = location
    line_index = linenumber - 1
    del linenumber
    # TODO: Avoid opening and reading the same file over and over again.
    with open(filepath, "r") as fp:
        lines_orig = fp.readlines()
    lines_modified = lines_orig.copy()
    lines_modified[line_index] = RE_SPELLCHECK_DISABLE.sub(
        "FIXME-REPLACED ", lines_modified[line_index]
    )
    assert lines_modified[line_index] != lines_orig[line_index], (
        filepath,
        linenumber,
        lines_orig[line_index],
    )
    with open(filepath, "w") as fp:
        fp.write("".join(lines_modified))
    if does_it_spellcheck():
        return 1
    else:
        with open(filepath, "w") as fp:
            fp.write("".join(lines_orig))
        return 0


def run():
    check_right_directory()
    assert does_it_spellcheck(), "Fix spellcheck issues first!"
    locations = list_locations()  # git grep -Frn 'spell-checker:disable-'
    print(f"Checking {len(locations)} occurrences ...")
    matches = 0
    for loc in locations:
        matches += try_replacing(loc)
    print(
        f"Found {len(locations)} occurrences of spell-checker disabling, replaced {matches} of them."
    )


if __name__ == "__main__":
    run()