update-pre-commit

#!/usr/bin/env python3

# Copyright © Michal Čihař <michal@weblate.org>
#
# SPDX-License-Identifier: CC0-1.0

"""Synchronizes pre-commit config for Weblate repositories."""

import re
import subprocess
import sys
from pathlib import Path

import ruamel.yaml

MD_REPOS = [
    "https://github.com/executablebooks/mdformat",
    "https://github.com/igorshubovych/markdownlint-cli",
]
REMOVE_REPOS = [
    "https://github.com/asottile/pyupgrade",
    "https://github.com/psf/black",
    "https://github.com/PyCQA/isort",
    "https://github.com/adamchainz/blacken-docs",
    "https://github.com/rhysd/actionlint",
    *MD_REPOS,
]
SYNC_REPOS = [
    "https://github.com/kjanat/actionlint",
    "https://github.com/zizmorcore/zizmor-pre-commit",
]
DEPENDENCY_PATTERNS = (
    ("python", re.compile(r"[^'\" ]+==[^'\" ,\s]+")),
    ("node", re.compile(r"[^'\" ]+@[^'\" ,\s]+")),
)

yaml = ruamel.yaml.YAML()
yaml.indent = 2
yaml.preserve_quotes = False
yaml.width = sys.maxsize

# Load template
template_path = Path(__file__).parent / ".pre-commit-config.yaml"
with template_path.open() as handle:
    template = yaml.load(handle)
template_repos = {repo["repo"]: repo for repo in template["repos"]}

# Load target
precommit_file = Path(sys.argv[1])
with precommit_file.open() as handle:
    data = yaml.load(handle)

changed = False

# Migrate markdown to rmdml
if [repo["repo"] in MD_REPOS for repo in data["repos"]]:
    SYNC_REPOS.append("https://github.com/rvben/rumdl-pre-commit")

# Remove no longer wanted
remove = [
    offset for offset, repo in enumerate(data["repos"]) if repo["repo"] in REMOVE_REPOS
]
if remove:
    for offset in reversed(remove):
        del data["repos"][offset]
    changed = True

# Apply fixups
missing = set(SYNC_REPOS)

# pyproject linter/formatter
pyproject = precommit_file.parent / "pyproject.toml"
if pyproject.exists():
    missing.add("https://github.com/abravalheri/validate-pyproject")
    missing.add("https://github.com/pappasam/toml-sort")

for offset, repo in enumerate(data["repos"]):
    if repo["repo"] in missing:
        missing.remove(repo["repo"])
        if repo["repo"] in template_repos and template_repos[repo["repo"]] != repo:
            data["repos"][offset] = template_repos[repo["repo"]]
            changed = True

    # This is checked by ruff as well
    if (
        repo["repo"] == "https://github.com/pre-commit/pre-commit-hooks"
        and {"id": "debug-statements"} in repo["hooks"]
    ):
        repo["hooks"] = [
            hook for hook in repo["hooks"] if hook["id"] != "debug-statements"
        ]
        changed = True

    # Migrate gitleaks to kingfisher
    if repo["repo"] == "https://github.com/gitleaks/gitleaks":
        repo["repo"] = "https://github.com/mongodb/kingfisher"
        repo["rev"] = template_repos[repo["repo"]]["rev"]
        repo["hooks"] = [{"id": "kingfisher-auto"}]
        changed = True

# Add missing
for name in missing:
    data["repos"].append(template_repos[name])
    changed = True

for repo in data["repos"]:
    for hook in repo["hooks"]:
        if (
            repo["repo"] == "https://github.com/mongodb/kingfisher"
            and hook["id"] == "kingfisher-auto"
        ):
            env = hook.setdefault("env", {})
            if env.get("KINGFISHER_VERSION") != repo["rev"]:
                env["KINGFISHER_VERSION"] = repo["rev"]
                changed = True
        # Add language tags so that Renovate can detect additional dependencies
        dependencies = hook.get("additional_dependencies", ())
        language = next(
            (
                language
                for language, pattern in DEPENDENCY_PATTERNS
                if any(pattern.search(dependency) for dependency in dependencies)
            ),
            None,
        )
        if language is not None and hook.get("language") != language:
            hook["language"] = language
            changed = True
    # Convert wrong float rev to str
    if "rev" in repo and isinstance(repo["rev"], float):
        repo["rev"] = str(repo["rev"])
        changed = True

# Update CI config
if "ci" not in data:
    data["ci"] = {}
    changed = True
if "autoupdate_schedule" not in data["ci"]:
    data["ci"]["autoupdate_schedule"] = "quarterly"
    changed = True

if any(
    repo["repo"] == "https://github.com/mongodb/kingfisher" for repo in data["repos"]
):
    if "kingfisher-auto" not in data["ci"]["skip"]:
        data["ci"]["skip"].append("kingfisher-auto")
        changed = True
elif "kingfisher-auto" in data["ci"]["skip"]:
    data["ci"]["skip"].remove("kingfisher-auto")
    changed = True

if any(repo["repo"] == "https://github.com/fsfe/reuse-tool" for repo in data["repos"]):
    if "reuse" not in data["ci"]["skip"]:
        data["ci"]["skip"].append("reuse")
        changed = True
elif "reuse" in data["ci"]["skip"]:
    data["ci"]["skip"].remove("reuse")
    changed = True

# Save changed file
if changed:
    with open(sys.argv[1], "w") as handle:
        yaml.dump(data, handle)

    subprocess.run(
        ["uvx", "prek", "run", "--files", sys.argv[1]], check=False, capture_output=True
    )