Script to get all files touched by PRs

files_in_pr.py

"""Script to get all files touched by PRs."""


import shutil
from pathlib import Path
from warnings import warn

import requests
from rich import print

USERNAME = "Remi-Gau"

# may require a token if run often
TOKEN_FILE = Path(__file__).parent.joinpath("token.txt")

# repo to check
GH_USERNAME = "nilearn"
GH_REPO = "nilearn"

DEBUG = False
USE_LOCAL = True

OUTPUT_FOLDER = Path(__file__).parent / "tmp"
OUTPUT_FILE = OUTPUT_FOLDER / "output.md"

EXLUDE_PR = {"title": ["format"]}


def print_to_output(output_file, all_files):
    unique_files = set(all_files)
    unique_files = sorted(unique_files)
    print_line_to_output(
        output_file=output_file, text="| file | nb of PR found in |"
    )
    print_line_to_output(
        output_file=output_file, text="| ---- | ----------------- |"
    )
    for file in unique_files:
        print_line_to_output(
            output_file=output_file,
            text=f"| {file} | {all_files.count(file)} |",
        )


def print_line_to_output(output_file, text):
    """Print to file or stdout."""
    if output_file is not None:
        with open(output_file, "a") as f:
            print(f"{text}", file=f)
        return
    else:
        print(text)


def save_diffs(pulls, auth, output_folder):
    for i, pull_ in enumerate(pulls):
        if DEBUG and i == 2:
            break
        print(f"\n{pull_['number']}, {pull_['title']}")
        diff = get_this_pr_diff(url=pull_["diff_url"], auth=auth)
        if diff is None:
            continue
        save_diff_to_file(
            number=pull_["number"],
            name=pull_["title"],
            diff=diff,
            output_folder=output_folder,
        )


def save_diff_to_file(number, name, diff, output_folder):
    filename = (
        output_folder
        / f"{number}_{name.replace(' ', '_').replace('/', '_')}.diff"
    )
    with open(filename, "w") as f:
        f.write(diff)


def get_list_of_prs(gh_username, gh_repo, auth=None):
    url = f"https://api.github.com/repos/{gh_username}/{gh_repo}/pulls?per_page=100"
    response = requests.get(url, auth=auth)
    if response.status_code != 200:
        warn(f"Error {response.status_code}: {response.text}")
        return None
    return response.json()


def get_this_pr_diff(url, auth=None):
    response = requests.get(url, auth=auth)
    if response.status_code != 200:
        warn(f"Error {response.status_code}: {response.text}")
        return
    return response.text


def list_all_files_in_prs(input_folder, exclude_pr):
    all_files = []
    pulls = input_folder.glob("*.diff")
    for pull_ in pulls:
        pr_number = pull_.stem.split("_")[0]
        pr_title = pull_.stem.split("_")[1:]
        if any(ex in pr_title for ex in exclude_pr["title"]):
            print(f"[red]skipping {pr_number}, {' '.join(pr_title)}[/red]")
            continue
        print(f"{pr_number}, {' '.join(pr_title)}")
        diff = Path(pull_).read_text()
        for line in diff.splitlines():
            if line.startswith("diff --git "):
                this_file = line.split(" ")[2][2:]
                all_files.append(this_file)
    return all_files


def main():
    if not USE_LOCAL:
        shutil.rmtree(OUTPUT_FOLDER, ignore_errors=True)

    OUTPUT_FILE.unlink(missing_ok=True)
    OUTPUT_FOLDER.mkdir(exist_ok=True)

    TOKEN = None
    if TOKEN_FILE.exists():
        with open(Path(__file__).parent.joinpath("token.txt")) as f:
            TOKEN = f.read().strip()

    auth = None if USERNAME is None or TOKEN is None else (USERNAME, TOKEN)
    if not USE_LOCAL:
        pulls = get_list_of_prs(
            gh_username=GH_USERNAME, gh_repo=GH_REPO, auth=auth
        )
        save_diffs(pulls=pulls, auth=auth, output_folder=OUTPUT_FOLDER)

    all_files = list_all_files_in_prs(
        input_folder=OUTPUT_FOLDER, exclude_pr=EXLUDE_PR
    )

    print_to_output(output_file=OUTPUT_FILE, all_files=all_files)


if __name__ == "__main__":
    main()