Visualize zh_TW translation progress of Python official docs

visualize_translation_progress.py

"""Visualize zh_TW translation progress of Python official docs
Run the script under the project root of the repo.
ref: https://github.com/python/python-docs-zh-tw
"""
import logging
from pathlib import Path

import pandas as pd
import polib
from git import Repo

root_path = Path(__file__)
repo = Repo(root_path)
current_head = repo.head.commit
repo.git.checkout('3.12')


def _get_po_paths(root: Path) -> list[Path]:
    """Find all .po files in given path"""
    if not root.exists():
        logging.error(f"The path '{root.absolute()}' does not exist!")

    if root.is_file():
        return [root.resolve()]

    po_paths = [p.resolve() for p in root.glob("**/*.po") if '.git' not in str(p)]
    return po_paths


def _get_progress() -> float:
    po_files = _get_po_paths(Path(root_path).resolve())
    errors = []
    num_translated = 0
    num_entries = 0
    for path in po_files:
        try:
            pofile = polib.pofile(path)
        except OSError:
            errors.append(f"{path} doesn't seem to be a .po file")
            continue

        num_entries += (len(pofile) - len(pofile.obsolete_entries()))
        num_translated += len(pofile.translated_entries())

    progress = round(100 * num_translated / num_entries, 2)
    if errors:
        print(errors)
    return progress


if __name__ == '__main__':
    memo = {}
    prev_commit_time = None
    commits = list(repo.iter_commits())
    for commit in commits:
        if prev_commit_time and commit.committed_datetime.date() == prev_commit_time.date():
            print(commit.committed_datetime, commit, 'skipped')
            continue
        else:
            print(commit.committed_datetime, commit)
            prev_commit_time = commit.committed_datetime

        repo.git.checkout(commit)
        try:
            progress = _get_progress()
            memo[commit] = (commit.committed_datetime.timestamp(), progress)
        except Exception as e:
            print(e)

    repo.git.checkout(current_head)

    # plot
    df = pd.DataFrame(memo).T
    df.columns = ['timestamp', 'progress']
    df['timestamp'] = pd.to_datetime(df['timestamp'], unit='s')
    df['progress'] = df['progress'].astype(float)

    df['date'] = df['timestamp'].dt.date
    ax = df.plot(x='date', y='progress', title='Translation progress based on commit history')
    ax.set_xlabel('Date')
    ax.set_ylabel('Progress (%)')
    ax.grid()