test de prova

parse_readme.py

import re
from pathlib import Path
import argparse
from typing import Dict, List, Optional
import json


def parse_doc_comments(readme_content: str) -> List[str]:
    """Parse DOC comments in the form <!-- [DOC-*] --> from the README.md content."""
    pattern = r"<!--\s*\[\s*(.*?)\s*\]\s*-->"
    matches = re.findall(pattern, readme_content)
    return matches


def parse_doc_comments_from_file(file_path: Path) -> List[str]:
    """Parse DOC comments in the form <!-- [DOC-*] --> from a file."""
    with open(str(file_path), "r") as file:
        content = file.read()
        parsed_contents = parse_doc_comments(content)
        return parsed_contents


def parse_and_save_extended_json(
    file_path: Path,
    repo_name: str,
    last_updated: str,
    last_commit_sha: str,
    readme_url: str,
    save_to_path: Optional[Path] = None,
) -> None:
    """Parse DOC comments from a file and save them to a JSON file."""

    parsed_contents = parse_doc_comments_from_file(file_path)

    final_json_output = craft_json_output(
        repo_name=repo_name,
        doc_comments=parsed_contents,
        last_updated=last_updated,
        last_commit_sha=last_commit_sha,
        readme_url=readme_url,
    )

    if save_to_path:
        print(f"Saving to {save_to_path}")
        save_to_path.write_text(json.dumps(final_json_output))
    return final_json_output


def craft_json_output(
    repo_name: str,
    doc_comments: List[str],
    last_updated: str,
    last_commit_sha: str,
    readme_url: str,
) -> Dict:
    """Craft a JSON output with the repository metadata."""

    base_doc_keys = [
        "DOC-PROJECT-SECRETS",
        "DOC-DEV-HOW-TO",
        "DOC-DEPLOY-HOW-TO",
        "DOC-HOW-TO-CONTRIBUTE",
        "DOC-BUSINESS-RULES",
        "DOC-PROCESS-HOW-TO",
        "DOC-PROJECT-SETTINGS",
        "DOC-DEV-TOOLS",
        "DOC-ARCH-FUNC",
        "DOC-API-DOCS",
        "DOC-TRANSLATE-HOW-TO",
        "DOC-ADR",
        "DOC-USER-GUIDE",
        "DOC-USER-TUTORIAL",
    ]

    doc_comments_dict = {
        key: True if key in doc_comments else False for key in base_doc_keys
    }

    json_output = {
        "repository_name": repo_name,
        "doc_comments": doc_comments_dict,
        "last_updated": last_updated,
        "last_commit_sha": last_commit_sha,
        "readme_url": readme_url,
    }

    return json_output



def parse_args():
    """Parse arg.sys inputs
    
    # Example
    
    python parse_readme.py  --input README_no_docs.md --output outputdb.json --repo-name "somenergia-exemple-documentacio" --last-updated "2024-08-12T10:53:57.000-04:00" --last-commit-sha "c9f82dcbe2e6c01806096ca18b711b461f039421" --readme-url "https://gitlab.somenergia.coop/projects/94/repository/files/README.md/raw"
    
    
    parser = argparse.ArgumentParser(
        description="Parse DOC comments from a README.md file"
    )
    parser.add_argument(
        "--input",
        type=Path,
        help="The input file path",
        required=True,
    )
    parser.add_argument(
        "--output",
        type=Path,
        help="The output file path. If not passed, the output will be printed to the console.",
        required=False,
    )
    parser.add_argument(
        "--repo-name",
        type=str,
        help="The repository name",
        required=False,
    )
    parser.add_argument(
        "--last-updated",
        type=str,
        help="The last updated date",
        required=False,
    )
    parser.add_argument(
        "--last-commit-sha",
        type=str,
        help="The last commit SHA",
        required=False,
    )
    parser.add_argument(
        "--readme-url",
        type=str,
        help="The README.md URL",
        required=False,
    )

    args = parser.parse_args()
    return args


def main():

    args = parse_args()
    file_input = args.input
    file_output = args.output
    repo_name = args.repo_name
    last_updated = args.last_updated
    last_commit_sha = args.last_commit_sha
    readme_url = args.readme_url


    doc_comments = parse_and_save_extended_json(
        file_path=file_input,
        save_to_path=file_output,
        repo_name=repo_name,
        last_updated=last_updated,
        last_commit_sha=last_commit_sha,
        readme_url=readme_url,
    )

if __name__ == "__main__":
    main()

poormansdb.json

[
        {
            "repository_name": "somenergia-exemple-documentacio",
            "doc_comments": {
                "DOC-PROJECT-SECRETS": true,
                "DOC-DEV-HOW-TO": true,
                "DOC-DEPLOY-HOW-TO": true,
                "DOC-HOW-TO-CONTRIBUTE": true,
                "DOC-BUSINESS-RULES": true,
                "DOC-PROCESS-HOW-TO": true,
                "DOC-PROJECT-SETTINGS": true,
                "DOC-DEV-TOOLS": true,
                "DOC-ARCH-FUNC": true,
                "DOC-API-DOCS": true,
                "DOC-TRANSLATE-HOW-TO": true,
                "DOC-ADR": true,
                "DOC-USER-GUIDE": true,
                "DOC-USER-TUTORIAL": true,
            },
            "last_updated": "2024-08-12T10:53:57.000-04:00",
            "last_commit_sha": "c9f82dcbe2e6c01806096ca18b711b461f039421",
            "readme_url": "https://gitlab.somenergia.coop/projects/94/repository/files/README.md/raw",
        }
    ]