formatting/format_as_md.py

import argparse
import json
import pathlib
import os

parser = argparse.ArgumentParser(
    description="Format the output of the data card tool as .md for the hub."
)
parser.add_argument("--input_path", "-i", type=pathlib.Path, required=True)
parser.add_argument("--output_path", "-o", type=pathlib.Path, required=True)
args = parser.parse_args()

def read_json_file(json_path: pathlib.Path):
    """Load a json file and return it as object."""
    with open(json_path, "r") as f:
        data = json.load(f)
    return data

def save_md_file(md_path: pathlib.Path, md_string: str):
    """Takes a string and saves it as .md file."""
    with open(md_path, 'w') as f:
        f.write(md_string)

def construct_md(data_card_data: dict, text_by_key: dict):
    """Constructs the markdown file

    This function iterates through text_by_key and extracts all answers from
    the data_card_data object. It uses the levels of hierarchy as indicator for
    the heading indentation and does not change the order in which anything
    appears.

    Args:
        data_card_data: Output from the data card tool
        text_by_key: configuration defined in key_to_question.json

    Returns:
        data_card_md_string: Markdown-formatted content
    """

    data_card_md_string = ""

    for main_key, main_content in text_by_key.items():
        section_header = main_content['section-title']
        # Remove it so that we don't iterate over it.
        del main_content['section-title']
        # Add it to string.
        data_card_md_string += f"## {section_header} \n\n"
        # Iterate over the subsections.
        for second_key, second_content in main_content.items():
            subsection_header = second_content['section-title']
            # Remove it so that we don't iterate over it.
            del second_content['section-title']
            # Add it to string.
            data_card_md_string += f"### {subsection_header} \n\n"
            # Finally, iterate over actual questions.
            for question_key, question_text in second_content.items():
                # Now grab the answer text.
                reply = data_card_data[main_key][second_key].get(question_key, "N/A")
                if reply and reply != "N/A":
                    data_card_md_string += f"#### {question_text} \n\n"
                    # Special parsing options here.
                    if question_key == question_key == "paper-bibtext":
                        data_card_md_string += f"```\n{reply}\n``` \n\n"
                    elif question_key == "structure-example":
                        data_card_md_string += f"```json\n{reply}\n``` \n\n"
                    elif isinstance(reply, list):
                        data_card_md_string += f"{' '.join(reply)} \n\n"
                    else:
                        data_card_md_string += f"{reply} \n\n"

    return data_card_md_string


if __name__ == "__main__":
    data_card_data = read_json_file(args.input_path)
    text_by_key = read_json_file(os.path.join(os.path.dirname(__file__), "key_to_question.json"))
    data_card_md_string = construct_md(data_card_data, text_by_key)
    save_md_file(args.output_path, data_card_md_string)