src/data_processing/taxonomy_processing.py

import os

from src.utils.data import load_json, save_json

AUTHORS_TAXONOMY = os.path.join("data", "authors_taxonomy.json")
AUTHORS_FILTER = os.path.join("data", "authors_filter.json")

DRAFT_CATEGORIES_TAXONOMY = os.path.join("data", "draftcat_taxonomy.json")
DRAFT_CATEGORIES_FILTER = os.path.join("data", "draftcat_taxonomy_filter.json")


def get_authors(taxonomy: dict) -> dict:
    countries = taxonomy["Members"]["Countries"]
    associations = taxonomy["Members"][
        "International and Regional State Associations"
    ]  # noqa: E501
    intergovernmental_negotiations = taxonomy[
        "Intergovernmental Negotiation Committee"
    ]  # noqa: E501
    observers = taxonomy["Observers and Other Participants"]  # noqa: E501
    return {
        "Members - Countries": countries,
        "Members - International and Regional State Associations": associations,  # noqa: E501
        "Intergovernmental Negotiation Committee": intergovernmental_negotiations,  # noqa: E501
        "Observers and Other Participants": observers,
    }


def get_draftcategories(taxonomy: dict) -> dict:
    taxonomy_filter = {}
    for draft_part, part_values in taxonomy.items():
        part = draft_part
        temp_values = []
        for part_name, part_value in part_values.items():
            temp_values.append(part_value)
        taxonomy_filter[part] = temp_values
    return taxonomy_filter


if __name__ == "__main__":
    authors_taxonomy = load_json(AUTHORS_TAXONOMY)
    authors_filter = get_authors(authors_taxonomy)
    save_json(file_path=AUTHORS_FILTER, data=authors_filter)

    draft_categories_taxonomy = load_json(DRAFT_CATEGORIES_TAXONOMY)
    draft_categories_filter = get_draftcategories(draft_categories_taxonomy)
    save_json(file_path=DRAFT_CATEGORIES_FILTER, data=draft_categories_filter)