extract_mixed_citations_from_json.py

import os
import json
import logging
import argparse


logging.basicConfig(format="%(asctime)s - %(message)s", level=logging.DEBUG)
logger = logging.getLogger(__name__)


def main():
    parser = argparse.ArgumentParser(
        description="Programa para agrupar as mixed citations a partir"
        " de arquivos de parágrafos"
    )

    parser.add_argument(
        "pids",
        help="Lista de pids que serão utilizados para extrair os parágrafos"
        " de arquivos JSON",
        type=argparse.FileType("r"),
    )
    parser.add_argument(
        "paragraphs",
        help="Diretório contendo os arquivos de parágrafos no formatos JSON.",
    )
    parser.add_argument(
        "output",
        help="Arquivo com o resultado do processamento.",
        type=argparse.FileType("w"),
    )

    args = parser.parse_args()

    for pid in args.pids:
        pid = pid.strip()
        paragraph_file = os.path.abspath(os.path.join(args.paragraphs, pid + ".json"))

        if not os.path.exists(paragraph_file):
            logger.error("Paragraph file '%s' does not exist.", paragraph_file)
            continue

        with open(paragraph_file, "r") as f:
            for line in f.readlines():
                paragraph: dict = json.loads(line)
                article_pid: str = paragraph.get("v880", [{}])[0].get("_")
                raw_mixed_citation: str = paragraph.get("v704", [{}])[0].get("_")
                raw_mixed_citation_index: str = paragraph.get("v888", [{}])[0].get("_")

                if (
                    article_pid is not None
                    and raw_mixed_citation is not None
                    and raw_mixed_citation_index is not None
                    and raw_mixed_citation_index.isdigit()
                ):

                    mixed_citation = {
                        "mixed": raw_mixed_citation,
                        "order": int(raw_mixed_citation_index) - 1,
                        "collection": "scl",
                    }

                    args.output.write(json.dumps(mixed_citation) + "\n")


if __name__ == "__main__":
    main()