clean_bib.py

from pybtex.database.input import bibtex
import argparse


def main():
    args = argparse.ArgumentParser()
    args.add_argument("path")
    args = args.parse_args()

    parser = bibtex.Parser()
    bib_data = parser.parse_file(args.path)

    def get_author_last_name(entry):
        authors = entry.persons.get("author", None)
        if authors is None or len(authors) == 0:
            return ""
        first_author = authors[0]
        return " ".join(first_author.last_names)

    entries = sorted(
        bib_data.entries.values(),
        key=lambda x: (x.type, get_author_last_name(x), x.fields.get("year", "")),
    )

    for entry in entries:
        fields_to_remove = ["abstract", "keywords"]
        if entry.type != "misc":
            fields_to_remove += ["url"]
        for field in fields_to_remove:
            if field in entry.fields:
                del entry.fields[field]

    output_str = ""
    for entry in entries:
        output_str += entry.to_string("bibtex")

    print(output_str)


if __name__ == "__main__":
    main()