cleans up my MAL export so thats its importable to anilist

anilist_413.py

#!/usr/bin/env python3
# cleans up my MAL export according to this thread:
# https://anilist.co/forum/thread/3291
# so that its importable
# otherwise the graphql error fails with a 413 request entity too large
# since the anilist page just inlines the XML contents into the gql query

import sys
import typing
from pathlib import Path

import lxml.etree as ET
import click

KEEP_ATTRS = set(
    [
        "series_animedb_id",
        "series_title",
        "my_watched_episodes",
        "my_start_date",
        "my_finish_date",
        "my_score",
        "my_status",
    ]
)


def fix(from_file: Path) -> str:
    tree = ET.parse(str(from_file))
    root = tree.getroot()
    root.remove(root.find("myinfo"))
    for anime_tag in root.findall("anime"):
        for anime_info in anime_tag:
            if anime_info.tag not in KEEP_ATTRS:
                anime_tag.remove(anime_info)
    return ET.tostring(root)


@click.command()
@click.argument("XML_FILE")
def main(xml_file: str):
    """
    pass the xml file from MAL as the first argument
    prints results to STDOUT
    """
    assert xml_file.endswith(".xml"), "Pass the xml file, not the .xml.gz file"
    click.echo(fix(Path(xml_file)))


if __name__ == "__main__":
    main(prog_name="anilist_413")

Stills time outs...

https://anilist.co/forum/thread/33798 bug report