personal-data/personal_data/fetchers/jellyfin_watch_history.py

import dataclasses
import logging
from collections.abc import Iterator
from typing import Any

from jellyfin_apiclient_python import JellyfinClient

from .. import _version, secrets
from ..data import DeduplicateMode, Scraper

logger = logging.getLogger(__name__)

URL_SITE_ROOT = 'https://steamcommunity.com/'

URL_GAME_ACHIVEMENTS = URL_SITE_ROOT + 'id/{username}/stats/appid/{appid}'

FORMAT_DATE_HEADER = '%d/%m/%YYYY'


def iterate_series(client):
    result = client.jellyfin.user_items(
        params={
            'includeItemTypes': 'Series',
            'parentId': 'a656b907eb3a73532e40e44b968d0225',
            'userId': 'dd95c1085c1b4e83ba8e8853fbc644ab',
        },
    )
    yield from result['Items']


def iterate_watched_episodes_of_series(client, series_id: str):
    result = client.jellyfin.user_items(
        params={
            'filters': 'IsPlayed',
            'recursive': True,
            'includeItemTypes': 'Episode',
            'parentId': series_id,
            'userId': 'dd95c1085c1b4e83ba8e8853fbc644ab',
            'fields': 'AirTime',
        },
    )
    yield from result['Items']


@dataclasses.dataclass(frozen=True)
class JellyfinWatchHistoryScraper(Scraper):
    dataset_name = 'show_episodes_watched'
    deduplicate_mode = DeduplicateMode.BY_ALL_COLUMNS

    def scrape(self) -> Iterator[dict[str, Any]]:
        client = JellyfinClient()

        client.config.app(
            'personal_data',
            _version.__version__,
            'test_machine',
            'unique_id_1',
        )

        client.config.data['auth.ssl'] = False
        client.auth.connect_to_address(secrets.JELLYFIN_URL)
        client.auth.login(
            secrets.JELLYFIN_URL,
            secrets.JELLYFIN_USERNAME,
            secrets.JELLYFIN_PASSWORD,
        )

        for series_data in iterate_series(client):
            series_id = series_data['Id']
            for episode_data in iterate_watched_episodes_of_series(client, series_id):
                episode_index = episode_data.get('IndexNumber')
                if episode_index is None:
                    continue
                yield {
                    'series.name': episode_data['SeriesName'],
                    'season.name': episode_data['SeasonName'],
                    'episode.index': int(episode_index),
                    'episode.name': episode_data['Name'],
                    'me.last_played_time': episode_data['UserData']['LastPlayedDate'],
                    'episode.duration_seconds': episode_data['RunTimeTicks'] / 10000000,
                    'episode.premiere_date': episode_data.get('PremiereDate'),
                }

                del episode_data
            del series_data, series_id