wikidata_sf_awards.py

import pywikibot

from collections import defaultdict

AWARDS = {
    "Hugo Best Novel": "Q255032",
    "Nebula Best Novel": "Q266012",
    "Locus Best SF Novel": "Q2576795",
    "Philip K. Dick Award": "Q1341487"
}

P_WINNER = "P1346"
P_FOR_WORK = "P1686"
P_TIME_POINT = "P585"

P_TITLE = "P1476"
P_AUTHOR = "P50"

site = pywikibot.Site("wikidata", "wikidata")
repo = site.data_repository()

award_winners = defaultdict(dict)


def all_award_winners():
    for award_name in AWARDS:
        award_id = AWARDS[award_name]
        winners_parsed = all_winners(award_id)
        for winner in winners_parsed:
            if winner:
                award_winners[winner[0]][award_id] = winner[1]


def all_winners(award_id):
    item = pywikibot.ItemPage(repo, award_id)
    item_dict = item.get()
    winners = item_dict["claims"][P_WINNER]
    winners_parsed = [winner_data(w) for w in winners]
    return winners_parsed


def winner_data(winner_claim):
    winner = winner_claim.toJSON()
    try:
        winner_qualifiers = winner["qualifiers"]
        winner_year = time_point_to_year(winner_qualifiers[P_TIME_POINT][0]["datavalue"]["value"]["time"])
        winner_work_claim_id = "Q" + str(winner_qualifiers[P_FOR_WORK][0]["datavalue"]["value"]["numeric-id"])
        return (winner_work_claim_id, winner_year)
    except KeyError:
        print(winner)
        return None


def time_point_to_year(time_point: str) -> int:
    return int(time_point.split("-")[0][-4:])


all_award_winners()
print(award_winners)