moegirl_id.py

# -*- coding: utf-8 -*-

from pre_wd import *
import mwclient
from tqdm import tqdm

MOEGIRL_P = 'P5737'
# NAMEDAS_P = "P1810"
MWPAGEDID_P = 'P9675'
LANG_P = 'P407'
LANG_ZH_CODE = 'Q7850'
SUMMARY_UPDATE = "Bot: Update metadata from moegirl API"

config_data = json.load(open("config.json"))
moe_token = config_data['moegirl_token']
ua = config_data['ua']
mwcsite = mwclient.Site('zh.moegirl.org.cn', path='/', clients_useragent=ua)
mwcsite.login(moe_token[0], moe_token[1])

QUERY = ''
with open('moegirl_id_gen.rq', 'r') as query_file:
    QUERY = query_file.read()
gen = pg.WikidataSPARQLPageGenerator(QUERY, site=pwbsite)

def moe_get_curid(title: str) -> str:
    info = mwcsite.api('query', prop='info', titles=title)
    curid = next(iter(info['query']['pages']))
    # print(curid)
    return curid

def get_wikipage_quals(repo, title):
    curid = moe_get_curid(title)

    quals = []
    LANG_CODE = 'Q7850'

    if title.startswith("en:"):
        LANG_CODE = parse_iso_lang("en")
    elif title.startswith("ja:"):
        LANG_CODE = parse_iso_lang("ja")

    curid_claim = pywikibot.Claim(repo, MWPAGEDID_P, is_qualifier=True)
    curid_claim.setTarget(curid)
    quals.append(curid_claim)

    lang_claim = pywikibot.Claim(repo, LANG_P, is_qualifier=True)
    LANG_ITEM = pywikibot.ItemPage(repo, LANG_CODE)
    lang_claim.setTarget(LANG_ITEM)
    quals.append(lang_claim)
    
    return quals


count = 0
for i in tqdm(gen):
    item = get_item(i)
    if not item:
        continue
    
    moe_cliams = get_valid_claims(item, MOEGIRL_P)
    if not moe_cliams:
        continue
    for moe_claim in moe_cliams:
        title = moe_claim.getTarget()
        if title:
            # print(title)
            quals = get_wikipage_quals(repo, title)
            if quals:
                count += 1
                # print(count)
                # print("===")
                update_qualifiers(repo, moe_claim, quals, comment=SUMMARY_UPDATE)

print(f"Updated {count} entries")