deadend-zhwiki.py

import pywikibot
from tqdm import tqdm
import re
from datetime import datetime

site = pywikibot.Site('zh', 'wikipedia', user='Crystal-bot')
site.login()

BLACKLIST = [
    '互联网档案馆',
    '机器翻译',
    '维基百科',
    '人称'
]
MIN = 5

s_curtime = datetime.utcnow()
s_curtime_ts = s_curtime.strftime('%Y-%m-%d %H:%M:%S')
print("==========")
print(f"[INFO] {s_curtime_ts} Searching articles with \"dead_end\" template.")

template = pywikibot.Page(pywikibot.Link('Dead_end', default_namespace=10, source=site))
gen = template.getReferences(only_template_inclusion=True, namespaces=0)

count = 0
matched = 0
for page in tqdm(gen):
    count += 1
    i_links = list(page.linkedPages(namespaces=0))
    for _page in i_links:
        for name in BLACKLIST:
            if _page.title() == name:
                i_links.remove(_page)
    for _page in i_links:
        if not _page.exists():
            i_links.remove(_page)

    link_count = len(i_links)
    if link_count == 0:
        continue
    if link_count >= MIN:
        matched += 1
        # print(f"{page.title()} -> {str(link_count)}")
        page.text = re.sub(r"\{\{([Dd]ead ?end|斷[鏈連聯]頁面)\|.*?\}\}", r"", page.text)
        #page.save("机器人: 移除含有内部链接条目中的[[Template:Dead end]]模板")

e_curtime = datetime.utcnow()
e_curtime_ts = e_curtime.strftime('%Y-%m-%d %H:%M:%S')
elapse = e_curtime - s_curtime

print(f"[INFO] {e_curtime_ts} Found {str(matched)} articles in {str(count)} articles, replaced \"dead_end\" template in {elapse.total_seconds():.2f} seconds, sleeping...")
print("==========\n")