jenkins-bot has submitted this change. ( https://gerrit.wikimedia.org/r/c/pywikibot/core/+/1062670?usp=email )

Change subject: delinker: use difflib to find the closest image match ......................................................................

delinker: use difflib to find the closest image match

Change-Id: Ib5604e2a533e433f0434b109d5d0d0ed53031efa --- M scripts/delinker.py 1 file changed, 25 insertions(+), 22 deletions(-)

Approvals: jenkins-bot: Verified Xqt: Looks good to me, approved

diff --git a/scripts/delinker.py b/scripts/delinker.py index a6dc176..a2b3c90 100755 --- a/scripts/delinker.py +++ b/scripts/delinker.py @@ -45,6 +45,7 @@ import configparser import heapq import re +from difflib import get_close_matches

import pywikibot from pywikibot.backports import removeprefix @@ -64,7 +65,7 @@ summary_key = 'delinker-delink'

def skip_page(self, page) -> bool: - """Skip pages which neither exists locally nor on shared repository.""" + """Skip pages which either exists locally or on shared repository.""" pywikibot.info('.', newline=False) if page.exists() or page.file_is_shared(): return True @@ -135,31 +136,33 @@ if super().skip_page(page): return True

+ title = page.title(with_ns=False) params = { 'logtype': 'delete', - 'page': 'File:' + page.title(underscore=True, with_ns=False), + 'page': 'File:' + title, + 'total': 1, } - try: - entry = next(self.site.logevents(**params)) - except StopIteration: - try: - entry = next(self.site.image_repository().logevents(**params)) - except StopIteration: - pywikibot.info() - pywikibot.warning( - f'unable to delink missing {page.title(as_link=True)}') - found = list(self.site.search( - page.title(), - namespaces=self.site.namespaces.MAIN, - total=1 - )) - if found: - pywikibot.info('probably <<lightblue>>' - f'{found[0].title(as_link=True)}' - '<<default>> is meant') - return True + entries = list(self.site.logevents(**params)) + if not entries: + entries = list(self.site.image_repository().logevents(**params))

- self.summary_parameters = dict(entry) + if not entries: + pywikibot.info() + pywikibot.warning( + f'unable to delink missing {page.title(as_link=True)}') + possibilities = [ + p.title(with_ns=False) + for p in self.site.search(page.title(), + namespaces=self.site.namespaces.MAIN, + total=5) + ] + found = get_close_matches(title, possibilities, n=1) + if found: + pywikibot.info( + f'probably <<lightblue>>{found[0]}<<default>> is meant') + return True + + self.summary_parameters = dict(entries[0]) return False