index_refresh.py
ActivePublic
Actions

Authored by jayvdb on Oct 3 2015, 8:38 AM.

Tags

Referenced Files

	F2655644: index_refresh.py
	Oct 3 2015, 8:38 AM

Subscribers

None

	# -- coding: utf-8 --
	from __future__ import unicode_literals

	import datetime

	import pywikibot

	from pywikibot.pagegenerators import RecentChangesPageGenerator

	from pywikibot.tools import filter_unique


	def main():
	pywikibot.handle_args()

	site = pywikibot.Site()
	assert site.family.name == 'wikisource'

	page_namespace = site.proofread_page_ns
	index_namespace = site.proofread_index_ns

	now = datetime.datetime.utcnow()
	five_hours_ago = now - datetime.timedelta(hours=5)

	print('** find %s namespace %d edits in last five hours'
	% (site, page_namespace))

	rcgen = RecentChangesPageGenerator(site=site,
	namespaces=[page_namespace],
	end=five_hours_ago,
	total=500,
	_filter_unique=filter_unique)

	pages = sorted(rcgen)

	indexes = set()

	for page in pages:
	no_ns_title = page._link.title
	try:
	no_pageno_title, page_no = no_ns_title.rsplit('/', 1)
	except:
	continue

	indexes.add(no_pageno_title)

	print('** refreshing %s namespace %d pages'
	% (site, index_namespace))

	for title in indexes:
	index_page = pywikibot.Page(site, title, index_namespace)

	if not index_page.exists():
	print('somehow %s doesnt exist' % index_page)
	else:
	index_page.purge()
	print('%s purged' % index_page)


	if __name__ == "__main__":
	main()

jayvdb changed the title of this paste from untitled to index_refresh.py.

jayvdb updated the paste's language from autodetect to python.

index_refresh.pyActivePublicActions