def recategorize_over_redirect(self, category_namespace=14):
# FIXME: the source_namespace parameter of redirects_map does not work,
# so we need to do manual filtering
redirects = self.api.redirects_map()
catredirs = dict((key, value) for key, value in redirects.items() if Title(self.api, key).namespace == "Category")
for source, target in catredirs.items():
ans = ask_yesno("Recategorize pages from '{}' to '{}'?".format(source, target))
if ans is False:
continue
catmembers = self.api.generator(generator="categorymembers", gcmtitle=source, gcmlimit="max", prop="revisions", rvprop="content|timestamp")
for page in catmembers:
# the same page might be yielded multiple times
if "revisions" in page:
self.recategorize_page(page, source, target)
# check again to see if the category is empty
catmembers = list(self.api.list(list="categorymembers", cmtitle=source, cmlimit="max"))
if len(catmembers) == 0:
self.flag_for_deletion(source)
else:
logger.warning("'{}' is still not empty: {}".format(source, sorted(page["title"] for page in catmembers)))
input("Press Enter to continue...")
print("""
Recategorization complete. Before deleting the unused categories, make sure to \
update interlanguage links. The unused categories are still redirects and are \
not listed under Special:UnusedCategories, but they can be found in \
Special:WhatLinksHere/Template:Deletion.
""")
def rename_non_english(self):
del self.allpages
# FIXME: starting with English pages is not very good:
# - some pages are omitted (e.g. when two pages link to the same English page, at least warning should be printed)
# - it suggests to move e.g. Xfce (Česky) to Xfwm (Česky) because multiple English pages link to it
# Therefore we limit it only to categories...
for page in self.allpages:
title = page["title"]
if lang.detect_language(
title)[1] == "English" and title.startswith("Category:"):
langlinks = self.get_langlinks(title)
for tag, localized_title in langlinks:
logger.info("Checking [[{}:{}]] for renaming...".format(
tag, localized_title))
if lang.is_internal_tag(tag) and localized_title != title:
source = "{} ({})".format(localized_title,
lang.langname_for_tag(tag))
target = "{} ({})".format(title,
lang.langname_for_tag(tag))
if self._page_exists(target):
logger.warning(
"Cannot move page [[{}]] to [[{}]]: target page already exists"
.format(source, target))
else:
# interactive mode is necessary because this assumes that all English pages are named correctly
ans = ask_yesno(
"Move page [[{}]] to [[{}]]?".format(
source, target))
if ans is True:
summary = "comply with [[Help:I18n#Page titles]] and match the title of the English page"
self.api.move(source, target, summary)
def filter_pages(self, pages):
for page in pages:
if "revisions" in page:
# skip truncated results (due to PHP's 8MiB limit)
if len(page["revisions"]) == 0:
continue
rev = page["revisions"][0]
else:
# empty prop in generator due to continuation
continue
content = rev["*"]
if self.is_spam(page["title"], content):
print("Detected spam:")
pprint({"title": page["title"], "content": content, "timestamp": rev["timestamp"]})
if self.interactive and ask_yesno("Proceed with user account blocking and deletion?") is False:
continue
# block the account
block_user(self.api, rev["user"])
if rev["parentid"] == 0:
# first revision, delete whole page
delete_page(self.api, page["title"], page["pageid"])
else:
# TODO: if all revisions of the page are spam, delete the whole page
print("Warning: deletion of individual revisions is not implemented!")
else:
print("Page '{}' is not a spam.".format(page["title"]))
def recategorize_over_redirect(self, category_namespace=14):
# FIXME: the source_namespace parameter of redirects.fetch does not work, so we need to do manual filtering
redirects = self.api.redirects.fetch()
catredirs = dict((key, value) for key, value in redirects.items()
if self.api.Title(key).namespace == "Category")
for source, target in catredirs.items():
ans = ask_yesno("Recategorize pages from [[{}]] to [[{}]]?".format(
source, target))
if ans is False:
continue
catmembers = self.api.generator(generator="categorymembers",
gcmtitle=source,
gcmlimit="max",
prop="revisions",
rvprop="content|timestamp",
rvslots="main")
for page in catmembers:
# the same page might be yielded multiple times
if "revisions" in page:
self.recategorize_page(page, source, target)
# check again to see if the category is empty
catmembers = list(
self.api.list(list="categorymembers",
cmtitle=source,
cmlimit="max"))
if len(catmembers) == 0:
self.flag_for_deletion(source)
else:
logger.warning("Page [[{}]] is still not empty: {}".format(
source, sorted(page["title"] for page in catmembers)))
input("Press Enter to continue...")
print("""
Recategorization complete. Before deleting the unused categories, make sure to \
update interlanguage links. The unused categories are still redirects and are \
not listed under Special:UnusedCategories, but they can be found in \
Special:WhatLinksHere/Template:Deletion.
""")