def fix_allpages(self):
pageids = self.find_broken()
if not pageids:
logger.info("All pages are categorized under correct language.")
return
for chunk in ws.utils.iter_chunks(pageids, self.api.max_ids_per_query):
pageids = "|".join(str(pageid) for pageid in chunk)
result = self.api.call_api(action="query",
pageids=pageids,
prop="revisions",
rvprop="content|timestamp")
pages = result["pages"]
for page in pages.values():
logger.info(
"Fixing language of categories on page [[{}]]...".format(
page["title"]))
timestamp = page["revisions"][0]["timestamp"]
text_old = page["revisions"][0]["*"]
text_new = self.fix_page(page["title"], text_old)
if text_old != text_new:
try:
edit_interactive(self.api,
page["title"],
page["pageid"],
text_old,
text_new,
timestamp,
self.edit_summary,
bot="")
# self.api.edit(page["title"], page["pageid"], text_new, timestamp, self.edit_summary, bot="")
except APIError:
pass
def _edit(self, title, pageid, text_new, text_old, timestamp,
edit_summary):
if text_old != text_new:
if self.dry_run:
diff = diff_highlighted(text_old, text_new, title + ".old",
title + ".new", timestamp, "<utcnow>")
print(diff)
print("Edit summary: " + edit_summary)
print("(edit discarded due to --dry-run)")
else:
try:
if self.interactive is False:
self.api.edit(title,
pageid,
text_new,
timestamp,
edit_summary,
bot="")
else:
edit_interactive(self.api,
title,
pageid,
text_old,
text_new,
timestamp,
edit_summary,
bot="")
except APIError as e:
pass
def check_allpages(self):
if not self.finder.refresh():
raise Exception("Failed to refresh package information.")
# ensure that we are authenticated
require_login(self.api)
namespaces = [0, 4, 14, 3000]
for ns in namespaces:
for page in self.api.generator(generator="allpages", gaplimit="100", gapfilterredir="nonredirects", gapnamespace=ns,
prop="revisions", rvprop="content|timestamp", rvslots="main"):
title = page["title"]
if title in self.blacklist_pages:
logger.info("skipping blacklisted page [[{}]]".format(title))
continue
timestamp = page["revisions"][0]["timestamp"]
text_old = page["revisions"][0]["slots"]["main"]["*"]
text_new = self.update_page(title, text_old)
if text_old != text_new:
try:
if self.interactive:
edit_interactive(self.api, title, page["pageid"], text_old, text_new, timestamp, self.edit_summary, bot="")
else:
self.api.edit(title, page["pageid"], text_new, timestamp, self.edit_summary, bot="")
except APIError:
pass
def _edit(self, title, pageid, text_new, text_old, timestamp, edit_summary):
if text_old != text_new:
# print the info message
print("\nSuggested edit for page [[{}]]. Please double-check all changes before accepting!".format(title))
try:
# TODO: set bot="" only when the logged-in user is a bot
edit_interactive(self.api, title, pageid, text_old, text_new, timestamp, edit_summary, bot="")
except APIError as e:
pass
def _edit(self, title, pageid, text_new, text_old, timestamp, edit_summary):
if text_old != text_new:
if self.dry_run:
diff = diff_highlighted(text_old, text_new, title + ".old", title + ".new", timestamp, "<utcnow>")
print(diff)
print("Edit summary: " + edit_summary)
print("(edit discarded due to --dry-run)")
else:
try:
if self.interactive is False:
self.api.edit(title, pageid, text_new, timestamp, edit_summary, bot="")
else:
edit_interactive(self.api, title, pageid, text_old, text_new, timestamp, edit_summary, bot="")
except APIError as e:
pass
def _edit(self, title, pageid, text_new, text_old, timestamp,
edit_summary):
if text_old == text_new:
return
if self.dry_run:
diff = diff_highlighted(text_old, text_new, title + ".old",
title + ".new", timestamp, "<utcnow>")
print(diff)
print("Edit summary: " + edit_summary)
print("(edit discarded due to --dry-run)")
return
interactive = self.interactive
# override interactive mode for edits which are very frequent and "always" correct
if "bot" in self.api.user.rights and edit_summary == "update http to https":
interactive = False
try:
if interactive is False:
self.api.edit(title,
pageid,
text_new,
timestamp,
edit_summary,
bot="")
else:
# print the info message
print(
"\nSuggested edit for page [[{}]]. Please double-check all changes before accepting!"
.format(title))
if "bot" in self.api.user.rights:
edit_interactive(self.api,
title,
pageid,
text_old,
text_new,
timestamp,
edit_summary,
bot="")
else:
edit_interactive(self.api, title, pageid, text_old,
text_new, timestamp, edit_summary)
except APIError:
pass
def fix_allpages(self):
pageids = self.find_categorized()
if not pageids:
logger.info("All pages are categorized under correct language.")
return
for chunk in ws.utils.iter_chunks(pageids, self.api.max_ids_per_query):
pageids = "|".join(str(pageid) for pageid in chunk)
result = self.api.call_api(action="query",
pageids=pageids,
prop="revisions",
rvprop="content|timestamp")
pages = result["pages"]
for page in pages.values():
logger.info("Decategorizing page [[{}]]...".format(
page["title"]))
timestamp = page["revisions"][0]["timestamp"]
text_old = page["revisions"][0]["*"]
text_new = self.decategorize(page["title"], text_old)
if text_old != text_new:
title = self.api.Title(page["title"])
if title.namespace == "User":
edit_summary = "user pages shall not be categorized, see [[Help:Style#User pages]]"
else:
edit_summary = "talk pages should not be categorized"
try:
edit_interactive(self.api,
page["title"],
page["pageid"],
text_old,
text_new,
timestamp,
edit_summary,
bot="")
# self.api.edit(page["title"], page["pageid"], text_new, timestamp, edit_summary, bot="")
except APIError:
pass
else:
logger.error("Failed to decategorize page [[{}]]".format(
page["title"]))
def fix_allpages(self):
pageids = self.find_broken()
if not pageids:
logger.info("All pages are categorized under correct language.")
return
for chunk in ws.utils.iter_chunks(pageids, self.api.max_ids_per_query):
pageids = "|".join(str(pageid) for pageid in chunk)
result = self.api.call_api(action="query", pageids=pageids, prop="revisions", rvprop="content|timestamp")
pages = result["pages"]
for page in pages.values():
logger.info("Fixing language of categories on page [[{}]]...".format(page["title"]))
timestamp = page["revisions"][0]["timestamp"]
text_old = page["revisions"][0]["*"]
text_new = self.fix_page(page["title"], text_old)
if text_old != text_new:
try:
edit_interactive(self.api, page["title"], page["pageid"], text_old, text_new, timestamp, self.edit_summary, bot="")
# self.api.edit(page["title"], page["pageid"], text_new, timestamp, self.edit_summary, bot="")
except APIError:
pass
def save(self, edit_summary, interactive=False, **kwargs):
"""
Saves the updated wikicode of the page to the wiki.
:param str edit_summary: Summary of the change.
:param bool interactive:
If ``True``, calls :py:func:`ws.interactive.edit_interactive` to ask
the user before making the change; otherwise calls
:py:meth:`API.edit <ws.client.api.API.edit>` directly.
:param kwargs: Additional keyword arguments passed to the API query.
"""
text_new = str(self.wikicode)
if self.contents[self.title] != text_new:
# use bot=1 iff it makes sense
kwargs["bot"] = "1"
if "bot" not in self.api.user.rights:
del kwargs["bot"]
if interactive is True:
edit_interactive(self.api, self.title, self.pageids[self.title], self.contents[self.title], text_new, self.timestamps[self.title], edit_summary, **kwargs)
else:
self.api.edit(self.title, self.pageids[self.title], text_new, self.timestamps[self.title], edit_summary, **kwargs)
else:
logger.info("Page [[{}]] is already up to date.".format(self.title))
