def test_captcha(self, mock_site):
# Check that Captcha results in EditError
mock_site.blocked = False
mock_site.rights = ['read', 'edit']
title = 'Norge'
mock_site.get.return_value = {
'query': {
'pages': {
'728': {
'protection': []
}
}
}
}
page = Page(mock_site, title)
mock_site.post.return_value = {
'edit': {
'result': 'Failure',
'captcha': {
'type': 'math',
'mime': 'text/tex',
'id': '509895952',
'question': '36 + 4 = '
}
}
}
# For now, mwclient will just raise an EditError.
# <https://github.com/mwclient/mwclient/issues/33>
with pytest.raises(mwclient.errors.EditError):
page.edit('Some text')
def process(self, page_name):
page = Page(self.client, page_name)
wikitext = page.text()
if BEGINNING:
wikitext = TEMPLATE + "\n" + wikitext
else:
wikitext = wikitext + "\n" + TEMPLATE
page.edit(wikitext,
summary=self.lang.t("scripts.addtemplate.summary").format(
template=TEMPLATE))
def process(self, article):
page = Page(self.client, article)
if not page.redirect:
return
self.logger.info(self.lang.t("scripts.cleanupdoubleredirects.processing").format(name=article))
redirection_chain = []
while page.redirect:
if page.name in redirection_chain:
self.logger.warning(self.lang.t("cleanupdoubleredirects.circular_reference").format(name=article))
return
redirection_chain.append(page.name)
page = Page(self.client, page.redirects_to())
if len(redirection_chain) > 1:
for redirect in redirection_chain[:-1]:
self.logger.info(self.lang.t("resolving").format(redirect=redirect, target=page.name))
redir = Page(self.client, redirect)
redir.edit("#REDIRECT [[" + page.name + "]]", self.lang.t("scripts.cleanupdoubleredirects.reason")
.format(target=page.name))
def do_recategorize(self, article: Page, additions, removals, replaces):
if self.query == "":
return
print("Guardando...")
wikitext = article.text()
wikitext = wikitext.replace("[[Category:", "[[" + self.category_name)
for removal in removals:
wikitext = re.sub(r'\[\[' + removal + r'(?:|.*)*\]\]', '', wikitext)
for addition in additions:
wikitext = wikitext + "\n[[{}]]".format(addition)
for replace in replaces:
search, rep = replace.split("%")
wikitext = re.sub(r'\[\[' + self.category_name + search + r'(?:|.*)*\]\]', '[[' + self.category_name + rep + ']]',
wikitext)
plainadditions = ",".join(map(lambda x: "+" + ",".join(x.split(':')[1:]), additions))
plainremovals = ",".join(map(lambda x: "-" + ",".join(x.split(':')[1:]), removals))
plainreplaces = ",".join(map(lambda x: "=>".join(x.split('%')), replaces))
query = plainadditions + "; " + plainremovals + "; " + plainreplaces
article.edit(wikitext, summary=self.lang.t("scripts.recategorize.reason") + query, minor=True, bot=True)
def save(self,
page: Page,
text,
summary=u'',
minor=False,
bot=True,
section=None,
**kwargs):
"""
Performs a page edit, retrying the login once if the edit fails due to the user being logged out
This function hopefully makes it easy to workaround the lag and frequent login timeouts
experienced on the Fandom UCP platform compared to Gamepedia Hydra.
:param page: mwclient Page object
:param text: as in mwclient.Page.edit
:param summary: as in mwclient.Page.edit
:param minor: as in mwclient.Page.edit
:param bot: as in mwclient.Page.edit
:param section: as in mwclient.Page.edit
:param kwargs: as in mwclient.Page.edit
:return: nothing
"""
try:
page.site = self.client
page.edit(text,
summary=summary,
minor=minor,
bot=bot,
section=section,
**kwargs)
except self.write_errors:
self._retry_login_action(self._retry_save,
'edit',
page=page,
text=text,
summary=summary,
minor=minor,
bot=bot,
section=section,
**kwargs)
class TestPageApiArgs(unittest.TestCase):
def setUp(self):
title = 'Some page'
self.page_text = 'Hello world'
MockSite = mock.patch('mwclient.client.Site').start()
self.site = MockSite()
self.site.get.return_value = {
'query': {
'pages': {
'1': {
'title': title
}
}
}
}
self.site.rights = ['read']
self.site.api_limit = 500
self.site.version = (1, 32, 0)
self.page = Page(self.site, title)
self.site.get.return_value = {
'query': {
'pages': {
'2': {
'ns':
0,
'pageid':
2,
'revisions': [{
'*': 'Hello world',
'timestamp': '2014-08-29T22:25:15Z'
}],
'title':
title
}
}
}
}
def get_last_api_call_args(self, http_method='POST'):
if http_method == 'GET':
args, kwargs = self.site.get.call_args
else:
args, kwargs = self.site.post.call_args
action = args[0]
args = args[1:]
kwargs.update(args)
return kwargs
def tearDown(self):
mock.patch.stopall()
def test_get_page_text(self):
# Check that page.text() works, and that a correct API call is made
text = self.page.text()
args = self.get_last_api_call_args(http_method='GET')
assert text == self.page_text
assert args == {
'prop': 'revisions',
'rvdir': 'older',
'titles': self.page.page_title,
'uselang': None,
'rvprop': 'content|timestamp',
'rvlimit': '1',
'rvslots': 'main',
}
def test_get_page_text_cached(self):
# Check page.text() caching
self.page.revisions = mock.Mock(return_value=iter([]))
self.page.text()
self.page.text()
# When cache is hit, revisions is not, so call_count should be 1
assert self.page.revisions.call_count == 1
self.page.text(cache=False)
# With cache explicitly disabled, we should hit revisions
assert self.page.revisions.call_count == 2
def test_get_section_text(self):
# Check that the 'rvsection' parameter is sent to the API
text = self.page.text(section=0)
args = self.get_last_api_call_args(http_method='GET')
assert args['rvsection'] == '0'
def test_get_text_expanded(self):
# Check that the 'rvexpandtemplates' parameter is sent to the API
text = self.page.text(expandtemplates=True)
args = self.get_last_api_call_args(http_method='GET')
assert self.site.expandtemplates.call_count == 1
assert args.get('rvexpandtemplates') is None
def test_assertuser_true(self):
# Check that assert=user is sent when force_login=True
self.site.blocked = False
self.site.rights = ['read', 'edit']
self.site.logged_in = True
self.site.force_login = True
self.site.api.return_value = {'edit': {'result': 'Ok'}}
self.page.edit('Some text')
args = self.get_last_api_call_args()
assert args['assert'] == 'user'
def test_assertuser_false(self):
# Check that assert=user is not sent when force_login=False
self.site.blocked = False
self.site.rights = ['read', 'edit']
self.site.logged_in = False
self.site.force_login = False
self.site.api.return_value = {'edit': {'result': 'Ok'}}
self.page.edit('Some text')
args = self.get_last_api_call_args()
assert 'assert' not in args
def test_handle_edit_error_assertuserfailed(self):
# Check that AssertUserFailedError is triggered
api_error = APIError(
'assertuserfailed', 'Assertion that the user is logged in failed',
'See https://en.wikipedia.org/w/api.php for API usage')
with pytest.raises(AssertUserFailedError):
self.page.handle_edit_error(api_error, 'n/a')
def test_handle_edit_error_protected(self):
# Check that ProtectedPageError is triggered
api_error = APIError(
'protectedpage',
'The "editprotected" right is required to edit this page',
'See https://en.wikipedia.org/w/api.php for API usage')
with pytest.raises(ProtectedPageError) as pp_error:
self.page.handle_edit_error(api_error, 'n/a')
assert pp_error.value.code == 'protectedpage'
assert str(
pp_error.value
) == 'The "editprotected" right is required to edit this page'
def test_get_page_categories(self):
# Check that page.categories() works, and that a correct API call is made
self.site.get.return_value = {
"batchcomplete": "",
"query": {
"pages": {
"1009371": {
"pageid": 1009371,
"ns": 14,
"title": "Category:1879 births",
},
"1005547": {
"pageid": 1005547,
"ns": 14,
"title": "Category:1955 deaths",
}
}
}
}
cats = list(self.page.categories())
args = self.get_last_api_call_args(http_method='GET')
assert {
'generator': 'categories',
'titles': self.page.page_title,
'iiprop':
'timestamp|user|comment|url|size|sha1|metadata|archivename',
'inprop': 'protection',
'prop': 'info|imageinfo',
'gcllimit': repr(self.page.site.api_limit),
} == args
assert set([c.name for c in cats]) == set([
'Category:1879 births',
'Category:1955 deaths',
])
