def test_partially_qualified_NS1_code(self):
"""Test 'wikipedia:Talk:Main Page' on enws is ns 1."""
config.mylang = 'en'
config.family = 'wikisource'
link = Link('wikipedia:Talk:Main Page')
link.parse()
self.assertEqual(link.site, self.get_site())
self.assertEqual(link.title, 'Main Page')
self.assertEqual(link.namespace, 1)
def test_via_local_non_local(self):
"""Test de:translatewiki:Main Page on English Wikipedia."""
link = Link('de:translatewiki:Main Page', self.get_site('wp'))
with self.assertRaisesRegex(
InvalidTitleError,
'de:translatewiki:Main Page links to a non local '
'site i18n:i18n '
'via an interwiki link to wikipedia:de'):
link.parse()
def test_partially_qualified_NS0_family(self):
"""Test that Link uses config.family for namespace 0."""
config.mylang = 'de'
config.family = 'wikipedia'
link = Link('en:Main Page')
link.parse()
self.assertEqual(link.site, self.get_site())
self.assertEqual(link.title, 'Main Page')
self.assertEqual(link.namespace, 0)
def test_fully_qualified_NS0_family(self):
"""Test 'wikipedia:en:Main Page' on enws is namespace 0."""
config.mylang = 'en'
config.family = 'wikisource'
link = Link('wikipedia:en:Main Page')
link.parse()
self.assertEqual(link.site, self.get_site('enwp'))
self.assertEqual(link.title, 'Main Page')
self.assertEqual(link.namespace, 0)
def test_partially_qualified_NS1_family(self):
"""Test 'en:Talk:Main Page' on dewp is namespace 1."""
config.mylang = 'de'
config.family = 'wikipedia'
link = Link('en:Talk:Main Page')
link.parse()
self.assertEqual(link.site, self.get_site())
self.assertEqual(link.title, 'Main Page')
self.assertEqual(link.namespace, 1)
def test_fully_qualified_NS1_family(self):
"""Test ':wikidata:testwiki:Talk:Q6' on enwp is namespace 1."""
config.mylang = 'en'
config.family = 'wikipedia'
link = Link(':wikidata:testwiki:Talk:Q6')
link.parse()
self.assertEqual(link.site, self.get_site('test.wp'))
self.assertEqual(link.title, 'Q6')
self.assertEqual(link.namespace, 1)
def test_fully_qualified_NS1_code(self):
"""Test ':species:species:Talk:Main Page' on species is namespace 1."""
config.mylang = 'en'
config.family = 'wikipedia'
link = Link(':species:species:Talk:Main Page')
link.parse()
self.assertEqual(link.site, self.get_site())
self.assertEqual(link.title, 'Main Page')
self.assertEqual(link.namespace, 1)
def test_fully_qualified_NS0_code(self):
"""Test ':testwiki:wikidata:Q6' on enwp is namespace 0."""
config.mylang = 'en'
config.family = 'wikipedia'
link = Link(':testwiki:wikidata:Q6')
link.parse()
self.assertEqual(link.site, self.get_site('wikidata'))
self.assertEqual(link.title, 'Q6')
self.assertEqual(link.namespace, 0)
def test_fully_qualified_NS1_family(self):
"""Test ':wikipedia:en:Talk:Main Page' on wikidata is namespace 1."""
config.mylang = 'wikidata'
config.family = 'wikidata'
link = Link(':wikipedia:en:Talk:Main Page')
link.parse()
self.assertEqual(link.site, self.get_site('enwp'))
self.assertEqual(link.title, 'Main Page')
self.assertEqual(link.namespace, 1)
def test_fully_qualified_NS0_family(self):
"""Test 'wikidata:testwiki:Q6' on enwp is namespace 0."""
config.mylang = 'en'
config.family = 'wikipedia'
link = Link('wikidata:testwiki:Q6')
link.parse()
self.assertEqual(link.site, pywikibot.Site('test', 'wikipedia'))
self.assertEqual(link.title, 'Q6')
self.assertEqual(link.namespace, 0)
def test_fully_qualified_NS1_code(self):
"""Test 'en:wikipedia:Talk:Main Page' on enwp is namespace 4."""
config.mylang = 'en'
config.family = 'wikipedia'
link = Link('en:wikipedia:Talk:Main Page')
link.parse()
self.assertEqual(link.site, self.get_site())
self.assertEqual(link.title, 'Talk:Main Page')
self.assertEqual(link.namespace, 4)
def test_fully_qualified_NS1(self):
"""Test prefixed links with 'Talk:Q6' on enwp is namespace 1."""
test = [('testwiki:wikidata', 'wikidata:wikidata'),
('wikidata:testwiki', 'wikipedia:test')]
for linkprefix, sitetitle in test:
with self.subTest(pattern=linkprefix):
link = Link(linkprefix + ':Talk:Q6')
link.parse()
self.assertEqual(link.site, pywikibot.Site(sitetitle))
self.assertEqual(link.title, 'Q6')
self.assertEqual(link.namespace, 1)
def test_py266_bug_exception(self):
"""Test Python issue 10254 causes an exception."""
pywikibot.page.unicodedata = __import__('unicodedata')
title = 'Li̍t-sṳ́'
with self.assertRaisesRegex(
UnicodeError,
re.escape('Link(%r, %s): combining characters detected, which '
'are not supported by Pywikibot on Python 2.6.6. '
'See https://phabricator.wikimedia.org/T102461' %
(title, self.site))):
Link(title, self.site)
def test_non_wiki_prefix(self):
"""Test that Link fails if the interwiki prefix is not a wiki."""
link = Link('bugzilla:1337', source=self.site)
# bugzilla does not return a json content but redirects to phab.
# api.Request._json_loads cannot detect this problem and raises
# a SiteDefinitionError. The site is created anyway but the title
# cannot be parsed
with self.assertRaises(SiteDefinitionError):
link.site
self.assertEqual(link.site.sitename, 'wikimedia:wikimedia')
self.assertTrue(link._is_interwiki)
def test_valid(self):
"""Test that valid titles are correctly normalized."""
title_tests = [
'Sandbox', 'A "B"', "A 'B'", '.com', '~', '"', "'",
'Foo/.../Sandbox', 'Sandbox/...', 'A~~', 'X' * 252
]
extended_title_tests = [
('Talk:Sandbox', 'Sandbox'),
('Talk:Foo:Sandbox', 'Foo:Sandbox'),
('File:Example.svg', 'Example.svg'),
('File_talk:Example.svg', 'Example.svg'),
(':A', 'A'),
# Length is 256 total, but only title part matters
('Category:' + 'X' * 248, 'X' * 248),
('A%20B', 'A B'),
('A é B', 'A é B'),
('A é B', 'A é B'),
('A é B', 'A é B'),
('A B', 'A B'),
('A   B', 'A B'),
]
site = self.get_site()
for title in title_tests:
with self.subTest(title=title):
self.assertEqual(Link(title, site).title, title)
for link, title in extended_title_tests:
with self.subTest(link=link, title=title):
self.assertEqual(Link(link, site).title, title)
anchor_link = Link('A | B', site)
self.assertEqual(anchor_link.title, 'A')
self.assertEqual(anchor_link.anchor, ' B')
section_link = Link('A%23B', site)
self.assertEqual(section_link.title, 'A')
self.assertEqual(section_link.section, 'B')
def test_non_wiki_prefix(self):
"""Test that Link fails if the interwiki prefix is not a wiki."""
link = Link('bugzilla:1337', source=self.site)
# bugzilla does not return a json content but redirects to phab.
# api.Request._json_loads cannot detect this problem and retries
# reloading due to 'the server may be down'
# ignore Timeout when trying to load siteninfo;
# the site is created anyway but the title cannot be parsed
with suppress(TimeoutError):
link.site
self.assertEqual(link.site.sitename, 'wikimedia:wikimedia')
self.assertTrue(link._is_interwiki)
def test_fully_qualified_NS1_family(self):
"""Test 'wikidata:testwiki:Talk:Q6' on enwp is namespace 1."""
config.mylang = 'en'
config.family = 'wikipedia'
link = Link('wikidata:testwiki:Talk:Q6')
link.parse()
if show_failures:
self.assertEqual(link.site, pywikibot.Site('test', 'wikipedia'))
self.assertEqual(link.title, 'Q6')
self.assertEqual(link.namespace, 1)
else:
self.assertEqual(link.site, pywikibot.Site('en', 'wikipedia'))
self.assertEqual(link.title, 'Wikidata:testwiki:Talk:Q6')
self.assertEqual(link.namespace, 0)
def test_fully_qualified_NS0_family(self):
"""Test ':wikidata:testwiki:Q6' on enwp is namespace 0."""
config.mylang = 'en'
config.family = 'wikipedia'
link = Link(':wikidata:testwiki:Q6')
link.parse()
if show_failures:
self.assertEqual(link.site, self.get_site('test.wp'))
self.assertEqual(link.title, 'Q6')
self.assertEqual(link.namespace, 0)
else:
self.assertEqual(link.site, self.get_site('enwp'))
self.assertEqual(link.title, 'Wikidata:testwiki:Q6')
self.assertEqual(link.namespace, 0)
def test_fully_qualified_NS1_code(self):
"""Test 'testwiki:wikidata:Talk:Q6' on enwp is namespace 1."""
config.mylang = 'en'
config.family = 'wikipedia'
link = Link('testwiki:wikidata:Talk:Q6')
if show_failures:
link.parse()
self.assertEqual(link.site, pywikibot.Site('wikidata', 'wikidata'))
self.assertEqual(link.title, 'Q6')
self.assertEqual(link.namespace, 1)
else:
self.assertRaisesRegex(
pywikibot.Error,
'Family testwiki does not exist',
link.parse) # very bad
def test_fully_qualified_NS0_family(self):
"""Test 'wikipedia:en:Main Page' on wikidata is namespace 0."""
config.mylang = 'wikidata'
config.family = 'wikidata'
link = Link('wikipedia:en:Main Page')
if show_failures:
link.parse()
self.assertEqual(link.site, self.get_site('enwp'))
self.assertEqual(link.namespace, 0)
self.assertEqual(link.title, 'Main Page')
else:
self.assertRaisesRegex(
pywikibot.NoSuchSite,
'Language wikidata does not exist in family wikipedia',
link.parse) # very bad
def test_fully_qualified_NS1_code(self):
"""Test 'en:wikipedia:Talk:Main Page' on wikidata is not namespace 1."""
config.mylang = 'wikidata'
config.family = 'wikidata'
link = Link('en:wikipedia:Talk:Main Page')
link.parse()
if show_failures:
self.assertEqual(link.site, self.get_site('enwp'))
else:
self.assertEqual(link.site, self.get_site('wikidata'))
if show_failures:
self.assertEqual(link.title, 'Talk:Main Page')
self.assertEqual(link.namespace, 4)
else:
self.assertEqual(link.title, 'En:wikipedia:Talk:Main Page')
self.assertEqual(link.namespace, 0)
def test_fully_qualified_NS1(self):
"""Test that fully qualified link is in namespace 1."""
family, code = 'wikipedia:en'.split(':')
for colon in ('', ':'): # with or without preleading colon
# switch code:family sequence en:wikipedia or wikipedia:en
for first, second in [(family, code), (code, family)]:
with self.subTest(colon=colon,
site='{}:{}'.format(first, second)):
link_title = self.PATTERN.format(colon=colon,
first=first,
second=second,
title='Talk:Main Page')
link = Link(link_title)
link.parse()
self.assertEqual(link.site, self.get_site('enwp'))
self.assertEqual(link.title, 'Main Page')
self.assertEqual(link.namespace, 1)
def test_fully_qualified_NS1_code(self):
"""Test 'en:wikipedia:Main Page' on enws is namespace 1."""
config.mylang = 'en'
config.family = 'wikisource'
link = Link('en:wikipedia:Talk:Main Page')
if show_failures:
link.parse()
else:
self.assertRaisesRegex(
pywikibot.Error,
"Improperly formatted interwiki link 'en:wikipedia:Talk:Main Page'",
link.parse)
if show_failures:
self.assertEqual(link.site, self.get_site('enwp'))
else:
self.assertEqual(link.site, self.get_site('enws'))
if show_failures:
self.assertEqual(link.title, 'Main Page')
self.assertEqual(link.namespace, 1)
else:
try:
link.title
except pywikibot.Error as e:
self.assertEqual(str(e), "Improperly formatted interwiki link 'en:wikipedia:Talk:Main Page'")
def test_no_text(self):
"""Test that Link doesn't allow empty."""
link = Link('', self.get_site())
self.assertRaisesRegex(InvalidTitle,
'The link does not contain a page title',
link.parse)
def test_interwiki_namespace_without_title(self):
"""Test that Link doesn't allow links without a title."""
link = Link('en:Help:', self.get_site())
self.assertRaisesRegex(InvalidTitle, "'en:Help:' has no title.",
link.parse)
def test_no_change(self):
"""Test T102461 (Python issue 10254) is not encountered."""
title = 'Li̍t-sṳ́'
link = Link(title, self.site)
self.assertEqual(link.title, 'Li̍t-sṳ́')
def test_invalid(self):
"""Test that invalid titles raise InvalidTitle exception."""
# Bad characters forbidden regardless of wgLegalTitleChars
def generate_contains_illegal_chars_exc_regex(text):
exc_regex = (
r'^(u|)\'{}\' contains illegal char\(s\) (u|)\'{}\'$'.format(
re.escape(text), re.escape(text[2])))
return exc_regex
# Directory navigation
def generate_contains_dot_combinations_exc_regex(text):
exc_regex = (
r'^\(contains \. / combinations\): (u|)\'{}\'$'.format(
re.escape(text)))
return exc_regex
# Tilde
def generate_contains_tilde_exc_regex(text):
exc_regex = r'^\(contains ~~~\): (u|)\'%s\'$' % re.escape(text)
return exc_regex
# Overlength
def generate_overlength_exc_regex(text):
exc_regex = r'^\(over 255 bytes\): (u|)\'%s\'$' % re.escape(text)
return exc_regex
# Namespace prefix without actual title
def generate_has_no_title_exc_regex(text):
exc_regex = r'^(u|)\'{}\' has no title\.$'.format(
re.escape(text.strip()))
return exc_regex
title_tests = [
# Empty title
(['', ':', '__ __',
' __ '], r'^The link does not contain a page title$'),
(['A [ B', 'A ] B', 'A { B', 'A } B', 'A < B',
'A > B'], generate_contains_illegal_chars_exc_regex),
# URL encoding
# %XX is understood by wikimedia but not %XXXX
(['A%2523B'],
r'^(u|)\'A%23B\' contains illegal char\(s\) (u|)\'%23\'$'),
# A link is invalid if their (non-)talk page would be in another
# namespace than the link's "other" namespace
(['Talk:File:Example.svg'],
r'The \(non-\)talk page of (u|)\'Talk:File:Example.svg\''
r' is a valid title in another namespace.'),
([
'.', '..', './Sandbox', '../Sandbox', 'Foo/./Sandbox',
'Foo/../Sandbox', 'Sandbox/.', 'Sandbox/..'
], generate_contains_dot_combinations_exc_regex),
(['A ~~~ Name', 'A ~~~~ Signature',
'A ~~~~~ Timestamp'], generate_contains_tilde_exc_regex),
([('x' * 256),
('Invalid:' + 'X' * 248)], generate_overlength_exc_regex),
(['Talk:', 'Category: ',
'Category: #bar'], generate_has_no_title_exc_regex),
]
for texts_to_test, exception_regex in title_tests:
for text in texts_to_test:
with self.subTest(title=text):
if callable(exception_regex):
regex = exception_regex(text)
else:
regex = exception_regex
with self.assertRaisesRegex(InvalidTitle, regex):
Link(text, self.get_site()).parse()
def test_invalid(self):
self.assertRaises(InvalidTitle, Link('', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link(':', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('__ __', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link(' __ ', self.get_site()).parse)
# Bad characters forbidden regardless of wgLegalTitleChars
self.assertRaises(InvalidTitle, Link('A [ B', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('A ] B', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('A { B', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('A } B', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('A < B', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('A > B', self.get_site()).parse)
# URL encoding
# %XX is understood by wikimedia but not %XXXX
self.assertRaises(InvalidTitle, Link('A%2523B', self.get_site()).parse)
# A link is invalid if their (non-)talk page would be in another
# namespace than the link's "other" namespace
self.assertRaises(InvalidTitle, Link('Talk:File:Example.svg', self.get_site()).parse)
# Directory navigation
self.assertRaises(InvalidTitle, Link('.', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('..', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('./Sandbox', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('../Sandbox', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('Foo/./Sandbox', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('Foo/../Sandbox', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('Sandbox/.', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('Sandbox/..', self.get_site()).parse)
# Tilde
self.assertRaises(InvalidTitle, Link('A ~~~ Name', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('A ~~~~ Signature', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('A ~~~~~ Timestamp', self.get_site()).parse)
# Overlength
self.assertRaises(InvalidTitle, Link('x' * 256, self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('Invalid:' + 'X' * 248, self.get_site()).parse)
# Namespace prefix without actual title
self.assertRaises(InvalidTitle, Link('Talk:', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('Category: ', self.get_site()).parse)
self.assertRaises(InvalidTitle, Link('Category: #bar', self.get_site()).parse)
def test_valid(self):
self.assertEqual(Link('Sandbox', self.get_site()).title, 'Sandbox')
self.assertEqual(Link('A "B"', self.get_site()).title, 'A "B"')
self.assertEqual(Link('A \'B\'', self.get_site()).title, 'A \'B\'')
self.assertEqual(Link('.com', self.get_site()).title, '.com')
self.assertEqual(Link('~', self.get_site()).title, '~')
self.assertEqual(Link('"', self.get_site()).title, '"')
self.assertEqual(Link('\'', self.get_site()).title, '\'')
self.assertEqual(Link('Talk:Sandbox', self.get_site()).title, 'Sandbox')
self.assertEqual(Link('Talk:Foo:Sandbox', self.get_site()).title, 'Foo:Sandbox')
self.assertEqual(Link('File:Example.svg', self.get_site()).title, 'Example.svg')
self.assertEqual(Link('File_talk:Example.svg', self.get_site()).title, 'Example.svg')
self.assertEqual(Link('Foo/.../Sandbox', self.get_site()).title, 'Foo/.../Sandbox')
self.assertEqual(Link('Sandbox/...', self.get_site()).title, 'Sandbox/...')
self.assertEqual(Link('A~~', self.get_site()).title, 'A~~')
self.assertEqual(Link(':A', self.get_site()).title, 'A')
# Length is 256 total, but only title part matters
self.assertEqual(Link('Category:' + 'X' * 248, self.get_site()).title, 'X' * 248)
self.assertEqual(Link('X' * 252, self.get_site()).title, 'X' * 252)
self.assertEqual(Link('A%20B', self.get_site()).title, 'A B')
self.assertEqual(Link('A é B', self.get_site()).title, u'A é B')
self.assertEqual(Link('A é B', self.get_site()).title, u'A é B')
self.assertEqual(Link('A é B', self.get_site()).title, u'A é B')
l = Link('A | B', self.get_site())
self.assertEqual(l.title, 'A')
self.assertEqual(l.anchor, ' B')
l = Link('A%23B', self.get_site())
self.assertEqual(l.title, 'A')
self.assertEqual(l.section, 'B')
def test_other_wiki_prefix(self):
"""Test that Link fails if the interwiki prefix is a unknown family."""
link = Link('bulba:title on auto-generated Site', source=self.site)
self.assertEqual(link.title, 'Title on auto-generated Site')
self.assertEqual(link.site.sitename, 'bulba:bulba')
self.assertTrue(link._is_interwiki)