def test_gather_ignores_links(client, root_doc, simple_user):
user_profile_url = simple_user.get_absolute_url()
content = """
<ul>
<li><a href="/en-US/docs/Absolute/Link">Absolute Link</a></li>
<li><a href="Relative/Link">Relative Link</a></li>
<li><a href="#later">Later in this page.</a></li>
<li><a href="%s">Profile Link</a></li>
</ul>
""" % user_profile_url
new_rev = Revision(document=root_doc,
creator=root_doc.current_revision.creator,
content=content,
created=datetime(2017, 6, 5))
new_rev.save()
base_path = root_doc.get_absolute_url()
html = client.get(base_path, HTTP_HOST=settings.WIKI_HOST).content
source = LinksSource(base_path)
requester = mock_requester(content=html)
storage = mock_storage()
resources = source.gather(requester, storage)
assert source.state == source.STATE_DONE
assert source.freshness == source.FRESH_YES
expected = ('document', '/en-US/docs/Absolute/Link', {})
assert expected in resources
for doc, path, options in resources:
assert "Relative/Link" not in path
assert "#later" not in path
assert user_profile_url not in path
def revision(save=False, **kwargs):
"""Return an empty revision with enough stuff filled out that it can be
saved.
Revision's is_approved=False unless you specify otherwise.
Requires a users fixture if no creator is provided.
"""
doc = None
if 'document' not in kwargs:
doc = document(save=True)
else:
doc = kwargs['document']
defaults = {
'summary': 'Some summary',
'content': 'Some content',
'comment': 'Some comment',
'creator': kwargs.get('creator', get_user()),
'document': doc,
'tags': '"some", "tags"',
'toc_depth': 1,
}
defaults.update(kwargs)
rev = Revision(**defaults)
if save:
rev.save()
return rev
def handle(self, *args, **options):
# get first user to be the creator
u = User.objects.all()[0]
loaded_docs = []
skipped_docs = []
for slug in KS_AUTOLOAD_MODULES:
template_response = requests.get(RAW_TEMPLATE_URL % slug)
doc = Document(title=slug,
slug=slug,
category=Document.CATEGORIES[0][0])
try:
doc.save()
loaded_docs.append(slug)
except SlugCollision:
# skip modules already in the db
skipped_docs.append(slug)
continue
rev = Revision(document=doc,
content=template_response.content,
creator=u)
rev.save()
print "Loaded docs:"
for slug in loaded_docs:
print "%s" % slug
print "\nSkipped docs that were already loaded:"
for slug in skipped_docs:
print "%s" % slug
