def test_repo_from_text_3(self):
text = """ This text mentions the same repo twice, once
on github.io (user.github.io/repo) and once on github.com at
github.com/user/repo. There's also a repo name with a dash: github.com/user-
name/repo.
"""
self.assertSetEqual(gh_repo_from_text(text), {"user/repo", "user-name/repo"})
def test_repo_from_text_1(self):
text = """Here's some text. It talks about GitHub.
There is one repo called https://github.com/user1/repo1.
There is a sentence that mentions two repos: here's one (https://github.com/user2/repo2) and one
with more subdirectories (https://github.com/user3/repo-3/master/etc). There's one
on github.io: user4.github.io/repo4.
"""
self.assertSetEqual(gh_repo_from_text(text), {"user1/repo1", "user2/repo2", "user3/repo-3", "user4/repo4"})
def test_text_without_repo(self):
text = """These are some malformed repo names.
github.com///
github.com/user.name/repo.name
github.com/user?/repo?name
github.com//repo
github.com/user//
"""
self.assertEqual(gh_repo_from_text(text), None)
def gh_repos_from_metadata(metadata):
abstract = metadata['abstract']
gh_from_abstract = gh_repo_from_text(abstract)
# If the abstract contains a GitHub repo, return that
if gh_from_abstract is not None:
return {'repos': gh_from_abstract, 'source': 'abstract'}
else:
# Look for a pdf
pdf = "%s/%s" % (pdf_dir, metadata['internal_pdf'])
if os.path.isfile(pdf):
# Try to get GitHub repo from pdf; will be None if not applicable
gh_from_pdf = gh_repo_from_pdf(pdf)
if gh_from_pdf is not None:
return {'repos': gh_from_pdf, 'source': 'pdf'}
else:
return None
else:
# Return None if there is no pdf
return None
def test_repo_from_text_2(self):
text = "This one only has github.io: user.github.io/repo."
self.assertSetEqual(gh_repo_from_text(text), {"user/repo"})