def repo_path_to_local_path(repo_path):
"""
Convert a repository path into a local file system path.
This repository path is extended (compared to the repository path
used by RepoInfo) by adding support for existing local directories.
The process typically includes checking for a local directory or
parsing a full or short repository URL and then cloning the repository.
Arguments:
repo_path {string} -- Path of a remote repository or a local directory.
Returns:
string -- Path of a local directory equivalent to the one
specified by the repository path.
In case of local directories, the paths are equal.
"""
if repo_path is None:
return None
if isdir(repo_path):
return repo_path
info = RepoInfo.parse_repo_info(repo_path)
if info and info.clone_or_pull():
return info.dir
else:
return None
def get_repo_analysis(repo_path):
"""
Get analysis of a repository given its path.
Only one of the possible return values will be returned.
Use `isinstance` to determine the type of the returned value.
Returns:
string -- Message describing the state of the repo's analysis.
DetectionResult -- Detection result retrieved from the database.
list[dict] -- List of repositories matching the repository path.
"""
# Strip leading and trailing whitespace from the repo path.
repo_path = repo_path.strip()
repo_id = None
try:
repo_info = RepoInfo.parse_repo_info(repo_path)
conn = pg_conn(db_url)
if not repo_info:
if re.fullmatch(r"^[\w\.\-]+$", repo_path):
repos = _find_repos_by_metadata(conn, repo_path)
# No repository matches the given repository path.
if not repos:
raise UserInputError(
"No matching repository found in the database")
# Exact one matching repository.
if len(repos) == 1:
return _get_repo_summary(conn, repos[0])
# Multiple repositories match the repository path.
else:
return repos
else:
raise UserInputError("Invalid Git repository path format")
repo_id = _try_insert_repo(conn, repo_info)
if repo_id is not None:
Thread(target=analyze_repo, args=(repo_info, repo_id)).start()
return "The repository has been added to the queue"
repo_dict = _get_repo_dict_from_repoinfo(conn, repo_info)
return _get_repo_summary(conn, repo_dict)
except PG_Error as ex:
handle_pg_error(ex, conn, repo_id)
return "Database error"
finally:
conn.close()
def test_repoinfo_clone(self):
"""Test the RepoInfo's ability to clone a repository."""
info = RepoInfo.parse_repo_info(
"https://github.com/calebdehaan/codeDuplicationParser")
assert info is not None
assert info.clone_or_pull()
self._check_repo_info(info)
def test_repoinfo_pull_no_change(self):
"""Test the RepoInfo's ability to pull a repo without any change."""
info = RepoInfo.parse_repo_info(
"https://github.com/calebdehaan/codeDuplicationParser")
assert info is not None
# Clone the repository for the first time.
assert info.clone_or_pull()
self._check_repo_info(info)
# Pull the repository because it already exists.
assert info.clone_or_pull()
self._check_repo_info(info)
def test_repoinfo_parse_attrib():
"""Test RepoInfo's attributes after parsing a repository path."""
info = RepoInfo.parse_repo_info("hTTpS://GiTHuB.CoM/username/repository")
assert info is not None
assert info.url == "https://:@github.com/username/repository"
assert info.server == "github.com"
assert info.user == "username"
assert info.name == "repository"
expected_dir = path_join("github.com", "username", "repository")
assert relpath(info.dir, clone_root_dir) == expected_dir
assert info.hash is None
def test_repoinfo_parse_valid():
"""Test RepoInfo parser's ability to parse valid repository paths."""
VALID_PATHS = [
"https://github.com/user/repo", "HTTPS://GITHUB.COM/user/repo",
"https://www.github.com/user/repo", "https://gitlab.com/user/repo",
"HTTPS://GITLAB.COM/user/repo", "https://www.gitlab.com/user/repo",
"http://github.com/user/repo",
"https://github.com/user_user-user.user/repo.repo-repo_repo",
"github.com/user/repo", "user/repo", "user//repo", "/user/repo",
"user/repo/", "/user/repo/", "////user/repo////",
"////user////repo////", "user_user-user.user/repo.repo-repo_repo",
"///user_user-user.user///repo.repo-repo_repo///"
]
for p in VALID_PATHS:
assert RepoInfo.parse_repo_info(p) is not None, \
f"\"{p}\" is a valid repository path"
def test_repoinfo_parse_invalid():
"""Test RepoInfo parser's ability to detect invalid repository paths."""
INVALID_PATHS = [
"",
"https://",
"https//github.com/user/repo",
"https/github.com/user/repo",
"://github.com/user/repo",
"//github.com/user/repo"
"/github.com/user/repo"
"https://github,com/user/repo",
"https://github;com/user/repo",
"https;//github.com/user/repo",
"https:\\\\github.com\\user\\repo",
]
for p in INVALID_PATHS:
assert RepoInfo.parse_repo_info(p) is None, \
f"\"{p}\" is not a valid repository path"