Fixes regexes to accept multipart git repository address. (cvat-ai#1655)

* Fixes regexes to accept multipart git repository address.

* Added test case for multi level path.

* Reduced complexity of SSH regex.

* Fixing unit tests.

* Fix SSH formatting.

Co-authored-by: kpawelczyk <[email protected]>

cvat/apps/git/git.py

@@ -78,9 +78,9 @@ def _parse_url(self):
             # Reference on URL formats accepted by Git:
             # https://github.com/git/git/blob/77bd3ea9f54f1584147b594abc04c26ca516d987/url.c
 
-            host_pattern = r"((?:(?:(?:\d{1,3}\.){3}\d{1,3})|(?:[a-zA-Z0-9._-]+.[a-zA-Z]+))(?::\d+)?)"
-            http_pattern = r"(?:http[s]?://)?" + host_pattern + r"((?:/[a-zA-Z0-9._-]+){2})"
-            ssh_pattern = r"([a-zA-Z0-9._-]+)@" + host_pattern + r":([a-zA-Z0-9._-]+)/([a-zA-Z0-9._-]+)"
+            host_pattern = r"((?:(?:(?:\d{1,3}\.){3}\d{1,3})|(?:[a-zA-Z0-9._-]+[.a-zA-Z]+))(?::\d+)?)"
+            http_pattern = r"(?:http[s]?://)?" + host_pattern + r"((?:/[a-zA-Z0-9._-]+){2,})"
+            ssh_pattern = r"([a-zA-Z0-9._-]+)@" + host_pattern + r":([a-zA-Z0-9._-]+)((?:/[a-zA-Z0-9._-]+)+)"
 
             http_match = re.match(http_pattern, self._url)
             ssh_match = re.match(ssh_pattern, self._url)
@@ -95,7 +95,7 @@ def _parse_url(self):
             elif ssh_match:
                 user = ssh_match.group(1)
                 host = ssh_match.group(2)
-                repos = "{}/{}".format(ssh_match.group(3), ssh_match.group(4))
+                repos = "{}{}".format(ssh_match.group(3), ssh_match.group(4))
             else:
                 raise Exception("Git repository URL does not satisfy pattern")
 

cvat/apps/git/tests.py

@@ -28,7 +28,7 @@ def _check_correct_urls(self, samples):
     def test_correct_urls_can_be_parsed(self):
         hosts = ['host.zone', '1.2.3.4']
         ports = ['', ':42']
-        repo_groups = ['repo', 'r4p0']
+        repo_groups = ['repo', 'r4p0', 'multi/group', 'multi/group/level']
         repo_repos = ['nkjl23', 'hewj']
         git_suffixes = ['', '.git']