fix(code-hosting): recognize SSH repository hosts with userinfo

Use parsed hostnames instead of raw netloc values when matching configured code-hosting domains. This keeps ssh://git@host and explicit-port repository URLs on the supported code-hosting path while preserving git clone handling for SSH repository sources.

Add focused regression coverage for SSH URL userinfo and explicit-port repository URL helpers.

Author: officialasishkumar

openviking/parse/parsers/code/code.py

@@ -121,7 +121,9 @@ async def parse(self, source: Union[str, Path], instruction: str = "", **kwargs)
                 )
             elif source_str.startswith(("http://", "https://", "git://", "ssh://")):
                 repo_url, branch, commit = self._parse_repo_source(source_str, **kwargs)
-                if self._is_github_url(repo_url):
+                if repo_url.startswith(("http://", "https://", "git://")) and self._is_github_url(
+                    repo_url
+                ):
                     # Use GitHub ZIP API: supports branch names, tags, and commit SHAs
                     local_dir, repo_name = await self._github_zip_download(
                         repo_url, branch or commit, temp_local_dir
@@ -265,10 +267,18 @@ def _normalize_repo_url(self, url: str) -> str:
                 base_parts = path_parts[: git_index + 1]
 
             config = get_openviking_config()
-            if (
-                parsed.netloc in config.code.github_domains + config.code.gitlab_domains
-                and len(path_parts) >= 2
-            ):
+            domains = {
+                domain.lower() for domain in config.code.github_domains + config.code.gitlab_domains
+            }
+            host = (parsed.hostname or "").lower()
+            host_candidates = {host} if host else set()
+            try:
+                port = parsed.port
+            except ValueError:
+                port = None
+            if host and port is not None:
+                host_candidates.add(f"{host}:{port}")
+            if host_candidates & domains and len(path_parts) >= 2:
                 base_parts = path_parts[:2]
             base_path = "/" + "/".join(base_parts)
             return parsed._replace(path=base_path, query="", fragment="").geturl()

openviking/utils/code_hosting_utils.py

@@ -8,11 +8,37 @@
 """
 
 from typing import Optional
-from urllib.parse import urlparse
+from urllib.parse import ParseResult, urlparse
 
 from openviking_cli.utils.config import get_openviking_config
 
 
+def _domain_matches(parsed: ParseResult, domains: list[str]) -> bool:
+    """Return True when parsed URL host matches configured domains.
+
+    ``urlparse().netloc`` includes optional userinfo and port values. Repository
+    clone URLs commonly use forms like ``ssh://git@github.com/org/repo.git``,
+    where the netloc is ``git@github.com`` but the actual host is
+    ``github.com``.
+    """
+    hostname = parsed.hostname
+    if not hostname:
+        return False
+
+    normalized_domains = {domain.lower() for domain in domains}
+    host = hostname.lower()
+    candidates = {host}
+
+    try:
+        port = parsed.port
+    except ValueError:
+        port = None
+    if port is not None:
+        candidates.add(f"{host}:{port}")
+
+    return any(candidate in normalized_domains for candidate in candidates)
+
+
 def parse_code_hosting_url(url: str) -> Optional[str]:
     """Parse code hosting platform URL to get org/repo path.
 
@@ -60,7 +86,7 @@ def parse_code_hosting_url(url: str) -> Optional[str]:
 
     # For GitHub/GitLab URLs with org/repo structure
     if (
-        parsed.netloc in config.code.github_domains + config.code.gitlab_domains
+        _domain_matches(parsed, config.code.github_domains + config.code.gitlab_domains)
         and len(path_parts) >= 2
     ):
         # Take first two parts: org/repo
@@ -86,7 +112,7 @@ def is_github_url(url: str) -> bool:
         True if the URL is a GitHub URL
     """
     config = get_openviking_config()
-    return urlparse(url).netloc in config.code.github_domains
+    return _domain_matches(urlparse(url), config.code.github_domains)
 
 
 def is_gitlab_url(url: str) -> bool:
@@ -99,7 +125,7 @@ def is_gitlab_url(url: str) -> bool:
         True if the URL is a GitLab URL
     """
     config = get_openviking_config()
-    return urlparse(url).netloc in config.code.gitlab_domains
+    return _domain_matches(urlparse(url), config.code.gitlab_domains)
 
 
 def is_code_hosting_url(url: str) -> bool:
@@ -127,7 +153,7 @@ def is_code_hosting_url(url: str) -> bool:
         host_part = url[4:].split(":", 1)[0]
         return host_part in all_domains
 
-    return urlparse(url).netloc in all_domains
+    return _domain_matches(urlparse(url), all_domains)
 
 
 def validate_git_ssh_uri(url: str) -> None:
@@ -173,7 +199,7 @@ def is_git_repo_url(url: str) -> bool:
             )
         )
         parsed = urlparse(url)
-        if parsed.netloc not in all_domains:
+        if not _domain_matches(parsed, all_domains):
             return False
         path_parts = [p for p in parsed.path.split("/") if p]
         # Strip .git suffix from last part for counting

tests/test_code_hosting_utils.py

@@ -43,6 +43,8 @@ def _mock_config():
 _spec.loader.exec_module(_module)
 
 parse_code_hosting_url = _module.parse_code_hosting_url
+is_github_url = _module.is_github_url
+is_gitlab_url = _module.is_gitlab_url
 is_code_hosting_url = _module.is_code_hosting_url
 is_git_repo_url = _module.is_git_repo_url
 validate_git_ssh_uri = _module.validate_git_ssh_uri
@@ -81,6 +83,18 @@ def test_parse_code_hosting_url_https_dotgit():
     assert parse_code_hosting_url("https://github.com/org/repo.git") == "org/repo"
 
 
+def test_parse_code_hosting_url_ssh_url_with_userinfo():
+    assert parse_code_hosting_url("ssh://git@github.com/org/repo.git") == "org/repo"
+
+
+def test_parse_code_hosting_url_gitlab_ssh_url_with_userinfo():
+    assert parse_code_hosting_url("ssh://git@gitlab.com/group/repo.git") == "group/repo"
+
+
+def test_parse_code_hosting_url_https_with_port():
+    assert parse_code_hosting_url("https://github.com:443/org/repo") == "org/repo"
+
+
 # --- validate_git_ssh_uri ---
 
 
@@ -118,6 +132,25 @@ def test_is_code_hosting_url_https():
     assert is_code_hosting_url("https://github.com/org/repo") is True
 
 
+def test_is_code_hosting_url_ssh_url_with_userinfo():
+    assert is_code_hosting_url("ssh://git@github.com/org/repo.git") is True
+
+
+def test_is_code_hosting_url_https_with_port():
+    assert is_code_hosting_url("https://github.com:443/org/repo") is True
+
+
+# --- is_github_url / is_gitlab_url ---
+
+
+def test_is_github_url_ssh_url_with_userinfo():
+    assert is_github_url("ssh://git@github.com/org/repo.git") is True
+
+
+def test_is_gitlab_url_ssh_url_with_userinfo():
+    assert is_gitlab_url("ssh://git@gitlab.com/group/repo.git") is True
+
+
 # --- is_git_repo_url ---
 
 
@@ -129,6 +162,14 @@ def test_is_git_repo_url_https_repo():
     assert is_git_repo_url("https://github.com/org/repo") is True
 
 
+def test_is_git_repo_url_ssh_url_with_userinfo():
+    assert is_git_repo_url("ssh://git@github.com/org/repo.git") is True
+
+
+def test_is_git_repo_url_https_with_port():
+    assert is_git_repo_url("https://github.com:443/org/repo") is True
+
+
 def test_is_git_repo_url_https_issues():
     assert is_git_repo_url("https://github.com/org/repo/issues/123") is False