OSVauco/opax-mcp/test_gitea_handler.py
Chris Christiansen 56d2768473
Some checks are pending
Check Python Version Consistency / Check Python Version (push) Waiting to run
feat(opax-mcp): harden branch-resolved Gitea reads and plan transitions
2026-09-23 01:58:05 +00:00

728 lines
28 KiB
Python

import unittest
from pathlib import Path
import sys
from unittest.mock import AsyncMock, MagicMock, patch
import os
import base64
import json
import httpx
REPO_ROOT = Path(__file__).resolve().parents[1]
sys.path.insert(0, str(REPO_ROOT / "opax-mcp"))
from gitea_handler import handle_list_repo_files, _validate_read_branch_ref, handle_get_file_content
class TestValidateReadBranchRef(unittest.TestCase):
def test_accepted_refs(self):
accepted = [
"main",
"feat/opax-domain-decouple",
"docs/mcp-diagnostics",
"fix/example",
"chore/example",
"release/1.2.3",
]
for ref in accepted:
with self.subTest(ref=ref):
self.assertEqual(_validate_read_branch_ref(ref), ref)
def test_rejected_refs(self):
rejected = [
"",
" main",
"main ",
"feature branch",
"main\tbranch",
"main\nbranch",
"main\x00branch",
"main\x7fbranch",
"HEAD",
"head",
"refs/heads/main",
"/main",
"main/",
"main//next",
"main/./next",
"main/../next",
"main\\next",
"main?x=1",
"main#fragment",
"https://example.invalid",
"user@host",
"feature*",
"feature~1",
"feature^",
"feature{a}",
"feature;cmd",
"feature|cmd",
"feature%2Fbranch",
"feature'quote",
'feature"quote',
"feature[abc]",
]
for ref in rejected:
with self.subTest(ref=ref):
with self.assertRaisesRegex(ValueError, r"^Invalid git reference\.$"):
_validate_read_branch_ref(ref)
class TestGiteaHandler(unittest.TestCase):
def setUp(self):
self.original_gitea_url = os.environ.get("GITEA_URL")
self.original_gitea_token = os.environ.get("GITEA_TOKEN")
os.environ["GITEA_URL"] = "https://gitea.example.com"
os.environ["GITEA_TOKEN"] = "test-token"
def tearDown(self):
if self.original_gitea_url is not None:
os.environ["GITEA_URL"] = self.original_gitea_url
else:
del os.environ["GITEA_URL"]
if self.original_gitea_token is not None:
os.environ["GITEA_TOKEN"] = self.original_gitea_token
else:
del os.environ["GITEA_TOKEN"]
def _prepare_mock_client(self, payload, is_file=False):
response = MagicMock()
response.raise_for_status = MagicMock()
if is_file:
response.json.return_value = {"name": "file.txt", "path": "file.txt", "type": "file"}
else:
response.json.return_value = payload
client = MagicMock()
client.get = AsyncMock(return_value=response)
async_client = MagicMock()
async_client.__aenter__ = AsyncMock(return_value=client)
async_client.__aexit__ = AsyncMock(return_value=None)
return async_client, client
@patch("httpx.AsyncClient")
def test_list_repo_files_root(self, mock_async_client_constructor):
payload = [
{"name": "README.md", "path": "README.md", "type": "file", "sha": "sha123", "size": 1024},
{"name": "docs", "path": "docs", "type": "dir"}
]
mock_async_client, _ = self._prepare_mock_client(payload)
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
result = await handle_list_repo_files({"path": None}, "test/repo")
self.assertEqual(result["path"], "")
self.assertEqual(len(result["files"]), 2)
self.assertEqual(result["files"][0]["name"], "README.md")
self.assertIn("sha", result["files"][0])
self.assertIn("size", result["files"][0])
self.assertEqual(result["files"][1]["name"], "docs")
self.assertNotIn("sha", result["files"][1])
self.assertNotIn("size", result["files"][1])
import asyncio
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_list_repo_files_empty_path(self, mock_async_client_constructor):
payload = []
mock_async_client, _ = self._prepare_mock_client(payload)
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
result = await handle_list_repo_files({"path": ""}, "test/repo")
self.assertEqual(result["path"], "")
import asyncio
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_nested_path(self, mock_async_client_constructor):
payload = []
mock_async_client, mock_client = self._prepare_mock_client(payload)
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
await handle_list_repo_files({"path": "docs/security"}, "test/repo")
mock_client.get.assert_called_once()
called_url = mock_client.get.call_args[0][0]
self.assertIn("docs/security", called_url)
import asyncio
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_path_is_file(self, mock_async_client_constructor):
mock_async_client, _ = self._prepare_mock_client(None, is_file=True)
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
with self.assertRaisesRegex(ValueError, "Path is a file, not a directory. Use get_file_content instead."):
await handle_list_repo_files({"path": "file.txt"}, "test/repo")
import asyncio
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_hardened_root_none(self, mock_async_client_constructor):
"""Verify path=None retains root-listing behavior and params."""
mock_async_client, mock_client = self._prepare_mock_client([])
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
await handle_list_repo_files({"path": None}, "test/repo")
mock_client.get.assert_called_once()
called_url = mock_client.get.call_args[0][0]
self.assertTrue(called_url.endswith("/contents/"))
self.assertEqual(mock_client.get.call_args[1].get("params"), {"ref": "main"})
import asyncio
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_hardened_root_empty(self, mock_async_client_constructor):
"""Verify path="" retains root-listing behavior and params."""
mock_async_client, mock_client = self._prepare_mock_client([])
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
await handle_list_repo_files({"path": ""}, "test/repo")
mock_client.get.assert_called_once()
called_url = mock_client.get.call_args[0][0]
self.assertTrue(called_url.endswith("/contents/"))
self.assertEqual(mock_client.get.call_args[1].get("params"), {"ref": "main"})
import asyncio
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_hardened_dot_path_denied(self, mock_async_client_constructor):
"""Verify path='.' is denied before making an HTTP request."""
mock_async_client, mock_client = self._prepare_mock_client([])
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
with self.assertRaises(ValueError) as cm:
await handle_list_repo_files({"path": "."}, "test/repo")
self.assertEqual(str(cm.exception), "PATH_NOT_ALLOWED")
mock_client.get.assert_not_called()
import asyncio
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_hardened_safe_nested_path(self, mock_async_client_constructor):
"""Verify a safe nested path reaches the client and preserves params."""
mock_async_client, mock_client = self._prepare_mock_client([])
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
await handle_list_repo_files({"path": "docs/sub"}, "test/repo")
mock_client.get.assert_called_once()
called_url = mock_client.get.call_args[0][0]
self.assertTrue(called_url.endswith("/contents/docs/sub"))
self.assertEqual(mock_client.get.call_args[1].get("params"), {"ref": "main"})
import asyncio
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_hardened_traversal_path_denied(self, mock_async_client_constructor):
"""Verify a traversal path is denied before any HTTP request."""
mock_async_client, mock_client = self._prepare_mock_client([])
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
with self.assertRaises(ValueError) as cm:
await handle_list_repo_files({"path": "a/../b"}, "test/repo")
self.assertEqual(str(cm.exception), "PATH_NOT_ALLOWED")
mock_client.get.assert_not_called()
import asyncio
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_hardened_secret_request_path_denied(self, mock_async_client_constructor):
"""Verify a secret requested path is denied before any HTTP request."""
mock_async_client, mock_client = self._prepare_mock_client([])
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
with self.assertRaises(ValueError) as cm:
await handle_list_repo_files({"path": ".env"}, "test/repo")
self.assertEqual(str(cm.exception), "SECRET_PATH_DENIED")
mock_client.get.assert_not_called()
import asyncio
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_hardened_output_filtering(self, mock_async_client_constructor):
"""Verify secret paths are omitted from the returned list."""
gitea_response_payload = [
{"name": "README.md", "path": "README.md", "type": "file", "sha": "safe_sha", "size": 100},
{"name": ".env", "path": ".env", "type": "file"},
{"name": "config", "path": ".git/config", "type": "file"},
{"name": "app.txt", "path": "secrets/app.txt", "type": "file"},
{"name": "secret.pem", "path": "keys/secret.pem", "type": "file"},
]
mock_async_client, _ = self._prepare_mock_client(gitea_response_payload)
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
result = await handle_list_repo_files({"path": ""}, "test/repo")
self.assertEqual(result["files"], [{"name": "README.md", "path": "README.md", "type": "file", "sha": "safe_sha", "size": 100}])
returned_paths = {f["path"] for f in result["files"]}
self.assertNotIn(".env", returned_paths)
self.assertNotIn(".git/config", returned_paths)
self.assertNotIn("secrets/app.txt", returned_paths)
self.assertNotIn("keys/secret.pem", returned_paths)
import asyncio
asyncio.run(run_test())
def test_hardened_unexpected_error_reraised(self):
"""Verify other ValueErrors are not suppressed during filtering."""
import gitea_handler
real_validator = gitea_handler._validate_safe_path
def validator_side_effect(path, *args, **kwargs):
if path == "trigger.txt":
raise ValueError("UNEXPECTED_ERROR")
return real_validator(path, *args, **kwargs)
async def run_test():
with patch("gitea_handler._validate_safe_path") as mock_validator:
mock_validator.side_effect = validator_side_effect
with patch("httpx.AsyncClient") as mock_async_client_constructor:
mock_async_client, mock_client = self._prepare_mock_client(
[
{
"name": "safe.txt",
"path": "safe.txt",
"type": "file",
"sha": "safe_sha",
"size": 1,
},
{
"name": "trigger.txt",
"path": "trigger.txt",
"type": "file",
"sha": "trigger_sha",
"size": 1,
},
]
)
mock_async_client_constructor.return_value = mock_async_client
with self.assertRaises(ValueError) as cm:
await handle_list_repo_files({"path": ""}, "test/repo")
self.assertEqual(str(cm.exception), "UNEXPECTED_ERROR")
mock_client.get.assert_called_once()
import asyncio
asyncio.run(run_test())
def test_invalid_paths(self):
invalid_paths = [
"../secrets",
"docs/../secrets",
"docs\\secrets",
"/etc/passwd",
"https://example.com",
"docs//security",
"docs/./security",
]
async def run_test():
for path in invalid_paths:
with self.subTest(path=path):
# Use patch here to ensure no real network call is made for paths
# that might otherwise be valid URLs.
with patch("httpx.AsyncClient") as mock_client:
mock_async_client, mocked_http_client = self._prepare_mock_client([])
mock_client.return_value = mock_async_client
with self.assertRaises(ValueError) as cm:
await handle_list_repo_files({"path": path}, "test/repo")
self.assertEqual(str(cm.exception), "PATH_NOT_ALLOWED")
# Ensure no network call was attempted for invalid paths
mocked_http_client.get.assert_not_called()
import asyncio
asyncio.run(run_test())
def test_invalid_input_types(self):
async def run_test():
with self.assertRaises(ValueError) as cm:
await handle_list_repo_files({"path": 123}, "test/repo")
self.assertEqual(str(cm.exception), "PATH_NOT_ALLOWED")
import asyncio
asyncio.run(run_test())
def test_unsupported_fields(self):
unsupported = [
{"repo": "x"},
{"ref": "y"},
{"branch": "z"},
{"extra": "key"},
]
async def run_test():
for p in unsupported:
with self.subTest(p=p):
with self.assertRaisesRegex(ValueError, "Unsupported list_repo_files input field."):
await handle_list_repo_files(p, "test/repo")
import asyncio
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_uses_trusted_repo(self, mock_async_client_constructor):
payload = []
mock_async_client, mock_client = self._prepare_mock_client(payload)
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
await handle_list_repo_files({"path": "docs", "repo": "user/bad-repo"}, "test/repo")
# The handler should raise an error due to the unsupported 'repo' field.
import asyncio
with self.assertRaisesRegex(ValueError, "Unsupported list_repo_files input field."):
asyncio.run(run_test())
@patch("httpx.AsyncClient")
def test_http_error_propagation(self, mock_async_client_constructor):
response = MagicMock()
response.raise_for_status.side_effect = httpx.HTTPStatusError("Error", request=MagicMock(), response=MagicMock())
client = MagicMock()
client.get = AsyncMock(return_value=response)
async_client = MagicMock()
async_client.__aenter__ = AsyncMock(return_value=client)
async_client.__aexit__ = AsyncMock(return_value=None)
mock_async_client_constructor.return_value = async_client
async def run_test():
with self.assertRaises(httpx.HTTPStatusError):
await handle_list_repo_files({"path": ""}, "test/repo")
import asyncio
asyncio.run(run_test())
class TestHandleGetFileContent(unittest.TestCase):
def setUp(self):
self.original_gitea_url = os.environ.get("GITEA_URL")
self.original_gitea_token = os.environ.get("GITEA_TOKEN")
os.environ["GITEA_URL"] = "https://gitea.example.com"
os.environ["GITEA_TOKEN"] = "test-token"
self.server_repo_id = "chris/OSVauco"
def tearDown(self):
if self.original_gitea_url is None:
os.environ.pop("GITEA_URL", None)
else:
os.environ["GITEA_URL"] = self.original_gitea_url
if self.original_gitea_token is None:
os.environ.pop("GITEA_TOKEN", None)
else:
os.environ["GITEA_TOKEN"] = self.original_gitea_token
def _mock_contents_client(self, text):
body = json.dumps({
"content": base64.b64encode(text.encode("utf-8")).decode("ascii"),
"encoding": "base64",
}).encode("utf-8")
async def chunks():
yield body
response = MagicMock()
response.headers = {"content-length": str(len(body))}
response.aiter_bytes = chunks
stream = MagicMock()
stream.__aenter__.return_value = response
client = MagicMock()
client.stream.return_value = stream
async_client = MagicMock()
async_client.__aenter__.return_value = client
return async_client, client
@patch("gitea_handler.resolve_branch_to_commit_sha", new_callable=AsyncMock)
@patch("httpx.AsyncClient")
def test_direct_lowercase_sha_success(
self, mock_async_client_constructor, mock_resolver
):
import asyncio
sha = "a" * 40
mock_async_client, client = self._mock_contents_client("file content")
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
return await handle_get_file_content(
{
"repo": self.server_repo_id,
"path": "README.md",
"ref": sha,
},
self.server_repo_id,
)
result = asyncio.run(run_test())
mock_resolver.assert_not_called()
client.stream.assert_called_once()
request_url = client.stream.call_args.args[1]
self.assertIn(f"ref={sha}", request_url)
self.assertEqual(result["requested_ref"], sha)
self.assertEqual(result["resolved_commit_sha"], sha)
self.assertEqual(result["content"], "file content")
self.assertEqual(result["encoding"], "utf-8")
@patch("gitea_handler.resolve_branch_to_commit_sha", new_callable=AsyncMock)
@patch("httpx.AsyncClient")
def test_direct_uppercase_sha_normalizes_for_request(
self, mock_async_client_constructor, mock_resolver
):
import asyncio
requested_sha = "A" * 40
resolved_sha = requested_sha.lower()
mock_async_client, client = self._mock_contents_client("file content")
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
return await handle_get_file_content(
{
"repo": self.server_repo_id,
"path": "README.md",
"ref": requested_sha,
},
self.server_repo_id,
)
result = asyncio.run(run_test())
mock_resolver.assert_not_called()
request_url = client.stream.call_args.args[1]
self.assertIn(f"ref={resolved_sha}", request_url)
self.assertNotIn(f"ref={requested_sha}", request_url)
self.assertEqual(result["requested_ref"], requested_sha)
self.assertEqual(result["resolved_commit_sha"], resolved_sha)
@patch("gitea_handler.resolve_branch_to_commit_sha", new_callable=AsyncMock)
@patch("httpx.AsyncClient")
def test_branch_resolves_to_sha_before_contents_request(
self, mock_async_client_constructor, mock_resolver
):
import asyncio
resolved_sha = "b" * 40
mock_resolver.return_value = resolved_sha
mock_async_client, client = self._mock_contents_client("file content")
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
return await handle_get_file_content(
{
"repo": self.server_repo_id,
"path": "README.md",
"ref": "main",
},
self.server_repo_id,
)
result = asyncio.run(run_test())
mock_resolver.assert_awaited_once_with(
branch_name="main",
repo_id=self.server_repo_id,
gitea_url="https://gitea.example.com",
)
request_url = client.stream.call_args.args[1]
self.assertIn(f"ref={resolved_sha}", request_url)
self.assertNotIn("ref=main", request_url)
self.assertEqual(result["repo"], self.server_repo_id)
self.assertEqual(result["path"], "README.md")
self.assertEqual(result["requested_ref"], "main")
self.assertEqual(result["resolved_commit_sha"], resolved_sha)
self.assertEqual(result["content"], "file content")
self.assertEqual(result["encoding"], "utf-8")
@patch("gitea_handler.resolve_branch_to_commit_sha", new_callable=AsyncMock)
@patch("httpx.AsyncClient")
def test_unresolved_branch_raises_sanitized_error(
self, mock_async_client_constructor, mock_resolver
):
import asyncio
mock_resolver.side_effect = httpx.HTTPStatusError(
"Not Found",
request=MagicMock(),
response=MagicMock(status_code=404),
)
mock_async_client, client = self._mock_contents_client("file content")
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
with self.assertRaisesRegex(
ValueError, r"^Unknown or inaccessible branch reference\.$"
):
await handle_get_file_content(
{
"repo": self.server_repo_id,
"path": "README.md",
"ref": "missing-branch",
},
self.server_repo_id,
)
asyncio.run(run_test())
client.stream.assert_not_called()
@patch("gitea_handler.resolve_branch_to_commit_sha", new_callable=AsyncMock)
@patch("httpx.AsyncClient")
def test_resolver_non_404_http_error_is_sanitized(
self, mock_async_client_constructor, mock_resolver
):
import asyncio
mock_resolver.side_effect = httpx.HTTPStatusError(
"Server Error",
request=MagicMock(),
response=MagicMock(status_code=500),
)
mock_async_client, client = self._mock_contents_client("file content")
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
with self.assertRaisesRegex(
ValueError, r"^Repository file is unavailable\.$"
):
await handle_get_file_content(
{
"repo": self.server_repo_id,
"path": "README.md",
"ref": "main",
},
self.server_repo_id,
)
asyncio.run(run_test())
client.stream.assert_not_called()
@patch("gitea_handler.resolve_branch_to_commit_sha", new_callable=AsyncMock)
@patch("httpx.AsyncClient")
def test_resolver_unexpected_exception_is_sanitized(
self, mock_async_client_constructor, mock_resolver
):
import asyncio
mock_resolver.side_effect = RuntimeError(
"upstream internal details must not reach the caller"
)
mock_async_client, client = self._mock_contents_client("file content")
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
with self.assertRaisesRegex(
ValueError, r"^Repository file is unavailable\.$"
):
await handle_get_file_content(
{
"repo": self.server_repo_id,
"path": "README.md",
"ref": "main",
},
self.server_repo_id,
)
asyncio.run(run_test())
client.stream.assert_not_called()
@patch("gitea_handler.resolve_branch_to_commit_sha", new_callable=AsyncMock)
@patch("httpx.AsyncClient")
def test_resolver_returns_malformed_sha_is_rejected(
self, mock_async_client_constructor, mock_resolver
):
import asyncio
mock_resolver.return_value = "not-a-40-character-commit-sha"
mock_async_client, client = self._mock_contents_client("file content")
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
with self.assertRaisesRegex(ValueError, r"^Invalid commit SHA$"):
await handle_get_file_content(
{
"repo": self.server_repo_id,
"path": "README.md",
"ref": "main",
},
self.server_repo_id,
)
asyncio.run(run_test())
client.stream.assert_not_called()
@patch("gitea_handler.resolve_branch_to_commit_sha", new_callable=AsyncMock)
@patch("httpx.AsyncClient")
def test_invalid_branch_ref_is_rejected(
self, mock_async_client_constructor, mock_resolver
):
import asyncio
mock_async_client, client = self._mock_contents_client("file content")
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
with self.assertRaisesRegex(ValueError, r"^Invalid git reference\.$"):
await handle_get_file_content(
{
"repo": self.server_repo_id,
"path": "README.md",
"ref": "main/../secret",
},
self.server_repo_id,
)
asyncio.run(run_test())
mock_resolver.assert_not_called()
client.stream.assert_not_called()
@patch("gitea_handler.resolve_branch_to_commit_sha", new_callable=AsyncMock)
@patch("httpx.AsyncClient")
def test_contents_api_404_is_sanitized(
self, mock_async_client_constructor, mock_resolver
):
import asyncio
resolved_sha = "c" * 40
mock_resolver.return_value = resolved_sha
mock_async_client, client = self._mock_contents_client("file content")
response_mock = client.stream.return_value.__aenter__.return_value
response_mock.raise_for_status.side_effect = httpx.HTTPStatusError(
"Not Found", request=MagicMock(), response=MagicMock(status_code=404)
)
mock_async_client_constructor.return_value = mock_async_client
async def run_test():
with self.assertRaisesRegex(
ValueError, r"^Repository file is unavailable\.$"
):
await handle_get_file_content(
{
"repo": self.server_repo_id,
"path": "README.md",
"ref": "main",
},
self.server_repo_id,
)
asyncio.run(run_test())
mock_resolver.assert_awaited_once_with(
branch_name="main",
repo_id=self.server_repo_id,
gitea_url="https://gitea.example.com",
)
client.stream.assert_called_once()
if __name__ == '__main__':
unittest.main()