OSVauco/opax-mcp/test_gitea_helpers.py

426 lines
19 KiB
Python

import unittest
from unittest.mock import AsyncMock, MagicMock, patch
import os
import sys
from pathlib import Path
import base64
import httpx
import json
import binascii
# Correction 4: Schema test isolation
# Before importing server, install only a minimal test stub in sys.modules for
# the unrelated runtime dependency `emma_adapter`.
emma_adapter_stub = MagicMock()
emma_adapter_stub.CanonicalEmma = MagicMock()
sys.modules['emma_adapter'] = emma_adapter_stub
REPO_ROOT = Path(__file__).resolve().parents[1]
sys.path.insert(0, str(REPO_ROOT / "opax-mcp"))
# Now that the path is set, we can import the modules
import gitea_handler
import server
def make_stream_context(response_content, headers=None):
async def aiter_bytes():
if isinstance(response_content, list):
for item in response_content:
yield item
else:
yield response_content
response = MagicMock()
response.aiter_bytes = aiter_bytes
response.raise_for_status = MagicMock()
response.headers = headers if headers is not None else {}
context = MagicMock()
context.__aenter__ = AsyncMock(return_value=response)
context.__aexit__ = AsyncMock(return_value=False)
return context
def configure_stream(client, response_content, headers=None):
client.stream = MagicMock(
return_value=make_stream_context(response_content, headers=headers)
)
def _get_mock_file_response(content=b"test content", size=None, content_b64=None):
"""Helper to create a default valid mock response for get_file tests."""
if content_b64 is None:
encoded_content = base64.b64encode(content).decode('ascii')
else:
encoded_content = content_b64
if size is None:
size = len(content)
mock_response = MagicMock()
mock_response.status_code = 200
mock_response.json.return_value = {"size": size, "content": encoded_content}
mock_response.raise_for_status = MagicMock()
return mock_response
class TestGiteaHelpers(unittest.IsolatedAsyncioTestCase):
def test_validate_branch_name_valid(self):
self.assertEqual(gitea_handler._validate_branch_name("main"), "main")
self.assertEqual(gitea_handler._validate_branch_name("feat/opax-domain-decouple"), "feat/opax-domain-decouple")
def test_validate_branch_name_invalid(self):
with self.assertRaises(ValueError):
gitea_handler._validate_branch_name("../../etc/passwd")
with self.assertRaises(ValueError):
gitea_handler._validate_branch_name("main@{bad}")
with self.assertRaises(ValueError):
gitea_handler._validate_branch_name("feat//bad")
def test_validate_commit_sha_valid(self):
self.assertEqual(gitea_handler._validate_commit_sha("f" * 40), "f" * 40)
def test_validate_commit_sha_invalid(self):
with self.assertRaises(ValueError):
gitea_handler._validate_commit_sha("main")
with self.assertRaises(ValueError):
gitea_handler._validate_commit_sha("f" * 39)
def test_validate_gitea_url_https_required(self):
with self.assertRaises(ValueError):
gitea_handler._validate_gitea_url("http://example.com")
@patch("gitea_handler.httpx.AsyncClient")
async def test_resolve_branch_to_commit_sha(self, mock_client):
mock_response = unittest.mock.Mock()
mock_response.status_code = 200
mock_response.json.return_value = {"commit": {"id": "f" * 40}}
mock_client.return_value.__aenter__.return_value.get.return_value = mock_response
sha = await gitea_handler.resolve_branch_to_commit_sha("main", "chris/OSVauco", "https://gitea.example.com")
self.assertEqual(sha, "f" * 40)
@patch("gitea_handler.httpx.AsyncClient")
async def test_download_repo_archive_uses_sha(self, mock_client):
stream_context = make_stream_context(b"archive-data")
client = mock_client.return_value.__aenter__.return_value
client.stream = MagicMock(return_value=stream_context)
archive = await gitea_handler.download_repo_archive(
"f" * 40,
"chris/OSVauco",
"https://gitea.example.com",
)
self.assertEqual(archive, b"archive-data")
client.stream.assert_called_once_with(
"GET",
(
f"https://gitea.example.com/api/v1/repos/"
f"chris/OSVauco/archive/{'f' * 40}.tar.gz"
),
headers=unittest.mock.ANY,
)
@patch("gitea_handler.httpx.AsyncClient")
async def test_download_repo_archive_size_limit(self, mock_client):
response = MagicMock()
response.headers = {
"content-length": str(
gitea_handler.MAX_SOURCE_ARCHIVE_BYTES + 1
)
}
response.raise_for_status = MagicMock()
stream_context = make_stream_context([]) # Empty stream
client = mock_client.return_value.__aenter__.return_value
client.stream = MagicMock(return_value=stream_context)
# The response is used to check headers, not the stream content
stream_context.__aenter__.return_value = response
with self.assertRaisesRegex(
ValueError,
"Source archive exceeds allowed size",
):
await gitea_handler.download_repo_archive(
"f" * 40,
"chris/OSVauco",
"https://gitea.example.com",
)
@patch("gitea_handler.httpx.AsyncClient")
async def test_download_repo_archive_no_redirects(self, mock_client):
stream_context = make_stream_context(b"")
client = mock_client.return_value.__aenter__.return_value
client.stream = MagicMock(return_value=stream_context)
await gitea_handler.download_repo_archive(
"f" * 40,
"chris/OSVauco",
"https://gitea.example.com",
)
self.assertFalse(
mock_client.call_args.kwargs["follow_redirects"]
)
class TestGiteaCatalog(unittest.IsolatedAsyncioTestCase):
def setUp(self):
self.mock_env = patch.dict(os.environ, {
"GITEA_URL": "https://gitea.example.com",
"GITEA_TOKEN": "fake-token",
})
self.mock_env.start()
def tearDown(self):
self.mock_env.stop()
@patch("gitea_handler.httpx.AsyncClient")
async def test_catalog_uses_fixed_namespace(self, mock_client):
mock_response = MagicMock()
mock_response.status_code = 200
mock_response.json.return_value = []
mock_client.return_value.__aenter__.return_value.get.return_value = mock_response
await gitea_handler.list_allowed_namespace_repositories()
mock_client.return_value.__aenter__.return_value.get.assert_called_once()
call_url = mock_client.return_value.__aenter__.return_value.get.call_args[0][0]
self.assertIn("/api/v1/users/chris/repos", call_url)
@patch("gitea_handler.httpx.AsyncClient")
async def test_catalog_pagination_is_bounded(self, mock_client):
async def get_response(url, headers):
mock_response = MagicMock()
mock_response.status_code = 200
if "page=1" in url:
mock_response.json.return_value = [{"full_name": f"chris/repo{i}", "name": f"repo{i}"} for i in range(50)]
elif "page=2" in url:
mock_response.json.return_value = [{"full_name": f"chris/repo{i}", "name": f"repo{i}"} for i in range(50, 100)]
else: # Should not be called for page 3
mock_response.json.return_value = []
return mock_response
mock_client.return_value.__aenter__.return_value.get.side_effect = get_response
result = await gitea_handler.list_allowed_namespace_repositories()
self.assertEqual(mock_client.return_value.__aenter__.return_value.get.call_count, 2)
self.assertEqual(len(result["repositories"]), 100)
@patch("gitea_handler.httpx.AsyncClient")
async def test_catalog_normalization_and_security(self, mock_client):
gitea_response = [
{"full_name": "chris/repo1", "name": "repo1", "clone_url": "sensitive", "permissions": {}},
{"full_name": "other/repo2", "name": "repo2"},
{"full_name": "chris/repo3"},
{"full_name": "chris/repo1", "name": "repo1"},
]
mock_response = MagicMock()
mock_response.status_code = 200
mock_response.json.return_value = gitea_response
mock_client.return_value.__aenter__.return_value.get.return_value = mock_response
result = await gitea_handler.list_allowed_namespace_repositories()
self.assertEqual(len(result["repositories"]), 1)
repo = result["repositories"][0]
self.assertEqual(repo["full_name"], "chris/repo1")
self.assertNotIn("clone_url", repo)
self.assertNotIn("permissions", repo)
@patch("gitea_handler.httpx.AsyncClient")
async def test_catalog_upstream_failure_500(self, mock_client):
mock_client.return_value.__aenter__.return_value.get.side_effect = httpx.HTTPStatusError(
"Internal Server Error", request=MagicMock(), response=MagicMock(status_code=500)
)
with self.assertRaisesRegex(ValueError, "Repository catalog unavailable."):
await gitea_handler.list_allowed_namespace_repositories()
@patch("gitea_handler.httpx.AsyncClient")
async def test_catalog_upstream_failure_timeout(self, mock_client):
mock_client.return_value.__aenter__.return_value.get.side_effect = httpx.TimeoutException("Timeout")
with self.assertRaisesRegex(ValueError, "Repository catalog unavailable."):
await gitea_handler.list_allowed_namespace_repositories()
@patch("gitea_handler.httpx.AsyncClient")
async def test_catalog_malformed_json(self, mock_client):
mock_response = MagicMock()
mock_response.status_code = 200
mock_response.json.side_effect = json.JSONDecodeError("err", "doc", 0)
mock_client.return_value.__aenter__.return_value.get.return_value = mock_response
with self.assertRaisesRegex(ValueError, "Repository catalog unavailable."):
await gitea_handler.list_allowed_namespace_repositories()
class TestGetFileHardening(unittest.IsolatedAsyncioTestCase):
def setUp(self):
self.mock_env_patcher = patch.dict(os.environ, {
"GITEA_URL": "https://gitea.example.com",
"GITEA_TOKEN": "fake-token",
"GITEA_REPO": "chris/OSVauco",
})
self.mock_env = self.mock_env_patcher.start()
self.valid_sha = "f" * 40
self.params = {"path": "README.md", "ref": self.valid_sha}
self.repo = "chris/OSVauco"
def tearDown(self):
self.mock_env_patcher.stop()
@patch("gitea_handler.httpx.AsyncClient")
async def test_get_file_repo_compatibility(self, mock_client):
client = mock_client.return_value.__aenter__.return_value
configure_stream(client, json.dumps({'content': 'YQ==', 'size': 1}).encode("utf-8"))
# Allowed cases
await gitea_handler.handle_get_file_content(
{"path": "README.md", "ref": self.valid_sha}, "chris/OSVauco"
)
await gitea_handler.handle_get_file_content(
{"path": "README.md", "ref": self.valid_sha, "repo": None}, "chris/OSVauco"
)
await gitea_handler.handle_get_file_content(
{"path": "README.md", "ref": self.valid_sha, "repo": "chris/OSVauco"}, "chris/OSVauco"
)
self.assertEqual(client.stream.call_count, 3)
# Rejected cases
rejected_repos = {
"empty_string": "",
"wrong_type_false": False,
"wrong_type_zero": 0,
"mismatch": "other/repo",
}
for name, repo_val in rejected_repos.items():
with self.subTest(rejected_case=name):
client.stream.reset_mock()
with self.assertRaisesRegex(ValueError, "Repository file request is not allowed."):
await gitea_handler.handle_get_file_content(
{"path": "README.md", "ref": self.valid_sha, "repo": repo_val}, "chris/OSVauco"
)
client.stream.assert_not_called()
async def test_get_file_ref_validation(self):
with self.assertRaisesRegex(ValueError, "Invalid commit SHA"):
await gitea_handler.handle_get_file_content({"path": "README.md", "ref": "main"}, "chris/OSVauco")
with self.assertRaisesRegex(ValueError, "Invalid commit SHA"):
await gitea_handler.handle_get_file_content({"path": "README.md", "ref": None}, "chris/OSVauco")
async def test_gitea_repo_validation(self):
with self.assertRaisesRegex(ValueError, "Invalid configured Gitea repository ID"):
await gitea_handler.handle_get_file_content({"path": "README.md", "ref": self.valid_sha}, "invalid-repo-id")
with self.assertRaisesRegex(ValueError, "Invalid configured Gitea repository ID"):
await gitea_handler.handle_get_file_content({"path": "README.md", "ref": self.valid_sha}, "")
@patch("gitea_handler.httpx.AsyncClient")
async def test_get_file_path_validation(self, mock_client):
client = mock_client.return_value.__aenter__.return_value
configure_stream(client, json.dumps({'content': 'YQ==', 'size': 1}).encode("utf-8"))
for path in ["../secrets.txt", "/etc/passwd", "src/main.py", "docs/.env", "file.json"]:
with self.subTest(path=path):
with self.assertRaisesRegex(ValueError, "Repository file request is not allowed."):
await gitea_handler.handle_get_file_content({"path": path, "ref": self.valid_sha}, "chris/OSVauco")
for path in ["README.md", "docs/ARCHITECTURE.md", ".gemini/GEMINI.md"]:
with self.subTest(path=path):
await gitea_handler.handle_get_file_content({"path": path, "ref": self.valid_sha}, "chris/OSVauco")
@patch("gitea_handler.httpx.AsyncClient")
async def test_get_file_size_limit(self, mock_client):
client = mock_client.return_value.__aenter__.return_value
# Declared size too large
response_body = json.dumps({"size": gitea_handler.MAX_READ_FILE_BYTES + 1, "content": ""}).encode("utf-8")
configure_stream(client, response_body)
with self.assertRaisesRegex(ValueError, "Repository file exceeds the allowed size."):
await gitea_handler.handle_get_file_content({"path": "README.md", "ref": self.valid_sha}, "chris/OSVauco")
# Transport size too large
large_body = b'{' * (gitea_handler.MAX_GET_FILE_RESPONSE_BYTES + 1)
configure_stream(client, large_body)
with self.assertRaisesRegex(ValueError, "Repository file exceeds the allowed size."):
await gitea_handler.handle_get_file_content({"path": "README.md", "ref": self.valid_sha}, "chris/OSVauco")
@patch("gitea_handler.httpx.AsyncClient")
async def test_get_file_content_validation(self, mock_client):
client = mock_client.return_value.__aenter__.return_value
# Malformed Base64
configure_stream(client, json.dumps({'content': 'invalid-b64!', 'size': 12}).encode("utf-8"))
with self.assertRaisesRegex(ValueError, "Repository file content is not readable text."):
await gitea_handler.handle_get_file_content({"path": "README.md", "ref": self.valid_sha}, "chris/OSVauco")
# Invalid UTF-8
invalid_utf8_b64 = base64.b64encode(b"\xff\xff\xff").decode('ascii')
configure_stream(client, json.dumps({'content': invalid_utf8_b64, 'size': 3}).encode("utf-8"))
with self.assertRaisesRegex(ValueError, "Repository file content is not readable text."):
await gitea_handler.handle_get_file_content({"path": "README.md", "ref": self.valid_sha}, "chris/OSVauco")
# NUL byte
nul_byte_b64 = base64.b64encode(b"hello\0world").decode('ascii')
configure_stream(client, json.dumps({'content': nul_byte_b64, 'size': 11}).encode("utf-8"))
with self.assertRaisesRegex(ValueError, "Repository file content is not readable text."):
await gitea_handler.handle_get_file_content({"path": "README.md", "ref": self.valid_sha}, "chris/OSVauco")
@patch("gitea_handler.httpx.AsyncClient")
async def test_content_length_validation(self, mock_client):
client = mock_client.return_value.__aenter__.return_value
test_cases = {
"absent": None,
"empty": "",
"non-numeric": "not-a-number",
"negative": "-100",
}
for name, length_val in test_cases.items():
with self.subTest(name=name):
headers = {"content-length": length_val} if length_val is not None else {}
configure_stream(client,
json.dumps({'content': 'YQ==', 'size': 1}).encode("utf-8"), headers=headers
)
# No ValueError should be raised, should proceed and succeed
result = await gitea_handler.handle_get_file_content(self.params, self.repo)
self.assertEqual(result["content"], "a")
# Test valid but too large
with self.subTest(name="valid_too_large"):
headers = {"content-length": str(gitea_handler.MAX_GET_FILE_RESPONSE_BYTES + 1)}
configure_stream(client, b"", headers=headers)
with self.assertRaisesRegex(ValueError, "Repository file exceeds the allowed size."):
await gitea_handler.handle_get_file_content(self.params, self.repo)
@patch("gitea_handler.httpx.AsyncClient")
async def test_json_size_validation(self, mock_client):
client = mock_client.return_value.__aenter__.return_value
test_cases = {
"True": True,
"False": False,
"negative": -1,
"string": "123",
}
for name, size_val in test_cases.items():
with self.subTest(name=name):
body = json.dumps({"content": "YQ==", "size": size_val}).encode("utf-8")
configure_stream(client, body)
# No error should be raised, should proceed and succeed
result = await gitea_handler.handle_get_file_content(self.params, self.repo)
self.assertEqual(result["content"], "a")
# Test valid int over the limit
with self.subTest(name="valid_too_large"):
body = json.dumps({
"content": "", "size": gitea_handler.MAX_READ_FILE_BYTES + 1
}).encode("utf-8")
configure_stream(client, body)
with self.assertRaisesRegex(ValueError, "Repository file exceeds the allowed size."):
await gitea_handler.handle_get_file_content(self.params, self.repo)
class TestMCPSchema(unittest.TestCase):
def test_get_file_schema(self):
_handler, _desc, schema = server.TOOLS["get_file"]
self.assertEqual(schema["properties"].keys(), {"path", "ref", "repo"})
self.assertEqual(schema["required"], ["path", "ref"])
def test_list_gitea_repositories_schema(self):
_handler, _desc, schema = server.TOOLS["list_gitea_repositories"]
self.assertEqual(schema, {})
if __name__ == "__main__":
unittest.main()