import pytest
import json
from mcp_stata.models import CommandResponse
pytestmark = pytest.mark.requires_stata
def test_mock_command_execution(client, monkeypatch):
"""Verify that run_command_structured works with the mock setup."""
# Force mock output directly to avoid file I/O synchronization issues in mock mode
monkeypatch.setattr(client, "_read_persistent_log_chunk", lambda offset: "{txt}Mock Stata output\n")
monkeypatch.setattr(client, "_read_smcl_file", lambda path: "{txt}Mock Stata output\n")
result = client.run_command_structured("display 1+1")
# Check that our mock output was captured
assert result.success is True
assert "Mock Stata output" in result.stdout
def test_mock_error_handling(client, monkeypatch):
"""Verify that we can simulate a Stata error via mocks."""
import sfi
# Force a non-zero RC for the next call
monkeypatch.setattr(sfi.Scalar, "getValue", lambda x: 111 if x == "c(rc)" else 0)
# Force mock error content
error_smcl = "{err}variable not found{search r(111)}\n"
monkeypatch.setattr(client, "_read_persistent_log_chunk", lambda offset: error_smcl)
monkeypatch.setattr(client, "_read_smcl_file", lambda path: error_smcl)
result = client.run_command_structured("bad command")
assert result.success is False
assert result.rc == 111
assert "variable not found" in result.error.message
def test_mock_data_retrieval(client):
"""Verify that get_data still passes through logic even if mocked."""
# For now, get_data might fail or return mock depending on how deep we mock sfi.Data
# Let's see if we can at least call it.
try:
data = client.get_data(count=5)
assert isinstance(data, list)
except Exception as e:
# If it fails due to deeper sfi calls we haven't mocked yet, that's fine for now
# but we know the path is being exercised.
pytest.skip(f"Data mock not fully implemented: {e}")
