Spice MCP

from __future__ import annotations from pathlib import Path import polars as pl from spice_mcp.adapters.dune import cache def _make_df(): return pl.DataFrame({"a": [1, 2, 3]}) def test_build_cache_path_hash_determinism(tmp_path, monkeypatch): exec_obj = {"execution_id": "exe-1", "timestamp": 1700000000} execute_kwargs = { "query_id": 42, "api_key": None, "parameters": {"x": 1, "y": 2}, "performance": "medium", } result_kwargs_a = { "limit": 10, "offset": 0, "sample_count": None, "sort_by": None, "columns": None, "extras": None, "types": None, "all_types": None, "verbose": False, } # Same dict content but different key order result_kwargs_b = dict(reversed(list(result_kwargs_a.items()))) p1 = cache._build_cache_path( exec_obj, execute_kwargs, result_kwargs_a, str(tmp_path) ) p2 = cache._build_cache_path( exec_obj, execute_kwargs, result_kwargs_b, str(tmp_path) ) assert p1 == p2 def test_save_and_load_from_cache_roundtrip(tmp_path, monkeypatch): # Patch latest execution resolution to avoid network from spice_mcp.adapters.dune import extract exec_obj = {"execution_id": "exe-2", "timestamp": 1700001234} monkeypatch.setattr(extract, "get_latest_execution", lambda *_args, **_kw: exec_obj) execute_kwargs = { "query_id": 77, "api_key": None, "parameters": None, "performance": "medium", } result_kwargs = { "limit": None, "offset": None, "sample_count": None, "sort_by": None, "columns": None, "extras": None, "types": None, "all_types": None, "verbose": False, } output_kwargs = { "execute_kwargs": execute_kwargs, "result_kwargs": result_kwargs, "cache": True, "save_to_cache": True, "cache_dir": str(tmp_path), "include_execution": False, } df = _make_df() cache.save_to_cache(df, exec_obj, execute_kwargs, result_kwargs, str(tmp_path)) loaded, returned_exec = cache.load_from_cache( execute_kwargs, result_kwargs, output_kwargs ) assert isinstance(loaded, pl.DataFrame) assert returned_exec == exec_obj assert loaded.shape == df.shape def test_load_from_cache_includes_execution_when_requested(tmp_path, monkeypatch): from spice_mcp.adapters.dune import extract exec_obj = {"execution_id": "exe-3", "timestamp": 1700005678} monkeypatch.setattr(extract, "get_latest_execution", lambda *_args, **_kw: exec_obj) execute_kwargs = { "query_id": 88, "api_key": None, "parameters": None, "performance": "medium", } result_kwargs = { "limit": None, "offset": None, "sample_count": None, "sort_by": None, "columns": None, "extras": None, "types": None, "all_types": None, "verbose": False, } output_kwargs = { "execute_kwargs": execute_kwargs, "result_kwargs": result_kwargs, "cache": True, "save_to_cache": True, "cache_dir": str(tmp_path), "include_execution": True, } path = cache._build_cache_path(exec_obj, execute_kwargs, result_kwargs, str(tmp_path)) Path(path).parent.mkdir(parents=True, exist_ok=True) pl.DataFrame({"a": [1]}).write_parquet(path) loaded, returned_exec = cache.load_from_cache(execute_kwargs, result_kwargs, output_kwargs) assert isinstance(loaded, tuple) df_loaded, exec_meta = loaded assert exec_meta == exec_obj assert returned_exec == exec_obj assert df_loaded.shape == (1, 1)