"Vibe-Trading: Your Personal Trading Agent"
Python ai-agentalgorithmic-tradingbacktestingfintechllmmcpmulti-agentpythonquantitative-financetrading
"""Local cache entries must identify the configured file and schema."""
import json
from pathlib import Path
import pandas as pd
import pytest
import yaml
from backtest.loaders import base, local_loader
def _source(path: Path) -> dict:
return {"symbol": "AAA.US", "type": "csv", "path": str(path)}
def _write(path: Path, value: float) -> None:
pd.DataFrame(
{
"date": ["2025-01-02", "2025-01-03"],
"open": [value, value],
"high": [value + 1, value + 1],
"low": [value - 1, value - 1],
"close": [value, value],
"volume": [100, 100],
}
).to_csv(path, index=False)
def test_local_cache_does_not_reuse_another_configured_file(
tmp_path: Path, monkeypatch: pytest.MonkeyPatch
) -> None:
"""Changing only a symbol's file must not serve the previous file's bars."""
monkeypatch.setattr(base, "loader_cache_enabled", lambda: True)
monkeypatch.setattr(base, "loader_cache_root", lambda: tmp_path / "cache")
config = tmp_path / "config.yaml"
monkeypatch.setattr(local_loader, "_CONFIG_PATH", config)
first, second = tmp_path / "first.csv", tmp_path / "second.csv"
_write(first, 10)
_write(second, 20)
def fetch(path: Path) -> pd.DataFrame:
config.write_text(yaml.safe_dump({"sources": [_source(path)]}))
return local_loader.DataLoader().fetch(["AAA.US"], "2025-01-02", "2025-01-03")[
"AAA.US"
]
assert fetch(first)["close"].tolist() == [10, 10]
assert list((tmp_path / "cache").rglob("*.parquet")), "exercise the real cache"
assert fetch(second)["close"].tolist() == [20, 20]
def test_identical_local_config_still_hits_the_cache(
tmp_path: Path, monkeypatch: pytest.MonkeyPatch
) -> None:
"""Keep the existing opt-in settled-data cache contract for one source."""
monkeypatch.setattr(base, "loader_cache_enabled", lambda: True)
monkeypatch.setattr(base, "loader_cache_root", lambda: tmp_path / "cache")
path = tmp_path / "bars.csv"
_write(path, 10)
monkeypatch.setattr(
local_loader, "_load_config", lambda: {"sources": [_source(path)]}
)
before = local_loader.DataLoader().fetch(["AAA.US"], "2025-01-02", "2025-01-03")
def unexpected_read(*args: object) -> None:
raise AssertionError("identical configured source should hit its cache")
monkeypatch.setitem(local_loader._READERS, "csv", unexpected_read)
after = local_loader.DataLoader().fetch(["AAA.US"], "2025-01-02", "2025-01-03")
pd.testing.assert_frame_equal(before["AAA.US"], after["AAA.US"])
def test_local_cache_separates_queries_on_the_same_database(
tmp_path: Path, monkeypatch: pytest.MonkeyPatch
) -> None:
"""Two query declarations can select different bars for the same symbol."""
import duckdb
monkeypatch.setattr(base, "loader_cache_enabled", lambda: True)
monkeypatch.setattr(base, "loader_cache_root", lambda: tmp_path / "cache")
database = tmp_path / "bars.duckdb"
with duckdb.connect(str(database)) as connection:
connection.execute(
"CREATE TABLE bars AS SELECT '2025-01-02' AS date, "
"10 AS open, 21 AS high, 9 AS low, 10 AS close, "
"20 AS alternate_close, 100 AS volume"
)
def fetch(close: str) -> pd.DataFrame:
entry = {
"symbol": "AAA.US",
"type": "duckdb",
"db_path": str(database),
"query": f"SELECT date, open, high, low, {close} AS close, volume FROM bars",
}
monkeypatch.setattr(local_loader, "_load_config", lambda: {"sources": [entry]})
return local_loader.DataLoader().fetch(["AAA.US"], "2025-01-02", "2025-01-03")[
"AAA.US"
]
assert fetch("close")["close"].tolist() == [10]
assert fetch("alternate_close")["close"].tolist() == [20]
@pytest.mark.parametrize("cache_enabled", [False, True])
def test_ignored_yaml_metadata_does_not_break_local_fetch(
tmp_path: Path, monkeypatch: pytest.MonkeyPatch, cache_enabled: bool
) -> None:
"""YAML dates in source annotations aren't input fields or cache selectors."""
monkeypatch.setattr(base, "loader_cache_enabled", lambda: cache_enabled)
monkeypatch.setattr(base, "loader_cache_root", lambda: tmp_path / "cache")
path = tmp_path / "bars.csv"
_write(path, 10)
config = tmp_path / "config.yaml"
config.write_text(
yaml.safe_dump({"sources": [_source(path)]}) + " updated_at: 2025-01-01\n"
)
monkeypatch.setattr(local_loader, "_CONFIG_PATH", config)
frame = local_loader.DataLoader().fetch(["AAA.US"], "2025-01-02", "2025-01-03")
assert frame["AAA.US"]["close"].tolist() == [10, 10]
@pytest.mark.parametrize(
"corruption",
[
"null", "array", "index_names_length", "unhashable_index_column",
"empty_object", "empty_index", "missing_index", "missing_names",
"wrong_version", "missing_attrs", "invalid_attrs",
],
)
def test_corrupt_local_cache_metadata_refetches_and_repairs(
tmp_path: Path, monkeypatch: pytest.MonkeyPatch, corruption: str
) -> None:
"""A corrupt optional cache must not hide an available local CSV source."""
monkeypatch.setattr(base, "loader_cache_enabled", lambda: True)
monkeypatch.setattr(base, "loader_cache_root", lambda: tmp_path / "cache")
path = tmp_path / "bars.csv"
_write(path, 10)
monkeypatch.setattr(
local_loader, "_load_config", lambda: {"sources": [_source(path)]}
)
def fetch() -> pd.DataFrame:
return local_loader.DataLoader().fetch(["AAA.US"], "2025-01-02", "2025-01-03")[
"AAA.US"
]
assert fetch()["close"].tolist() == [10, 10]
metadata_paths = list((tmp_path / "cache").rglob("*.json"))
assert len(metadata_paths) == 1, "exercise one real persisted cache entry"
metadata_path = metadata_paths[0]
original = json.loads(metadata_path.read_text())
assert original["index_columns"], "exercise actual persisted index metadata"
malformed = dict(original)
if corruption == "null":
malformed = None
elif corruption == "array":
malformed = []
elif corruption == "index_names_length":
malformed["index_names"] = ["date", "unexpected_extra_level"]
elif corruption == "empty_object":
malformed = {}
elif corruption == "empty_index":
malformed["index_columns"] = []
elif corruption == "missing_index":
malformed.pop("index_columns")
elif corruption == "missing_names":
malformed.pop("index_names")
elif corruption == "wrong_version":
malformed["version"] = -1
elif corruption == "missing_attrs":
malformed.pop("frame_attrs")
elif corruption == "invalid_attrs":
malformed["frame_attrs"] = {"adjustment": ["qfq"]}
else:
malformed["index_columns"] = [{"not": "a column name"}]
metadata_path.write_text(json.dumps(malformed))
_write(path, 20)
repaired = fetch()
assert repaired["close"].tolist() == [20, 20]
assert isinstance(repaired.index, pd.DatetimeIndex)
assert repaired.index.tolist() == [
pd.Timestamp("2025-01-02"),
pd.Timestamp("2025-01-03"),
]
restored = json.loads(metadata_path.read_text())
assert isinstance(restored, dict)
assert restored["index_columns"] == original["index_columns"]
assert restored["index_names"] == original["index_names"]
def unexpected_read(*args: object) -> None:
raise AssertionError("repaired valid cache should avoid reading the CSV again")
monkeypatch.setitem(local_loader._READERS, "csv", unexpected_read)
pd.testing.assert_frame_equal(repaired, fetch())