microsoft/qlib · error · TypeError

Unsupported order file type: {order_file}

Error message

Unsupported order file type: {order_file}

What it means

read_order_file (qlib/rl/contrib/utils.py:22) loads an order list for RL backtests. It accepts an already-built pd.DataFrame directly, or a path whose suffix is .pkl (read via pd.read_pickle + reset_index) or .csv (read via pd.read_csv). Any other suffix (.parquet, .txt, .json, no suffix) raises TypeError('Unsupported order file type: <path>').

Source

Thrown at qlib/rl/contrib/utils.py:22

from __future__ import annotations

from pathlib import Path

import pandas as pd


def read_order_file(order_file: Path | pd.DataFrame) -> pd.DataFrame:
    if isinstance(order_file, pd.DataFrame):
        return order_file

    order_file = Path(order_file)

    if order_file.suffix == ".pkl":
        order_df = pd.read_pickle(order_file).reset_index()
    elif order_file.suffix == ".csv":
        order_df = pd.read_csv(order_file)
    else:
        raise TypeError(f"Unsupported order file type: {order_file}")

    if "date" in order_df.columns:
        # legacy dataframe columns
        order_df = order_df.rename(columns={"date": "datetime", "order_type": "direction"})
    order_df["datetime"] = order_df["datetime"].astype(str)

    return order_df

View on GitHub (pinned to 79633dd950)

Solutions

1. Save orders as .csv or .pkl and pass that path
2. Load the data yourself and pass the pd.DataFrame directly, which bypasses the suffix check
3. Convert parquet/json orders to CSV: df.to_csv('orders.csv', index=False)

Example fix

# before
order_df = read_order_file('orders.parquet')  # TypeError

# after
order_df = read_order_file(pd.read_parquet('orders.parquet'))
# or convert once:
# pd.read_parquet('orders.parquet').to_csv('orders.csv', index=False)

from pathlib import Path
if not isinstance(order_file, pd.DataFrame):
    assert Path(order_file).suffix in ('.pkl', '.csv'), f'order file must be .pkl/.csv or DataFrame: {order_file}'

def is_supported_order_source(src) -> bool:
    return isinstance(src, pd.DataFrame) or Path(src).suffix in ('.pkl', '.csv')

try:
    df = read_order_file(order_file)
except TypeError as e:
    df = read_order_file(pd.read_parquet(order_file))  # manual load fallback

When it happens

Trigger: Calling read_order_file('orders.parquet'), read_order_file('orders.json'), or passing a Path with an unexpected/empty suffix; also triggered by files whose name merely contains '.csv' not at the end.

Common situations: Users exporting orders from their own systems as parquet/json; renaming exported order files; passing a directory-like Path with no suffix.

Related errors

• Only py/yml/yaml/json type are supported now!
• This type of `limit_threshold` is not supported
• stock data from resam_ts_data must be a number, pd.Series or
• provider_uri does not support {type(provider_uri)}
• Unsupported data type: {type(data)}.

AI-assisted analysis of microsoft/qlib@79633dd950 (2026-08-15). Data as JSON: /api/errors/fd88cffd45ecffd6. Report an issue: GitHub.