fix: possible bug causing missing calendar_list data

Co-authored-by: abc <a@b.com>

.gitignore

@@ -22,6 +22,7 @@ dist/
 qlib/VERSION.txt
 qlib/data/_libs/expanding.cpp
 qlib/data/_libs/rolling.cpp
+qlib/_version.py
 examples/estimator/estimator_example/
 examples/rl/data/
 examples/rl/checkpoints/

Makefile

@@ -74,34 +74,37 @@ prerequisite:
 
 # Install the package in editable mode.
 dependencies:
-	python -m pip install -e .
+	python -m pip install --no-cache-dir -e .
 
 lightgbm:
-	python -m pip install lightgbm --prefer-binary
+	python -m pip install --no-cache-dir lightgbm --prefer-binary
 
 rl:
-	python -m pip install -e .[rl]
+	python -m pip install --no-cache-dir -e .[rl]
 
 develop:
-	python -m pip install -e .[dev]
+	python -m pip install --no-cache-dir -e .[dev]
 
 lint:
-	python -m pip install -e .[lint]
+	python -m pip install --no-cache-dir -e .[lint]
 
 docs:
-	python -m pip install -e .[docs]
+	python -m pip install --no-cache-dir -e .[docs]
 
 package:
-	python -m pip install -e .[package]
+	python -m pip install --no-cache-dir -e .[package]
 
 test:
-	python -m pip install -e .[test]
+	python -m pip install --no-cache-dir -e .[test]
 
 analysis:
-	python -m pip install -e .[analysis]
+	python -m pip install --no-cache-dir -e .[analysis]
+
+client:
+	python -m pip install --no-cache-dir -e .[client]
 
 all:
-	python -m pip install -e .[pywinpty,dev,lint,docs,package,test,analysis,rl]
+	python -m pip install --no-cache-dir -e .[pywinpty,dev,lint,docs,package,test,analysis,rl]
 
 install: prerequisite dependencies
 
@@ -113,7 +116,7 @@ dev: prerequisite all
 
 # Check lint with black.
 black:
-	black . -l 120 --check --diff
+	black . -l 120 --check --diff --exclude qlib/_version.py
 
 # Check code folder with pylint.
 # TODO: These problems we will solve in the future. Important among them are: W0221, W0223, W0237, E1102

docs/advanced/alpha.rst

@@ -42,7 +42,7 @@ Example
 
     .. math::
 
-        DEA = \frac{EMA(DIF, 9)}{CLOSE}
+        DEA = EMA(DIF, 9)
 
 Users can use ``Data Handler`` to build formulaic alphas `MACD` in qlib:
 
@@ -51,7 +51,7 @@ Users can use ``Data Handler`` to build formulaic alphas `MACD` in qlib:
 .. code-block:: python
 
     >> from qlib.data.dataset.loader import QlibDataLoader
-    >> MACD_EXP = '(EMA($close, 12) - EMA($close, 26))/$close - EMA((EMA($close, 12) - EMA($close, 26))/$close, 9)/$close'
+    >> MACD_EXP = '2 * ((EMA($close, 12) - EMA($close, 26))/$close - EMA((EMA($close, 12) - EMA($close, 26))/$close, 9))'
     >> fields = [MACD_EXP] # MACD
     >> names = ['MACD']
     >> labels = ['Ref($close, -2)/Ref($close, -1) - 1'] # label
@@ -66,17 +66,17 @@ Users can use ``Data Handler`` to build formulaic alphas `MACD` in qlib:
                             feature     label
                                MACD     LABEL
     datetime   instrument
-    2010-01-04 SH600000   -0.011547 -0.019672
+    2010-01-04 SH600000    0.008781 -0.019672
-               SH600004    0.002745 -0.014721
+               SH600004    0.006699 -0.014721
-               SH600006    0.010133  0.002911
+               SH600006    0.005714  0.002911
-               SH600008   -0.001113  0.009818
+               SH600008    0.000798  0.009818
-               SH600009    0.025878 -0.017758
+               SH600009    0.017015 -0.017758
     ...                         ...       ...
-    2017-12-29 SZ300124    0.007306 -0.005074
+    2017-12-29 SZ300124    0.015071 -0.005074
-               SZ300136   -0.013492  0.056352
+               SZ300136   -0.015466  0.056352
-               SZ300144   -0.000966  0.011853
+               SZ300144    0.013082  0.011853
-               SZ300251    0.004383  0.021739
+               SZ300251   -0.001026  0.021739
-               SZ300315   -0.030557  0.012455
+               SZ300315   -0.007559  0.012455
 
 Reference
 =========

docs/start/getdata.rst

@@ -129,7 +129,7 @@ For example, it looks quite long and complicated:
 
 
 But using string is not the only way to implement the expression. You can also implement expression by code.
-Here is an exmaple which does the same thing as above examples.
+Here is an example which does the same thing as above examples.
 
 
 .. code-block:: python

docs/start/integration.rst

@@ -71,7 +71,7 @@ The Custom models need to inherit `qlib.model.base.Model <../reference/api.html#
                 )
 
 - Override the `predict` method
-    - The parameters must include the parameter `dataset`, which will be userd to get the test dataset.
+    - The parameters must include the parameter `dataset`, which will be used to get the test dataset.
     - Return the `prediction score`.
     - Please refer to `Model API <../reference/api.html#module-qlib.model.base>`_ for the parameter types of the fit method.
     - Code Example: In the following example, users need to use `LightGBM` to predict the label(such as `preds`) of test data `x_test` and return it.

examples/rl_order_execution/scripts/gen_training_orders.py

@@ -17,11 +17,11 @@ def generate_order(stock: str, start_idx: int, end_idx: int) -> bool:
     if len(df) == 0 or df.isnull().values.any() or min(df["$volume0"]) < 1e-5:
         return False
 
-    df["date"] = df["datetime"].dt.date.astype("datetime64[ns]")
+    df["date"] = df["datetime"].dt.date.astype("datetime64")
     df = df.set_index(["instrument", "datetime", "date"])
-    df = df.groupby("date", group_keys=True).take(range(start_idx, end_idx)).droplevel(level=0)
+    df = df.groupby("date", group_keys=False).take(range(start_idx, end_idx)).droplevel(level=0)
 
-    order_all = pd.DataFrame(df.groupby(level=(2, 0), group_keys=True).mean().dropna())
+    order_all = pd.DataFrame(df.groupby(level=(2, 0), group_keys=False).mean().dropna())
     order_all["amount"] = np.random.lognormal(-3.28, 1.14) * order_all["$volume0"]
     order_all = order_all[order_all["amount"] > 0.0]
     order_all["order_type"] = 0

pyproject.toml

@@ -69,8 +69,10 @@ rl = [
   "torch",
   "numpy<2.0.0",
 ]
+# We exclude black version 26.1.0 due to known issues with nbqa when formatting Jupyter notebooks,
+# which can cause false-positive --check results and inconsistent notebook formatting.
 lint = [
-  "black",
+  "black!=26.1.0",
   "pylint",
   "mypy<1.5.0",
   "flake8",
@@ -101,6 +103,10 @@ analysis = [
   "plotly",
   "statsmodels",
 ]
+client = [
+  "python-socketio<6",
+  "tables",
+]
 
 # In the process of releasing a new version, when checking the manylinux package with twine, an error is reported:
 # InvalidDistribution: Invalid distribution metadata: unrecognized or malformed field 'license-file'
@@ -117,3 +123,4 @@ qrun = "qlib.cli.run:run"
 [tool.setuptools_scm]
 local_scheme = "no-local-version"
 version_scheme = "guess-next-dev"
+write_to = "qlib/_version.py"

qlib/__init__.py

@@ -4,7 +4,10 @@ from pathlib import Path
 
 from setuptools_scm import get_version
 
-__version__ = get_version(root="..", relative_to=__file__)
+try:
+    from ._version import version as __version__
+except ImportError:
+    __version__ = get_version(root="..", relative_to=__file__)
 __version__bak = __version__  # This version is backup for QlibConfig.reset_qlib_version
 import logging
 import os

qlib/cli/run.py

@@ -87,7 +87,7 @@ def workflow(config_path, experiment_name="workflow", uri_folder="mlruns"):
     """
     This is a Qlib CLI entrance.
     User can run the whole Quant research workflow defined by a configure file
-    - the code is located here ``qlib/cli/run.py`
+    - the code is located here ``qlib/cli/run.py``
 
     User can specify a base_config file in your workflow.yml file by adding "BASE_CONFIG_PATH".
     Qlib will load the configuration in BASE_CONFIG_PATH first, and the user only needs to update the custom fields

qlib/contrib/model/gbdt.py

@@ -51,7 +51,7 @@ class LGBModel(ModelFT, LightGBMFInt):
                     w = reweighter.reweight(df)
                 else:
                     raise ValueError("Unsupported reweighter type.")
-                ds_l.append((lgb.Dataset(x.values, label=y, weight=w), key))
+                ds_l.append((lgb.Dataset(x.values, label=y, weight=w, free_raw_data=False), key))
         return ds_l
 
     def fit(
@@ -109,8 +109,10 @@ class LGBModel(ModelFT, LightGBMFInt):
             verbose level
         """
         # Based on existing model and finetune by train more rounds
-        dtrain, _ = self._prepare_data(dataset, reweighter)  # pylint: disable=W0632
+        ds_l = self._prepare_data(dataset, reweighter)
-        if dtrain.empty:
+        dtrain, _ = ds_l[0]
+
+        if dtrain.construct().num_data() == 0:
             raise ValueError("Empty data from dataset, please check your dataset config.")
         verbose_eval_callback = lgb.log_evaluation(period=verbose_eval)
         self.model = lgb.train(

qlib/contrib/model/pytorch_nn.py

@@ -10,6 +10,7 @@ import os
 import gc
 import numpy as np
 import pandas as pd
+from packaging import version
 from typing import Callable, Optional, Text, Union
 from sklearn.metrics import roc_auc_score, mean_squared_error
 
@@ -148,7 +149,7 @@ class DNNModelPytorch(Model):
         if scheduler == "default":
             # In torch version 2.7.0, the verbose parameter has been removed. Reference Link:
             # https://github.com/pytorch/pytorch/pull/147301/files#diff-036a7470d5307f13c9a6a51c3a65dd014f00ca02f476c545488cd856bea9bcf2L1313
-            if str(torch.__version__).split("+", maxsplit=1)[0] <= "2.6.0":
+            if version.parse(str(torch.__version__).split("+", maxsplit=1)[0]) <= version.parse("2.6.0"):
                 # Reduce learning rate when loss has stopped decrease
                 self.scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(  # pylint: disable=E1123
                     self.train_optimizer,

qlib/contrib/rolling/ddgda.py

@@ -14,6 +14,7 @@ from qlib.model.meta.task import MetaTask
 from qlib.model.trainer import TrainerR
 from qlib.typehint import Literal
 from qlib.utils import init_instance_by_config
+from qlib.utils.pickle_utils import restricted_pickle_load
 from qlib.workflow import R
 from qlib.workflow.task.utils import replace_task_handler_with_cache
 
@@ -298,7 +299,7 @@ class DDGDA(Rolling):
         # but their task test segment are not aligned! It worked in my previous experiment.
         # So the misalignment will not affect the effectiveness of the method.
         with self._internal_data_path.open("rb") as f:
-            internal_data = pickle.load(f)
+            internal_data = restricted_pickle_load(f)
 
         md = MetaDatasetDS(exp_name=internal_data, **kwargs)
 
@@ -360,7 +361,7 @@ class DDGDA(Rolling):
         )
 
         with self._internal_data_path.open("rb") as f:
-            internal_data = pickle.load(f)
+            internal_data = restricted_pickle_load(f)
         mds = MetaDatasetDS(exp_name=internal_data, **kwargs)
 
         # 3) meta model make inference and get new qlib task

qlib/contrib/tuner/tuner.py

@@ -8,7 +8,6 @@ import os
 import yaml
 import json
 import copy
-import pickle
 import logging
 import importlib
 import subprocess
@@ -18,6 +17,7 @@ import numpy as np
 from abc import abstractmethod
 
 from ...log import get_module_logger, TimeInspector
+from ...utils.pickle_utils import restricted_pickle_load
 from hyperopt import fmin, tpe
 from hyperopt import STATUS_OK, STATUS_FAIL
 
@@ -136,7 +136,7 @@ class QLibTuner(Tuner):
         exp_result_dir = os.path.join(self.ex_dir, QLibTuner.EXP_RESULT_DIR.format(estimator_ex_id))
         exp_result_path = os.path.join(exp_result_dir, QLibTuner.EXP_RESULT_NAME)
         with open(exp_result_path, "rb") as fp:
-            analysis_df = pickle.load(fp)
+            analysis_df = restricted_pickle_load(fp)
 
         # 4. Get the backtest factor which user want to optimize, if user want to maximize the factor, then reverse the result
         res = analysis_df.loc[self.optim_config.report_type].loc[self.optim_config.report_factor]

qlib/data/cache.py

@@ -30,6 +30,7 @@ from ..utils import (
     normalize_cache_fields,
     normalize_cache_instruments,
 )
+from ..utils.pickle_utils import restricted_pickle_load
 
 from ..log import get_module_logger
 from .base import Feature
@@ -225,7 +226,7 @@ class CacheUtils:
             cache_path = Path(cache_path)
             meta_path = cache_path.with_suffix(".meta")
             with meta_path.open("rb") as f:
-                d = pickle.load(f)
+                d = restricted_pickle_load(f)
             with meta_path.open("wb") as f:
                 try:
                     d["meta"]["last_visit"] = str(time.time())
@@ -592,7 +593,7 @@ class DiskExpressionCache(ExpressionCache):
 
         with CacheUtils.writer_lock(self.r, f"{str(C.dpm.get_data_uri())}:expression-{cache_uri}"):
             with meta_path.open("rb") as f:
-                d = pickle.load(f)
+                d = restricted_pickle_load(f)
             instrument = d["info"]["instrument"]
             field = d["info"]["field"]
             freq = d["info"]["freq"]
@@ -959,7 +960,7 @@ class DiskDatasetCache(DatasetCache):
         im = DiskDatasetCache.IndexManager(cp_cache_uri)
         with CacheUtils.writer_lock(self.r, f"{str(C.dpm.get_data_uri())}:dataset-{cache_uri}"):
             with meta_path.open("rb") as f:
-                d = pickle.load(f)
+                d = restricted_pickle_load(f)
             instruments = d["info"]["instruments"]
             fields = d["info"]["fields"]
             freq = d["info"]["freq"]

qlib/data/client.py

@@ -2,15 +2,15 @@
 # Licensed under the MIT License.
 
 
-from __future__ import division
+from __future__ import division, print_function
-from __future__ import print_function
+
+import json
 
 import socketio
 
 import qlib
-from ..config import C
+
 from ..log import get_module_logger
-import pickle
 
 
 class Client:
@@ -96,7 +96,7 @@ class Client:
         self.logger.debug("connected")
         # The pickle is for passing some parameters with special type(such as
         # pd.Timestamp)
-        request_content = {"head": head_info, "body": pickle.dumps(request_content, protocol=C.dump_protocol_version)}
+        request_content = {"head": head_info, "body": json.dumps(request_content, default=str)}
         self.sio.on(request_type + "_response", request_callback)
         self.logger.debug("try sending")
         self.sio.emit(request_type + "_request", request_content)

qlib/data/dataset/loader.py

@@ -2,7 +2,6 @@
 # Licensed under the MIT License.
 
 import abc
-import pickle
 from pathlib import Path
 import warnings
 import pandas as pd
@@ -11,6 +10,7 @@ from typing import Tuple, Union, List, Dict
 
 from qlib.data import D
 from qlib.utils import load_dataset, init_instance_by_config, time_to_slc_point
+from qlib.utils.pickle_utils import restricted_pickle_load
 from qlib.log import get_module_logger
 from qlib.utils.serial import Serializable
 
@@ -283,7 +283,7 @@ class StaticDataLoader(DataLoader, Serializable):
                 self._data = pd.read_parquet(self._config, engine="pyarrow")
             else:
                 with Path(self._config).open("rb") as f:
-                    self._data = pickle.load(f)
+                    self._data = restricted_pickle_load(f)
         elif isinstance(self._config, pd.DataFrame):
             self._data = self._config
 

qlib/data/filter.py

@@ -168,7 +168,7 @@ class SeriesDFilter(BaseDFilter):
         for _ts, _bool in timestamp_series.items():
             # there is likely to be NAN when the filter series don't have the
             # bool value, so we just change the NAN into False
-            if _bool == np.nan:
+            if np.isnan(_bool):
                 _bool = False
             if _lbool is None:
                 _cur_start = _ts

qlib/rl/data/native.py

@@ -2,17 +2,18 @@
 # Licensed under the MIT License.
 from __future__ import annotations
 
+import os
 from pathlib import Path
-from typing import cast, List
+from typing import List, cast
 
 import cachetools
 import pandas as pd
-import pickle
-import os
 
 from qlib.backtest import Exchange, Order
 from qlib.backtest.decision import TradeRange, TradeRangeByTime
 from qlib.constant import EPS_T
+from qlib.utils.pickle_utils import restricted_pickle_load
+
 from .base import BaseIntradayBacktestData, BaseIntradayProcessedData, ProcessedDataProvider
 
 
@@ -162,7 +163,7 @@ class HandlerIntradayProcessedData(BaseIntradayProcessedData):
         path = os.path.join(data_dir, "backtest" if backtest else "feature", f"{stock_id}.pkl")
         start_time, end_time = date.replace(hour=0, minute=0, second=0), date.replace(hour=23, minute=59, second=59)
         with open(path, "rb") as fstream:
-            dataset = pickle.load(fstream)
+            dataset = restricted_pickle_load(fstream)
         data = dataset.handler.fetch(pd.IndexSlice[stock_id, start_time:end_time], level=None)
 
         if index_only:

qlib/utils/mod.py

@@ -11,7 +11,6 @@ import contextlib
 import importlib
 import os
 from pathlib import Path
-import pickle
 import pkgutil
 import re
 import sys
@@ -20,6 +19,7 @@ from typing import Any, Dict, List, Tuple, Union
 from urllib.parse import urlparse
 
 from qlib.typehint import InstConf
+from qlib.utils.pickle_utils import restricted_pickle_load
 
 
 def get_module_by_module_path(module_path: Union[str, ModuleType]):
@@ -168,10 +168,10 @@ def init_instance_by_config(
 
                 pr_path = os.path.join(pr.netloc, path) if bool(pr.path) else pr.netloc
                 with open(os.path.normpath(pr_path), "rb") as f:
-                    return pickle.load(f)
+                    return restricted_pickle_load(f)
         else:
             with config.open("rb") as f:
-                return pickle.load(f)
+                return restricted_pickle_load(f)
 
     klass, cls_kwargs = get_callable_kwargs(config, default_module=default_module)
 

qlib/utils/objm.py

@@ -6,6 +6,7 @@ import tempfile
 from pathlib import Path
 
 from qlib.config import C
+from qlib.utils.pickle_utils import restricted_pickle_load
 
 
 class ObjManager:
@@ -116,7 +117,7 @@ class FileManager(ObjManager):
 
     def load_obj(self, name):
         with (self.path / name).open("rb") as f:
-            return pickle.load(f)
+            return restricted_pickle_load(f)
 
     def exists(self, name):
         return (self.path / name).exists()

qlib/utils/pickle_utils.py

@@ -0,0 +1,171 @@
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.
+"""
+Secure pickle utilities to prevent arbitrary code execution through deserialization.
+
+This module provides a secure alternative to pickle.load() and pickle.loads()
+that restricts deserialization to a whitelist of safe classes.
+"""
+
+import io
+import pickle
+from typing import Any, BinaryIO, Set, Tuple
+
+# Whitelist of safe classes that are allowed to be unpickled
+# These are common data types used in qlib that should be safe to deserialize
+SAFE_PICKLE_CLASSES: Set[Tuple[str, str]] = {
+    # python builtins
+    ("builtins", "slice"),
+    ("builtins", "range"),
+    ("builtins", "dict"),
+    ("builtins", "list"),
+    ("builtins", "tuple"),
+    ("builtins", "set"),
+    ("builtins", "frozenset"),
+    ("builtins", "bytearray"),
+    ("builtins", "bytes"),
+    ("builtins", "str"),
+    ("builtins", "int"),
+    ("builtins", "float"),
+    ("builtins", "bool"),
+    ("builtins", "complex"),
+    ("builtins", "type"),
+    ("builtins", "property"),
+    # common utility classes
+    ("datetime", "datetime"),
+    ("datetime", "date"),
+    ("datetime", "time"),
+    ("datetime", "timedelta"),
+    ("datetime", "timezone"),
+    ("decimal", "Decimal"),
+    ("collections", "OrderedDict"),
+    ("collections", "defaultdict"),
+    ("collections", "Counter"),
+    ("collections", "namedtuple"),
+    ("enum", "Enum"),
+    ("pathlib", "Path"),
+    ("pathlib", "PosixPath"),
+    ("pathlib", "WindowsPath"),
+    ("qlib.data.dataset.handler", "DataHandler"),
+    ("qlib.data.dataset.handler", "DataHandlerLP"),
+    ("qlib.data.dataset.loader", "StaticDataLoader"),
+}
+
+
+TRUSTED_MODULE_PREFIXES = (
+    "pandas",
+    "numpy",
+)
+
+
+class RestrictedUnpickler(pickle.Unpickler):
+    """Custom unpickler that only allows safe classes to be deserialized.
+
+    This prevents arbitrary code execution through malicious pickle files by
+    restricting deserialization to a whitelist of safe classes.
+
+    Example:
+        >>> with open("data.pkl", "rb") as f:
+        ...     data = RestrictedUnpickler(f).load()
+    """
+
+    def find_class(self, module: str, name: str):
+        """Override find_class to restrict allowed classes.
+
+        Args:
+            module: Module name of the class
+            name: Class name
+
+        Returns:
+            The class object if it's in the whitelist
+
+        Raises:
+            pickle.UnpicklingError: If the class is not in the whitelist
+        """
+        if module.startswith(TRUSTED_MODULE_PREFIXES):
+            return super().find_class(module, name)
+
+        # 2. explicit whitelist (qlib internal)
+        if (module, name) in SAFE_PICKLE_CLASSES:
+            return super().find_class(module, name)
+
+        raise pickle.UnpicklingError(
+            f"Forbidden class: {module}.{name}. "
+            f"Only whitelisted classes are allowed for security reasons. "
+            f"This is to prevent arbitrary code execution through pickle deserialization."
+        )
+
+
+def restricted_pickle_load(file: BinaryIO) -> Any:
+    """Safely load a pickle file with restricted classes.
+
+    This is a drop-in replacement for pickle.load() that prevents
+    arbitrary code execution by only allowing whitelisted classes.
+
+    Args:
+        file: An opened file object in binary mode
+
+    Returns:
+        The unpickled Python object
+
+    Raises:
+        pickle.UnpicklingError: If the pickle contains forbidden classes
+
+    Example:
+        >>> with open("data.pkl", "rb") as f:
+        ...     data = restricted_pickle_load(f)
+    """
+    return RestrictedUnpickler(file).load()
+
+
+def restricted_pickle_loads(data: bytes) -> Any:
+    """Safely load a pickle from bytes with restricted classes.
+
+    This is a drop-in replacement for pickle.loads() that prevents
+    arbitrary code execution by only allowing whitelisted classes.
+
+    Args:
+        data: Bytes object containing pickled data
+
+    Returns:
+        The unpickled Python object
+
+    Raises:
+        pickle.UnpicklingError: If the pickle contains forbidden classes
+
+    Example:
+        >>> data = b'\\x80\\x04\\x95...'
+        >>> obj = restricted_pickle_loads(data)
+    """
+    file_like = io.BytesIO(data)
+    return RestrictedUnpickler(file_like).load()
+
+
+def add_safe_class(module: str, name: str) -> None:
+    """Add a class to the whitelist of safe classes for unpickling.
+
+    Use this function to extend the whitelist if your code needs to deserialize
+    additional classes. However, be very careful when adding classes, as this
+    could potentially introduce security vulnerabilities.
+
+    Args:
+        module: Module name of the class (e.g., 'my_package.my_module')
+        name: Class name (e.g., 'MyClass')
+
+    Warning:
+        Only add classes that you fully control and trust. Adding arbitrary
+        classes from external packages could introduce security risks.
+
+    Example:
+        >>> add_safe_class('my_package.models', 'CustomModel')
+    """
+    SAFE_PICKLE_CLASSES.add((module, name))
+
+
+def get_safe_classes() -> Set[Tuple[str, str]]:
+    """Get a copy of the current whitelist of safe classes.
+
+    Returns:
+        A set of (module, name) tuples representing allowed classes
+    """
+    return SAFE_PICKLE_CLASSES.copy()

qlib/workflow/task/gen.py

@@ -106,15 +106,13 @@ def handler_mod(task: dict, rolling_gen):
         rg (RollingGen): an instance of RollingGen
     """
     try:
-        interval = rolling_gen.ta.cal_interval(
+        handler_kwargs = task["dataset"]["kwargs"]["handler"]["kwargs"]
-            task["dataset"]["kwargs"]["handler"]["kwargs"]["end_time"],
+        handler_end_time = handler_kwargs.get("end_time")
-            task["dataset"]["kwargs"]["segments"][rolling_gen.test_key][1],
+        test_seg_end_time = task["dataset"]["kwargs"]["segments"][rolling_gen.test_key][1]
-        )
+        # if the end of test_segments is None (open-ended segment, i.e., "until now") or end_time < the end of test_segments,
-        # if end_time < the end of test_segments, then change end_time to allow load more data
+        # then change end_time to allow load more data
-        if interval < 0:
+        if test_seg_end_time is None or rolling_gen.ta.cal_interval(handler_end_time, test_seg_end_time) < 0:
-            task["dataset"]["kwargs"]["handler"]["kwargs"]["end_time"] = copy.deepcopy(
+            handler_kwargs["end_time"] = copy.deepcopy(test_seg_end_time)
-                task["dataset"]["kwargs"]["segments"][rolling_gen.test_key][1]
-            )
     except KeyError:
         # Maybe dataset do not have handler, then do nothing.
         pass

qlib/workflow/task/manage.py

@@ -28,6 +28,7 @@ from tqdm.cli import tqdm
 
 from .utils import get_mongodb
 from ...config import C
+from ...utils.pickle_utils import restricted_pickle_loads
 
 
 class TaskManager:
@@ -131,7 +132,7 @@ class TaskManager:
         for prefix in self.ENCODE_FIELDS_PREFIX:
             for k in list(task.keys()):
                 if k.startswith(prefix):
-                    task[k] = pickle.loads(task[k])
+                    task[k] = restricted_pickle_loads(task[k])
         return task
 
     def _dict_to_str(self, flt):

scripts/check_data_health.py

@@ -1,12 +1,12 @@
-from loguru import logger
 import os
 from typing import Optional
 
 import fire
 import pandas as pd
-import qlib
+from loguru import logger
 from tqdm import tqdm
 
+import qlib
 from qlib.data import D
 
 
@@ -36,6 +36,7 @@ class DataHealthChecker:
         self.large_step_threshold_price = large_step_threshold_price
         self.large_step_threshold_volume = large_step_threshold_volume
         self.missing_data_num = missing_data_num
+        self.qlib_dir = os.path.abspath(os.path.expanduser(qlib_dir))
 
         if csv_path:
             assert os.path.isdir(csv_path), f"{csv_path} should be a directory."
@@ -68,6 +69,43 @@ class DataHealthChecker:
             self.data[instrument] = df
         print(df)
 
+    # NOTE:
+    # This check is added due to a known issue in Qlib where feature paths
+    # are constructed using lowercased instrument names. On case-sensitive
+    # file systems (e.g. Linux), uppercase directory names under `features/`
+    # will cause data loading failures.
+    #
+    # See: https://github.com/microsoft/qlib/issues/2053
+    def check_features_dir_lowercase(self) -> Optional[pd.DataFrame]:
+        """
+        Check whether all subdirectories under `<qlib_dir>/features` are named in lowercase.
+
+        This validation helps prevent data loading issues on case-sensitive
+        file systems caused by uppercase instrument directory names.
+        """
+        if not self.qlib_dir:
+            return None
+
+        features_dir = os.path.join(self.qlib_dir, "features")
+        if not os.path.isdir(features_dir):
+            logger.warning(f"`features` directory not found under {self.qlib_dir}")
+            return None
+
+        bad_dirs = []
+        for name in os.listdir(features_dir):
+            full_path = os.path.join(features_dir, name)
+            if os.path.isdir(full_path) and name != name.lower():
+                bad_dirs.append(name)
+
+        if bad_dirs:
+            result_df = pd.DataFrame({"non_lowercase_dir": bad_dirs})
+            return result_df
+        else:
+            logger.info(
+                f"✅ All subdirectories under `{os.path.join(self.qlib_dir, 'features')}` are named in lowercase."
+            )
+            return None
+
     def check_missing_data(self) -> Optional[pd.DataFrame]:
         """Check if any data is missing in the DataFrame."""
         result_dict = {
@@ -177,11 +215,13 @@ class DataHealthChecker:
         check_large_step_changes_result = self.check_large_step_changes()
         check_required_columns_result = self.check_required_columns()
         check_missing_factor_result = self.check_missing_factor()
+        check_features_dir_case_result = self.check_features_dir_lowercase()
         if (
             check_large_step_changes_result is not None
             or check_large_step_changes_result is not None
             or check_required_columns_result is not None
             or check_missing_factor_result is not None
+            or check_features_dir_case_result is not None
         ):
             print(f"\nSummary of data health check ({len(self.data)} files checked):")
             print("-------------------------------------------------")
@@ -197,6 +237,11 @@ class DataHealthChecker:
             if isinstance(check_missing_factor_result, pd.DataFrame):
                 logger.warning(f"The factor column does not exist or is empty")
                 print(check_missing_factor_result)
+            if isinstance(check_features_dir_case_result, pd.DataFrame):
+                logger.warning(
+                    f"Some subdirectories under `{os.path.join(self.qlib_dir, 'features')}` contain uppercase letters, please rename them to lowercase manually."
+                )
+                print(check_features_dir_case_result)
 
 
 if __name__ == "__main__":

scripts/data_collector/us_index/collector.py

@@ -7,12 +7,14 @@ import sys
 from pathlib import Path
 from concurrent.futures import ThreadPoolExecutor
 from typing import List
+from io import StringIO
 
 import fire
 import requests
 import pandas as pd
 from tqdm import tqdm
 from loguru import logger
+from fake_useragent import UserAgent
 
 
 CUR_DIR = Path(__file__).resolve().parent
@@ -51,6 +53,7 @@ class WIKIIndex(IndexBase):
         )
 
         self._target_url = f"{WIKI_URL}/{WIKI_INDEX_NAME_MAP[self.index_name.upper()]}"
+        self._ua = UserAgent()
 
     @property
     @abc.abstractmethod
@@ -112,7 +115,8 @@ class WIKIIndex(IndexBase):
         return _calendar_list
 
     def _request_new_companies(self) -> requests.Response:
-        resp = requests.get(self._target_url, timeout=None)
+        headers = {"User-Agent": self._ua.random}
+        resp = requests.get(self._target_url, timeout=None, headers=headers)
         if resp.status_code != 200:
             raise ValueError(f"request error: {self._target_url}")
 
@@ -128,7 +132,7 @@ class WIKIIndex(IndexBase):
     def get_new_companies(self):
         logger.info(f"get new companies {self.index_name} ......")
         _data = deco_retry(retry=self._request_retry, retry_sleep=self._retry_sleep)(self._request_new_companies)()
-        df_list = pd.read_html(_data.text)
+        df_list = pd.read_html(StringIO(_data.text))
         for _df in df_list:
             _df = self.filter_df(_df)
             if (_df is not None) and (not _df.empty):
@@ -226,7 +230,11 @@ class SP500Index(WIKIIndex):
     def get_changes(self) -> pd.DataFrame:
         logger.info(f"get sp500 history changes......")
         # NOTE: may update the index of the table
-        changes_df = pd.read_html(self.WIKISP500_CHANGES_URL)[-1]
+        # Add headers to avoid 403 Forbidden error from Wikipedia
+        headers = {"User-Agent": self._ua.random}
+        response = requests.get(self.WIKISP500_CHANGES_URL, headers=headers, timeout=None)
+        response.raise_for_status()
+        changes_df = pd.read_html(StringIO(response.text))[-1]
         changes_df = changes_df.iloc[:, [0, 1, 3]]
         changes_df.columns = [self.DATE_FIELD_NAME, self.ADD, self.REMOVE]
         changes_df[self.DATE_FIELD_NAME] = pd.to_datetime(changes_df[self.DATE_FIELD_NAME])

scripts/data_collector/us_index/requirements.txt

@@ -3,3 +3,4 @@ requests
 pandas
 lxml
 loguru
+fake-useragent

scripts/data_collector/utils.py

@@ -3,6 +3,7 @@
 
 import re
 import copy
+import datetime
 import importlib
 import time
 import bisect
@@ -24,7 +25,7 @@ from bs4 import BeautifulSoup
 
 HS_SYMBOLS_URL = "http://app.finance.ifeng.com/hq/list.php?type=stock_a&class={s_type}"
 
-CALENDAR_URL_BASE = "http://push2his.eastmoney.com/api/qt/stock/kline/get?secid={market}.{bench_code}&fields1=f1%2Cf2%2Cf3%2Cf4%2Cf5&fields2=f51%2Cf52%2Cf53%2Cf54%2Cf55%2Cf56%2Cf57%2Cf58&klt=101&fqt=0&beg=19900101&end=20991231"
+CALENDAR_URL_BASE = "http://push2his.eastmoney.com/api/qt/stock/kline/get?secid={market}.{bench_code}&fields1=f1%2Cf2%2Cf3%2Cf4%2Cf5&fields2=f51%2Cf52%2Cf53%2Cf54%2Cf55%2Cf56%2Cf57%2Cf58&klt=101&fqt=0"
 SZSE_CALENDAR_URL = "http://www.szse.cn/api/report/exchange/onepersistenthour/monthList?month={month}&random={random}"
 
 CALENDAR_BENCH_URL_MAP = {
@@ -39,6 +40,24 @@ CALENDAR_BENCH_URL_MAP = {
     "BR_ALL": "^BVSP",
 }
 
+CHROME_UA_POOL = [
+    # Windows
+    "Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
+    "AppleWebKit/537.36 (KHTML, like Gecko) "
+    "Chrome/120.0.0.0 Safari/537.36",
+    "Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
+    "AppleWebKit/537.36 (KHTML, like Gecko) "
+    "Chrome/121.0.6167.85 Safari/537.36",
+    # macOS
+    "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) "
+    "AppleWebKit/537.36 (KHTML, like Gecko) "
+    "Chrome/121.0.0.0 Safari/537.36",
+    # Linux
+    "Mozilla/5.0 (X11; Linux x86_64) "  # pylint: disable=W1404
+    "AppleWebKit/537.36 (KHTML, like Gecko) "
+    "Chrome/120.0.0.0 Safari/537.36",
+]
+
 _BENCH_CALENDAR_LIST = None
 _ALL_CALENDAR_LIST = None
 _HS_SYMBOLS = None
@@ -52,6 +71,16 @@ _CALENDAR_MAP = {}
 MINIMUM_SYMBOLS_NUM = 3900
 
 
+def build_headers():
+    return {
+        "User-Agent": random.choice(CHROME_UA_POOL),
+        "Accept": "application/json,text/plain,*/*",
+        "Accept-Language": "zh-CN,zh;q=0.9",
+        "Referer": "https://quote.eastmoney.com/",
+        "Connection": "keep-alive",
+    }
+
+
 def get_calendar_list(bench_code="CSI300") -> List[pd.Timestamp]:
     """get SH/SZ history calendar list
 
@@ -67,9 +96,38 @@ def get_calendar_list(bench_code="CSI300") -> List[pd.Timestamp]:
 
     logger.info(f"get calendar list: {bench_code}......")
 
-    def _get_calendar(url):
+    def _get_calendar(url, max_retry=3):
-        _value_list = requests.get(url, timeout=None).json()["data"]["klines"]
+        session = requests.Session()
-        return sorted(map(lambda x: pd.Timestamp(x.split(",")[0]), _value_list))
+        session.headers.update(build_headers())
+        current_datetime = datetime.datetime.now()
+        cur_year = current_datetime.year
+        res_list = []
+        failed_years = []
+        for year in range(2000, cur_year + 1):
+            start = f"{year}0101"
+            end = f"{year}1231"
+            formatted_url = url + f"&beg={start}&end={end}".format(start=start, end=end)
+            for attempt in range(max_retry):
+                try:
+                    resp = session.get(formatted_url, timeout=10)
+                    resp.raise_for_status()
+
+                    data = resp.json().get("data")
+                    if not data or "klines" not in data:
+                        raise ValueError("missing klines")
+
+                    res_list.extend(pd.Timestamp(x.split(",")[0]) for x in data["klines"])
+                    break
+
+                except Exception as e:
+                    time.sleep(random.uniform(0.8, 1.5))
+            else:
+                failed_years.append(year)
+
+        if failed_years:
+            logger.warning(f"Calendar incomplete, failed years: {failed_years}")
+
+        return sorted(set(res_list))
 
     calendar = _CALENDAR_MAP.get(bench_code, None)
     if calendar is None:
@@ -80,28 +138,14 @@ def get_calendar_list(bench_code="CSI300") -> List[pd.Timestamp]:
             calendar = df.index.get_level_values(level="date").map(pd.Timestamp).unique().tolist()
         else:
             if bench_code.upper() == "ALL":
+                import akshare as ak  # pylint: disable=C0415
 
-                @deco_retry
+                trade_date_df = ak.tool_trade_date_hist_sina()
-                def _get_calendar(month):
+                trade_date_list = trade_date_df["trade_date"].tolist()
-                    _cal = []
+                trade_date_list = [pd.Timestamp(d) for d in trade_date_list]
-                    try:
+                dates = pd.DatetimeIndex(trade_date_list)
-                        resp = requests.get(
+                filtered_dates = dates[(dates >= "2000-01-04") & (dates <= pd.Timestamp.today().normalize())]
-                            SZSE_CALENDAR_URL.format(month=month, random=random.random), timeout=None
+                calendar = filtered_dates.tolist()
-                        ).json()
-                        for _r in resp["data"]:
-                            if int(_r["jybz"]):
-                                _cal.append(pd.Timestamp(_r["jyrq"]))
-                    except Exception as e:
-                        raise ValueError(f"{month}-->{e}") from e
-                    return _cal
-
-                month_range = pd.date_range(start="2000-01", end=pd.Timestamp.now() + pd.Timedelta(days=31), freq="M")
-                calendar = []
-                for _m in month_range:
-                    cal = _get_calendar(_m.strftime("%Y-%m"))
-                    if cal:
-                        calendar += cal
-                calendar = list(filter(lambda x: x <= pd.Timestamp.now(), calendar))
             else:
                 calendar = _get_calendar(CALENDAR_BENCH_URL_MAP[bench_code])
         _CALENDAR_MAP[bench_code] = calendar
@@ -261,7 +305,11 @@ def get_hs_stock_symbols() -> list:
 
         # Add suffix after the stock code to conform to yahooquery standard, otherwise the data will not be fetched.
         _symbols = [
-            _symbol + ".ss" if _symbol.startswith("6") else _symbol + ".sz" if _symbol.startswith(("0", "3")) else None
+            (
+                _symbol + ".ss"
+                if _symbol.startswith("6")
+                else _symbol + ".sz" if _symbol.startswith(("0", "3")) else None
+            )
             for _symbol in _symbols
         ]
         _symbols = [_symbol for _symbol in _symbols if _symbol is not None]
@@ -372,7 +420,14 @@ def get_us_stock_symbols(qlib_data_path: [str, Path] = None) -> list:
             s_ = s_.strip("*")
             return s_
 
-        _US_SYMBOLS = sorted(set(map(_format, filter(lambda x: len(x) < 8 and not x.endswith("WS"), _all_symbols))))
+        _US_SYMBOLS = sorted(
+            set(
+                map(
+                    _format,
+                    filter(lambda x: len(x) < 8 and not x.endswith("WS"), _all_symbols),
+                )
+            )
+        )
 
     return _US_SYMBOLS
 
@@ -486,7 +541,10 @@ def get_en_fund_symbols(qlib_data_path: [str, Path] = None) -> list:
             raise ValueError("request error")
         try:
             _symbols = []
-            for sub_data in re.findall(r"[\[](.*?)[\]]", resp.content.decode().split("= [")[-1].replace("];", "")):
+            for sub_data in re.findall(
+                r"[\[](.*?)[\]]",
+                resp.content.decode().split("= [")[-1].replace("];", ""),
+            ):
                 data = sub_data.replace('"', "").replace("'", "")
                 # TODO: do we need other information, like fund_name from ['000001', 'HXCZHH', '华夏成长混合', '混合型', 'HUAXIACHENGZHANGHUNHE']
                 _symbols.append(data.split(",")[0])
@@ -667,7 +725,11 @@ def get_instruments(
     """
     _cur_module = importlib.import_module("data_collector.{}.collector".format(market_index))
     obj = getattr(_cur_module, f"{index_name.upper()}Index")(
-        qlib_dir=qlib_dir, index_name=index_name, freq=freq, request_retry=request_retry, retry_sleep=retry_sleep
+        qlib_dir=qlib_dir,
+        index_name=index_name,
+        freq=freq,
+        request_retry=request_retry,
+        retry_sleep=retry_sleep,
     )
     getattr(obj, method)()
 
@@ -675,7 +737,10 @@ def get_instruments(
 def _get_all_1d_data(_date_field_name: str, _symbol_field_name: str, _1d_data_all: pd.DataFrame):
     df = copy.deepcopy(_1d_data_all)
     df.reset_index(inplace=True)
-    df.rename(columns={"datetime": _date_field_name, "instrument": _symbol_field_name}, inplace=True)
+    df.rename(
+        columns={"datetime": _date_field_name, "instrument": _symbol_field_name},
+        inplace=True,
+    )
     df.columns = list(map(lambda x: x[1:] if x.startswith("$") else x, df.columns))
     return df
 

scripts/data_collector/yahoo/collector.py

@@ -613,10 +613,6 @@ class YahooNormalize1min(YahooNormalize, ABC):
     def symbol_to_yahoo(self, symbol):
         raise NotImplementedError("rewrite symbol_to_yahoo")
 
-    @abc.abstractmethod
-    def _get_1d_calendar_list(self) -> Iterable[pd.Timestamp]:
-        raise NotImplementedError("rewrite _get_1d_calendar_list")
-
 
 class YahooNormalizeUS:
     def _get_calendar_list(self) -> Iterable[pd.Timestamp]:

scripts/data_collector/yahoo/requirements.txt

@@ -10,3 +10,4 @@ joblib
 beautifulsoup4
 bs4
 soupsieve
+akshare

setup.py

@@ -2,22 +2,12 @@ import os
 
 import numpy
 from setuptools import Extension, setup
-from setuptools_scm import get_version
-
-
-def read(rel_path: str) -> str:
-    here = os.path.abspath(os.path.dirname(__file__))
-    with open(os.path.join(here, rel_path), encoding="utf-8") as fp:
-        return fp.read()
 
 
 NUMPY_INCLUDE = numpy.get_include()
 
 
-VERSION = get_version(root=".", relative_to=__file__)
-
 setup(
-    version=VERSION,
     ext_modules=[
         Extension(
             "qlib.data._libs.rolling",

tests/data_mid_layer_tests/test_handler.py

@@ -1,10 +1,10 @@
 import os
-import pickle
-import shutil
 import unittest
-from qlib.tests import TestAutoData
+
 from qlib.data import D
 from qlib.data.dataset.handler import DataHandlerLP
+from qlib.tests import TestAutoData
+from qlib.utils.pickle_utils import restricted_pickle_load
 
 
 class HandlerTests(TestAutoData):
@@ -23,7 +23,7 @@ class HandlerTests(TestAutoData):
         dh.to_pickle(fname, dump_all=True)
 
         with open(fname, "rb") as f:
-            dh_d = pickle.load(f)
+            dh_d = restricted_pickle_load(f)
 
         self.assertTrue(dh_d._data.equals(df))
         self.assertTrue(dh_d._infer is dh_d._data)