Reworked the project structure and fixed import cycling issues.

src/data/__init__.py

@@ -4,7 +4,8 @@
 from pathlib import Path
 from typing import Optional
 
-from .provider import _PROVIDER_REGISTRY
+from src.data.core import *
+from src.data.core.provider import _PROVIDER_REGISTRY
 
 def get_default_cache_root() -> str:
     """Returns the default cache root directory."""
@@ -36,4 +37,4 @@ def initialize_cache_directories(cache_root: Optional[str] = None):
 # Run the initialization function when the module is imported
 initialize_cache_directories()
 
-from .manager import DataManager
+from .managers.data_manager import DataManager

src/data/core/operations/__init__.py

@@ -2,10 +2,7 @@
 
 import math
 import numpy as np
-import pandas as pd
-import xarray as xr
 import jax
-import xarray_jax as xj
 import jax.numpy as jnp
 import equinox as eqx
 from typing import Union, Dict, List, Optional, Tuple, Any, Callable

src/data/provider.py → src/data/core/provider.py

@@ -1,15 +1,14 @@
 # src/data/provider.py
 
-from typing import Dict, Type
-from .abstracts.base import BaseDataSource
+from typing import Dict, Any
 
 class Provider:
     def __init__(
         self,
         name: str,
         website: str,
         description: str,
-        fetcher_dict: Dict[str, Type[BaseDataSource]],
+        fetcher_dict: Dict[str, Any],
         repr_name: str = None,
     ):
         self.name = name

src/data/loaders/abstracts/__init__.py

@@ -0,0 +1,5 @@
+"""
+Abstract base classes for data loaders.
+"""
+
+from .base import BaseDataSource

src/data/abstracts/base.py → src/data/loaders/abstracts/base.py

@@ -1,4 +1,4 @@
-# src/data/loaders/base.py
+# src/data/loaders/abstracts/base.py
 
 import os
 import pandas as pd
@@ -9,11 +9,9 @@
 import hashlib
 from abc import ABC, abstractmethod
 
-from ..core.struct import DatasetDateTimeAccessor
-from ..core.util import FrequencyType
-from ..core.util import Loader as load
-
-from ..processors_registry import post_processor
+from src.data.core.util import FrequencyType
+from src.data.core.util import Loader as load
+from src.data.processors.registry import post_processor
 
 class BaseDataSource(ABC):
     """
@@ -192,7 +190,7 @@ def load_from_cache(
         try:
             ds = xr.load_dataset(netcdf_path)  # or xr.open_dataset, either is OK
 
-            from ..core.util import TimeSeriesIndex
+            from src.data.core import TimeSeriesIndex
 
             time_coord = ds.coords['time']
             ts_index = TimeSeriesIndex(time_coord)

src/data/loaders/open_bb/__init__.py

@@ -1,6 +1,10 @@
 # data/loaders/openbb/__init__.py
 
-from ...provider import Provider, register_provider
+"""
+OpenBB data loaders and provider registration.
+"""
+
+from src.data.core.provider import Provider, register_provider
 from .generic import GenericOpenBBDataFetcher
 from openbb import obb
 

src/data/loaders/open_bb/generic.py

@@ -1,6 +1,10 @@
 # data/loaders/openbb/generic.py
 
-from src.data.abstracts.base import BaseDataSource
+"""
+Generic OpenBB data fetcher implementation.
+"""
+
+from src.data.loaders.abstracts.base import BaseDataSource
 from typing import Dict, Any, List
 import xarray as xr
 import pandas as pd

src/data/loaders/wrds/__init__.py

@@ -1,6 +1,10 @@
 # data/loaders/wrds/__init__.py
 
-from ...provider import Provider, register_provider
+"""
+WRDS data loaders and provider registration.
+"""
+
+from src.data.core.provider import Provider, register_provider
 
 # Fetchers
 from .compustat import CompustatDataFetcher

src/data/loaders/wrds/crsp.py

@@ -136,5 +136,6 @@ def _preprocess_df(self, df: pd.DataFrame, **config) -> pd.DataFrame:
             )
         except Exception as e:
             print(f"Warning: Could not merge msedelist data due to error: {e}")
-
+
+
         return df

src/data/loaders/wrds/generic.py

@@ -1,13 +1,17 @@
 # data/loaders/wrds/generic.py
 
+"""
+Generic WRDS data loader implementation.
+"""
+
 import os
 from typing import Any, Dict, List, Optional, Union
 import pandas as pd
 import xarray as xr
 import pyreadstat
 
-from src.data.abstracts.base import BaseDataSource
-from src.data.core import FrequencyType
+from src.data.loaders.abstracts.base import BaseDataSource
+from src.data.core.util import FrequencyType
 
 class GenericWRDSDataLoader(BaseDataSource):
     """

src/data/managers/__init__.py

@@ -0,0 +1,5 @@
+"""
+Managers for centralized control over data operations.
+"""
+
+from .data_manager import DataManager

src/data/manager.py → src/data/managers/data_manager.py

@@ -2,13 +2,12 @@
 
 import xarray as xr
 import xarray_jax
-from xarray import DataTree
 
 from typing import Union, List, Dict, Any
 import os
 import yaml  
 
-from .provider import _PROVIDER_REGISTRY
+from src.data.core.provider import _PROVIDER_REGISTRY
 from src.data.loaders import *
 
 class DataManager:

src/data/processors/__init__.py

@@ -0,0 +1,6 @@
+"""
+Data processing utilities and registries.
+"""
+
+from .registry import post_processor
+from .processors import *

src/data/processors.py → src/data/processors/processors.py

@@ -1,7 +1,8 @@
 # src/data/processors.py
 
-from processors_registry import post_processor
+from src.data.processors.registry import post_processor
 import xarray
+from typing import *
 
 @post_processor
 def processor_test(funcs: Dict[str, Any], ds: xarray.Dataset) -> xarray.Dataset: