featurebyte 0.2.2__tar.gz → 0.3.1__tar.gz

{featurebyte-0.2.2 → featurebyte-0.3.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: featurebyte
-Version: 0.2.2
+Version: 0.3.1
 Summary: Python Library for FeatureOps
 Home-page: https://featurebyte.com
 License: Elastic License 2.0
@@ -16,11 +16,6 @@ Classifier: Programming Language :: Python :: 3.8
 Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
-Classifier: Programming Language :: Python :: 3
-Classifier: Programming Language :: Python :: 3.10
-Classifier: Programming Language :: Python :: 3.11
-Classifier: Programming Language :: Python :: 3.8
-Classifier: Programming Language :: Python :: 3.9
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Provides-Extra: server
 Requires-Dist: PyYAML (>=6.0,<7.0)
@@ -40,7 +35,7 @@ Requires-Dist: fastapi (>=0.95.1,<0.96.0) ; extra == "server"
 Requires-Dist: featurebyte-freeware (>=0.2.14,<0.3.0) ; extra == "server"
 Requires-Dist: gevent (>=22.10.2,<23.0.0) ; extra == "server"
 Requires-Dist: humanize (>=4.4.0,<5.0.0)
-Requires-Dist: importlib_metadata (>=4.5.0,<5.0.0) ; python_version >= "3.8" and python_version < "4.0"
+Requires-Dist: importlib_metadata ; python_version >= "3.8" and python_version < "4.0"
 Requires-Dist: jinja2 (>=3.1.2,<4.0.0)
 Requires-Dist: lazy-object-proxy (>=1.7.1,<2.0.0)
 Requires-Dist: motor (>=3.0.0,<4.0.0) ; extra == "server"
@@ -57,8 +52,8 @@ Requires-Dist: redis (>=5.0.0b1,<6.0.0) ; extra == "server"
 Requires-Dist: requests (>=2.27.1,<3.0.0)
 Requires-Dist: rich (>=13.3.4,<14.0.0)
 Requires-Dist: sasl (>=0.3.1,<0.4.0) ; extra == "server"
-Requires-Dist: smart-open (>=6.3.0,<7.0.0) ; extra == "server"
-Requires-Dist: snowflake-connector-python (>=3.0.3,<4.0.0) ; extra == "server"
+Requires-Dist: smart-open[azure,gcs] (>=6.3.0,<7.0.0) ; extra == "server"
+Requires-Dist: snowflake-connector-python (>=3.0.3,<4.0.0,!=3.0.4) ; extra == "server"
 Requires-Dist: sqlglot (>=10.1.3,<10.4)
 Requires-Dist: thrift-sasl (>=0.4.3,<0.5.0) ; extra == "server"
 Requires-Dist: typeguard (>=2.13.3,<3.0.0)
@@ -105,19 +100,16 @@ Feature Engineering and management doesn’t have to be complicated. Take charge
 
 ``` python
 # Get view from catalog
-invoices = catalog.get_view("INVOICES")
-# Customer average spend over past 5 weeks
-features = invoices.groupby(
-    "CustomerId"
-).aggregate_over(
+invoice_view = catalog.get_view("GROCERYINVOICE")
+# Declare features of total spent by customer in the past 7 and 28 days
+customer_purchases = invoice_view.groupby("GroceryCustomerGuid").aggregate_over(
     "Amount",
-    method="avg",
-    feature_names=["AvgSpend5w"],
+    method="sum",
+    feature_names=["CustomerTotalSpent_7d", "CustomerTotalSpent_28d"],
     fill_value=0,
-    windows=["5w"]
+    windows=['7d', '28d']
 )
-# Save feature
-features["AvgSpend5w"].save()
+customer_purchases.save()
 ```
 
 ### Experiment
@@ -307,33 +299,33 @@ items_view = catalog.get_view("INVOICEITEMS")
 product_view = catalog.get_view("GROCERYPRODUCT")
 # Join product view to items view
 items_view = items_view.join(product_view)
-# Get Customer purchases across product group over the past 4 weeks
-customer_inventory_28d = items_view.groupby(
+# Get Customer purchases across product group in the past 4 weeks
+customer_basket_28d = items_view.groupby(
     by_keys = "GroceryCustomerGuid", category=”ProductGroup”
 ).aggregate_over(
    "TotalCost",
     method=fb.AggFunc.SUM,
-    feature_names=["CustomerInventory_28d"],
+    feature_names=["CustomerBasket_28d"],
     windows=['28d']
 )
 # Get customer view and join it to items view
 customer_view = catalog.get_view("GROCERYCUSTOMER")
 items_view = items_view.join(customer_view)
 # Get Purchases of Customers living in the same state
-# across product group over the past 4 weeks
-state_inventory_28d = items_view.groupby(
+# across product group in the past 4 weeks
+state_basket_28d = items_view.groupby(
     by_keys="State", category="ProductGroup"
 ).aggregate_over(
    "TotalCost",
     method=fb.AggFunc.SUM,
-    feature_names=["StateInventory_28d"],
+    feature_names=["StateBasket_28d"],
     windows=['28d']
 )
 # Create a feature that measures the similarity of a customer purchases
 # and purchases of customers living in the same state
 customer_state_similarity_28d = \
-    customer_inventory_28d["CustomerInventory_28d"].cd.cosine_similarity(
-        state_inventory_28d["StateInventory_28d"]
+    customer_basket_28d["CustomerBasket_28d"].cd.cosine_similarity(
+        state_basket_28d["StateBasket_28d"]
     )
 # save the new feature
 customer_state_similarity_28d.name = \

{featurebyte-0.2.2 → featurebyte-0.3.1}/README.md

@@ -33,19 +33,16 @@ Feature Engineering and management doesn’t have to be complicated. Take charge
 
 ``` python
 # Get view from catalog
-invoices = catalog.get_view("INVOICES")
-# Customer average spend over past 5 weeks
-features = invoices.groupby(
-    "CustomerId"
-).aggregate_over(
+invoice_view = catalog.get_view("GROCERYINVOICE")
+# Declare features of total spent by customer in the past 7 and 28 days
+customer_purchases = invoice_view.groupby("GroceryCustomerGuid").aggregate_over(
     "Amount",
-    method="avg",
-    feature_names=["AvgSpend5w"],
+    method="sum",
+    feature_names=["CustomerTotalSpent_7d", "CustomerTotalSpent_28d"],
     fill_value=0,
-    windows=["5w"]
+    windows=['7d', '28d']
 )
-# Save feature
-features["AvgSpend5w"].save()
+customer_purchases.save()
 ```
 
 ### Experiment
@@ -235,33 +232,33 @@ items_view = catalog.get_view("INVOICEITEMS")
 product_view = catalog.get_view("GROCERYPRODUCT")
 # Join product view to items view
 items_view = items_view.join(product_view)
-# Get Customer purchases across product group over the past 4 weeks
-customer_inventory_28d = items_view.groupby(
+# Get Customer purchases across product group in the past 4 weeks
+customer_basket_28d = items_view.groupby(
     by_keys = "GroceryCustomerGuid", category=”ProductGroup”
 ).aggregate_over(
    "TotalCost",
     method=fb.AggFunc.SUM,
-    feature_names=["CustomerInventory_28d"],
+    feature_names=["CustomerBasket_28d"],
     windows=['28d']
 )
 # Get customer view and join it to items view
 customer_view = catalog.get_view("GROCERYCUSTOMER")
 items_view = items_view.join(customer_view)
 # Get Purchases of Customers living in the same state
-# across product group over the past 4 weeks
-state_inventory_28d = items_view.groupby(
+# across product group in the past 4 weeks
+state_basket_28d = items_view.groupby(
     by_keys="State", category="ProductGroup"
 ).aggregate_over(
    "TotalCost",
     method=fb.AggFunc.SUM,
-    feature_names=["StateInventory_28d"],
+    feature_names=["StateBasket_28d"],
     windows=['28d']
 )
 # Create a feature that measures the similarity of a customer purchases
 # and purchases of customers living in the same state
 customer_state_similarity_28d = \
-    customer_inventory_28d["CustomerInventory_28d"].cd.cosine_similarity(
-        state_inventory_28d["StateInventory_28d"]
+    customer_basket_28d["CustomerBasket_28d"].cd.cosine_similarity(
+        state_basket_28d["StateBasket_28d"]
     )
 # save the new feature
 customer_state_similarity_28d.name = \

{featurebyte-0.2.2 → featurebyte-0.3.1}/featurebyte/__init__.py

@@ -1,7 +1,6 @@
 """Python Library for FeatureOps"""
 from typing import Any, List, Optional
 
-import inspect
 import sys
 from http import HTTPStatus
 
@@ -51,9 +50,12 @@ from featurebyte.exception import (
     InvalidSettingsError,
     RecordRetrievalException,
 )
+from featurebyte.feature_utility import list_unsaved_features
 from featurebyte.logging import get_logger
 from featurebyte.models.credential import (
     AccessTokenCredential,
+    AzureBlobStorageCredential,
+    GCSStorageCredential,
     S3StorageCredential,
     UsernamePasswordCredential,
 )
@@ -365,97 +367,6 @@ def list_deployments(
     return output_df.drop(columns=["id"])
 
 
-def list_unsaved_features() -> pd.DataFrame:
-    """
-    List all unsaved features in the current session.
-
-    Returns
-    -------
-    pd.DataFrame
-        List of unsaved features
-
-    Examples
-    --------
-    >>> customer_gender = catalog.get_view("GROCERYCUSTOMER")["Gender"].as_feature(
-    ...     feature_name="Customer Gender"
-    ... )
-    >>> fb.list_unsaved_features()[["variable_name", "name", "catalog", "active_catalog"]]
-         variable_name             name  catalog  active_catalog
-    0  customer_gender  Customer Gender  grocery            True
-    """
-    processed_variables = set()
-    unsaved_features = []
-    client = Configurations().get_client()
-
-    def _is_saved(feature: Feature) -> bool:
-        """
-        Check if a feature is saved.
-
-        Parameters
-        ----------
-        feature: Feature
-            Feature to check
-
-        Returns
-        -------
-        bool
-        """
-        response = client.get(
-            url=f"/feature/{feature.id}", headers={"active-catalog-id": str(feature.catalog_id)}
-        )
-        if response.status_code == HTTPStatus.OK:
-            return True
-        return False
-
-    # get list of frame info from the current call stack
-    call_stack = inspect.stack()
-    # skip the first frame, which is the current function, to get the caller's frame
-    caller_frame_info = call_stack[1]
-    # check caller's local variables first, then global variables for unsaved features
-    caller_variables = [caller_frame_info.frame.f_locals, caller_frame_info.frame.f_globals]
-    for variables in caller_variables:
-        for var_name, var_obj in variables.items():
-            # global variable may be overriden by local variable
-            if var_name in processed_variables:
-                continue
-            if isinstance(var_obj, Feature) and not _is_saved(var_obj):
-                unsaved_features.append(
-                    {
-                        "object_id": str(var_obj.id),
-                        "variable_name": var_name,
-                        "name": var_obj.name,
-                        "catalog_id": var_obj.catalog_id,
-                    }
-                )
-            elif isinstance(var_obj, BaseFeatureGroup):
-                for name, feature in var_obj.feature_objects.items():
-                    if not _is_saved(feature):
-                        unsaved_features.append(
-                            {
-                                "object_id": str(feature.id),
-                                "variable_name": f'{var_name}["{name}"]',
-                                "name": feature.name,
-                                "catalog_id": feature.catalog_id,
-                            }
-                        )
-            processed_variables.add(var_name)
-
-    if unsaved_features:
-        catalogs = Catalog.list(include_id=True)
-        return (
-            pd.DataFrame(unsaved_features)
-            .merge(
-                catalogs.rename({"name": "catalog", "active": "active_catalog"}, axis=1),
-                left_on="catalog_id",
-                right_on="id",
-                how="left",
-            )[["object_id", "variable_name", "name", "catalog", "active_catalog"]]
-            .sort_values("object_id")
-            .reset_index(drop=True)
-        )
-    return pd.DataFrame(columns=["object_id", "variable_name", "name", "catalog", "active_catalog"])
-
-
 __all__ = [
     # API objects
     "BatchFeatureTable",
@@ -491,8 +402,10 @@ __all__ = [
     "Table",
     "TableFeatureJobSetting",
     # credentials
+    "AzureBlobStorageCredential",
     "AccessTokenCredential",
     "Credential",
+    "GCSStorageCredential",
     "S3StorageCredential",
     "UsernamePasswordCredential",
     # enums
@@ -518,6 +431,8 @@ __all__ = [
     "start",
     "stop",
     "playground",
+    # utility
+    "list_unsaved_features",
 ]
 
 

{featurebyte-0.2.2 → featurebyte-0.3.1}/featurebyte/api/api_object.py

@@ -21,15 +21,14 @@ from pandas import DataFrame
 from requests.models import Response
 from typeguard import typechecked
 
-from featurebyte.api.api_object_util import PrettyDict, ProgressThread
+from featurebyte.api.api_object_util import ProgressThread
 from featurebyte.common.env_util import get_alive_bar_additional_params
+from featurebyte.common.formatting_util import InfoDict
 from featurebyte.common.utils import construct_repr_string
 from featurebyte.config import Configurations
 from featurebyte.exception import (
     DuplicatedRecordException,
-    ObjectHasBeenSavedError,
     RecordCreationException,
-    RecordDeletionException,
     RecordRetrievalException,
     RecordUpdateException,
 )
@@ -114,13 +113,26 @@ class ApiObject(FeatureByteBaseDocumentModel):
     # global api object cache shared by all the ApiObject class & its child classes
     _cache: Any = TTLCache(maxsize=1024, ttl=1)
 
+    def _repr_html_(self) -> str:
+        """
+        HTML representation of the object
+
+        Returns
+        -------
+        str
+        """
+        try:
+            return InfoDict(self.info()).to_html()
+        except (RecordCreationException, RecordRetrievalException):
+            # object has not been saved yet
+            return repr(self)
+
     def __repr__(self) -> str:
-        info_repr = ""
         try:
             info_repr = repr(self.info())
         except (RecordCreationException, RecordRetrievalException):
             # object has not been saved yet
-            pass
+            info_repr = super().__repr__()
 
         return construct_repr_string(self, info_repr)
 
@@ -720,7 +732,9 @@ class ApiObject(FeatureByteBaseDocumentModel):
         client = Configurations().get_client()
         response = client.get(url=f"{self._route}/{self.id}/info", params={"verbose": verbose})
         if response.status_code == HTTPStatus.OK:
-            return PrettyDict(response.json())
+            info = response.json()
+            info["class_name"] = self.__class__.__name__
+            return InfoDict(info)
         raise RecordRetrievalException(response, "Failed to retrieve object info.")
 
     @classmethod
@@ -869,117 +883,3 @@ class ApiObject(FeatureByteBaseDocumentModel):
             self._poll_async_task(task_response=update_response, delay=delay, retrieve_result=False)
         # call get to update the object cache as retrieve result is False
         self.get_by_id(self.id)
-
-
-class SavableApiObject(ApiObject):
-    """
-    ApiObject contains common methods used to interact with API routes
-    """
-
-    def _get_create_payload(self) -> dict[str, Any]:
-        """
-        Construct payload used for post route
-
-        Returns
-        -------
-        dict[str, Any]
-        """
-        return self.json_dict(exclude_none=True)
-
-    def _pre_save_operations(self, conflict_resolution: ConflictResolution) -> None:
-        """
-        Operations to be executed before saving the api object
-
-        Parameters
-        ----------
-        conflict_resolution: ConflictResolution
-            "raise" raises error when then counters conflict error (default)
-            "retrieve" handle conflict error by retrieving object with the same name
-        """
-        _ = conflict_resolution
-
-    @typechecked
-    def save(self, conflict_resolution: ConflictResolution = "raise") -> None:
-        """
-        Save an object to the persistent data store.
-
-        A conflict could be triggered when the object being saved has violated a uniqueness check at the persistent
-        data store. For example, the same object ID could have been used by another record that is already stored.
-
-        In these scenarios, we can either raise an error or retrieve the object with the same name, depending on the
-        conflict resolution parameter passed in. The default behavior is to raise an error.
-
-        Parameters
-        ----------
-        conflict_resolution: ConflictResolution
-            "raise" will raise an error when we encounter a conflict error.
-            "retrieve" will handle the conflict error by retrieving the object with the same name.
-
-        Raises
-        ------
-        ObjectHasBeenSavedError
-            If the object has been saved before.
-        DuplicatedRecordException
-            When a record with the same key exists at the persistent data store.
-        RecordCreationException
-            When we fail to save the new object (general failure).
-
-        Examples
-        --------
-        Note that the examples below are not exhaustive.
-
-        Save a new Entity object.
-
-        >>> entity = fb.Entity(name="grocerycustomer_example", serving_names=["GROCERYCUSTOMERGUID"])  # doctest: +SKIP
-        >>> entity.save()  # doctest: +SKIP
-        None
-
-        Calling save again returns an error.
-
-        >>> entity = fb.Entity(name="grocerycustomer", serving_names=["GROCERYCUSTOMERGUID"])  # doctest: +SKIP
-        >>> entity.save()  # doctest: +SKIP
-        >>> entity.save()  # doctest: +SKIP
-        Entity (id: <entity.id>) has been saved before.
-        """
-        if self.saved and conflict_resolution == "raise":
-            raise ObjectHasBeenSavedError(
-                f'{type(self).__name__} (id: "{self.id}") has been saved before.'
-            )
-
-        self._pre_save_operations(conflict_resolution=conflict_resolution)
-        client = Configurations().get_client()
-        response = client.post(url=self._route, json=self._get_create_payload())
-        retrieve_object = False
-        if response.status_code != HTTPStatus.CREATED:
-            if response.status_code == HTTPStatus.CONFLICT:
-                if conflict_resolution == "retrieve":
-                    retrieve_object = True
-                else:
-                    raise DuplicatedRecordException(response=response)
-            if not retrieve_object:
-                raise RecordCreationException(response=response)
-
-        if retrieve_object:
-            assert self.name is not None
-            object_dict = self._get_object_dict_by_name(name=self.name)
-        else:
-            object_dict = response.json()
-
-        self._update_cache(object_dict)  # update api object cache store
-        type(self).__init__(
-            self,
-            **object_dict,
-            **self._get_init_params_from_object(),
-        )
-
-
-class DeletableApiObject(ApiObject):
-    """
-    DeleteMixin contains common methods used to delete an object
-    """
-
-    def _delete(self) -> None:
-        client = Configurations().get_client()
-        response = client.delete(url=f"{self._route}/{self.id}")
-        if response.status_code != HTTPStatus.OK:
-            raise RecordDeletionException(response, "Failed to delete the specified object.")

{featurebyte-0.2.2 → featurebyte-0.3.1}/featurebyte/api/api_object_util.py

@@ -1,13 +1,11 @@
 """
 API Object Util
 """
-from typing import Any, Dict, Optional
+from typing import Any, Optional
 
 import ctypes
 import threading
 
-from rich.pretty import pretty_repr
-
 from featurebyte.config import Configurations
 from featurebyte.exception import RecordRetrievalException
 from featurebyte.logging import get_logger
@@ -84,15 +82,6 @@ class ProgressThread(threading.Thread):
                 logger.warning("Exception raise failure")
 
 
-class PrettyDict(Dict[str, Any]):
-    """
-    Dict with prettified representation
-    """
-
-    def __repr__(self) -> str:
-        return pretty_repr(dict(self), expand_all=True, indent_size=2)
-
-
 class NameAttributeUpdatableMixin:
     """
     This mixin is used to handle the case when name of the api object is updatable.

{featurebyte-0.2.2 → featurebyte-0.3.1}/featurebyte/api/asat_aggregator.py

@@ -140,9 +140,6 @@ class AsAtAggregator(BaseAggregator):
         if feature_name is None:
             raise ValueError("feature_name is required")
 
-        if self.category is not None:
-            raise ValueError("category is not supported for aggregate_asat")
-
         view = cast(SCDView, self.view)
         for key in self.keys:
             if key == view.natural_key_column: