Better error messages (#304)

errors: Better errors

docs/examples/featuresets/overview.py

@@ -103,7 +103,7 @@ def e2(cls, ts: pd.Series, durations: pd.Series) -> pd.Series:
         view._get_sync_request_proto()
     assert (
         str(e.value)
-        == "Feature `over_3hrs` is extracted by multiple extractors including `e2`."
+        == "Feature `over_3hrs` is extracted by multiple extractors including `e2` in featureset `Movies`."
     )
 
 

fennel/datasets/datasets.py

@@ -228,7 +228,10 @@ def join(
         within: Tuple[Duration, Duration] = ("forever", "0s"),
     ) -> Join:
         if not isinstance(other, Dataset) and isinstance(other, _Node):
-            raise ValueError("Cannot join with an intermediate dataset")
+            raise ValueError(
+                "Cannot join with an intermediate dataset, i.e something defined inside a pipeline."
+                " Only joining against keyed datasets is permitted."
+            )
         if not isinstance(other, _Node):
             raise TypeError("Cannot join with a non-dataset object")
         return Join(self, other, within, how, on, left_on, right_on)
@@ -622,18 +625,14 @@ def make_types_optional(types: Dict[str, Type]) -> Dict[str, Type]:
         for col in common_cols:
             if self.lsuffix != "" and (col + self.lsuffix) in left_schema:
                 raise ValueError(
-                    "Column name collision. `{}` already exists in schema of left input {}".format(
-                        col + self.lsuffix, left_dsschema.name
-                    )
+                    f"Column name collision. `{col + self.lsuffix}` already exists in schema of left input {left_dsschema.name}, while joining with {self.dataset.dsschema().name}"
                 )
             if (
                 self.rsuffix != ""
                 and (col + self.rsuffix) in right_value_schema
             ):
                 raise ValueError(
-                    "Column name collision. `{}` already exists in schema of right input {}".format(
-                        col + self.rsuffix, self.dataset.dsschema().name
-                    )
+                    f"Column name collision. `{col + self.rsuffix}` already exists in schema of right input {self.dataset.dsschema().name}, while joining with {self.dataset.dsschema().name}"
                 )
             left_dsschema.rename_column(col, col + self.lsuffix)
             left_schema[col + self.lsuffix] = left_schema.pop(col)
@@ -648,9 +647,7 @@ def make_types_optional(types: Dict[str, Type]) -> Dict[str, Type]:
         for col, dtype in right_value_schema.items():
             if col in left_schema:
                 raise ValueError(
-                    "Column name collision. `{}` already exists in schema of left input {}".format(
-                        col, left_dsschema.name
-                    )
+                    f"Column name collision. `{col}` already exists in schema of left input {left_dsschema.name}, while joining with {self.dataset.dsschema().name}"
                 )
             joined_dsschema.append_value_column(col, dtype)
 
@@ -787,7 +784,7 @@ def lookup(
         ) -> Tuple[pd.DataFrame, pd.Series]:
             if len(key_fields) == 0:
                 raise Exception(
-                    f"Trying to lookup dataset `{cls_name} with no keys defined.\n"
+                    f"Trying to lookup dataset `{cls_name}` with no keys defined.\n"
                     f"Please define one or more keys using field(key=True) to perform a lookup."
                 )
             if len(args) > 0:
@@ -1273,7 +1270,7 @@ def _set_timestamp_field(self):
                 if timestamp_field_set:
                     raise ValueError(
                         f"Multiple timestamp fields are not supported in "
-                        f"dataset `{self._name}`."
+                        f"dataset `{self._name}`. Please set one of the datetime fields to be the timestamp field."
                     )
                 timestamp_field_set = True
 
@@ -1292,7 +1289,7 @@ def _set_timestamp_field(self):
             else:
                 raise ValueError(
                     f"Multiple timestamp fields are not "
-                    f"supported in dataset `{self._name}`."
+                    f"supported in dataset `{self._name}`. Please set one of the datetime fields to be the timestamp field."
                 )
         if not timestamp_field_set:
             raise ValueError(

fennel/datasets/test_invalid_dataset.py

@@ -38,7 +38,7 @@ class UserInfoDataset:
     _ = InternalTestClient()
     assert (
         str(e.value) == "Multiple timestamp fields are not supported in "
-        "dataset `UserInfoDataset`."
+        "dataset `UserInfoDataset`. Please set one of the datetime fields to be the timestamp field."
     )
 
 
@@ -647,7 +647,10 @@ def create_pipeline(cls, a: Dataset):
                 b = a.transform(lambda x: x)
                 return a.join(b, how="left", on=["user_id"])  # type: ignore
 
-    assert str(e.value) == "Cannot join with an intermediate dataset"
+    assert (
+        str(e.value)
+        == "Cannot join with an intermediate dataset, i.e something defined inside a pipeline. Only joining against keyed datasets is permitted."
+    )
 
 
 def test_dataset_incorrect_join_bounds():

fennel/featuresets/featureset.py

@@ -178,11 +178,14 @@ def _create_extractor(extractor_func: Callable, version: int):
                 elif hasattr(inp, "__name__"):
                     name = inp.__name__
                 else:
-                    name = str(inp)
+                    if hasattr(inp, "fqn"):
+                        name = inp.fqn()
+                    else:
+                        name = str(inp)
                 raise TypeError(
-                    f"Parameter `{name}` is not a feature of but a "
-                    f"`{type(inp)}`. Please note "
-                    f"that Featuresets are mutable and hence not supported."
+                    f"Parameter `{name}` is not a feature, but a "
+                    f"`{type(inp)}`, and hence not supported as an input for the extractor "
+                    f"`{extractor_name}`"
                 )
             params.append(inp)
         if hasattr(extractor_func, FENNEL_OUTPUTS):
@@ -748,6 +751,6 @@ def sync_validation_for_extractors(extractors: List[Extractor]):
             if feature in extracted_features:
                 raise TypeError(
                     f"Feature `{feature}` is "
-                    f"extracted by multiple extractors including `{extractor.name}`."
+                    f"extracted by multiple extractors including `{extractor.name}` in featureset `{extractor.featureset}`."
                 )
             extracted_features.add(feature)

fennel/featuresets/test_featureset.py

@@ -398,7 +398,7 @@ def get_user_income(cls, ts: pd.Series, user_id: pd.Series):
         view._get_sync_request_proto()
     assert (
         str(e.value)
-        == "Feature `income` is extracted by multiple extractors including `get_user_income`."
+        == "Feature `income` is extracted by multiple extractors including `get_user_income` in featureset `UserInfo`."
     )
 
     sync_request = view._get_sync_request_proto("prod")

fennel/featuresets/test_invalid_derived_extractors.py

@@ -96,7 +96,7 @@ def get_age(cls, ts: pd.Series, user_id: pd.Series):
         view._get_sync_request_proto()
     assert (
         str(e.value)
-        == "Feature `age` is extracted by multiple extractors including `get_age`."
+        == "Feature `age` is extracted by multiple extractors including `get_age` in featureset `UserInfo3`."
     )
 
 

fennel/featuresets/test_invalid_featureset.py

@@ -49,7 +49,7 @@ def get_user_info1(cls, ts: pd.Series, user: pd.Series):
 
     assert (
         str(e.value)
-        == "Parameter `User` is not a feature of but a `<class 'fennel.featuresets.featureset.Featureset'>`. Please note that Featuresets are mutable and hence not supported."
+        == "Parameter `User` is not a feature, but a `<class 'fennel.featuresets.featureset.Featureset'>`, and hence not supported as an input for the extractor `get_user_info1`"
     )
 
 
@@ -90,7 +90,7 @@ def get_user_info3(cls, ts: pd.Series, user_id: pd.Series):
         view._get_sync_request_proto()
     assert (
         str(e.value)
-        == "Feature `gender` is extracted by multiple extractors including `get_user_info3`."
+        == "Feature `gender` is extracted by multiple extractors including `get_user_info3` in featureset `UserInfo`."
     )
 
 

fennel/lib/to_proto/to_proto.py

@@ -202,7 +202,9 @@ def fields_to_dsschema(fields: List[Field]) -> schema_proto.DSSchema:
             keys.append(_field_to_proto(field))
         elif field.timestamp:
             if ts is not None:
-                raise ValueError("Multiple timestamp fields are not supported")
+                raise ValueError(
+                    "Multiple timestamp fields are not supported. Please set one of the datetime fields to be the timestamp field."
+                )
             ts = field.name
         else:
             values.append(_field_to_proto(field))
@@ -1199,6 +1201,7 @@ def to_extractor_pycode(
             f"extractor {extractor.name} has type PyFunc but no function defined"
         )
     dependencies = []
+    extractor_fqn = f"{featureset._name}.{extractor.name}"
     gen_code = ""
     if hasattr(extractor.func, FENNEL_INCLUDED_MOD):
         for f in getattr(extractor.func, FENNEL_INCLUDED_MOD):
@@ -1214,8 +1217,8 @@ def to_extractor_pycode(
     for input in extractor.inputs:
         if not isinstance(input, Feature):
             raise ValueError(
-                f"Extractor {extractor.name} must have inputs "
-                f"of type Feature, but got {type(input)}"
+                f"Extractor `{extractor_fqn}` must have inputs "
+                f"of type Feature, but got `{type(input)}`"
             )
         # Dont add the featureset of the extractor itself
         if input.featureset_name == featureset._name:
@@ -1224,9 +1227,10 @@ def to_extractor_pycode(
             input_fs_added.add(input.featureset_name)
             if input.featureset_name not in fs_obj_map:
                 raise ValueError(
-                    f"Extractor {extractor.name} has an input "
-                    f"feature {input.name} from featureset "
-                    f"{input.featureset_name} which is not synced"
+                    f"Extractor `{extractor_fqn}` has an input "
+                    f"feature `{input.name}` from featureset "
+                    f"`{input.featureset_name}` which is not synced."
+                    f"Please add the featureset to the sync call."
                 )
             gen_code = (
                 gen_code

fennel/test_lib/mock_client.py

@@ -867,7 +867,7 @@ def _check_schema_exceptions(
         if len(exceptions) > 0:
             raise Exception(
                 f"Extractor `{extractor_name}` returned "
-                f"invalid schema: {exceptions}"
+                f"invalid schema for data: {exceptions}"
             )
 
     def _compute_lookup_extractor(

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "fennel-ai"
-version = "0.18.19"
+version = "0.18.20"
 description = "The modern realtime feature engineering platform"
 authors = ["Fennel AI <[email protected]>"]
 packages = [{ include = "fennel" }]