feat(sdk): auto-fix bad entity type casing (#13218)

metadata-ingestion/src/datahub/emitter/mcp.py

@@ -1,9 +1,11 @@
 import dataclasses
 import json
+import warnings
 from typing import TYPE_CHECKING, Dict, List, Optional, Sequence, Tuple, Union
 
 from datahub.emitter.aspect import ASPECT_MAP, JSON_CONTENT_TYPE
 from datahub.emitter.serialization_helper import post_json_transform, pre_json_transform
+from datahub.errors import DataHubDeprecationWarning
 from datahub.metadata.schema_classes import (
     ChangeTypeClass,
     DictWrapper,
@@ -75,13 +77,22 @@ class MetadataChangeProposalWrapper:
         if self.entityUrn and self.entityType == _ENTITY_TYPE_UNSET:
             self.entityType = guess_entity_type(self.entityUrn)
         elif self.entityUrn and self.entityType:
-            guessed_entity_type = guess_entity_type(self.entityUrn).lower()
+            guessed_entity_type = guess_entity_type(self.entityUrn)
-            # Entity type checking is actually case insensitive.
+            if self.entityType.lower() != guessed_entity_type.lower():
-            # Note that urns are case sensitive, but entity types are not.
+                # If they aren't a case-ignored match, raise an error.
-            if self.entityType.lower() != guessed_entity_type:
                 raise ValueError(
                     f"entityType {self.entityType} does not match the entity type {guessed_entity_type} from entityUrn {self.entityUrn}",
                 )
+            elif self.entityType != guessed_entity_type:
+                # If they only differ in case, normalize and print a warning.
+                self.entityType = guessed_entity_type
+                warnings.warn(
+                    f"The passed entityType {self.entityType} differs in case from the expected entity type {guessed_entity_type}. "
+                    "This will be automatically corrected for now, but will become an error in a future release. "
+                    "Note that the entityType field is optional and will be automatically inferred from the entityUrn.",
+                    DataHubDeprecationWarning,
+                    stacklevel=3,
+                )
         elif self.entityType == _ENTITY_TYPE_UNSET:
             raise ValueError("entityType must be set if entityUrn is not set")
 

metadata-ingestion/src/datahub/errors.py

@@ -41,3 +41,7 @@ class ExperimentalWarning(Warning):
 
 class APITracingWarning(Warning):
     pass
+
+
+class DataHubDeprecationWarning(DeprecationWarning):
+    pass

metadata-ingestion/tests/unit/sdk/test_mcp_wrapper.py

@@ -2,9 +2,10 @@ import pytest
 
 import datahub.metadata.schema_classes as models
 from datahub.emitter.mcp import MetadataChangeProposalWrapper
+from datahub.errors import DataHubDeprecationWarning
 
 
-def test_mcpw_inference():
+def test_mcpw_inference() -> None:
     mcpw = MetadataChangeProposalWrapper(
         entityUrn="urn:li:dataset:(urn:li:dataPlatform:bigquery,harshal-playground-306419.test_schema.excess_deaths_derived,PROD)",
         aspect=models.DomainsClass(domains=["urn:li:domain:health"]),
@@ -20,7 +21,17 @@ def test_mcpw_inference():
         )
 
 
-def test_mcpw_from_obj():
+def test_mcpw_case_coercion() -> None:
+    with pytest.warns(DataHubDeprecationWarning):
+        mcpw = MetadataChangeProposalWrapper(
+            entityUrn="urn:li:dataset:(urn:li:dataPlatform:bigquery,harshal-playground-306419.test_schema.excess_deaths_derived,PROD)",
+            entityType="DATASET",
+            aspect=models.DomainsClass(domains=["urn:li:domain:health"]),
+        )
+    assert mcpw.entityType == "dataset"
+
+
+def test_mcpw_from_obj() -> None:
     # Checks that the MCPW from_obj() method returns a MCPW instead
     # of an MCP with a serialized inner aspect object.