minor changes + change deprecated fields in examples

docs/how/add-custom-data-platform.md

@@ -77,7 +77,7 @@ datahub put platform --name MyCustomDataPlatform --display_name "My Custom Data
 source:
   type: "file"
   config:
-    filename: "./my-custom-data-platform.json"
+    path: "./my-custom-data-platform.json"
 
 # see https://datahubproject.io/docs/metadata-ingestion/sink_docs/datahub for complete documentation
 sink:

docs/how/add-user-data.md

@@ -57,7 +57,7 @@ Define an [ingestion recipe](https://datahubproject.io/docs/metadata-ingestion/#
 source:
   type: "file"
   config:
-    filename: "./my-user.json"
+    path: "./my-user.json"
 
 # see https://datahubproject.io/docs/metadata-ingestion/sink_docs/datahub for complete documentation
 sink:

docs/ownership/ownership-types.md

@@ -85,7 +85,7 @@ source:
   type: "file"
   config:
     # path to json file
-    filename: "metadata-ingestion/examples/ownership/ownership_type.json"
+    path: "metadata-ingestion/examples/ownership/ownership_type.json"
 
 # see https://datahubproject.io/docs/metadata-ingestion/sink_docs/datahub for complete documentation
 sink:

metadata-ingestion/src/datahub/ingestion/transformer/extract_ownership_from_tags.py

@@ -50,23 +50,25 @@ def create(
         config = ExtractOwnersFromTagsConfig.parse_obj(config_dict)
         return cls(config, ctx)
 
+    def get_owner_urn(self, owner_str: str):
+        if self.config.email_domain is not None:
+            return owner_str + "@" + self.config.email_domain
+        return owner_str
+
     def transform_aspect(
         self, entity_urn: str, aspect_name: str, aspect: Optional[Aspect]
     ) -> Optional[Aspect]:
         in_tags_aspect: Optional[GlobalTagsClass] = cast(GlobalTagsClass, aspect)
         if in_tags_aspect is None:
             return None
-        tags_str = in_tags_aspect.tags
+        tags = in_tags_aspect.tags
         owners: List[OwnerClass] = []
-        for tag in tags_str:
-            tag_urn = TagUrn.create_from_string(tag)
+        for tag_class in tags:
+            tag_urn = TagUrn.create_from_string(tag_class.tag)
             tag_str = tag_urn.get_entity_id()[0]
             if tag_str.startswith(self.config.tag_prefix):
                 owner_str = tag_str[len(self.config.tag_prefix) :]
-                if self.config.email_domain is not None:
-                    owner_urn_str = owner_str + "@" + self.config.email_domain
-                else:
-                    owner_urn_str = owner_str
+                owner_urn_str = self.get_owner_urn(owner_str)
                 if self.config.is_user:
                     owner_urn = CorpuserUrn.create_from_id(owner_urn_str)
                 else:

metadata-ingestion/tests/unit/test_transform_dataset.py

@@ -597,7 +597,13 @@ def _test_owner(
         expected_owner: str,
         expected_owner_type: Optional[OwnershipTypeClass] = None,
     ) -> None:
-        dataset = make_generic_dataset(aspects=[models.GlobalTagsClass(tags=[tag])])
+        dataset = make_generic_dataset(
+            aspects=[
+                models.GlobalTagsClass(
+                    tags=[TagAssociationClass(tag=builder.make_tag_urn(tag))]
+                )
+            ]
+        )
         transformer = ExtractOwnersFromTagsTransformer.create(
             config,
             PipelineContext(run_id="test"),
@@ -617,30 +623,30 @@ def _test_owner(
         assert owner.owner == expected_owner
 
     _test_owner(
-        tag="urn:li:tag:owner:foo",
+        tag="owner:foo",
         config={
             "tag_prefix": "owner:",
         },
         expected_owner="urn:li:corpuser:foo",
     )
     _test_owner(
-        tag="urn:li:tag:owner:foo",
+        tag="owner:foo",
         config={
             "tag_prefix": "owner:",
             "is_user": False,
         },
         expected_owner="urn:li:corpGroup:foo",
     )
     _test_owner(
-        tag="urn:li:tag:owner:foo",
+        tag="owner:foo",
         config={
             "tag_prefix": "owner:",
             "email_domain": "example.com",
         },
         expected_owner="urn:li:corpuser:[email protected]",
     )
     _test_owner(
-        tag="urn:li:tag:owner:foo",
+        tag="owner:foo",
         config={
             "tag_prefix": "owner:",
             "email_domain": "example.com",
@@ -650,7 +656,7 @@ def _test_owner(
         expected_owner_type=OwnershipTypeClass.TECHNICAL_OWNER,
     )
     _test_owner(
-        tag="urn:li:tag:owner:foo",
+        tag="owner:foo",
         config={
             "tag_prefix": "owner:",
             "email_domain": "example.com",