fix(py-sdk): DataJobPatchBuilder handling timestamps, output edges

Author: shirshanka

metadata-ingestion/src/datahub/specific/datajob.py

@@ -102,7 +102,7 @@ def add_input_datajob(self, input: Union[Edge, Urn, str]) -> "DataJobPatchBuilde
 
         Notes:
             If `input` is an Edge object, it is used directly. If `input` is a Urn object or string,
-            it is converted to an Edge object and added with default audit stamps.
+            it is converted to an Edge object and added without any audit stamps.
         """
         if isinstance(input, Edge):
             input_urn: str = input.destinationUrn
@@ -114,8 +114,6 @@ def add_input_datajob(self, input: Union[Edge, Urn, str]) -> "DataJobPatchBuilde
 
             input_edge = Edge(
                 destinationUrn=input_urn,
-                created=self._mint_auditstamp(),
-                lastModified=self._mint_auditstamp(),
             )
 
         self._ensure_urn_type("dataJob", [input_edge], "add_input_datajob")
@@ -185,7 +183,7 @@ def add_input_dataset(self, input: Union[Edge, Urn, str]) -> "DataJobPatchBuilde
 
         Notes:
             If `input` is an Edge object, it is used directly. If `input` is a Urn object or string,
-            it is converted to an Edge object and added with default audit stamps.
+            it is converted to an Edge object and added without any audit stamps.
         """
         if isinstance(input, Edge):
             input_urn: str = input.destinationUrn
@@ -197,8 +195,6 @@ def add_input_dataset(self, input: Union[Edge, Urn, str]) -> "DataJobPatchBuilde
 
             input_edge = Edge(
                 destinationUrn=input_urn,
-                created=self._mint_auditstamp(),
-                lastModified=self._mint_auditstamp(),
             )
 
         self._ensure_urn_type("dataset", [input_edge], "add_input_dataset")
@@ -270,7 +266,7 @@ def add_output_dataset(
 
         Notes:
             If `output` is an Edge object, it is used directly. If `output` is a Urn object or string,
-            it is converted to an Edge object and added with default audit stamps.
+            it is converted to an Edge object and added without any audit stamps.
         """
         if isinstance(output, Edge):
             output_urn: str = output.destinationUrn
@@ -282,15 +278,13 @@ def add_output_dataset(
 
             output_edge = Edge(
                 destinationUrn=output_urn,
-                created=self._mint_auditstamp(),
-                lastModified=self._mint_auditstamp(),
             )
 
         self._ensure_urn_type("dataset", [output_edge], "add_output_dataset")
         self._add_patch(
             DataJobInputOutput.ASPECT_NAME,
             "add",
-            path=f"/outputDatasetEdges/{self.quote(str(output))}",
+            path=f"/outputDatasetEdges/{self.quote(output_urn)}",
             value=output_edge,
         )
         return self

smoke-test/tests/patch/test_datajob_patches.py

@@ -1,5 +1,7 @@
+import time
 import uuid
 
+import datahub.metadata.schema_classes as models
 from datahub.emitter.mce_builder import make_data_job_urn, make_dataset_urn
 from datahub.emitter.mcp import MetadataChangeProposalWrapper
 from datahub.metadata.schema_classes import (
@@ -136,3 +138,95 @@ def test_datajob_inputoutput_dataset_patch(graph_client):
         inputoutput_lineage_read.inputDatasetEdges[0].destinationUrn
         == other_dataset_urn
     )
+
+
+def test_datajob_multiple_inputoutput_dataset_patch(graph_client):
+    """Test creating a data job with multiple input and output datasets and verifying the aspects."""
+    # Create the data job
+    datajob_urn = "urn:li:dataJob:(urn:li:dataFlow:(airflow,training,default),training)"
+
+    # Create input and output dataset URNs
+    input_datasets = ["input_data_1", "input_data_2"]
+    output_datasets = ["output_data_1", "output_data_2"]
+
+    input_dataset_urns = [
+        make_dataset_urn(platform="s3", name=f"test_patch_{dataset}", env="PROD")
+        for dataset in input_datasets
+    ]
+    output_dataset_urns = [
+        make_dataset_urn(platform="s3", name=f"test_patch_{dataset}", env="PROD")
+        for dataset in output_datasets
+    ]
+
+    # Create edges for datasets
+    def make_edge(urn, generate_auditstamp=False):
+        audit_stamp = models.AuditStampClass(
+            time=int(time.time() * 1000.0),
+            actor="urn:li:corpuser:datahub",
+        )
+        return EdgeClass(
+            destinationUrn=str(urn),
+            lastModified=audit_stamp if generate_auditstamp else None,
+        )
+
+    # Initialize empty input/output lineage
+    initial_lineage = DataJobInputOutputClass(
+        inputDatasets=[], outputDatasets=[], inputDatasetEdges=[], outputDatasetEdges=[]
+    )
+
+    # Emit initial lineage
+    mcpw = MetadataChangeProposalWrapper(entityUrn=datajob_urn, aspect=initial_lineage)
+    graph_client.emit_mcp(mcpw)
+
+    # Create patches for input and output datasets
+    patch_builder = DataJobPatchBuilder(datajob_urn)
+    for input_urn in input_dataset_urns:
+        patch_builder.add_input_dataset(make_edge(input_urn))
+    for output_urn in output_dataset_urns:
+        patch_builder.add_output_dataset(make_edge(output_urn))
+
+    # Apply patches
+    for patch_mcp in patch_builder.build():
+        graph_client.emit_mcp(patch_mcp)
+
+    # Verify the lineage was correctly applied
+    lineage_aspect = graph_client.get_aspect(
+        entity_urn=datajob_urn,
+        aspect_type=DataJobInputOutputClass,
+    )
+
+    # Assert lineage was created
+    assert lineage_aspect is not None
+    assert lineage_aspect.inputDatasetEdges is not None
+    assert lineage_aspect.outputDatasetEdges is not None
+
+    # Verify input datasets
+    assert len(lineage_aspect.inputDatasetEdges) == len(input_datasets)
+    input_urns = {edge.destinationUrn for edge in lineage_aspect.inputDatasetEdges}
+    expected_input_urns = {str(urn) for urn in input_dataset_urns}
+    assert input_urns == expected_input_urns
+
+    # Verify output datasets
+    assert len(lineage_aspect.outputDatasetEdges) == len(output_datasets)
+    output_urns = {edge.destinationUrn for edge in lineage_aspect.outputDatasetEdges}
+    expected_output_urns = {str(urn) for urn in output_dataset_urns}
+    assert output_urns == expected_output_urns
+
+    # Test updating the same datasets again (idempotency)
+    patch_builder = DataJobPatchBuilder(datajob_urn)
+    for input_urn in input_dataset_urns:
+        patch_builder.add_input_dataset(make_edge(input_urn))
+    for output_urn in output_dataset_urns:
+        patch_builder.add_output_dataset(make_edge(output_urn))
+
+    for patch_mcp in patch_builder.build():
+        graph_client.emit_mcp(patch_mcp)
+
+    # Verify the aspect hasn't changed
+    updated_lineage_aspect = graph_client.get_aspect(
+        entity_urn=datajob_urn,
+        aspect_type=DataJobInputOutputClass,
+    )
+
+    assert updated_lineage_aspect is not None
+    assert updated_lineage_aspect.to_obj() == lineage_aspect.to_obj()