Add artifact example recipes

metadata-ingestion/examples/affirm_artifact/3rd-party.dataplatform.json

@@ -0,0 +1,22 @@
+[
+  {
+    "auditHeader": null,
+    "proposedSnapshot": {
+      "com.linkedin.pegasus2avro.metadata.snapshot.DataPlatformSnapshot": {
+        "urn": "urn:li:dataPlatform:affirm3rdParty",
+        "aspects": [
+          {
+            "com.linkedin.pegasus2avro.dataplatform.DataPlatformInfo": {
+              "datasetNameDelimiter": "/",
+              "name": "affirm3rdParty",
+              "displayName": "3rd-party Artifact",
+              "type": "OTHERS",
+              "logoUrl": "https://cdn-assets.affirm.com/images/black_logo-white_bg.jpg"
+            }
+          }
+        ]
+      }
+    },
+    "proposedDelta": null
+  }
+]

metadata-ingestion/examples/affirm_artifact/3rd-party.dataplatform.recipe.yml

@@ -0,0 +1,9 @@
+source:
+    type: file
+    config:
+        filename: './3rd-party.dataplatform.json'
+
+sink:
+    type: 'datahub-rest'
+    config:
+        server: 'http://localhost:8080'

metadata-ingestion/examples/affirm_artifact/3rd-party.recipe.yml

@@ -0,0 +1,12 @@
+source:
+    type: affirm-artifact
+    config:
+        directory: '/Users/tao.sun/tao/datahub-metadata/artifacts/3rd_party'
+        platform: 'affirm3rdParty'
+        platform_instance: ''
+        env: 'PROD'
+
+sink:
+    type: 'datahub-rest'
+    config:
+        server: 'http://localhost:8080'

metadata-ingestion/examples/affirm_artifact/infra.dataplatform.json

@@ -0,0 +1,22 @@
+[
+  {
+    "auditHeader": null,
+    "proposedSnapshot": {
+      "com.linkedin.pegasus2avro.metadata.snapshot.DataPlatformSnapshot": {
+        "urn": "urn:li:dataPlatform:affirmInfra",
+        "aspects": [
+          {
+            "com.linkedin.pegasus2avro.dataplatform.DataPlatformInfo": {
+              "datasetNameDelimiter": "/",
+              "name": "affirmInfra",
+              "displayName": "Infra Artifact",
+              "type": "OTHERS",
+              "logoUrl": "https://cdn-assets.affirm.com/images/black_logo-white_bg.jpg"
+            }
+          }
+        ]
+      }
+    },
+    "proposedDelta": null
+  }
+]

metadata-ingestion/examples/affirm_artifact/infra.dataplatform.recipe.yml

@@ -0,0 +1,9 @@
+source:
+    type: file
+    config:
+        filename: './infra.dataplatform.json'
+
+sink:
+    type: 'datahub-rest'
+    config:
+        server: 'http://localhost:8080'

metadata-ingestion/examples/affirm_artifact/infra.recipe.yml

@@ -0,0 +1,12 @@
+source:
+    type: affirm-artifact
+    config:
+        directory: '/Users/tao.sun/tao/datahub-metadata/artifacts/infra'
+        platform: 'affirmInfra'
+        platform_instance: ''
+        env: 'PROD'
+
+sink:
+    type: 'datahub-rest'
+    config:
+        server: 'http://localhost:8080'

metadata-ingestion/examples/affirm_artifact/unstructured-s3-recipe.yml

@@ -0,0 +1,12 @@
+source:
+    type: affirm-artifact
+    config:
+        directory: '/Users/tao.sun/tao/datahub-metadata/artifacts/unstructured_s3'
+        platform: 's3'
+        platform_instance: 'unstructured_s3'
+        env: 'PROD'
+
+sink:
+    type: 'datahub-rest'
+    config:
+        server: 'http://localhost:8080'

metadata-ingestion/src/datahub/classification/classifier.py

@@ -4,7 +4,6 @@
 from typing import Dict, Set
 
 import pandas as pd
-import spacy
 
 from datahub.classification.privacy.privacy.api import PIIEngine
 from datahub.ingestion.api.common import RecordEnvelope

metadata-ingestion/src/datahub/ingestion/source/affirm/artifact.py

@@ -20,6 +20,7 @@
 
 
 yaml = YAML(typ='rt')
+logging.basicConfig(stream=sys.stderr, level=logging.INFO)
 
 
 @dataclass
@@ -37,6 +38,8 @@ def __post_init__(self):
             self.privacy_entrypoint = ''
         if self.processing_purposes is None:
             self.processing_purposes = []
+        if self.schema_name is None:
+            self.schema_name = ''
 
 
 class AffirmArtifactSourceConfig(ConfigModel):
@@ -45,6 +48,7 @@ class AffirmArtifactSourceConfig(ConfigModel):
     '''
     directory: str
     platform: str
+    platform_instance: str
     env: str
 
 
@@ -106,11 +110,21 @@ def create(cls, config_dict, ctx):
     def get_workunits(self) -> Iterable[MetadataWorkUnit]:
         directory = self.config.directory
         platform = self.config.platform
+        platform_instance = self.config.platform_instance
         env = self.config.env
         for artifact in iterate_artifact(directory):
-            dataset_name = f'{artifact.schema_name}.{artifact.name}' if len(artifact.schema_name) > 0 else artifact.name
+            dataset_name = (
+                f'{artifact.schema_name}.{artifact.name}'
+                if len(artifact.schema_name) > 0
+                else artifact.name
+            )
             logging.info(f'> Processing dataset {dataset_name}')
-            dataset_urn = builder.make_dataset_urn(platform, dataset_name, env)
+            dataset_urn = builder.make_dataset_urn_with_platform_instance(
+                platform=platform,
+                name=dataset_name,
+                platform_instance=platform_instance,
+                env=env
+            )
             dataset_snapshot = DatasetSnapshot(
                 urn=dataset_urn,
                 aspects=[],