feat(ingest/looker): include project name in model/explore properties

metadata-ingestion/src/datahub/ingestion/source/looker/looker_common.py

@@ -1207,15 +1207,19 @@ def _to_metadata_events(  # noqa: C901
         dataset_snapshot.aspects.append(browse_paths)
         dataset_snapshot.aspects.append(StatusClass(removed=False))
 
-        custom_properties = {}
-        if self.label is not None:
-            custom_properties["looker.explore.label"] = str(self.label)
-        if self.source_file is not None:
-            custom_properties["looker.explore.file"] = str(self.source_file)
+        custom_properties = {
+            "project": self.project_name,
+            "model": self.model_name,
+            "looker.explore.label": self.label,
+            "looker.explore.name": self.name,
+            "looker.explore.file": self.source_file,
+        }
         dataset_props = DatasetPropertiesClass(
             name=str(self.label) if self.label else LookerUtil._display_name(self.name),
             description=self.description,
-            customProperties=custom_properties,
+            customProperties={
+                k: v for k, v in custom_properties.items() if v is not None
+            },
         )
         dataset_props.externalUrl = self._get_url(base_url)
 

metadata-ingestion/src/datahub/ingestion/source/looker/looker_source.py

@@ -139,26 +139,21 @@ class LookerDashboardSource(TestableSource, StatefulIngestionSourceBase):
     """
 
     platform = "looker"
-    source_config: LookerDashboardSourceConfig
-    reporter: LookerDashboardSourceReport
-    user_registry: LookerUserRegistry
-    reachable_look_registry: Set[
-        str
-    ]  # Keep track of look-id which are reachable from Dashboard
 
     def __init__(self, config: LookerDashboardSourceConfig, ctx: PipelineContext):
         super().__init__(config, ctx)
-        self.source_config = config
-        self.reporter = LookerDashboardSourceReport()
+        self.source_config: LookerDashboardSourceConfig = config
+        self.reporter: LookerDashboardSourceReport = LookerDashboardSourceReport()
         self.looker_api: LookerAPI = LookerAPI(self.source_config)
-        self.user_registry = LookerUserRegistry(self.looker_api)
-        self.explore_registry = LookerExploreRegistry(
+        self.user_registry: LookerUserRegistry = LookerUserRegistry(self.looker_api)
+        self.explore_registry: LookerExploreRegistry = LookerExploreRegistry(
             self.looker_api, self.reporter, self.source_config
         )
         self.reporter._looker_explore_registry = self.explore_registry
         self.reporter._looker_api = self.looker_api
 
-        self.reachable_look_registry = set()
+        # Keep track of look-id which are reachable from Dashboard
+        self.reachable_look_registry: Set[str] = set()
 
         # (model, explore) -> list of charts/looks/dashboards that reference this explore
         # The list values are used purely for debugging purposes.
@@ -868,21 +863,31 @@ def _make_explore_metadata_events(
     ) -> Iterable[
         Union[MetadataChangeEvent, MetadataChangeProposalWrapper, MetadataWorkUnit]
     ]:
-        if self.source_config.emit_used_explores_only:
-            explores_to_fetch = list(self.reachable_explores.keys())
-        else:
+        if not self.source_config.emit_used_explores_only:
             explores_to_fetch = list(self.list_all_explores())
+        else:
+            # We don't keep track of project names for each explore right now.
+            # Because project names are just used for a custom property, it's
+            # fine to set them to None.
+            # TODO: Track project names for each explore.
+            explores_to_fetch = [
+                (None, model, explore)
+                for (model, explore) in self.reachable_explores.keys()
+            ]
         explores_to_fetch.sort()
 
         processed_models: List[str] = []
 
-        for model, _ in explores_to_fetch:
+        for project_name, model, _ in explores_to_fetch:
             if model not in processed_models:
                 model_key = gen_model_key(self.source_config, model)
                 yield from gen_containers(
                     container_key=model_key,
                     name=model,
                     sub_types=[BIContainerSubTypes.LOOKML_MODEL],
+                    extra_properties=(
+                        {"project": project_name} if project_name is not None else None
+                    ),
                 )
                 yield MetadataChangeProposalWrapper(
                     entityUrn=model_key.as_urn(),
@@ -896,7 +901,7 @@ def _make_explore_metadata_events(
         self.reporter.total_explores = len(explores_to_fetch)
         for future in BackpressureAwareExecutor.map(
             self.fetch_one_explore,
-            ((model, explore) for (model, explore) in explores_to_fetch),
+            ((model, explore) for (_project, model, explore) in explores_to_fetch),
             max_workers=self.source_config.max_threads,
         ):
             events, explore_id, start_time, end_time = future.result()
@@ -907,7 +912,7 @@ def _make_explore_metadata_events(
                 f"Running time of fetch_one_explore for {explore_id}: {(end_time - start_time).total_seconds()}"
             )
 
-    def list_all_explores(self) -> Iterable[Tuple[str, str]]:
+    def list_all_explores(self) -> Iterable[Tuple[Optional[str], str, str]]:
         # returns a list of (model, explore) tuples
 
         for model in self.looker_api.all_lookml_models():
@@ -916,7 +921,7 @@ def list_all_explores(self) -> Iterable[Tuple[str, str]]:
             for explore in model.explores:
                 if explore.name is None:
                     continue
-                yield (model.name, explore.name)
+                yield (model.project_name, model.name, explore.name)
 
     def fetch_one_explore(
         self, model: str, explore: str

metadata-ingestion/tests/integration/looker/golden_looker_mces.json

@@ -11,6 +11,7 @@
                         "description": "lorem ipsum",
                         "charts": [],
                         "datasets": [],
+                        "dashboards": [],
                         "lastModified": {
                             "created": {
                                 "time": 1586847600000,
@@ -440,7 +441,10 @@
                 {
                     "com.linkedin.pegasus2avro.dataset.DatasetProperties": {
                         "customProperties": {
+                            "project": "lkml_samples",
+                            "model": "bogus data",
                             "looker.explore.label": "My Explore View",
+                            "looker.explore.name": "my_view",
                             "looker.explore.file": "test_source_file.lkml"
                         },
                         "externalUrl": "https://looker.company.com/explore/bogus data/my_view",
@@ -616,7 +620,10 @@
                 {
                     "com.linkedin.pegasus2avro.dataset.DatasetProperties": {
                         "customProperties": {
+                            "project": "lkml_samples",
+                            "model": "data",
                             "looker.explore.label": "My Explore View",
+                            "looker.explore.name": "my_view",
                             "looker.explore.file": "test_source_file.lkml"
                         },
                         "externalUrl": "https://looker.company.com/explore/data/my_view",

metadata-ingestion/tests/integration/looker/golden_test_allow_ingest.json

@@ -11,6 +11,7 @@
                         "description": "lorem ipsum",
                         "charts": [],
                         "datasets": [],
+                        "dashboards": [],
                         "lastModified": {
                             "created": {
                                 "time": 1586847600000,
@@ -282,7 +283,10 @@
                 {
                     "com.linkedin.pegasus2avro.dataset.DatasetProperties": {
                         "customProperties": {
+                            "project": "lkml_samples",
+                            "model": "data",
                             "looker.explore.label": "My Explore View",
+                            "looker.explore.name": "my_view",
                             "looker.explore.file": "test_source_file.lkml"
                         },
                         "externalUrl": "https://looker.company.com/explore/data/my_view",

metadata-ingestion/tests/integration/looker/golden_test_external_project_view_mces.json

@@ -202,6 +202,7 @@
                             "urn:li:chart:(looker,dashboard_elements.2)"
                         ],
                         "datasets": [],
+                        "dashboards": [],
                         "lastModified": {
                             "created": {
                                 "time": 1586847600000,
@@ -520,7 +521,10 @@
                 {
                     "com.linkedin.pegasus2avro.dataset.DatasetProperties": {
                         "customProperties": {
+                            "project": "looker_hub",
+                            "model": "data",
                             "looker.explore.label": "My Explore View",
+                            "looker.explore.name": "my_view",
                             "looker.explore.file": "test_source_file.lkml"
                         },
                         "externalUrl": "https://looker.company.com/explore/data/my_view",

metadata-ingestion/tests/integration/looker/golden_test_file_path_ingest.json

@@ -202,6 +202,7 @@
                             "urn:li:chart:(looker,dashboard_elements.2)"
                         ],
                         "datasets": [],
+                        "dashboards": [],
                         "lastModified": {
                             "created": {
                                 "time": 1586847600000,
@@ -520,7 +521,10 @@
                 {
                     "com.linkedin.pegasus2avro.dataset.DatasetProperties": {
                         "customProperties": {
+                            "project": "looker_hub",
+                            "model": "data",
                             "looker.explore.label": "My Explore View",
+                            "looker.explore.name": "my_view",
                             "looker.explore.file": "test_source_file.lkml"
                         },
                         "externalUrl": "https://looker.company.com/explore/data/my_view",

metadata-ingestion/tests/integration/looker/golden_test_folder_path_pattern_ingest.json

@@ -287,6 +287,7 @@
                         "description": "third",
                         "charts": [],
                         "datasets": [],
+                        "dashboards": [],
                         "lastModified": {
                             "created": {
                                 "time": 1586847600000,
@@ -613,7 +614,10 @@
                 {
                     "com.linkedin.pegasus2avro.dataset.DatasetProperties": {
                         "customProperties": {
+                            "project": "lkml_samples",
+                            "model": "data",
                             "looker.explore.label": "My Explore View",
+                            "looker.explore.name": "my_view",
                             "looker.explore.file": "test_source_file.lkml"
                         },
                         "externalUrl": "https://looker.company.com/explore/data/my_view",