acryl-datahub 1.2.0.3rc2__py3-none-any.whl → 1.2.0.4rc2__py3-none-any.whl

datahub/ingestion/source/unity/source.py

@@ -7,12 +7,14 @@ from urllib.parse import urljoin
 from datahub.api.entities.external.external_entities import PlatformResourceRepository
 from datahub.api.entities.external.unity_catalog_external_entites import UnityCatalogTag
 from datahub.emitter.mce_builder import (
+    UNKNOWN_USER,
     make_data_platform_urn,
     make_dataplatform_instance_urn,
     make_dataset_urn_with_platform_instance,
     make_domain_urn,
     make_group_urn,
     make_schema_field_urn,
+    make_ts_millis,
     make_user_urn,
 )
 from datahub.emitter.mcp import MetadataChangeProposalWrapper
@@ -111,6 +113,7 @@ from datahub.metadata.com.linkedin.pegasus2avro.dataset import (
     ViewProperties,
 )
 from datahub.metadata.schema_classes import (
+    AuditStampClass,
     BrowsePathsClass,
     DataPlatformInstanceClass,
     DatasetLineageTypeClass,
@@ -203,6 +206,7 @@ class UnityCatalogSource(StatefulIngestionSourceBase, TestableSource):
             config.warehouse_id,
             report=self.report,
             hive_metastore_proxy=self.hive_metastore_proxy,
+            lineage_data_source=config.lineage_data_source,
         )
 
         self.external_url_base = urljoin(self.config.workspace_url, "/explore/data")
@@ -410,12 +414,12 @@ class UnityCatalogSource(StatefulIngestionSourceBase, TestableSource):
                         self.config.workspace_url, f"#notebook/{notebook.id}"
                     ),
                     created=(
-                        TimeStampClass(int(notebook.created_at.timestamp() * 1000))
+                        TimeStampClass(make_ts_millis(notebook.created_at))
                         if notebook.created_at
                         else None
                     ),
                     lastModified=(
-                        TimeStampClass(int(notebook.modified_at.timestamp() * 1000))
+                        TimeStampClass(make_ts_millis(notebook.modified_at))
                         if notebook.modified_at
                         else None
                     ),
@@ -434,17 +438,20 @@ class UnityCatalogSource(StatefulIngestionSourceBase, TestableSource):
         if not notebook.upstreams:
             return None
 
+        upstreams = []
+        for upstream_ref in notebook.upstreams:
+            timestamp = make_ts_millis(upstream_ref.last_updated)
+            upstreams.append(
+                self._create_upstream_class(
+                    self.gen_dataset_urn(upstream_ref),
+                    DatasetLineageTypeClass.COPY,
+                    timestamp,
+                )
+            )
+
         return MetadataChangeProposalWrapper(
             entityUrn=self.gen_notebook_urn(notebook),
-            aspect=UpstreamLineageClass(
-                upstreams=[
-                    UpstreamClass(
-                        dataset=self.gen_dataset_urn(upstream_ref),
-                        type=DatasetLineageTypeClass.COPY,
-                    )
-                    for upstream_ref in notebook.upstreams
-                ]
-            ),
+            aspect=UpstreamLineageClass(upstreams=upstreams),
         ).as_workunit()
 
     def process_metastores(self) -> Iterable[MetadataWorkUnit]:
@@ -463,14 +470,15 @@ class UnityCatalogSource(StatefulIngestionSourceBase, TestableSource):
         self, metastore: Optional[Metastore]
     ) -> Iterable[MetadataWorkUnit]:
         for catalog in self._get_catalogs(metastore):
-            if not self.config.catalog_pattern.allowed(catalog.id):
-                self.report.catalogs.dropped(catalog.id)
-                continue
+            with self.report.new_stage(f"Ingest catalog {catalog.id}"):
+                if not self.config.catalog_pattern.allowed(catalog.id):
+                    self.report.catalogs.dropped(catalog.id)
+                    continue
 
-            yield from self.gen_catalog_containers(catalog)
-            yield from self.process_schemas(catalog)
+                yield from self.gen_catalog_containers(catalog)
+                yield from self.process_schemas(catalog)
 
-            self.report.catalogs.processed(catalog.id)
+                self.report.catalogs.processed(catalog.id)
 
     def _get_catalogs(self, metastore: Optional[Metastore]) -> Iterable[Catalog]:
         if self.config.catalogs:
@@ -647,9 +655,21 @@ class UnityCatalogSource(StatefulIngestionSourceBase, TestableSource):
         ]
 
     def ingest_lineage(self, table: Table) -> Optional[UpstreamLineageClass]:
+        # Calculate datetime filters for lineage
+        lineage_start_time = None
+        lineage_end_time = self.config.end_time
+
+        if self.config.ignore_start_time_lineage:
+            lineage_start_time = None  # Ignore start time to get all lineage
+        else:
+            lineage_start_time = self.config.start_time
+
         if self.config.include_table_lineage:
             self.unity_catalog_api_proxy.table_lineage(
-                table, include_entity_lineage=self.config.include_notebooks
+                table,
+                include_entity_lineage=self.config.include_notebooks,
+                start_time=lineage_start_time,
+                end_time=lineage_end_time,
             )
 
         if self.config.include_column_lineage and table.upstreams:
@@ -661,7 +681,11 @@ class UnityCatalogSource(StatefulIngestionSourceBase, TestableSource):
                 for column in table.columns[: self.config.column_lineage_column_limit]
             ]
             self.unity_catalog_api_proxy.get_column_lineage(
-                table, column_names, max_workers=self.config.lineage_max_workers
+                table,
+                column_names,
+                max_workers=self.config.lineage_max_workers,
+                start_time=lineage_start_time,
+                end_time=lineage_end_time,
             )
 
         return self._generate_lineage_aspect(self.gen_dataset_urn(table.ref), table)
@@ -690,18 +714,22 @@ class UnityCatalogSource(StatefulIngestionSourceBase, TestableSource):
                 for d_col, u_cols in sorted(downstream_to_upstream_cols.items())
             )
 
+            timestamp = make_ts_millis(upstream_ref.last_updated)
             upstreams.append(
-                UpstreamClass(
-                    dataset=upstream_urn,
-                    type=DatasetLineageTypeClass.TRANSFORMED,
+                self._create_upstream_class(
+                    upstream_urn,
+                    DatasetLineageTypeClass.TRANSFORMED,
+                    timestamp,
                 )
             )
 
-        for notebook in table.upstream_notebooks:
+        for notebook in table.upstream_notebooks.values():
+            timestamp = make_ts_millis(notebook.last_updated)
             upstreams.append(
-                UpstreamClass(
-                    dataset=self.gen_notebook_urn(notebook),
-                    type=DatasetLineageTypeClass.TRANSFORMED,
+                self._create_upstream_class(
+                    self.gen_notebook_urn(notebook.id),
+                    DatasetLineageTypeClass.TRANSFORMED,
+                    timestamp,
                 )
             )
 
@@ -771,6 +799,31 @@ class UnityCatalogSource(StatefulIngestionSourceBase, TestableSource):
             instance=self.config.platform_instance,
         ).as_urn()
 
+    def _create_upstream_class(
+        self,
+        dataset_urn: str,
+        lineage_type: Union[str, DatasetLineageTypeClass],
+        timestamp: Optional[int],
+    ) -> UpstreamClass:
+        """
+        Helper method to create UpstreamClass with optional audit stamp.
+        If timestamp is None, audit stamp is omitted.
+        """
+        if timestamp is not None:
+            return UpstreamClass(
+                dataset=dataset_urn,
+                type=lineage_type,
+                auditStamp=AuditStampClass(
+                    time=timestamp,
+                    actor=UNKNOWN_USER,
+                ),
+            )
+        else:
+            return UpstreamClass(
+                dataset=dataset_urn,
+                type=lineage_type,
+            )
+
     def gen_schema_containers(self, schema: Schema) -> Iterable[MetadataWorkUnit]:
         domain_urn = self._gen_domain_urn(f"{schema.catalog.name}.{schema.name}")
         schema_tags = []
@@ -961,16 +1014,20 @@ class UnityCatalogSource(StatefulIngestionSourceBase, TestableSource):
         created: Optional[TimeStampClass] = None
         if table.created_at:
             custom_properties["created_at"] = str(table.created_at)
-            created = TimeStampClass(
-                int(table.created_at.timestamp() * 1000),
-                make_user_urn(table.created_by) if table.created_by else None,
-            )
+            created_ts = make_ts_millis(table.created_at)
+            if created_ts is not None:
+                created = TimeStampClass(
+                    created_ts,
+                    make_user_urn(table.created_by) if table.created_by else None,
+                )
         last_modified = created
         if table.updated_at:
-            last_modified = TimeStampClass(
-                int(table.updated_at.timestamp() * 1000),
-                table.updated_by and make_user_urn(table.updated_by),
-            )
+            updated_ts = make_ts_millis(table.updated_at)
+            if updated_ts is not None:
+                last_modified = TimeStampClass(
+                    updated_ts,
+                    table.updated_by and make_user_urn(table.updated_by),
+                )
 
         return DatasetPropertiesClass(
             name=table.name,

datahub/ingestion/source/unity/tag_entities.py

@@ -36,7 +36,7 @@ class UnityCatalogTagPlatformResourceId(BaseModel, ExternalEntityId):
 
     tag_key: str
     tag_value: Optional[str] = None
-    platform_instance: Optional[str]
+    platform_instance: Optional[str] = None
     exists_in_unity_catalog: bool = False
     persisted: bool = False
 
@@ -77,13 +77,13 @@ class UnityCatalogTagPlatformResourceId(BaseModel, ExternalEntityId):
         )
         if existing_platform_resource:
             logger.info(
-                f"Found existing UnityCatalogTagPlatformResourceId for tag {tag.key.original}: {existing_platform_resource}"
+                f"Found existing UnityCatalogTagPlatformResourceId for tag {tag.key.raw_text}: {existing_platform_resource}"
             )
             return existing_platform_resource
 
         return UnityCatalogTagPlatformResourceId(
-            tag_key=tag.key.original,
-            tag_value=tag.value.original if tag.value is not None else None,
+            tag_key=tag.key.raw_text,
+            tag_value=tag.value.raw_text if tag.value is not None else None,
             platform_instance=platform_instance,
             exists_in_unity_catalog=exists_in_unity_catalog,
             persisted=False,
@@ -218,7 +218,7 @@ class UnityCatalogTagPlatformResource(BaseModel, ExternalEntity):
     datahub_urns: LinkedResourceSet
     managed_by_datahub: bool
     id: UnityCatalogTagPlatformResourceId
-    allowed_values: Optional[List[str]]
+    allowed_values: Optional[List[str]] = None
 
     def get_id(self) -> ExternalEntityId:
         return self.id

datahub/ingestion/source/usage/starburst_trino_usage.py

@@ -60,7 +60,7 @@ AggregatedDataset = GenericAggregatedDataset[TrinoTableRef]
 
 class TrinoConnectorInfo(BaseModel):
     partitionIds: List[str]
-    truncated: Optional[bool]
+    truncated: Optional[bool] = None
 
 
 class TrinoAccessedMetadata(BaseModel):
@@ -80,7 +80,7 @@ class TrinoJoinedAccessEvent(BaseModel):
     table: Optional[str] = None
     accessed_metadata: List[TrinoAccessedMetadata]
     starttime: datetime = Field(alias="create_time")
-    endtime: Optional[datetime] = Field(alias="end_time")
+    endtime: Optional[datetime] = Field(None, alias="end_time")
 
 
 class EnvBasedSourceBaseConfig:

datahub/sdk/entity_client.py

@@ -1,11 +1,12 @@
 from __future__ import annotations
 
 import warnings
-from typing import TYPE_CHECKING, Union, overload
+from typing import TYPE_CHECKING, Optional, Union, overload
 
 import datahub.metadata.schema_classes as models
 from datahub.emitter.mcp import MetadataChangeProposalWrapper
 from datahub.emitter.mcp_patch_builder import MetadataPatchProposal
+from datahub.emitter.rest_emitter import EmitMode
 from datahub.errors import IngestionAttributionWarning, ItemNotFoundError, SdkUsageError
 from datahub.ingestion.graph.client import DataHubGraph
 from datahub.metadata.urns import (
@@ -133,7 +134,7 @@ class EntityClient:
 
         return entity
 
-    def create(self, entity: Entity) -> None:
+    def create(self, entity: Entity, *, emit_mode: Optional[EmitMode] = None) -> None:
         mcps = []
 
         if self._graph.exists(str(entity.urn)):
@@ -152,9 +153,12 @@ class EntityClient:
         )
         mcps.extend(entity.as_mcps(models.ChangeTypeClass.CREATE))
 
-        self._graph.emit_mcps(mcps)
+        if emit_mode:
+            self._graph.emit_mcps(mcps, emit_mode=emit_mode)
+        else:
+            self._graph.emit_mcps(mcps)
 
-    def upsert(self, entity: Entity) -> None:
+    def upsert(self, entity: Entity, *, emit_mode: Optional[EmitMode] = None) -> None:
         if entity._prev_aspects is None and self._graph.exists(str(entity.urn)):
             warnings.warn(
                 f"The entity {entity.urn} already exists. This operation will partially overwrite the existing entity.",
@@ -164,9 +168,17 @@ class EntityClient:
             # TODO: If there are no previous aspects but the entity exists, should we delete aspects that are not present here?
 
         mcps = entity.as_mcps(models.ChangeTypeClass.UPSERT)
-        self._graph.emit_mcps(mcps)
+        if emit_mode:
+            self._graph.emit_mcps(mcps, emit_mode=emit_mode)
+        else:
+            self._graph.emit_mcps(mcps)
 
-    def update(self, entity: Union[Entity, MetadataPatchProposal]) -> None:
+    def update(
+        self,
+        entity: Union[Entity, MetadataPatchProposal],
+        *,
+        emit_mode: Optional[EmitMode] = None,
+    ) -> None:
         if isinstance(entity, MetadataPatchProposal):
             return self._update_patch(entity)
 
@@ -179,7 +191,10 @@ class EntityClient:
         # -> probably add a "mode" parameter that can be "update" (e.g. if not modified) or "update_force"
 
         mcps = entity.as_mcps(models.ChangeTypeClass.UPSERT)
-        self._graph.emit_mcps(mcps)
+        if emit_mode:
+            self._graph.emit_mcps(mcps, emit_mode=emit_mode)
+        else:
+            self._graph.emit_mcps(mcps)
 
     def _update_patch(
         self, updater: MetadataPatchProposal, check_exists: bool = True

datahub/utilities/mapping.py

@@ -83,7 +83,7 @@ class Constants:
     MATCH = "match"
     USER_OWNER = "user"
     GROUP_OWNER = "group"
-    OPERAND_DATATYPE_SUPPORTED = [int, bool, str, float]
+    OPERAND_DATATYPE_SUPPORTED = [int, bool, str, float, list]
     TAG_PARTITION_KEY = "PARTITION_KEY"
     TAG_DIST_KEY = "DIST_KEY"
     TAG_SORT_KEY = "SORT_KEY"
@@ -455,7 +455,34 @@ class OperationProcessor:
         # function to check if a match clause is satisfied to a value.
         if not any(
             isinstance(raw_props_value, t) for t in Constants.OPERAND_DATATYPE_SUPPORTED
-        ) or not isinstance(raw_props_value, type(match_clause)):
+        ):
+            return None
+
+        # Handle list values by checking if any item in the list matches
+        if isinstance(raw_props_value, list):
+            # For lists, we need to find at least one matching item
+            # Return a match with the concatenated values of all matching items
+            matching_items = []
+            for item in raw_props_value:
+                if isinstance(item, str):
+                    match = re.match(match_clause, item)
+                    if match:
+                        matching_items.append(item)
+                elif isinstance(match_clause, type(item)):
+                    match = re.match(str(match_clause), str(item))
+                    if match:
+                        matching_items.append(str(item))
+
+            if matching_items:
+                # Create a synthetic match object with all matching items joined
+                combined_value = ",".join(matching_items)
+                return re.match(
+                    ".*", combined_value
+                )  # Always matches, returns combined value
+            return None
+
+        # Handle scalar values (existing logic)
+        elif not isinstance(raw_props_value, type(match_clause)):
             return None
         elif isinstance(raw_props_value, str):
             return re.match(match_clause, raw_props_value)