openmetadata-ingestion 1.3.2.0rc2__py3-none-any.whl → 1.3.3.0rc1__py3-none-any.whl

metadata/ingestion/api/parser.py

@@ -69,6 +69,28 @@ from metadata.generated.schema.metadataIngestion.databaseServiceQueryUsagePipeli
     DatabaseServiceQueryUsagePipeline,
     DatabaseUsageConfigType,
 )
+from metadata.generated.schema.metadataIngestion.dbtconfig.dbtAzureConfig import (
+    DbtAzureConfig,
+)
+from metadata.generated.schema.metadataIngestion.dbtconfig.dbtCloudConfig import (
+    DbtCloudConfig,
+)
+from metadata.generated.schema.metadataIngestion.dbtconfig.dbtGCSConfig import (
+    DbtGcsConfig,
+)
+from metadata.generated.schema.metadataIngestion.dbtconfig.dbtHttpConfig import (
+    DbtHttpConfig,
+)
+from metadata.generated.schema.metadataIngestion.dbtconfig.dbtLocalConfig import (
+    DbtLocalConfig,
+)
+from metadata.generated.schema.metadataIngestion.dbtconfig.dbtS3Config import (
+    DbtS3Config,
+)
+from metadata.generated.schema.metadataIngestion.dbtPipeline import (
+    DbtConfigType,
+    DbtPipeline,
+)
 from metadata.generated.schema.metadataIngestion.messagingServiceMetadataPipeline import (
     MessagingMetadataConfigType,
     MessagingServiceMetadataPipeline,
@@ -125,6 +147,16 @@ SOURCE_CONFIG_CLASS_MAP = {
     DatabaseMetadataConfigType.DatabaseMetadata.value: DatabaseServiceMetadataPipeline,
     StorageMetadataConfigType.StorageMetadata.value: StorageServiceMetadataPipeline,
     SearchMetadataConfigType.SearchMetadata.value: SearchServiceMetadataPipeline,
+    DbtConfigType.DBT.value: DbtPipeline,
+}
+
+DBT_CONFIG_TYPE_MAP = {
+    "cloud": DbtCloudConfig,
+    "local": DbtLocalConfig,
+    "http": DbtHttpConfig,
+    "s3": DbtS3Config,
+    "gcs": DbtGcsConfig,
+    "azure": DbtAzureConfig,
 }
 
 
@@ -171,6 +203,7 @@ def get_source_config_class(
     Type[PipelineServiceMetadataPipeline],
     Type[MlModelServiceMetadataPipeline],
     Type[DatabaseServiceMetadataPipeline],
+    Type[DbtPipeline],
 ]:
     """
     Return the source config type for a source string
@@ -179,7 +212,7 @@ def get_source_config_class(
     """
     source_config_class = SOURCE_CONFIG_CLASS_MAP.get(source_config_type)
 
-    if source_config_type:
+    if source_config_class:
         return source_config_class
 
     raise ValueError(f"Cannot find the service type of {source_config_type}")
@@ -266,6 +299,27 @@ def _unsafe_parse_config(config: dict, cls: Type[T], message: str) -> None:
         raise err
 
 
+def _unsafe_parse_dbt_config(config: dict, cls: Type[T], message: str) -> None:
+    """
+    Given a config dictionary and the class it should match,
+    try to parse it or log the given message
+    """
+    logger.debug(f"Parsing message: [{message}]")
+    try:
+        # Parse the oneOf config types of dbt to check
+        dbt_config_type = config["dbtConfigSource"]["dbtConfigType"]
+        dbt_config_class = DBT_CONFIG_TYPE_MAP.get(dbt_config_type)
+        dbt_config_class.parse_obj(config["dbtConfigSource"])
+
+        # Parse the entire dbtPipeline object
+        cls.parse_obj(config)
+    except ValidationError as err:
+        logger.debug(
+            f"The supported properties for {cls.__name__} are {list(cls.__fields__.keys())}"
+        )
+        raise err
+
+
 def _parse_inner_connection(config_dict: dict, source_type: str) -> None:
     """
     Parse the inner connection of the flagged connectors
@@ -291,32 +345,35 @@ def parse_service_connection(config_dict: dict) -> None:
     :param config_dict: JSON configuration
     """
     # Unsafe access to the keys. Allow a KeyError if the config is not well formatted
-    source_type = config_dict["source"]["serviceConnection"]["config"].get("type")
-    if source_type is None:
-        raise InvalidWorkflowException("Missing type in the serviceConnection config")
+    if config_dict["source"].get("serviceConnection"):
+        source_type = config_dict["source"]["serviceConnection"]["config"].get("type")
+        if source_type is None:
+            raise InvalidWorkflowException(
+                "Missing type in the serviceConnection config"
+            )
 
-    logger.debug(
-        f"Error parsing the Workflow Configuration for {source_type} ingestion"
-    )
+        logger.debug(
+            f"Error parsing the Workflow Configuration for {source_type} ingestion"
+        )
 
-    service_type = get_service_type(source_type)
-    connection_class = get_connection_class(source_type, service_type)
+        service_type = get_service_type(source_type)
+        connection_class = get_connection_class(source_type, service_type)
 
-    if source_type in HAS_INNER_CONNECTION:
-        # We will first parse the inner `connection` configuration
-        _parse_inner_connection(
-            config_dict["source"]["serviceConnection"]["config"]["connection"][
-                "config"
-            ]["connection"],
-            source_type,
-        )
+        if source_type in HAS_INNER_CONNECTION:
+            # We will first parse the inner `connection` configuration
+            _parse_inner_connection(
+                config_dict["source"]["serviceConnection"]["config"]["connection"][
+                    "config"
+                ]["connection"],
+                source_type,
+            )
 
-    # Parse the service connection dictionary with the scoped class
-    _unsafe_parse_config(
-        config=config_dict["source"]["serviceConnection"]["config"],
-        cls=connection_class,
-        message="Error parsing the service connection",
-    )
+        # Parse the service connection dictionary with the scoped class
+        _unsafe_parse_config(
+            config=config_dict["source"]["serviceConnection"]["config"],
+            cls=connection_class,
+            message="Error parsing the service connection",
+        )
 
 
 def parse_source_config(config_dict: dict) -> None:
@@ -334,6 +391,13 @@ def parse_source_config(config_dict: dict) -> None:
 
     source_config_class = get_source_config_class(source_config_type)
 
+    if source_config_class == DbtPipeline:
+        _unsafe_parse_dbt_config(
+            config=config_dict["source"]["sourceConfig"]["config"],
+            cls=source_config_class,
+            message="Error parsing the dbt source config",
+        )
+
     _unsafe_parse_config(
         config=config_dict["source"]["sourceConfig"]["config"],
         cls=source_config_class,

metadata/ingestion/lineage/parser.py

@@ -217,6 +217,11 @@ class LineageParser:
         """
         aliases = self.table_aliases
         values = identifier.value.split(".")
+
+        if len(values) > 4:
+            logger.debug(f"Invalid comparison element from identifier: {identifier}")
+            return None, None
+
         database_name, schema_name, table_or_alias, column_name = (
             [None] * (4 - len(values))
         ) + values
@@ -307,32 +312,39 @@ class LineageParser:
                 comparisons.append(sub)
 
         for comparison in comparisons:
-            if "." not in comparison.left.value or "." not in comparison.right.value:
-                logger.debug(f"Ignoring comparison {comparison}")
-                continue
-
-            table_left, column_left = self.get_comparison_elements(
-                identifier=comparison.left
-            )
-            table_right, column_right = self.get_comparison_elements(
-                identifier=comparison.right
-            )
-
-            if not table_left or not table_right:
-                logger.warning(
-                    f"Can't extract table names when parsing JOIN information from {comparison}"
+            try:
+                if (
+                    "." not in comparison.left.value
+                    or "." not in comparison.right.value
+                ):
+                    logger.debug(f"Ignoring comparison {comparison}")
+                    continue
+
+                table_left, column_left = self.get_comparison_elements(
+                    identifier=comparison.left
+                )
+                table_right, column_right = self.get_comparison_elements(
+                    identifier=comparison.right
                 )
-                logger.debug(f"Query: {sql_statement}")
-                continue
 
-            left_table_column = TableColumn(table=table_left, column=column_left)
-            right_table_column = TableColumn(table=table_right, column=column_right)
+                if not table_left or not table_right:
+                    logger.warning(
+                        f"Can't extract table names when parsing JOIN information from {comparison}"
+                    )
+                    logger.debug(f"Query: {sql_statement}")
+                    continue
+
+                left_table_column = TableColumn(table=table_left, column=column_left)
+                right_table_column = TableColumn(table=table_right, column=column_right)
 
-            # We just send the info once, from Left -> Right.
-            # The backend will prepare the symmetric information.
-            self.stateful_add_table_joins(
-                join_data, left_table_column, right_table_column
-            )
+                # We just send the info once, from Left -> Right.
+                # The backend will prepare the symmetric information.
+                self.stateful_add_table_joins(
+                    join_data, left_table_column, right_table_column
+                )
+            except Exception as exc:
+                logger.debug(f"Cannot process comparison {comparison}: {exc}")
+                logger.debug(traceback.format_exc())
 
     @cached_property
     def table_joins(self) -> Dict[str, List[TableColumnJoin]]:

metadata/ingestion/ometa/routes.py

@@ -90,6 +90,16 @@ from metadata.generated.schema.api.tests.createTestDefinition import (
 from metadata.generated.schema.api.tests.createTestSuite import CreateTestSuiteRequest
 from metadata.generated.schema.dataInsight.dataInsightChart import DataInsightChart
 from metadata.generated.schema.dataInsight.kpi.kpi import Kpi
+from metadata.generated.schema.entity.applications.app import App
+from metadata.generated.schema.entity.applications.createAppRequest import (
+    CreateAppRequest,
+)
+from metadata.generated.schema.entity.applications.marketplace.appMarketPlaceDefinition import (
+    AppMarketPlaceDefinition,
+)
+from metadata.generated.schema.entity.applications.marketplace.createAppMarketPlaceDefinitionReq import (
+    CreateAppMarketPlaceDefinitionRequest,
+)
 from metadata.generated.schema.entity.automations.workflow import Workflow
 from metadata.generated.schema.entity.bot import Bot
 from metadata.generated.schema.entity.classification.classification import (
@@ -232,4 +242,9 @@ ROUTES = {
     # Suggestions
     Suggestion.__name__: "/suggestions",
     CreateSuggestionRequest.__name__: "/suggestions",
+    # Apps
+    App.__name__: "/apps",
+    CreateAppRequest.__name__: "/apps",
+    AppMarketPlaceDefinition.__name__: "/apps/marketplace",
+    CreateAppMarketPlaceDefinitionRequest.__name__: "/apps/marketplace",
 }

metadata/ingestion/source/database/dbt/dbt_config.py

@@ -173,7 +173,7 @@ def _(config: DbtCloudConfig):  # pylint: disable=too-many-locals
             params_data["job_definition_id"] = job_id
 
         response = client.get(f"/accounts/{account_id}/runs", data=params_data)
-        if not response and not response.get("data"):
+        if not response or not response.get("data"):
             raise DBTConfigException(
                 "Unable to get the dbt job runs information.\n"
                 "Please check if the auth token is correct and has the necessary scopes to fetch dbt runs"

metadata/ingestion/source/database/oracle/connection.py

@@ -38,6 +38,7 @@ from metadata.ingestion.connections.builders import (
 )
 from metadata.ingestion.connections.test_connections import test_connection_db_common
 from metadata.ingestion.ometa.ometa_api import OpenMetadata
+from metadata.ingestion.source.database.oracle.queries import CHECK_ACCESS_TO_DBA
 from metadata.utils.logger import ingestion_logger
 
 CX_ORACLE_LIB_VERSION = "8.3.0"
@@ -136,9 +137,13 @@ def test_connection(
     Test connection. This can be executed either as part
     of a metadata workflow or during an Automation Workflow
     """
+
+    test_conn_queries = {"CheckAccess": CHECK_ACCESS_TO_DBA}
+
     test_connection_db_common(
         metadata=metadata,
         engine=engine,
         service_connection=service_connection,
         automation_workflow=automation_workflow,
+        queries=test_conn_queries,
     )

metadata/ingestion/source/database/oracle/queries.py

@@ -88,7 +88,7 @@ WHERE
     type = 'PROCEDURE' and owner = '{schema}'
 """
 )
-
+CHECK_ACCESS_TO_DBA = "SELECT table_name FROM DBA_TABLES where ROWNUM < 2"
 ORACLE_GET_STORED_PROCEDURE_QUERIES = textwrap.dedent(
     """
 WITH SP_HISTORY AS (SELECT

metadata/ingestion/source/database/unitycatalog/connection.py

@@ -68,16 +68,20 @@ def test_connection(
             break
 
     def get_schemas(connection: WorkspaceClient, table_obj: DatabricksTable):
-        for schema in connection.schemas.list(catalog_name=table_obj.catalog_name):
-            table_obj.schema_name = schema.name
-            break
+        for catalog in connection.catalogs.list():
+            for schema in connection.schemas.list(catalog_name=catalog.name):
+                if schema.name:
+                    table_obj.schema_name = schema.name
+                    table_obj.catalog_name = catalog.name
+                    return
 
     def get_tables(connection: WorkspaceClient, table_obj: DatabricksTable):
-        for table in connection.tables.list(
-            catalog_name=table_obj.catalog_name, schema_name=table_obj.schema_name
-        ):
-            table_obj.name = table.name
-            break
+        if table_obj.catalog_name and table_obj.schema_name:
+            for table in connection.tables.list(
+                catalog_name=table_obj.catalog_name, schema_name=table_obj.schema_name
+            ):
+                table_obj.name = table.name
+                break
 
     test_fn = {
         "CheckAccess": connection.catalogs.list,

metadata/ingestion/source/pipeline/dagster/metadata.py

@@ -214,7 +214,9 @@ class DagsterSource(PipelineServiceSource):
             service_name=self.context.pipeline_service,
             pipeline_name=self.context.pipeline,
         )
-        pipeline_entity = self.metadata.get_by_name(entity=Pipeline, fqn=pipeline_fqn)
+        pipeline_entity = self.metadata.get_by_name(
+            entity=Pipeline, fqn=pipeline_fqn, fields=["tasks"]
+        )
         for task in pipeline_entity.tasks or []:
             try:
                 runs = self.client.get_task_runs(