openmetadata-ingestion 1.9.2.1__py3-none-any.whl → 1.9.4.0__py3-none-any.whl

metadata/generated/schema/type/tableUsageCount.py

@@ -1,6 +1,6 @@
 # generated by datamodel-codegen:
 #   filename:  type/tableUsageCount.json
-#   timestamp: 2025-08-22T15:30:10+00:00
+#   timestamp: 2025-08-27T02:29:56+00:00
 
 from __future__ import annotations
 

metadata/generated/schema/type/tagLabel.py

@@ -1,6 +1,6 @@
 # generated by datamodel-codegen:
 #   filename:  type/tagLabel.json
-#   timestamp: 2025-08-22T15:30:10+00:00
+#   timestamp: 2025-08-27T02:29:56+00:00
 
 from __future__ import annotations
 

metadata/generated/schema/type/usageDetails.py

@@ -1,6 +1,6 @@
 # generated by datamodel-codegen:
 #   filename:  type/usageDetails.json
-#   timestamp: 2025-08-22T15:30:10+00:00
+#   timestamp: 2025-08-27T02:29:56+00:00
 
 from __future__ import annotations
 

metadata/generated/schema/type/usageRequest.py

@@ -1,6 +1,6 @@
 # generated by datamodel-codegen:
 #   filename:  type/usageRequest.json
-#   timestamp: 2025-08-22T15:30:10+00:00
+#   timestamp: 2025-08-27T02:29:56+00:00
 
 from __future__ import annotations
 

metadata/generated/schema/type/votes.py

@@ -1,6 +1,6 @@
 # generated by datamodel-codegen:
 #   filename:  type/votes.json
-#   timestamp: 2025-08-22T15:30:10+00:00
+#   timestamp: 2025-08-27T02:29:56+00:00
 
 from __future__ import annotations
 

metadata/ingestion/lineage/sql_lineage.py

@@ -63,9 +63,16 @@ def get_column_fqn(table_entity: Table, column: str) -> Optional[str]:
     """
     if not table_entity:
         return None
-    for tbl_column in table_entity.columns:
-        if column.lower() == tbl_column.name.root.lower():
-            return tbl_column.fullyQualifiedName.root
+    for tbl_column in table_entity.columns or []:
+        try:
+            if column.lower() == tbl_column.name.root.lower():
+                return tbl_column.fullyQualifiedName.root
+        except Exception as e:
+            logger.debug(traceback.format_exc())
+            logger.debug(
+                f"Error getting column FQN for column [{column}] in"
+                f"table [{table_entity.fullyQualifiedName}] for column [{tbl_column}]: {e}"
+            )
 
     return None
 

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

@@ -14,6 +14,7 @@ Source connection handler
 """
 import os
 import sys
+from copy import deepcopy
 from typing import Optional
 from urllib.parse import quote_plus
 
@@ -26,7 +27,9 @@ from metadata.generated.schema.entity.automations.workflow import (
     Workflow as AutomationWorkflow,
 )
 from metadata.generated.schema.entity.services.connections.database.oracleConnection import (
-    OracleConnection,
+    OracleConnection as OracleConnectionConfig,
+)
+from metadata.generated.schema.entity.services.connections.database.oracleConnection import (
     OracleDatabaseSchema,
     OracleServiceName,
     OracleTNSConnection,
@@ -39,6 +42,8 @@ from metadata.ingestion.connections.builders import (
     get_connection_args_common,
     get_connection_options_dict,
 )
+from metadata.ingestion.connections.connection import BaseConnection
+from metadata.ingestion.connections.secrets import connection_with_options_secrets
 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 (
@@ -55,123 +60,174 @@ LD_LIB_ENV = "LD_LIBRARY_PATH"
 logger = ingestion_logger()
 
 
-def get_connection_url(connection: OracleConnection) -> str:
-    """
-    Build the URL and handle driver version at system level
-    """
-
-    oracledb.version = CX_ORACLE_LIB_VERSION
-    sys.modules["cx_Oracle"] = oracledb
+class OracleConnection(BaseConnection[OracleConnectionConfig, Engine]):
+    def __init__(self, connection: OracleConnectionConfig):
+        super().__init__(connection)
 
-    url = f"{connection.scheme.value}://"
-    if connection.username:
-        url += f"{quote_plus(connection.username)}"
-        if not connection.password:
-            connection.password = SecretStr("")
-        url += f":{quote_plus(connection.password.get_secret_value())}"
-        url += "@"
-
-    url = _handle_connection_type(url=url, connection=connection)
-
-    options = get_connection_options_dict(connection)
-    if options:
-        params = "&".join(
-            f"{key}={quote_plus(value)}" for (key, value) in options.items() if value
+    def _get_client(self) -> Engine:
+        """
+        Create connection
+        """
+        try:
+            if self.service_connection.instantClientDirectory:
+                logger.info(
+                    f"Initializing Oracle thick client at {self.service_connection.instantClientDirectory}"
+                )
+                os.environ[LD_LIB_ENV] = self.service_connection.instantClientDirectory
+                oracledb.init_oracle_client(
+                    lib_dir=self.service_connection.instantClientDirectory
+                )
+        except DatabaseError as err:
+            logger.info(f"Could not initialize Oracle thick client: {err}")
+
+        return create_generic_db_connection(
+            connection=self.service_connection,
+            get_connection_url_fn=self.get_connection_url,
+            get_connection_args_fn=get_connection_args_common,
         )
-        if isinstance(connection.oracleConnectionType, OracleServiceName):
-            url = f"{url}&{params}"
-        else:
-            url = f"{url}?{params}"
-
-    return url
-
-
-def _handle_connection_type(url: str, connection: OracleConnection) -> str:
-    """
-    Depending on the oracle connection type, we need to handle the URL differently
-    """
-
-    if isinstance(connection.oracleConnectionType, OracleTNSConnection):
-        # ref https://stackoverflow.com/questions/14140902/using-oracle-service-names-with-sqlalchemy
-        url += connection.oracleConnectionType.oracleTNSConnection
-        return url
-
-    # If not TNS, we add the hostPort
-    url += connection.hostPort
 
-    if isinstance(connection.oracleConnectionType, OracleDatabaseSchema):
-        url += (
-            f"/{connection.oracleConnectionType.databaseSchema}"
-            if connection.oracleConnectionType.databaseSchema
-            else ""
+    def test_connection(
+        self,
+        metadata: OpenMetadata,
+        automation_workflow: Optional[AutomationWorkflow] = None,
+        timeout_seconds: Optional[int] = THREE_MIN,
+    ) -> TestConnectionResult:
+        """
+        Test connection. This can be executed either as part
+        of a metadata workflow or during an Automation Workflow
+        """
+
+        def test_oracle_package_access(engine):
+            try:
+                schema_name = engine.execute(ORACLE_GET_SCHEMA).scalar()
+                return ORACLE_GET_STORED_PACKAGES.format(schema=schema_name)
+            except Exception as e:
+                raise OraclePackageAccessError(
+                    f"Failed to access Oracle stored packages: {e}"
+                )
+
+        test_conn_queries = {
+            "CheckAccess": CHECK_ACCESS_TO_ALL,
+            "PackageAccess": test_oracle_package_access(self.client),
+        }
+
+        return test_connection_db_common(
+            metadata=metadata,
+            engine=self.client,
+            service_connection=self.service_connection,
+            automation_workflow=automation_workflow,
+            queries=test_conn_queries,
+            timeout_seconds=timeout_seconds,
         )
-        return url
-
-    if isinstance(connection.oracleConnectionType, OracleServiceName):
-        url = f"{url}/?service_name={connection.oracleConnectionType.oracleServiceName}"
-        return url
 
-    raise ValueError(f"Unknown connection type {connection.oracleConnectionType}")
+    def get_connection_dict(self) -> dict:
+        """
+        Return the connection dictionary for this service.
+        """
+        url = self.client.url
+        connection_copy = deepcopy(self.service_connection)
+
+        connection_dict = {
+            "driver": url.drivername,
+            "host": f"{url.host}:{url.port}",  # This is the format expected by data-diff. If we start using this for something else, we need to change it and modify the data-diff code.
+            "user": url.username,
+        }
+
+        # Add password if present in the connection
+        if connection_copy.password:
+            connection_dict["password"] = connection_copy.password.get_secret_value()
+
+        # Add connection type specific information
+        if isinstance(connection_copy.oracleConnectionType, OracleDatabaseSchema):
+            connection_dict[
+                "database"
+            ] = connection_copy.oracleConnectionType.databaseSchema
+        elif isinstance(connection_copy.oracleConnectionType, OracleServiceName):
+            connection_dict[
+                "database"
+            ] = connection_copy.oracleConnectionType.oracleServiceName
+        elif isinstance(connection_copy.oracleConnectionType, OracleTNSConnection):
+            connection_dict[
+                "host"
+            ] = connection_copy.oracleConnectionType.oracleTNSConnection
+
+        # Add connection options if present
+        if connection_copy.connectionOptions and connection_copy.connectionOptions.root:
+            connection_with_options_secrets(lambda: connection_copy)
+            connection_dict.update(connection_copy.connectionOptions.root)
+
+        # Add connection arguments if present
+        if (
+            connection_copy.connectionArguments
+            and connection_copy.connectionArguments.root
+        ):
+            connection_dict.update(get_connection_args_common(connection_copy))
+
+        return connection_dict
+
+    @staticmethod
+    def get_connection_url(connection: OracleConnectionConfig) -> str:
+        """
+        Build the URL and handle driver version at system level
+        """
+
+        oracledb.version = CX_ORACLE_LIB_VERSION
+        sys.modules["cx_Oracle"] = oracledb
+
+        url = f"{connection.scheme.value}://"
+        if connection.username:
+            url += f"{quote_plus(connection.username)}"
+            if not connection.password:
+                connection.password = SecretStr("")
+            url += f":{quote_plus(connection.password.get_secret_value())}"
+            url += "@"
+
+        url = OracleConnection._handle_connection_type(url=url, connection=connection)
+
+        options = get_connection_options_dict(connection)
+        if options:
+            params = "&".join(
+                f"{key}={quote_plus(value)}"
+                for (key, value) in options.items()
+                if value
+            )
+            if isinstance(connection.oracleConnectionType, OracleServiceName):
+                url = f"{url}&{params}"
+            else:
+                url = f"{url}?{params}"
 
+        return url
 
-def get_connection(connection: OracleConnection) -> Engine:
-    """
-    Create connection
-    """
-    try:
-        if connection.instantClientDirectory:
-            logger.info(
-                f"Initializing Oracle thick client at {connection.instantClientDirectory}"
+    @staticmethod
+    def _handle_connection_type(url: str, connection: OracleConnectionConfig) -> str:
+        """
+        Depending on the oracle connection type, we need to handle the URL differently
+        """
+
+        if isinstance(connection.oracleConnectionType, OracleTNSConnection):
+            # ref https://stackoverflow.com/questions/14140902/using-oracle-service-names-with-sqlalchemy
+            url += connection.oracleConnectionType.oracleTNSConnection
+            return url
+
+        # If not TNS, we add the hostPort
+        url += connection.hostPort
+
+        if isinstance(connection.oracleConnectionType, OracleDatabaseSchema):
+            url += (
+                f"/{connection.oracleConnectionType.databaseSchema}"
+                if connection.oracleConnectionType.databaseSchema
+                else ""
             )
-            os.environ[LD_LIB_ENV] = connection.instantClientDirectory
-            oracledb.init_oracle_client()
-    except DatabaseError as err:
-        logger.info(f"Could not initialize Oracle thick client: {err}")
+            return url
+
+        if isinstance(connection.oracleConnectionType, OracleServiceName):
+            url = f"{url}/?service_name={connection.oracleConnectionType.oracleServiceName}"
+            return url
 
-    return create_generic_db_connection(
-        connection=connection,
-        get_connection_url_fn=get_connection_url,
-        get_connection_args_fn=get_connection_args_common,
-    )
+        raise ValueError(f"Unknown connection type {connection.oracleConnectionType}")
 
 
 class OraclePackageAccessError(Exception):
     """
     Raised when unable to access Oracle stored packages
     """
-
-
-def test_connection(
-    metadata: OpenMetadata,
-    engine: Engine,
-    service_connection: OracleConnection,
-    automation_workflow: Optional[AutomationWorkflow] = None,
-    timeout_seconds: Optional[int] = THREE_MIN,
-) -> TestConnectionResult:
-    """
-    Test connection. This can be executed either as part
-    of a metadata workflow or during an Automation Workflow
-    """
-
-    def test_oracle_package_access(engine):
-        try:
-            schema_name = engine.execute(ORACLE_GET_SCHEMA).scalar()
-            return ORACLE_GET_STORED_PACKAGES.format(schema=schema_name)
-        except Exception as e:
-            raise OraclePackageAccessError(
-                f"Failed to access Oracle stored packages: {e}"
-            )
-
-    test_conn_queries = {
-        "CheckAccess": CHECK_ACCESS_TO_ALL,
-        "PackageAccess": test_oracle_package_access(engine),
-    }
-
-    return test_connection_db_common(
-        metadata=metadata,
-        engine=engine,
-        service_connection=service_connection,
-        automation_workflow=automation_workflow,
-        queries=test_conn_queries,
-        timeout_seconds=timeout_seconds,
-    )

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

@@ -1,3 +1,4 @@
+from metadata.ingestion.source.database.oracle.connection import OracleConnection
 from metadata.ingestion.source.database.oracle.lineage import OracleLineageSource
 from metadata.ingestion.source.database.oracle.metadata import OracleSource
 from metadata.ingestion.source.database.oracle.usage import OracleUsageSource
@@ -7,4 +8,5 @@ ServiceSpec = DefaultDatabaseSpec(
     metadata_source_class=OracleSource,
     lineage_source_class=OracleLineageSource,
     usage_source_class=OracleUsageSource,
+    connection_class=OracleConnection,
 )