feat/drop use of old file data (#446)

* drop use of old file data

* fix missing imports

* fix unit tests

* fix import

Author: rbiseck3

CHANGELOG.md

@@ -1,3 +1,9 @@
+## 0.6.0
+
+### Features
+
+* **Isolate FileData to limit dependencies**
+
 ## 0.5.25
 
 ### Features

test/integration/connectors/databricks/test_volumes_native.py

@@ -21,7 +21,6 @@
 )
 from test.integration.utils import requires_env
 from unstructured_ingest.v2.errors import UserAuthError, UserError
-from unstructured_ingest.v2.interfaces import FileData, SourceIdentifiers
 from unstructured_ingest.v2.processes.connectors.databricks.volumes_native import (
     CONNECTOR_TYPE,
     DatabricksNativeVolumesAccessConfig,
@@ -33,6 +32,7 @@
     DatabricksNativeVolumesUploader,
     DatabricksNativeVolumesUploaderConfig,
 )
+from unstructured_ingest.v2.types.file_data import FileData, SourceIdentifiers
 
 
 @dataclass

test/integration/connectors/duckdb/test_duckdb.py

@@ -10,14 +10,14 @@
     StagerValidationConfigs,
     stager_validation,
 )
-from unstructured_ingest.v2.interfaces.file_data import FileData, SourceIdentifiers
 from unstructured_ingest.v2.processes.connectors.duckdb.duckdb import (
     CONNECTOR_TYPE,
     DuckDBConnectionConfig,
     DuckDBUploader,
     DuckDBUploaderConfig,
     DuckDBUploadStager,
 )
+from unstructured_ingest.v2.types.file_data import FileData, SourceIdentifiers
 
 
 @pytest.fixture

test/integration/connectors/duckdb/test_motherduck.py

@@ -9,7 +9,6 @@
 
 from test.integration.connectors.utils.constants import DESTINATION_TAG, SQL_TAG
 from test.integration.utils import requires_env
-from unstructured_ingest.v2.interfaces.file_data import FileData, SourceIdentifiers
 from unstructured_ingest.v2.processes.connectors.duckdb.motherduck import (
     CONNECTOR_TYPE,
     MotherDuckAccessConfig,
@@ -18,6 +17,7 @@
     MotherDuckUploaderConfig,
     MotherDuckUploadStager,
 )
+from unstructured_ingest.v2.types.file_data import FileData, SourceIdentifiers
 
 
 @pytest.fixture

test/integration/connectors/elasticsearch/test_elasticsearch.py

@@ -22,7 +22,7 @@
     source_connector_validation,
 )
 from unstructured_ingest.error import DestinationConnectionError, SourceConnectionError
-from unstructured_ingest.v2.interfaces import FileData, SourceIdentifiers
+from unstructured_ingest.v2.types.file_data import FileData, SourceIdentifiers
 from unstructured_ingest.v2.processes.connectors.elasticsearch.elasticsearch import (
     CONNECTOR_TYPE,
     ElasticsearchAccessConfig,

test/integration/connectors/elasticsearch/test_opensearch.py

@@ -24,7 +24,6 @@
     DestinationConnectionError,
     SourceConnectionError,
 )
-from unstructured_ingest.v2.interfaces import FileData, SourceIdentifiers
 from unstructured_ingest.v2.processes.connectors.elasticsearch.opensearch import (
     CONNECTOR_TYPE,
     OpenSearchAccessConfig,
@@ -38,6 +37,7 @@
     OpenSearchUploadStager,
     OpenSearchUploadStagerConfig,
 )
+from unstructured_ingest.v2.types.file_data import FileData, SourceIdentifiers
 
 SOURCE_INDEX_NAME = "movies"
 DESTINATION_INDEX_NAME = "elements"

test/integration/connectors/kafka/test_kafka_cloud.py

@@ -19,7 +19,6 @@
     source_connector_validation,
 )
 from test.integration.utils import requires_env
-from unstructured_ingest.v2.interfaces import FileData, SourceIdentifiers
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connectors.kafka.cloud import (
     CloudKafkaAccessConfig,
@@ -32,6 +31,7 @@
     CloudKafkaUploaderConfig,
 )
 from unstructured_ingest.v2.processes.connectors.kafka.local import CONNECTOR_TYPE
+from unstructured_ingest.v2.types.file_data import FileData, SourceIdentifiers
 
 
 @pytest.fixture

test/integration/connectors/kafka/test_kafka_local.py

@@ -24,7 +24,6 @@
     source_connector_validation,
 )
 from unstructured_ingest.error import DestinationConnectionError, SourceConnectionError
-from unstructured_ingest.v2.interfaces import FileData, SourceIdentifiers
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connectors.kafka.local import (
     CONNECTOR_TYPE,
@@ -36,6 +35,7 @@
     LocalKafkaUploader,
     LocalKafkaUploaderConfig,
 )
+from unstructured_ingest.v2.types.file_data import FileData, SourceIdentifiers
 
 SEED_MESSAGES = 10
 

test/integration/connectors/sql/test_databricks_delta_tables.py

@@ -14,7 +14,6 @@
 
 from test.integration.connectors.utils.constants import DESTINATION_TAG, SQL_TAG, env_setup_path
 from test.integration.utils import requires_env
-from unstructured_ingest.v2.interfaces import FileData, SourceIdentifiers
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connectors.sql.databricks_delta_tables import (
     CONNECTOR_TYPE,
@@ -24,6 +23,7 @@
     DatabricksDeltaTablesUploaderConfig,
     DatabricksDeltaTablesUploadStager,
 )
+from unstructured_ingest.v2.types.file_data import FileData, SourceIdentifiers
 
 CATALOG = "utic-dev-tech-fixtures"
 

test/integration/connectors/sql/test_postgres.py

@@ -20,7 +20,6 @@
     SourceValidationConfigs,
     source_connector_validation,
 )
-from unstructured_ingest.v2.interfaces import FileData, SourceIdentifiers
 from unstructured_ingest.v2.processes.connectors.sql.postgres import (
     CONNECTOR_TYPE,
     PostgresAccessConfig,
@@ -32,6 +31,7 @@
     PostgresUploader,
     PostgresUploadStager,
 )
+from unstructured_ingest.v2.types.file_data import FileData, SourceIdentifiers
 
 SEED_DATA_ROWS = 10