unstructured-ingest 0.5.23__py3-none-any.whl → 0.6.0__py3-none-any.whl

unstructured_ingest/v2/processes/connectors/fsspec/box.py

@@ -11,7 +11,6 @@ from pydantic.functional_validators import BeforeValidator
 
 from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.v2.errors import ProviderError, UserAuthError, UserError
-from unstructured_ingest.v2.interfaces import FileDataSourceMetadata
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
@@ -32,6 +31,7 @@ from unstructured_ingest.v2.processes.utils.blob_storage import (
     BlobStoreUploadStager,
     BlobStoreUploadStagerConfig,
 )
+from unstructured_ingest.v2.types.file_data import FileDataSourceMetadata
 
 if TYPE_CHECKING:
     from boxfs import BoxFileSystem

unstructured_ingest/v2/processes/connectors/fsspec/dropbox.py

@@ -15,7 +15,6 @@ from unstructured_ingest.v2.errors import (
 from unstructured_ingest.v2.errors import (
     RateLimitError as CustomRateLimitError,
 )
-from unstructured_ingest.v2.interfaces import FileDataSourceMetadata
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
@@ -35,6 +34,7 @@ from unstructured_ingest.v2.processes.utils.blob_storage import (
     BlobStoreUploadStager,
     BlobStoreUploadStagerConfig,
 )
+from unstructured_ingest.v2.types.file_data import FileDataSourceMetadata
 
 if TYPE_CHECKING:
     pass

unstructured_ingest/v2/processes/connectors/fsspec/fsspec.py

@@ -18,16 +18,18 @@ from unstructured_ingest.v2.interfaces import (
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
-    FileDataSourceMetadata,
     Indexer,
     IndexerConfig,
-    SourceIdentifiers,
     Uploader,
     UploaderConfig,
 )
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connectors.fsspec.utils import sterilize_dict
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 
 if TYPE_CHECKING:
     from fsspec import AbstractFileSystem

unstructured_ingest/v2/processes/connectors/fsspec/gcs.py

@@ -12,7 +12,6 @@ from pydantic import Field, Secret
 from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.utils.string_and_date_utils import json_to_dict
 from unstructured_ingest.v2.errors import ProviderError, UserError
-from unstructured_ingest.v2.interfaces import FileDataSourceMetadata
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
@@ -32,6 +31,7 @@ from unstructured_ingest.v2.processes.utils.blob_storage import (
     BlobStoreUploadStager,
     BlobStoreUploadStagerConfig,
 )
+from unstructured_ingest.v2.types.file_data import FileDataSourceMetadata
 
 if TYPE_CHECKING:
     from gcsfs import GCSFileSystem

unstructured_ingest/v2/processes/connectors/fsspec/s3.py

@@ -8,9 +8,6 @@ from pydantic import Field, Secret
 
 from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.v2.errors import ProviderError, UserAuthError, UserError
-from unstructured_ingest.v2.interfaces import (
-    FileDataSourceMetadata,
-)
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
@@ -30,9 +27,15 @@ from unstructured_ingest.v2.processes.utils.blob_storage import (
     BlobStoreUploadStager,
     BlobStoreUploadStagerConfig,
 )
+from unstructured_ingest.v2.types.file_data import (
+    FileDataSourceMetadata,
+)
 
 CONNECTOR_TYPE = "s3"
 
+# https://docs.aws.amazon.com/AmazonS3/latest/userguide/object-keys.html#object-key-guidelines-avoid-characters
+CHARACTERS_TO_AVOID = ["\\", "{", "^", "}", "%", "`", "]", '"', ">", "[", "~", "<", "#", "|"]
+
 if TYPE_CHECKING:
     from s3fs import S3FileSystem
 
@@ -91,7 +94,7 @@ class S3ConnectionConfig(FsspecConnectionConfig):
         if isinstance(e, PermissionError):
             return UserAuthError(e)
         if isinstance(e, FileNotFoundError):
-            return UserError(e)
+            return UserError(f"File not found: {e}")
         if cause := getattr(e, "__cause__", None):
             error_response = cause.response
             error_meta = error_response["ResponseMetadata"]
@@ -140,6 +143,12 @@ class S3Indexer(FsspecIndexer):
         }
         if metadata:
             record_locator["metadata"] = metadata
+        issue_characters = [char for char in CHARACTERS_TO_AVOID if char in path]
+        if issue_characters:
+            logger.warning(
+                f"File path {path} contains characters "
+                f"that can cause issues with S3: {issue_characters}"
+            )
         return FileDataSourceMetadata(
             date_created=date_created,
             date_modified=date_modified,

unstructured_ingest/v2/processes/connectors/fsspec/sftp.py

@@ -11,7 +11,6 @@ from urllib.parse import urlparse
 from pydantic import Field, Secret
 
 from unstructured_ingest.utils.dep_check import requires_dependencies
-from unstructured_ingest.v2.interfaces import FileData, FileDataSourceMetadata
 from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
     SourceRegistryEntry,
@@ -30,6 +29,7 @@ from unstructured_ingest.v2.processes.utils.blob_storage import (
     BlobStoreUploadStager,
     BlobStoreUploadStagerConfig,
 )
+from unstructured_ingest.v2.types.file_data import FileData, FileDataSourceMetadata
 
 if TYPE_CHECKING:
     from fsspec.implementations.sftp import SFTPFileSystem

unstructured_ingest/v2/processes/connectors/gitlab.py

@@ -16,14 +16,16 @@ from unstructured_ingest.v2.interfaces import (
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
-    FileDataSourceMetadata,
     Indexer,
     IndexerConfig,
-    SourceIdentifiers,
 )
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import SourceRegistryEntry
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 
 CONNECTOR_TYPE = "gitlab"
 if TYPE_CHECKING:

unstructured_ingest/v2/processes/connectors/google_drive.py

@@ -21,15 +21,17 @@ from unstructured_ingest.v2.interfaces import (
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
-    FileDataSourceMetadata,
     Indexer,
     IndexerConfig,
-    SourceIdentifiers,
 )
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import SourceRegistryEntry
 from unstructured_ingest.v2.processes.connectors.utils import conform_string_to_dict
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 
 CONNECTOR_TYPE = "google_drive"
 

unstructured_ingest/v2/processes/connectors/ibm_watsonx/ibm_watsonx_s3.py

@@ -15,7 +15,6 @@ from unstructured_ingest.v2.errors import ProviderError, UserAuthError, UserErro
 from unstructured_ingest.v2.interfaces import (
     AccessConfig,
     ConnectionConfig,
-    FileData,
     UploaderConfig,
 )
 from unstructured_ingest.v2.logger import logger
@@ -27,6 +26,7 @@ from unstructured_ingest.v2.processes.connectors.sql.sql import (
     SQLUploadStager,
     SQLUploadStagerConfig,
 )
+from unstructured_ingest.v2.types.file_data import FileData
 
 if TYPE_CHECKING:
     from pyarrow import Table as ArrowTable

unstructured_ingest/v2/processes/connectors/jira.py

@@ -15,16 +15,18 @@ from unstructured_ingest.v2.interfaces import (
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
-    FileDataSourceMetadata,
     Indexer,
     IndexerConfig,
-    SourceIdentifiers,
 )
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     SourceRegistryEntry,
 )
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 
 if TYPE_CHECKING:
     from atlassian import Jira

unstructured_ingest/v2/processes/connectors/kafka/kafka.py

@@ -21,15 +21,17 @@ from unstructured_ingest.v2.interfaces import (
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
-    FileDataSourceMetadata,
     Indexer,
     IndexerConfig,
-    SourceIdentifiers,
     Uploader,
     UploaderConfig,
 )
 from unstructured_ingest.v2.logger import logger
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 
 if TYPE_CHECKING:
     from confluent_kafka import Consumer, Producer

unstructured_ingest/v2/processes/connectors/kdbai.py

@@ -11,7 +11,6 @@ from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.v2.interfaces import (
     AccessConfig,
     ConnectionConfig,
-    FileData,
     Uploader,
     UploaderConfig,
     UploadStager,
@@ -21,6 +20,7 @@ from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
 )
+from unstructured_ingest.v2.types.file_data import FileData
 from unstructured_ingest.v2.utils import get_enhanced_element_id
 
 if TYPE_CHECKING:

unstructured_ingest/v2/processes/connectors/lancedb/lancedb.py

@@ -15,10 +15,14 @@ from unstructured_ingest.logger import logger
 from unstructured_ingest.utils.data_prep import flatten_dict
 from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.v2.constants import RECORD_ID_LABEL
-from unstructured_ingest.v2.interfaces.connector import ConnectionConfig
-from unstructured_ingest.v2.interfaces.file_data import FileData
-from unstructured_ingest.v2.interfaces.upload_stager import UploadStager, UploadStagerConfig
-from unstructured_ingest.v2.interfaces.uploader import Uploader, UploaderConfig
+from unstructured_ingest.v2.interfaces import (
+    ConnectionConfig,
+    Uploader,
+    UploaderConfig,
+    UploadStager,
+    UploadStagerConfig,
+)
+from unstructured_ingest.v2.types.file_data import FileData
 
 CONNECTOR_TYPE = "lancedb"
 

unstructured_ingest/v2/processes/connectors/local.py

@@ -14,11 +14,8 @@ from unstructured_ingest.v2.interfaces import (
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
-    FileDataSourceMetadata,
     Indexer,
     IndexerConfig,
-    SourceIdentifiers,
     Uploader,
     UploaderConfig,
 )
@@ -31,6 +28,11 @@ from unstructured_ingest.v2.processes.utils.blob_storage import (
     BlobStoreUploadStager,
     BlobStoreUploadStagerConfig,
 )
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 
 CONNECTOR_TYPE = "local"
 

unstructured_ingest/v2/processes/connectors/milvus.py

@@ -13,7 +13,6 @@ from unstructured_ingest.v2.constants import RECORD_ID_LABEL
 from unstructured_ingest.v2.interfaces import (
     AccessConfig,
     ConnectionConfig,
-    FileData,
     Uploader,
     UploaderConfig,
     UploadStager,
@@ -23,6 +22,7 @@ from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
 )
+from unstructured_ingest.v2.types.file_data import FileData
 
 if TYPE_CHECKING:
     from pymilvus import MilvusClient

unstructured_ingest/v2/processes/connectors/mongodb.py

@@ -13,17 +13,12 @@ from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.v2.constants import RECORD_ID_LABEL
 from unstructured_ingest.v2.interfaces import (
     AccessConfig,
-    BatchFileData,
-    BatchItem,
     ConnectionConfig,
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
-    FileDataSourceMetadata,
     Indexer,
     IndexerConfig,
-    SourceIdentifiers,
     Uploader,
     UploaderConfig,
     download_responses,
@@ -33,6 +28,13 @@ from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
     SourceRegistryEntry,
 )
+from unstructured_ingest.v2.types.file_data import (
+    BatchFileData,
+    BatchItem,
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 
 if TYPE_CHECKING:
     from pymongo import MongoClient

unstructured_ingest/v2/processes/connectors/neo4j.py

@@ -18,7 +18,6 @@ from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.v2.interfaces import (
     AccessConfig,
     ConnectionConfig,
-    FileData,
     Uploader,
     UploaderConfig,
     UploadStager,
@@ -28,6 +27,7 @@ from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
 )
 from unstructured_ingest.v2.processes.connectors.utils import format_and_truncate_orig_elements
+from unstructured_ingest.v2.types.file_data import FileData
 
 SimilarityFunction = Literal["cosine"]
 

unstructured_ingest/v2/processes/connectors/notion/connector.py

@@ -12,14 +12,16 @@ from unstructured_ingest.v2.interfaces import (
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
-    FileDataSourceMetadata,
     Indexer,
     IndexerConfig,
-    SourceIdentifiers,
 )
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import SourceRegistryEntry
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 
 if TYPE_CHECKING:
     from unstructured_ingest.v2.processes.connectors.notion.client import Client

unstructured_ingest/v2/processes/connectors/onedrive.py

@@ -22,11 +22,8 @@ from unstructured_ingest.v2.interfaces import (
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
-    FileDataSourceMetadata,
     Indexer,
     IndexerConfig,
-    SourceIdentifiers,
     Uploader,
     UploaderConfig,
 )
@@ -39,6 +36,11 @@ from unstructured_ingest.v2.processes.utils.blob_storage import (
     BlobStoreUploadStager,
     BlobStoreUploadStagerConfig,
 )
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 
 if TYPE_CHECKING:
     from office365.graph_client import GraphClient

unstructured_ingest/v2/processes/connectors/outlook.py

@@ -16,12 +16,15 @@ from unstructured_ingest.v2.interfaces import (
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
     Indexer,
     IndexerConfig,
 )
-from unstructured_ingest.v2.interfaces.file_data import FileDataSourceMetadata, SourceIdentifiers
 from unstructured_ingest.v2.processes.connector_registry import SourceRegistryEntry
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 
 MAX_EMAILS_PER_FOLDER = 1_000_000  # Maximum number of emails per folder
 

unstructured_ingest/v2/processes/connectors/pinecone.py

@@ -13,7 +13,6 @@ from unstructured_ingest.v2.errors import UserError
 from unstructured_ingest.v2.interfaces import (
     AccessConfig,
     ConnectionConfig,
-    FileData,
     UploaderConfig,
     UploadStager,
     UploadStagerConfig,
@@ -21,6 +20,7 @@ from unstructured_ingest.v2.interfaces import (
 )
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import DestinationRegistryEntry
+from unstructured_ingest.v2.types.file_data import FileData
 from unstructured_ingest.v2.utils import get_enhanced_element_id
 
 if TYPE_CHECKING:

unstructured_ingest/v2/processes/connectors/qdrant/qdrant.py

@@ -13,13 +13,13 @@ from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.v2.interfaces import (
     AccessConfig,
     ConnectionConfig,
-    FileData,
     Uploader,
     UploaderConfig,
     UploadStager,
     UploadStagerConfig,
 )
 from unstructured_ingest.v2.logger import logger
+from unstructured_ingest.v2.types.file_data import FileData
 from unstructured_ingest.v2.utils import get_enhanced_element_id
 
 if TYPE_CHECKING:

unstructured_ingest/v2/processes/connectors/redisdb.py

@@ -11,12 +11,12 @@ from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.v2.interfaces import (
     AccessConfig,
     ConnectionConfig,
-    FileData,
     Uploader,
     UploaderConfig,
 )
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import DestinationRegistryEntry
+from unstructured_ingest.v2.types.file_data import FileData
 
 if TYPE_CHECKING:
     from redis.asyncio import Redis

unstructured_ingest/v2/processes/connectors/salesforce.py

@@ -28,16 +28,18 @@ from unstructured_ingest.v2.interfaces import (
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
-    FileDataSourceMetadata,
     Indexer,
     IndexerConfig,
-    SourceIdentifiers,
 )
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     SourceRegistryEntry,
 )
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 
 
 class MissingCategoryError(Exception):

unstructured_ingest/v2/processes/connectors/sharepoint.py

@@ -11,9 +11,6 @@ from unstructured_ingest.error import (
     SourceConnectionNetworkError,
 )
 from unstructured_ingest.utils.dep_check import requires_dependencies
-from unstructured_ingest.v2.interfaces import (
-    FileData,
-)
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     SourceRegistryEntry,
@@ -26,6 +23,9 @@ from unstructured_ingest.v2.processes.connectors.onedrive import (
     OnedriveIndexer,
     OnedriveIndexerConfig,
 )
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+)
 
 if TYPE_CHECKING:
     from office365.onedrive.driveitems.driveItem import DriveItem

unstructured_ingest/v2/processes/connectors/slack.py

@@ -20,12 +20,12 @@ from unstructured_ingest.v2.interfaces import (
     Indexer,
     IndexerConfig,
 )
-from unstructured_ingest.v2.interfaces.file_data import (
+from unstructured_ingest.v2.processes.connector_registry import SourceRegistryEntry
+from unstructured_ingest.v2.types.file_data import (
     FileData,
     FileDataSourceMetadata,
     SourceIdentifiers,
 )
-from unstructured_ingest.v2.processes.connector_registry import SourceRegistryEntry
 
 if TYPE_CHECKING:
     from slack_sdk import WebClient

unstructured_ingest/v2/processes/connectors/sql/databricks_delta_tables.py

@@ -7,7 +7,6 @@ from pydantic import Field, Secret
 
 from unstructured_ingest.utils.data_prep import split_dataframe
 from unstructured_ingest.utils.dep_check import requires_dependencies
-from unstructured_ingest.v2.interfaces import FileData
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
@@ -20,6 +19,7 @@ from unstructured_ingest.v2.processes.connectors.sql.sql import (
     SQLUploadStager,
     SQLUploadStagerConfig,
 )
+from unstructured_ingest.v2.types.file_data import FileData
 
 if TYPE_CHECKING:
     from databricks.sdk.core import oauth_service_principal

unstructured_ingest/v2/processes/connectors/sql/snowflake.py

@@ -7,7 +7,6 @@ from pydantic import Field, Secret
 
 from unstructured_ingest.utils.data_prep import split_dataframe
 from unstructured_ingest.utils.dep_check import requires_dependencies
-from unstructured_ingest.v2.interfaces.file_data import FileData
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
@@ -28,6 +27,7 @@ from unstructured_ingest.v2.processes.connectors.sql.sql import (
     SQLUploadStagerConfig,
     parse_date_string,
 )
+from unstructured_ingest.v2.types.file_data import FileData
 
 if TYPE_CHECKING:
     from pandas import DataFrame

unstructured_ingest/v2/processes/connectors/sql/sql.py

@@ -16,17 +16,12 @@ from unstructured_ingest.utils.data_prep import get_data, get_data_df, split_dat
 from unstructured_ingest.v2.constants import RECORD_ID_LABEL
 from unstructured_ingest.v2.interfaces import (
     AccessConfig,
-    BatchFileData,
-    BatchItem,
     ConnectionConfig,
     Downloader,
     DownloaderConfig,
     DownloadResponse,
-    FileData,
-    FileDataSourceMetadata,
     Indexer,
     IndexerConfig,
-    SourceIdentifiers,
     Uploader,
     UploaderConfig,
     UploadStager,
@@ -34,6 +29,13 @@ from unstructured_ingest.v2.interfaces import (
     download_responses,
 )
 from unstructured_ingest.v2.logger import logger
+from unstructured_ingest.v2.types.file_data import (
+    BatchFileData,
+    BatchItem,
+    FileData,
+    FileDataSourceMetadata,
+    SourceIdentifiers,
+)
 from unstructured_ingest.v2.utils import get_enhanced_element_id
 
 if TYPE_CHECKING:
@@ -251,8 +253,9 @@ class SQLUploadStager(UploadStager):
             df[column] = df[column].apply(str)
         return df
 
-    def write_output(self, output_path: Path, data: list[dict]) -> None:
+    def write_output(self, output_path: Path, data: list[dict]) -> Path:
         write_data(path=output_path, data=data)
+        return output_path
 
     def run(
         self,
@@ -278,8 +281,10 @@ class SQLUploadStager(UploadStager):
         output_filename = f"{Path(output_filename).stem}{output_filename_suffix}"
         output_path = self.get_output_path(output_filename=output_filename, output_dir=output_dir)
 
-        self.write_output(output_path=output_path, data=df.to_dict(orient="records"))
-        return output_path
+        final_output_path = self.write_output(
+            output_path=output_path, data=df.to_dict(orient="records")
+        )
+        return final_output_path
 
 
 class SQLUploaderConfig(UploaderConfig):

unstructured_ingest/v2/processes/connectors/sql/vastdb.py

@@ -8,9 +8,6 @@ from unstructured_ingest.error import DestinationConnectionError
 from unstructured_ingest.utils.data_prep import split_dataframe
 from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.v2.constants import RECORD_ID_LABEL
-from unstructured_ingest.v2.interfaces import (
-    FileData,
-)
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
@@ -29,6 +26,9 @@ from unstructured_ingest.v2.processes.connectors.sql.sql import (
     SQLUploadStager,
     SQLUploadStagerConfig,
 )
+from unstructured_ingest.v2.types.file_data import (
+    FileData,
+)
 from unstructured_ingest.v2.utils import get_enhanced_element_id
 
 if TYPE_CHECKING:

unstructured_ingest/v2/processes/connectors/vectara.py

@@ -14,7 +14,6 @@ from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.v2.interfaces import (
     AccessConfig,
     ConnectionConfig,
-    FileData,
     Uploader,
     UploaderConfig,
     UploadStager,
@@ -22,6 +21,7 @@ from unstructured_ingest.v2.interfaces import (
 )
 from unstructured_ingest.v2.logger import logger
 from unstructured_ingest.v2.processes.connector_registry import DestinationRegistryEntry
+from unstructured_ingest.v2.types.file_data import FileData
 
 BASE_URL = "https://api.vectara.io/v2"