unstructured-ingest 1.0.53__py3-none-any.whl → 1.0.55__py3-none-any.whl

unstructured_ingest/__version__.py

@@ -1 +1 @@
-__version__ = "1.0.53"  # pragma: no cover
+__version__ = "1.0.55"  # pragma: no cover

unstructured_ingest/processes/connectors/onedrive.py

@@ -20,6 +20,7 @@ from unstructured_ingest.error import (
     SourceConnectionError,
     SourceConnectionNetworkError,
 )
+from unstructured_ingest.errors_v2 import UserAuthError
 from unstructured_ingest.interfaces import (
     AccessConfig,
     ConnectionConfig,
@@ -114,12 +115,27 @@ class OnedriveConnectionConfig(ConnectionConfig):
             except ValueError as exc:
                 logger.error("Couldn't set up credentials.")
                 raise exc
+            
             if "error" in token:
-                raise SourceConnectionNetworkError(
-                    "failed to fetch token, {}: {}".format(
-                        token["error"], token["error_description"]
+                error_codes = token.get("error_codes", [])
+                error_type = token.get("error", "")
+                error_description = token.get("error_description", "")
+                
+                # 7000215: Invalid client secret provided
+                # 7000218: Invalid client id provided
+                # 700016: Application not found in directory
+                # 90002: Tenant not found
+                auth_error_codes = [7000215, 7000218, 700016, 90002]
+                
+                if (any(code in error_codes for code in auth_error_codes) or 
+                    error_type in ["invalid_client", "unauthorized_client", "invalid_grant"]):
+                    raise UserAuthError(
+                        f"Authentication failed: {error_type}: {error_description}"
+                    )
+                else:
+                    raise SourceConnectionNetworkError(
+                        f"Failed to fetch token: {error_type}: {error_description}"
                     )
-                )
             return token
 
     @requires_dependencies(["office365"], extras="onedrive")

unstructured_ingest/processes/connectors/sharepoint.py

@@ -13,6 +13,7 @@ from unstructured_ingest.error import (
     SourceConnectionError,
     SourceConnectionNetworkError,
 )
+from unstructured_ingest.errors_v2 import UserAuthError, UserError
 from unstructured_ingest.logger import logger
 from unstructured_ingest.processes.connector_registry import (
     SourceRegistryEntry,
@@ -30,6 +31,7 @@ from unstructured_ingest.utils.dep_check import requires_dependencies
 if TYPE_CHECKING:
     from office365.onedrive.driveitems.driveItem import DriveItem
     from office365.onedrive.sites.site import Site
+    from office365.runtime.client_request_exception import ClientRequestException
 
 CONNECTOR_TYPE = "sharepoint"
 LEGACY_DEFAULT_PATH = "Shared Documents"
@@ -82,8 +84,8 @@ class SharepointConnectionConfig(OnedriveConnectionConfig):
 
 
 class SharepointIndexerConfig(OnedriveIndexerConfig):
-    pass
-
+    # TODO: We can probably make path non-optional on OnedriveIndexerConfig once tested
+    path: str = Field(default="")
 
 @dataclass
 class SharepointIndexer(OnedriveIndexer):
@@ -91,6 +93,85 @@ class SharepointIndexer(OnedriveIndexer):
     index_config: SharepointIndexerConfig
     connector_type: str = CONNECTOR_TYPE
 
+    def _handle_client_request_exception(self, e: ClientRequestException, context: str) -> None:
+        """Convert ClientRequestException to appropriate user-facing error based on HTTP status."""
+        if hasattr(e, "response") and e.response is not None and hasattr(e.response, "status_code"):
+            status_code = e.response.status_code
+            if status_code == 401:
+                raise UserAuthError(
+                    f"Unauthorized access to {context}. Check client credentials and permissions"
+                )
+            elif status_code == 403:
+                raise UserAuthError(
+                    f"Access forbidden to {context}. "
+                    f"Check app permissions (Sites.Read.All required)"
+                )
+            elif status_code == 404:
+                raise UserError(f"Not found: {context}")
+
+        raise UserError(f"Failed to access {context}: {str(e)}")
+
+    def _is_root_path(self, path: str) -> bool:
+        """Check if the path represents root access (empty string or legacy default)."""
+        return not path or not path.strip() or path == LEGACY_DEFAULT_PATH
+    
+    def _get_target_drive_item(self, site_drive_item: DriveItem, path: str) -> DriveItem:
+        """Get the drive item to search in based on the path."""
+        if self._is_root_path(path):
+            return site_drive_item
+        else:
+            return site_drive_item.get_by_path(path).get().execute_query()
+    
+    def _validate_folder_path(self, site_drive_item: DriveItem, path: str) -> None:
+        """Validate that a specific folder path exists and is accessible."""
+        from office365.runtime.client_request_exception import ClientRequestException
+
+        try:
+            path_item = site_drive_item.get_by_path(path).get().execute_query()
+            if path_item is None or not hasattr(path_item, "is_folder"):
+                raise UserError(
+                    f"SharePoint path '{path}' not found in site {self.connection_config.site}. "
+                    f"Check that the path exists and you have access to it"
+                )
+            logger.info(f"SharePoint folder path '{path}' validated successfully")
+        except ClientRequestException as e:
+            logger.error(f"Failed to access SharePoint path '{path}': {e}")
+            self._handle_client_request_exception(e, f"SharePoint path '{path}'")
+        except Exception as e:
+            logger.error(f"Unexpected error accessing SharePoint path '{path}': {e}")
+            raise UserError(f"Failed to validate SharePoint path '{path}': {str(e)}")
+
+    @requires_dependencies(["office365"], extras="sharepoint")
+    def precheck(self) -> None:
+        """Validate SharePoint connection before indexing."""
+        from office365.runtime.client_request_exception import ClientRequestException
+
+        # Validate authentication - this call will raise UserAuthError if invalid
+        self.connection_config.get_token()
+
+        try:
+            client = self.connection_config.get_client()
+            client_site = client.sites.get_by_url(self.connection_config.site).get().execute_query()
+            site_drive_item = self.connection_config._get_drive_item(client_site)
+
+            path = self.index_config.path
+            if not self._is_root_path(path):
+                self._validate_folder_path(site_drive_item, path)
+
+            logger.info(
+                f"SharePoint connection validated successfully for site: "
+                f"{self.connection_config.site}"
+            )
+
+        except ClientRequestException as e:
+            logger.error(f"SharePoint precheck failed for site: {self.connection_config.site}")
+            self._handle_client_request_exception(
+                e, f"SharePoint site {self.connection_config.site}"
+            )
+        except Exception as e:
+            logger.error(f"Unexpected error during SharePoint precheck: {e}", exc_info=True)
+            raise UserError(f"Failed to validate SharePoint connection: {str(e)}")
+
     @requires_dependencies(["office365"], extras="sharepoint")
     async def run_async(self, **kwargs: Any) -> AsyncIterator[FileData]:
         from office365.runtime.client_request_exception import ClientRequestException
@@ -113,11 +194,11 @@ class SharepointIndexer(OnedriveIndexer):
             )
 
         path = self.index_config.path
-        # Deprecated sharepoint sdk needed a default path. Microsoft Graph SDK does not.
-        if path and path != LEGACY_DEFAULT_PATH:
-            site_drive_item = site_drive_item.get_by_path(path).get().execute_query()
+        target_drive_item = await asyncio.to_thread(
+            self._get_target_drive_item, site_drive_item, path
+        )
 
-        for drive_item in site_drive_item.get_files(
+        for drive_item in target_drive_item.get_files(
             recursive=self.index_config.recursive
         ).execute_query():
             file_data = await self.drive_item_to_file_data(drive_item=drive_item)

{unstructured_ingest-1.0.53.dist-info → unstructured_ingest-1.0.55.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: unstructured_ingest
-Version: 1.0.53
+Version: 1.0.55
 Summary: Local ETL data pipeline to get data RAG ready
 Author-email: Unstructured Technologies <devops@unstructuredai.io>
 License-Expression: Apache-2.0

{unstructured_ingest-1.0.53.dist-info → unstructured_ingest-1.0.55.dist-info}/RECORD

@@ -1,5 +1,5 @@
 unstructured_ingest/__init__.py,sha256=U4S_2y3zgLZVfMenHRaJFBW8yqh2mUBuI291LGQVOJ8,35
-unstructured_ingest/__version__.py,sha256=R6_J5XcJv3o4se9ZaxG3ld-O1m1WwU2CyxpqODyEhmg,43
+unstructured_ingest/__version__.py,sha256=WcrHy96lfPCfFMicHHTxBEY2M7zSC_2LVKoyMsYUTrI,43
 unstructured_ingest/error.py,sha256=qDncnJgbf5ils956RcO2CGlAKYDT5OaEM9Clv1JVTNc,1448
 unstructured_ingest/errors_v2.py,sha256=9RuRCi7lbDxCguDz07y5RiHoQiFIOWwOD7xqzJ2B3Yw,436
 unstructured_ingest/logger.py,sha256=7e_7UeK6hVOd5BQ6i9NzRUAPCS_DF839Y8TjUDywraY,1428
@@ -79,12 +79,12 @@ unstructured_ingest/processes/connectors/local.py,sha256=CesMduUiSPqdJpqIyW28icG
 unstructured_ingest/processes/connectors/milvus.py,sha256=L-PM5osheNyNsLGYZmiF3rRmeulp7Ejk92JCoaQ_F9Y,12075
 unstructured_ingest/processes/connectors/mongodb.py,sha256=idjolwS5TXShcIz2jR_socSgh8HOzJwyOnzE1qLUPBw,15362
 unstructured_ingest/processes/connectors/neo4j.py,sha256=ztxvI9KY8RF5kYUuMGSzzN5mz7Fu_4Ai9P7dqCpJLc0,20267
-unstructured_ingest/processes/connectors/onedrive.py,sha256=nZt6hsFMlURgB5-BioFBzJySieRVU8xi99QhOCtorxQ,19343
+unstructured_ingest/processes/connectors/onedrive.py,sha256=fGwa-x9D3gyLQtaSXbz6pfiFiLpnO2GVtJmU5kb-qd0,20197
 unstructured_ingest/processes/connectors/outlook.py,sha256=6HHubZI_zttEfYp0XNd4Y1vhjsS8uSg7aZ2LBrTjfHk,9376
 unstructured_ingest/processes/connectors/pinecone.py,sha256=jCabAqKQyBFzaGjphxLMr57y7P0Z15Jd9Jj-JM40YnU,15090
 unstructured_ingest/processes/connectors/redisdb.py,sha256=rTihbfv0Mlk1eo5Izn-JXRu5Ad5C-KD58nSqeKsaZJ8,8024
 unstructured_ingest/processes/connectors/salesforce.py,sha256=N_UoebrhzXZNWw-X7lg8_qAziXx5L_d8XHnHWKNNYR8,11767
-unstructured_ingest/processes/connectors/sharepoint.py,sha256=Wgv9Pih9S9FmQJud1bg7kj_qqi55d7QZ48LqlTU_mk0,6509
+unstructured_ingest/processes/connectors/sharepoint.py,sha256=IV6gs4vx4q-QEDwA-Rm6yYCwzopuVl8bKC8CcBU1Lkk,10677
 unstructured_ingest/processes/connectors/slack.py,sha256=oboIfX7ayBMK0te5Nv50iyL3FQJFXJbRxZSQaCMp3kM,9318
 unstructured_ingest/processes/connectors/utils.py,sha256=TAd0hb1f291N-q7-TUe6JKSCGkhqDyo7Ij8zmliBZUc,2071
 unstructured_ingest/processes/connectors/vectara.py,sha256=xrC6jkgW8BII4UjdzUelDu122xT484cpfMTK2wl-sko,12292
@@ -232,8 +232,8 @@ unstructured_ingest/utils/ndjson.py,sha256=nz8VUOPEgAFdhaDOpuveknvCU4x82fVwqE01q
 unstructured_ingest/utils/pydantic_models.py,sha256=BT_j15e4rX40wQbt8LUXbqfPhA3rJn1PHTI_G_A_EHY,1720
 unstructured_ingest/utils/string_and_date_utils.py,sha256=oXOI6rxXq-8ncbk7EoJK0WCcTXWj75EzKl8pfQMID3U,2522
 unstructured_ingest/utils/table.py,sha256=WZechczgVFvlodUWFcsnCGvBNh1xRm6hr0VbJTPxKAc,3669
-unstructured_ingest-1.0.53.dist-info/METADATA,sha256=FTZfBOPcfl-pbxoBAIK-d0MwQKfN5nbJXQh92H1QwMs,8842
-unstructured_ingest-1.0.53.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-unstructured_ingest-1.0.53.dist-info/entry_points.txt,sha256=gUAAFnjFPnBgThJSEbw0N5ZjxtaKlT1s9e05_arQrNw,70
-unstructured_ingest-1.0.53.dist-info/licenses/LICENSE.md,sha256=SxkKP_62uIAKb9mb1eH7FH4Kn2aYT09fgjKpJt5PyTk,11360
-unstructured_ingest-1.0.53.dist-info/RECORD,,
+unstructured_ingest-1.0.55.dist-info/METADATA,sha256=x-w3d3LQjOuPDVtLQbGgyeCzfMlgVF4OyhGWwm91o8w,8842
+unstructured_ingest-1.0.55.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+unstructured_ingest-1.0.55.dist-info/entry_points.txt,sha256=gUAAFnjFPnBgThJSEbw0N5ZjxtaKlT1s9e05_arQrNw,70
+unstructured_ingest-1.0.55.dist-info/licenses/LICENSE.md,sha256=SxkKP_62uIAKb9mb1eH7FH4Kn2aYT09fgjKpJt5PyTk,11360
+unstructured_ingest-1.0.55.dist-info/RECORD,,