unstructured-ingest 1.0.34__py3-none-any.whl → 1.0.37__py3-none-any.whl

unstructured_ingest/__version__.py

@@ -1 +1 @@
-__version__ = "1.0.34"  # pragma: no cover
+__version__ = "1.0.37"  # pragma: no cover

unstructured_ingest/processes/connectors/databricks/volumes.py

@@ -196,9 +196,10 @@ class DatabricksVolumesUploader(Uploader, ABC):
     connection_config: DatabricksVolumesConnectionConfig
 
     def get_output_path(self, file_data: FileData) -> str:
-        if file_data.source_identifiers.fullpath:
+        if file_data.source_identifiers.relative_path:
             return os.path.join(
-                self.upload_config.path, f"{file_data.source_identifiers.fullpath}.json"
+                self.upload_config.path,
+                f"{file_data.source_identifiers.relative_path.lstrip('/')}.json",
             )
         else:
             return os.path.join(

unstructured_ingest/processes/connectors/fsspec/fsspec.py

@@ -345,7 +345,7 @@ class FsspecUploader(Uploader):
     def get_upload_path(self, file_data: FileData) -> Path:
         upload_path = Path(
             self.upload_config.path_without_protocol
-        ) / file_data.source_identifiers.fullpath.lstrip("/")
+        ) / file_data.source_identifiers.relative_path.lstrip("/")
         updated_upload_path = upload_path.parent / f"{upload_path.name}.json"
         return updated_upload_path
 

unstructured_ingest/processes/connectors/notion/types/blocks/synced_block.py

@@ -18,6 +18,12 @@ class OriginalSyncedBlock(BlockBase):
 
     @classmethod
     def from_dict(cls, data: dict):
+        """Create OriginalSyncedBlock from dictionary data.
+        
+        Original blocks contain children content.
+        """
+        if "children" not in data:
+             raise ValueError(f"OriginalSyncedBlock data missing 'children': {data}")
         return cls(children=data["children"])
 
     def get_html(self) -> Optional[HtmlTag]:
@@ -31,27 +37,74 @@ class DuplicateSyncedBlock(BlockBase):
 
     @staticmethod
     def can_have_children() -> bool:
+        """Check if duplicate synced blocks can have children.
+        
+        Duplicate blocks themselves don't have children directly fetched here,
+        but they represent content that does, so Notion API might report has_children=True
+        on the parent block object. The actual children are fetched from the original block.
+        """
         return True
 
     @classmethod
     def from_dict(cls, data: dict):
-        return cls(**data)
+        """Create DuplicateSyncedBlock from dictionary data.
+        
+        Duplicate blocks contain a 'synced_from' reference.
+        """
+        synced_from_data = data.get("synced_from")
+        if not synced_from_data or not isinstance(synced_from_data, dict):
+            raise ValueError(f"Invalid data structure for DuplicateSyncedBlock: {data}")
+        # Ensure required keys are present in the nested dictionary
+        if "type" not in synced_from_data or "block_id" not in synced_from_data:
+            raise ValueError(
+                f"Missing 'type' or 'block_id' in synced_from data: {synced_from_data}"
+            )
+        return cls(type=synced_from_data["type"], block_id=synced_from_data["block_id"])
 
     def get_html(self) -> Optional[HtmlTag]:
+        """Get HTML representation of the duplicate synced block.
+        
+        HTML representation might need fetching the original block's content,
+        which is outside the scope of this simple data class.
+        """
         return None
 
 
 class SyncBlock(BlockBase):
     @staticmethod
     def can_have_children() -> bool:
+        """Check if synced blocks can have children.
+        
+        Synced blocks (both original and duplicate) can conceptually have children.
+        """
         return True
 
     @classmethod
     def from_dict(cls, data: dict):
-        if "synced_from" in data:
+        """Create appropriate SyncedBlock subclass from dictionary data.
+        
+        Determine if it's a duplicate (has 'synced_from') or original (has 'children').
+        """
+        if data.get("synced_from") is not None:
+            # It's a duplicate block containing a reference
+            return DuplicateSyncedBlock.from_dict(data)
+        elif "children" in data:
+            # It's an original block containing children
             return OriginalSyncedBlock.from_dict(data)
         else:
-            return DuplicateSyncedBlock.from_dict(data)
+            # Handle cases where neither 'synced_from' nor 'children' are present.
+            # Notion API might return this for an empty original synced block.
+            # Let's treat it as an empty OriginalSyncedBlock.
+            # If this assumption is wrong, errors might occur later.
+            # Consider logging a warning here if strictness is needed.
+            return OriginalSyncedBlock(children=[])
+
 
     def get_html(self) -> Optional[HtmlTag]:
+        """Get HTML representation of the synced block.
+        
+        The specific instance returned by from_dict (Original or Duplicate)
+        will handle its own get_html logic.
+        This method on the base SyncBlock might not be directly called.
+        """
         return None

unstructured_ingest/processes/connectors/onedrive.py

@@ -369,15 +369,14 @@ class OnedriveUploader(Uploader):
 
         # Use the remote_url from upload_config as the base destination folder
         base_destination_folder = self.upload_config.url
-
-        # Use the file's full path to maintain directory structure, if needed
-        if file_data.source_identifiers and file_data.source_identifiers.fullpath:
-            # Combine the base destination folder with the file's full path
+        # Use the file's relative path to maintain directory structure, if needed
+        if file_data.source_identifiers and file_data.source_identifiers.relative_path:
+            # Combine the base destination folder with the file's relative path
             destination_path = Path(base_destination_folder) / Path(
-                f"{file_data.source_identifiers.fullpath}.json"
+                f"{file_data.source_identifiers.relative_path}.json"
             )
         else:
-            # If no full path is provided, upload directly to the base destination folder
+            # If no relative path is provided, upload directly to the base destination folder
             destination_path = Path(base_destination_folder) / f"{path.name}.json"
 
         destination_folder = destination_path.parent

unstructured_ingest/processes/connectors/sharepoint.py

@@ -29,6 +29,7 @@ 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
 
 CONNECTOR_TYPE = "sharepoint"
 LEGACY_DEFAULT_PATH = "Shared Documents"
@@ -51,6 +52,33 @@ class SharepointConnectionConfig(OnedriveConnectionConfig):
                     https://[tenant]-admin.sharepoint.com.\
                     This requires the app to be registered at a tenant level"
     )
+    library: Optional[str] = Field(
+        default=None,
+        description="Sharepoint library name. If not provided, the default \
+                    drive will be used.",
+    )
+
+    def _get_drive_item(self, client_site: Site) -> DriveItem:
+        """Helper method to get the drive item for the specified library or default drive."""
+        site_drive_item = None
+        if self.library:
+            for drive in client_site.drives.get().execute_query():
+                if drive.name == self.library:
+                    logger.info(f"Found the requested library: {self.library}")
+                    site_drive_item = drive.get().execute_query().root
+                    break
+
+        # If no specific library was found or requested, use the default drive
+        if not site_drive_item:
+            if self.library:
+                logger.warning(
+                    f"Library '{self.library}' not found in site '{self.site}'. "
+                    "Using the default drive instead."
+                )
+
+            site_drive_item = client_site.drive.get().execute_query().root
+
+        return site_drive_item
 
 
 class SharepointIndexerConfig(OnedriveIndexerConfig):
@@ -76,8 +104,8 @@ class SharepointIndexer(OnedriveIndexer):
 
         client = await asyncio.to_thread(self.connection_config.get_client)
         try:
-            site = client.sites.get_by_url(self.connection_config.site).get().execute_query()
-            site_drive_item = site.drive.get().execute_query().root
+            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)
         except ClientRequestException:
             logger.info("Site not found")
 
@@ -118,8 +146,8 @@ class SharepointDownloader(OnedriveDownloader):
         client = self.connection_config.get_client()
 
         try:
-            site = client.sites.get_by_url(self.connection_config.site).get().execute_query()
-            site_drive_item = site.drive.get().execute_query().root
+            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)
         except ClientRequestException:
             logger.info("Site not found")
         file = site_drive_item.get_by_path(server_relative_path).get().execute_query()

{unstructured_ingest-1.0.34.dist-info → unstructured_ingest-1.0.37.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: unstructured_ingest
-Version: 1.0.34
+Version: 1.0.37
 Summary: Local ETL data pipeline to get data RAG ready
 Author-email: Unstructured Technologies <devops@unstructuredai.io>
 License-Expression: Apache-2.0

{unstructured_ingest-1.0.34.dist-info → unstructured_ingest-1.0.37.dist-info}/RECORD

@@ -1,5 +1,5 @@
 unstructured_ingest/__init__.py,sha256=U4S_2y3zgLZVfMenHRaJFBW8yqh2mUBuI291LGQVOJ8,35
-unstructured_ingest/__version__.py,sha256=S3Vgmk2V2EWfbef_sUbnJb_d5x0m64Z8D_xx-_9kXOM,43
+unstructured_ingest/__version__.py,sha256=De73lzt6X-hjX65lK6tF1Rs23QRJQqTCx5Zn-JyPtFI,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=LluTLKv4g7FbJb4A6vuSxI9
 unstructured_ingest/processes/connectors/milvus.py,sha256=Jr9cul7By03tGAPFnFBoqncnNWwbhKd-qbmkuqnin8U,8908
 unstructured_ingest/processes/connectors/mongodb.py,sha256=1g_5bfbS6lah3nsOXqLAanR3zNYJ47_Njw_uV-uj3_U,14324
 unstructured_ingest/processes/connectors/neo4j.py,sha256=ztxvI9KY8RF5kYUuMGSzzN5mz7Fu_4Ai9P7dqCpJLc0,20267
-unstructured_ingest/processes/connectors/onedrive.py,sha256=k0bhQCCSIgmHAk3lQd4CMA3dc4fPAjegNlLxlDWGowc,19284
+unstructured_ingest/processes/connectors/onedrive.py,sha256=JIADpc31PI9Yzr0raF6bSqzes2jhfcniUzew1aKVWeI,19305
 unstructured_ingest/processes/connectors/outlook.py,sha256=zHM5frO7CqQG0-KcTyX49aZeSlsvVrl8kh_lR_ESgQw,9275
 unstructured_ingest/processes/connectors/pinecone.py,sha256=pSREUNsQqel6q1EFZsFWelg-uZgGubQY5m_6nVnBFKs,15090
 unstructured_ingest/processes/connectors/redisdb.py,sha256=rTihbfv0Mlk1eo5Izn-JXRu5Ad5C-KD58nSqeKsaZJ8,8024
 unstructured_ingest/processes/connectors/salesforce.py,sha256=OaKEWCqZrirHqFJ650K5jSPwYlWefPOapas8Y-4D9oc,11661
-unstructured_ingest/processes/connectors/sharepoint.py,sha256=jI-erp4YUfHxPeUTcfHSPEG3w0wjSBYfAnMg1WT6lfw,4996
+unstructured_ingest/processes/connectors/sharepoint.py,sha256=vIfLIactYXcdetccHvKlYOay6NOzGj2X0CkXbY0KuRo,6213
 unstructured_ingest/processes/connectors/slack.py,sha256=EkFj9PcAu5_gF2xLogikKDADLbJYq-_jvchzYrTdLO4,9224
 unstructured_ingest/processes/connectors/utils.py,sha256=TAd0hb1f291N-q7-TUe6JKSCGkhqDyo7Ij8zmliBZUc,2071
 unstructured_ingest/processes/connectors/vectara.py,sha256=xrC6jkgW8BII4UjdzUelDu122xT484cpfMTK2wl-sko,12292
@@ -92,7 +92,7 @@ unstructured_ingest/processes/connectors/assets/__init__.py,sha256=47DEQpj8HBSa-
 unstructured_ingest/processes/connectors/assets/databricks_delta_table_schema.sql,sha256=8a9HTcRWA6IuswSD632b_uZSO6Dax_0rUYnflqktcek,226
 unstructured_ingest/processes/connectors/assets/weaviate_collection_config.json,sha256=SJlIO0kXxy866tWQ8bEzvwLwflsoUMIS-OKlxMvHIuE,504
 unstructured_ingest/processes/connectors/databricks/__init__.py,sha256=RtKAPyNtXh6fzEsOQ08pA0-vC1uMr3KqYG6cqiBoo70,2133
-unstructured_ingest/processes/connectors/databricks/volumes.py,sha256=fZeXRozTUM3JeZlmsxhn_glqRhxr8CGG-8I8QRhRcP8,8232
+unstructured_ingest/processes/connectors/databricks/volumes.py,sha256=yT5JFbVzAEOJsKjfGH8KG3eQfKaTNFEsg_FVDPVK7Xs,8271
 unstructured_ingest/processes/connectors/databricks/volumes_aws.py,sha256=WhGTp6aRTLSdc4GChCL4mz2b-IanderW8j1IqezX6YA,2958
 unstructured_ingest/processes/connectors/databricks/volumes_azure.py,sha256=pF2d6uAIbwJJUeOIG5xknUMCGc5d9Aztmc2776wp-a0,3740
 unstructured_ingest/processes/connectors/databricks/volumes_gcp.py,sha256=y9AvVl6PtnIxlTlrPj_wyHBDBRJNq3uoTOuZwTryNg8,2994
@@ -109,7 +109,7 @@ unstructured_ingest/processes/connectors/fsspec/__init__.py,sha256=3HTdw4L4mdN4W
 unstructured_ingest/processes/connectors/fsspec/azure.py,sha256=31VNiG5YnXfhrFX7QJ2O1ubeWHxbe1sYVIztefbscAQ,7148
 unstructured_ingest/processes/connectors/fsspec/box.py,sha256=1gLS7xR2vbjgKBrQ4ZpI1fKTsJuIDfXuAzx_a4FzxG4,5873
 unstructured_ingest/processes/connectors/fsspec/dropbox.py,sha256=HwwKjQmjM7yFk9Esh_F20xDisRPXGUkFduzaasByRDE,8355
-unstructured_ingest/processes/connectors/fsspec/fsspec.py,sha256=4K8Q2D_6_HCqTVM3HBJv3SNz9gjbQhk44nzeSheDpzA,14462
+unstructured_ingest/processes/connectors/fsspec/fsspec.py,sha256=NbId5WMq6M5kF3fYAwSUuaL2e_gutgmTATrE_X8okGY,14467
 unstructured_ingest/processes/connectors/fsspec/gcs.py,sha256=ouxISCKpZTAj3T6pWGYbASu93wytJjl5WSICvQcrgfE,7172
 unstructured_ingest/processes/connectors/fsspec/s3.py,sha256=2ZV6b2E2pIsf_ab1Lty74FwpMnJZhpQUdamPgpwcKsQ,7141
 unstructured_ingest/processes/connectors/fsspec/sftp.py,sha256=pR_a2SgLjt8ffNkariHrPB1E0HVSTj5h3pt7KxTU3TI,6371
@@ -166,7 +166,7 @@ unstructured_ingest/processes/connectors/notion/types/blocks/numbered_list.py,sh
 unstructured_ingest/processes/connectors/notion/types/blocks/paragraph.py,sha256=qvc4orjP2XcbaeBWor-a3xAEglLkyb-epknm7SXgU1E,992
 unstructured_ingest/processes/connectors/notion/types/blocks/pdf.py,sha256=St43RmpefAzDwJKTwz2CdGVm-xeUwHkYgtQtLYQbnw0,1661
 unstructured_ingest/processes/connectors/notion/types/blocks/quote.py,sha256=yl7npmdcO6oFNgTNGVN_Ihvzexv12Xwg1r4NWAOjILQ,1176
-unstructured_ingest/processes/connectors/notion/types/blocks/synced_block.py,sha256=Rc3xyKtnOwovx-O-dzmS9pX0h4-s41YnWmmEz5TYxdU,1333
+unstructured_ingest/processes/connectors/notion/types/blocks/synced_block.py,sha256=aHu4yg8N1EDqZmMEHK7dd7fiQ8Mc8otHQLJPRDbkaT8,4049
 unstructured_ingest/processes/connectors/notion/types/blocks/table.py,sha256=eYUlRp4uCwjy_eB0mLh7MGMe1qrr_hnOxXS5RfUM2DQ,1724
 unstructured_ingest/processes/connectors/notion/types/blocks/table_of_contents.py,sha256=bR5DdecXFz468okM5WOs10DK8_14Dj7OCLSRusMZzsk,534
 unstructured_ingest/processes/connectors/notion/types/blocks/template.py,sha256=bq2Vh2X7ptpofs9OZnATHySZe2DzbOLsNNfpEI70NgM,968
@@ -231,8 +231,8 @@ 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-1.0.34.dist-info/METADATA,sha256=Pw-KP4al9gteAFj6lqY7xkFRjWj1rTAgN960UsAUZAM,8747
-unstructured_ingest-1.0.34.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-unstructured_ingest-1.0.34.dist-info/entry_points.txt,sha256=gUAAFnjFPnBgThJSEbw0N5ZjxtaKlT1s9e05_arQrNw,70
-unstructured_ingest-1.0.34.dist-info/licenses/LICENSE.md,sha256=SxkKP_62uIAKb9mb1eH7FH4Kn2aYT09fgjKpJt5PyTk,11360
-unstructured_ingest-1.0.34.dist-info/RECORD,,
+unstructured_ingest-1.0.37.dist-info/METADATA,sha256=wct0um6qunVNGSNozJ0a3UatsfCHDyXG7p9XMNBCTcU,8747
+unstructured_ingest-1.0.37.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+unstructured_ingest-1.0.37.dist-info/entry_points.txt,sha256=gUAAFnjFPnBgThJSEbw0N5ZjxtaKlT1s9e05_arQrNw,70
+unstructured_ingest-1.0.37.dist-info/licenses/LICENSE.md,sha256=SxkKP_62uIAKb9mb1eH7FH4Kn2aYT09fgjKpJt5PyTk,11360
+unstructured_ingest-1.0.37.dist-info/RECORD,,