unstructured-ingest 1.0.35__tar.gz → 1.0.40__tar.gz

{unstructured_ingest-1.0.35 → unstructured_ingest-1.0.40}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: unstructured_ingest
-Version: 1.0.35
+Version: 1.0.40
 Summary: Local ETL data pipeline to get data RAG ready
 Author-email: Unstructured Technologies <devops@unstructuredai.io>
 License-Expression: Apache-2.0

unstructured_ingest-1.0.40/unstructured_ingest/__version__.py

@@ -0,0 +1 @@
+__version__ = "1.0.40"  # pragma: no cover

{unstructured_ingest-1.0.35 → unstructured_ingest-1.0.40}/unstructured_ingest/processes/connectors/confluence.py

@@ -33,6 +33,8 @@ from unstructured_ingest.utils.string_and_date_utils import fix_unescaped_unicod
 
 if TYPE_CHECKING:
     from atlassian import Confluence
+    from bs4 import BeautifulSoup
+    from bs4.element import Tag
 
 CONNECTOR_TYPE = "confluence"
 
@@ -235,11 +237,28 @@ class ConfluenceIndexer(Indexer):
                 yield file_data
 
 
-class ConfluenceDownloaderConfig(DownloaderConfig, HtmlMixin):
+class ConfluenceDownloaderConfig(HtmlMixin, DownloaderConfig):
     max_num_metadata_permissions: int = Field(
         250, description="Approximate maximum number of permissions included in metadata"
     )
 
+    @requires_dependencies(["bs4"])
+    def _find_hyperlink_tags(self, html_soup: "BeautifulSoup") -> list["Tag"]:
+        from bs4.element import Tag
+
+        return [
+            element
+            for element in html_soup.find_all(
+                "a",
+                attrs={
+                    "class": "confluence-embedded-file",
+                    "data-linked-resource-type": "attachment",
+                    "href": True,
+                },
+            )
+            if isinstance(element, Tag)
+        ]
+
 
 @dataclass
 class ConfluenceDownloader(Downloader):

{unstructured_ingest-1.0.35 → unstructured_ingest-1.0.40}/unstructured_ingest/processes/connectors/fsspec/fsspec.py

@@ -343,10 +343,9 @@ class FsspecUploader(Uploader):
             raise self.wrap_error(e=e)
 
     def get_upload_path(self, file_data: FileData) -> Path:
-        upload_path = (
-            Path(self.upload_config.path_without_protocol)
-            / file_data.source_identifiers.relative_path.lstrip("/")
-        )
+        upload_path = Path(
+            self.upload_config.path_without_protocol
+        ) / file_data.source_identifiers.relative_path.lstrip("/")
         updated_upload_path = upload_path.parent / f"{upload_path.name}.json"
         return updated_upload_path
 

{unstructured_ingest-1.0.35 → unstructured_ingest-1.0.40}/unstructured_ingest/processes/connectors/milvus.py

@@ -1,7 +1,7 @@
 import json
 from contextlib import contextmanager
 from dataclasses import dataclass, field
-from typing import TYPE_CHECKING, Any, Generator, Optional, Union
+from typing import TYPE_CHECKING, Any, Generator, Optional
 
 from dateutil import parser
 from pydantic import Field, Secret
@@ -97,10 +97,16 @@ class MilvusUploadStager(UploadStager):
 
     def conform_dict(self, element_dict: dict, file_data: FileData) -> dict:
         working_data = element_dict.copy()
-        if self.upload_stager_config.flatten_metadata and (
-            metadata := working_data.pop("metadata", None)
-        ):
-            working_data.update(flatten_dict(metadata, keys_to_omit=["data_source_record_locator"]))
+
+        if self.upload_stager_config.flatten_metadata:
+            metadata: dict[str, Any] = working_data.pop("metadata", {})
+            flattened_metadata = flatten_dict(
+                metadata,
+                separator="_",
+                flatten_lists=False,
+                remove_none=True,
+            )
+            working_data.update(flattened_metadata)
 
         # TODO: milvus sdk doesn't seem to support defaults via the schema yet,
         #  remove once that gets updated
@@ -154,6 +160,23 @@ class MilvusUploader(Uploader):
     upload_config: MilvusUploaderConfig
     connector_type: str = CONNECTOR_TYPE
 
+    def has_dynamic_fields_enabled(self) -> bool:
+        """Check if the target collection has dynamic fields enabled."""
+        try:
+            with self.get_client() as client:
+                collection_info = client.describe_collection(self.upload_config.collection_name)
+
+                # Check if dynamic field is enabled
+                # The schema info should contain enable_dynamic_field or enableDynamicField
+                schema_info = collection_info.get(
+                    "enable_dynamic_field",
+                    collection_info.get("enableDynamicField", False),
+                )
+                return bool(schema_info)
+        except Exception as e:
+            logger.warning(f"Could not determine if collection has dynamic fields enabled: {e}")
+            return False
+
     @DestinationConnectionError.wrap
     def precheck(self):
         from pymilvus import MilvusException
@@ -164,6 +187,7 @@ class MilvusUploader(Uploader):
                     raise DestinationConnectionError(
                         f"Collection '{self.upload_config.collection_name}' does not exist"
                     )
+
         except MilvusException as milvus_exception:
             raise DestinationConnectionError(
                 f"failed to precheck Milvus: {str(milvus_exception.message)}"
@@ -193,16 +217,66 @@ class MilvusUploader(Uploader):
             )
 
     @requires_dependencies(["pymilvus"], extras="milvus")
-    def insert_results(self, data: Union[dict, list[dict]]):
+    def _prepare_data_for_insert(self, data: list[dict]) -> list[dict]:
+        """
+        Conforms the provided data to the schema of the target Milvus collection.
+        - If dynamic fields are enabled, it ensures JSON-stringified fields are decoded.
+        - If dynamic fields are disabled, it filters out any fields not present in the schema.
+        """
+
+        dynamic_fields_enabled = self.has_dynamic_fields_enabled()
+
+        # If dynamic fields are enabled, 'languages' field needs to be a list
+        if dynamic_fields_enabled:
+            logger.debug("Dynamic fields enabled, ensuring 'languages' field is a list.")
+            prepared_data = []
+            for item in data:
+                new_item = item.copy()
+                if "languages" in new_item and isinstance(new_item["languages"], str):
+                    try:
+                        new_item["languages"] = json.loads(new_item["languages"])
+                    except (json.JSONDecodeError, TypeError):
+                        logger.warning(
+                            f"Could not JSON decode languages field: {new_item['languages']}. "
+                            "Leaving as string.",
+                        )
+                prepared_data.append(new_item)
+            return prepared_data
+
+        # If dynamic fields are not enabled, we need to filter out the metadata fields
+        # to avoid insertion errors for fields not defined in the schema
+        with self.get_client() as client:
+            collection_info = client.describe_collection(
+                self.upload_config.collection_name,
+            )
+        schema_fields = {
+            field["name"]
+            for field in collection_info.get("fields", [])
+            if not field.get("auto_id", False)
+        }
+        # Remove metadata fields that are not part of the base schema
+        filtered_data = []
+        for item in data:
+            filtered_item = {key: value for key, value in item.items() if key in schema_fields}
+            filtered_data.append(filtered_item)
+        return filtered_data
+
+    @requires_dependencies(["pymilvus"], extras="milvus")
+    def insert_results(self, data: list[dict]):
         from pymilvus import MilvusException
 
         logger.info(
             f"uploading {len(data)} entries to {self.connection_config.db_name} "
             f"db in collection {self.upload_config.collection_name}"
         )
+
+        prepared_data = self._prepare_data_for_insert(data=data)
+
         with self.get_client() as client:
             try:
-                res = client.insert(collection_name=self.upload_config.collection_name, data=data)
+                res = client.insert(
+                    collection_name=self.upload_config.collection_name, data=prepared_data
+                )
             except MilvusException as milvus_exception:
                 raise WriteError(
                     f"failed to upload records to Milvus: {str(milvus_exception.message)}"

unstructured_ingest-1.0.40/unstructured_ingest/processes/connectors/notion/types/blocks/synced_block.py

@@ -0,0 +1,109 @@
+# https://developers.notion.com/reference/block#synced-block
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.tags import HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+
+
+@dataclass
+class OriginalSyncedBlock(BlockBase):
+    synced_from: Optional[str] = None
+    children: List[dict] = field(default_factory=list)
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return True
+
+    @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]:
+        return None
+
+
+@dataclass
+class DuplicateSyncedBlock(BlockBase):
+    type: str
+    block_id: str
+
+    @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):
+        """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):
+        """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:
+            # 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-1.0.35 → unstructured_ingest-1.0.40}/unstructured_ingest/processes/connectors/pinecone.py

@@ -240,7 +240,7 @@ class PineconeUploader(VectorDBUploader):
         destination_name: str = "unstructuredautocreated",
         destination_type: Literal["pod", "serverless"] = "serverless",
         serverless_cloud: str = "aws",
-        serverless_region: str = "us-west-2",
+        serverless_region: str = "us-east-1",
         pod_environment: str = "us-east1-gcp",
         pod_type: str = "p1.x1",
         pod_count: int = 1,

{unstructured_ingest-1.0.35 → unstructured_ingest-1.0.40}/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-1.0.35 → unstructured_ingest-1.0.40}/unstructured_ingest/utils/html.py

@@ -12,6 +12,7 @@ from unstructured_ingest.logger import logger
 from unstructured_ingest.utils.dep_check import requires_dependencies
 
 if TYPE_CHECKING:
+    from bs4 import BeautifulSoup
     from bs4.element import Tag
     from requests import Session
 
@@ -96,7 +97,7 @@ class HtmlMixin(BaseModel):
         from bs4 import BeautifulSoup
 
         soup = BeautifulSoup(html, "html.parser")
-        tags = soup.find_all("a", href=True)
+        tags = self._find_hyperlink_tags(soup)
         hrefs = [
             tag["href"]
             for tag in tags
@@ -158,3 +159,15 @@ class HtmlMixin(BaseModel):
             )
             for url_to_download in urls_to_download
         ]
+
+    @requires_dependencies(["bs4"])
+    def _find_hyperlink_tags(self, html_soup: "BeautifulSoup") -> list["Tag"]:
+        """Find hyperlink tags in the HTML.
+
+        Overwrite this method to customize the tag search.
+        """
+        from bs4.element import Tag
+
+        return [
+            element for element in html_soup.find_all("a", href=True) if isinstance(element, Tag)
+        ]

unstructured_ingest-1.0.35/unstructured_ingest/__version__.py

@@ -1 +0,0 @@
-__version__ = "1.0.35"  # pragma: no cover

unstructured_ingest-1.0.35/unstructured_ingest/processes/connectors/notion/types/blocks/synced_block.py

@@ -1,57 +0,0 @@
-# https://developers.notion.com/reference/block#synced-block
-from dataclasses import dataclass, field
-from typing import List, Optional
-
-from htmlBuilder.tags import HtmlTag
-
-from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
-
-
-@dataclass
-class OriginalSyncedBlock(BlockBase):
-    synced_from: Optional[str] = None
-    children: List[dict] = field(default_factory=list)
-
-    @staticmethod
-    def can_have_children() -> bool:
-        return True
-
-    @classmethod
-    def from_dict(cls, data: dict):
-        return cls(children=data["children"])
-
-    def get_html(self) -> Optional[HtmlTag]:
-        return None
-
-
-@dataclass
-class DuplicateSyncedBlock(BlockBase):
-    type: str
-    block_id: str
-
-    @staticmethod
-    def can_have_children() -> bool:
-        return True
-
-    @classmethod
-    def from_dict(cls, data: dict):
-        return cls(**data)
-
-    def get_html(self) -> Optional[HtmlTag]:
-        return None
-
-
-class SyncBlock(BlockBase):
-    @staticmethod
-    def can_have_children() -> bool:
-        return True
-
-    @classmethod
-    def from_dict(cls, data: dict):
-        if "synced_from" in data:
-            return OriginalSyncedBlock.from_dict(data)
-        else:
-            return DuplicateSyncedBlock.from_dict(data)
-
-    def get_html(self) -> Optional[HtmlTag]:
-        return None