unstructured-ingest 0.5.17__py3-none-any.whl → 0.5.18__py3-none-any.whl

unstructured_ingest/__version__.py

@@ -1 +1 @@
-__version__ = "0.5.17"  # pragma: no cover
+__version__ = "0.5.18"  # pragma: no cover

unstructured_ingest/utils/data_prep.py

@@ -192,6 +192,16 @@ def get_data(path: Union[Path, str]) -> list[dict]:
             logger.warning(f"failed to read {path} as parquet: {e}")
 
 
+def get_json_data(path: Path) -> list[dict]:
+    with path.open() as f:
+        if path.suffix == ".json":
+            return json.load(f)
+        elif path.suffix == ".ndjson":
+            return ndjson.load(f)
+        else:
+            raise ValueError(f"Unsupported file type: {path}")
+
+
 def get_data_df(path: Path) -> pd.DataFrame:
     with path.open() as f:
         if path.suffix == ".json":

unstructured_ingest/v2/processes/connectors/delta_table.py

@@ -1,4 +1,3 @@
-import json
 import os
 import traceback
 from dataclasses import dataclass, field
@@ -11,7 +10,7 @@ import pandas as pd
 from pydantic import Field, Secret
 
 from unstructured_ingest.error import DestinationConnectionError
-from unstructured_ingest.utils.data_prep import get_data_df
+from unstructured_ingest.utils.data_prep import get_data_df, get_json_data
 from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.utils.table import convert_to_pandas_dataframe
 from unstructured_ingest.v2.interfaces import (
@@ -86,9 +85,7 @@ class DeltaTableUploadStager(UploadStager):
         output_filename: str,
         **kwargs: Any,
     ) -> Path:
-        with open(elements_filepath) as elements_file:
-            elements_contents = json.load(elements_file)
-
+        elements_contents = get_json_data(elements_filepath)
         output_path = Path(output_dir) / Path(f"{output_filename}.parquet")
 
         df = convert_to_pandas_dataframe(elements_dict=elements_contents)

unstructured_ingest/v2/processes/connectors/neo4j.py

@@ -1,7 +1,6 @@
 from __future__ import annotations
 
 import asyncio
-import json
 import uuid
 from collections import defaultdict
 from contextlib import asynccontextmanager
@@ -14,7 +13,7 @@ from pydantic import BaseModel, ConfigDict, Field, Secret, field_validator
 
 from unstructured_ingest.error import DestinationConnectionError
 from unstructured_ingest.logger import logger
-from unstructured_ingest.utils.data_prep import batch_generator
+from unstructured_ingest.utils.data_prep import batch_generator, get_json_data
 from unstructured_ingest.utils.dep_check import requires_dependencies
 from unstructured_ingest.v2.interfaces import (
     AccessConfig,
@@ -97,8 +96,7 @@ class Neo4jUploadStager(UploadStager):
         output_filename: str,
         **kwargs: Any,
     ) -> Path:
-        with elements_filepath.open() as file:
-            elements = json.load(file)
+        elements = get_json_data(elements_filepath)
 
         nx_graph = self._create_lexical_graph(
             elements, self._create_document_node(file_data=file_data)
@@ -294,8 +292,7 @@ class Neo4jUploader(Uploader):
         return True
 
     async def run_async(self, path: Path, file_data: FileData, **kwargs) -> None:  # type: ignore
-        with path.open() as file:
-            staged_data = json.load(file)
+        staged_data = get_json_data(path)
 
         graph_data = _GraphData.model_validate(staged_data)
         async with self.connection_config.get_client() as client:

unstructured_ingest/v2/processes/connectors/onedrive.py

@@ -35,6 +35,10 @@ from unstructured_ingest.v2.processes.connector_registry import (
     DestinationRegistryEntry,
     SourceRegistryEntry,
 )
+from unstructured_ingest.v2.processes.utils.blob_storage import (
+    BlobStoreUploadStager,
+    BlobStoreUploadStagerConfig,
+)
 
 if TYPE_CHECKING:
     from office365.graph_client import GraphClient
@@ -428,4 +432,6 @@ onedrive_destination_entry = DestinationRegistryEntry(
     connection_config=OnedriveConnectionConfig,
     uploader=OnedriveUploader,
     uploader_config=OnedriveUploaderConfig,
+    upload_stager_config=BlobStoreUploadStagerConfig,
+    upload_stager=BlobStoreUploadStager,
 )

{unstructured_ingest-0.5.17.dist-info → unstructured_ingest-0.5.18.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: unstructured-ingest
-Version: 0.5.17
+Version: 0.5.18
 Summary: A library that prepares raw documents for downstream ML tasks.
 Home-page: https://github.com/Unstructured-IO/unstructured-ingest
 Author: Unstructured Technologies
@@ -25,10 +25,10 @@ License-File: LICENSE.md
 Requires-Dist: python-dateutil
 Requires-Dist: opentelemetry-sdk
 Requires-Dist: pandas
-Requires-Dist: click
 Requires-Dist: dataclasses_json
-Requires-Dist: pydantic>=2.7
 Requires-Dist: tqdm
+Requires-Dist: click
+Requires-Dist: pydantic>=2.7
 Provides-Extra: remote
 Requires-Dist: unstructured-client>=0.30.0; extra == "remote"
 Provides-Extra: csv
@@ -66,8 +66,8 @@ Requires-Dist: pyairtable; extra == "airtable"
 Provides-Extra: astradb
 Requires-Dist: astrapy; extra == "astradb"
 Provides-Extra: azure
-Requires-Dist: fsspec; extra == "azure"
 Requires-Dist: adlfs; extra == "azure"
+Requires-Dist: fsspec; extra == "azure"
 Provides-Extra: azure-ai-search
 Requires-Dist: azure-search-documents; extra == "azure-ai-search"
 Provides-Extra: biomed
@@ -81,29 +81,29 @@ Requires-Dist: chromadb; extra == "chroma"
 Provides-Extra: clarifai
 Requires-Dist: clarifai; extra == "clarifai"
 Provides-Extra: confluence
-Requires-Dist: atlassian-python-api; extra == "confluence"
 Requires-Dist: requests; extra == "confluence"
+Requires-Dist: atlassian-python-api; extra == "confluence"
 Provides-Extra: couchbase
 Requires-Dist: couchbase; extra == "couchbase"
 Provides-Extra: delta-table
-Requires-Dist: boto3; extra == "delta-table"
 Requires-Dist: deltalake; extra == "delta-table"
+Requires-Dist: boto3; extra == "delta-table"
 Provides-Extra: discord
 Requires-Dist: discord.py; extra == "discord"
 Provides-Extra: dropbox
-Requires-Dist: dropboxdrivefs; extra == "dropbox"
 Requires-Dist: fsspec; extra == "dropbox"
+Requires-Dist: dropboxdrivefs; extra == "dropbox"
 Provides-Extra: duckdb
 Requires-Dist: duckdb; extra == "duckdb"
 Provides-Extra: elasticsearch
 Requires-Dist: elasticsearch[async]; extra == "elasticsearch"
 Provides-Extra: gcs
 Requires-Dist: bs4; extra == "gcs"
-Requires-Dist: gcsfs; extra == "gcs"
 Requires-Dist: fsspec; extra == "gcs"
+Requires-Dist: gcsfs; extra == "gcs"
 Provides-Extra: github
-Requires-Dist: requests; extra == "github"
 Requires-Dist: pygithub>1.58.0; extra == "github"
+Requires-Dist: requests; extra == "github"
 Provides-Extra: gitlab
 Requires-Dist: python-gitlab; extra == "gitlab"
 Provides-Extra: google-drive
@@ -128,19 +128,19 @@ Requires-Dist: networkx; extra == "neo4j"
 Requires-Dist: neo4j-rust-ext; extra == "neo4j"
 Requires-Dist: cymple; extra == "neo4j"
 Provides-Extra: notion
-Requires-Dist: htmlBuilder; extra == "notion"
-Requires-Dist: notion-client; extra == "notion"
 Requires-Dist: httpx; extra == "notion"
 Requires-Dist: backoff; extra == "notion"
+Requires-Dist: notion-client; extra == "notion"
+Requires-Dist: htmlBuilder; extra == "notion"
 Provides-Extra: onedrive
 Requires-Dist: bs4; extra == "onedrive"
-Requires-Dist: msal; extra == "onedrive"
 Requires-Dist: Office365-REST-Python-Client; extra == "onedrive"
+Requires-Dist: msal; extra == "onedrive"
 Provides-Extra: opensearch
 Requires-Dist: opensearch-py; extra == "opensearch"
 Provides-Extra: outlook
-Requires-Dist: msal; extra == "outlook"
 Requires-Dist: Office365-REST-Python-Client; extra == "outlook"
+Requires-Dist: msal; extra == "outlook"
 Provides-Extra: pinecone
 Requires-Dist: pinecone-client>=3.7.1; extra == "pinecone"
 Provides-Extra: postgres
@@ -155,8 +155,8 @@ Provides-Extra: s3
 Requires-Dist: fsspec; extra == "s3"
 Requires-Dist: s3fs; extra == "s3"
 Provides-Extra: sharepoint
-Requires-Dist: msal; extra == "sharepoint"
 Requires-Dist: Office365-REST-Python-Client; extra == "sharepoint"
+Requires-Dist: msal; extra == "sharepoint"
 Provides-Extra: salesforce
 Requires-Dist: simple-salesforce; extra == "salesforce"
 Provides-Extra: sftp
@@ -165,8 +165,8 @@ Requires-Dist: fsspec; extra == "sftp"
 Provides-Extra: slack
 Requires-Dist: slack_sdk[optional]; extra == "slack"
 Provides-Extra: snowflake
-Requires-Dist: snowflake-connector-python; extra == "snowflake"
 Requires-Dist: psycopg2-binary; extra == "snowflake"
+Requires-Dist: snowflake-connector-python; extra == "snowflake"
 Provides-Extra: wikipedia
 Requires-Dist: wikipedia; extra == "wikipedia"
 Provides-Extra: weaviate
@@ -178,17 +178,17 @@ Requires-Dist: databricks-sql-connector; extra == "databricks-delta-tables"
 Provides-Extra: singlestore
 Requires-Dist: singlestoredb; extra == "singlestore"
 Provides-Extra: vectara
+Requires-Dist: aiofiles; extra == "vectara"
 Requires-Dist: httpx; extra == "vectara"
 Requires-Dist: requests; extra == "vectara"
-Requires-Dist: aiofiles; extra == "vectara"
 Provides-Extra: vastdb
-Requires-Dist: pyarrow; extra == "vastdb"
 Requires-Dist: vastdb; extra == "vastdb"
+Requires-Dist: pyarrow; extra == "vastdb"
 Requires-Dist: ibis; extra == "vastdb"
 Provides-Extra: zendesk
+Requires-Dist: aiofiles; extra == "zendesk"
 Requires-Dist: httpx; extra == "zendesk"
 Requires-Dist: bs4; extra == "zendesk"
-Requires-Dist: aiofiles; extra == "zendesk"
 Provides-Extra: embed-huggingface
 Requires-Dist: sentence-transformers; extra == "embed-huggingface"
 Provides-Extra: embed-octoai

{unstructured_ingest-0.5.17.dist-info → unstructured_ingest-0.5.18.dist-info}/RECORD

@@ -111,7 +111,7 @@ test/unit/v2/partitioners/test_partitioner.py,sha256=iIYg7IpftV3LusoO4H8tr1IHY1U
 test/unit/v2/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 test/unit/v2/utils/data_generator.py,sha256=UoYVNjG4S4wlaA9gceQ82HIpF9_6I1UTHD1_GrQBHp0,973
 unstructured_ingest/__init__.py,sha256=U4S_2y3zgLZVfMenHRaJFBW8yqh2mUBuI291LGQVOJ8,35
-unstructured_ingest/__version__.py,sha256=KerQQWRY3fKuaApvHFeNECgY6_9Sn7bl2FfaUuLCr4c,43
+unstructured_ingest/__version__.py,sha256=QYn6GUOSyCz_KH2wi4yg_FlUU4SE844Xhf0hR6-jv8s,43
 unstructured_ingest/error.py,sha256=qDncnJgbf5ils956RcO2CGlAKYDT5OaEM9Clv1JVTNc,1448
 unstructured_ingest/interfaces.py,sha256=7DOnDpGvUNlCoFR7UPRGmOarqH5sFtuUOO5vf8X3oTM,31489
 unstructured_ingest/logger.py,sha256=S5nSqGcABoQyeicgRnBQFjDScCaTvFVivOCvbo-laL0,4479
@@ -370,7 +370,7 @@ unstructured_ingest/runner/writers/fsspec/s3.py,sha256=kHJq2O3864QBd_tL2SKb0mdyw
 unstructured_ingest/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 unstructured_ingest/utils/chunking.py,sha256=9b3sXMA6L8RW5xAkKQbwdtVudGLAcj_sgT6Grh5tyYM,1870
 unstructured_ingest/utils/compression.py,sha256=NNiY-2S2Gf3at7zC1PYxMijaEza9vVSzRn5mdFf6mHo,4434
-unstructured_ingest/utils/data_prep.py,sha256=AKtsdu9stYA63CV1C5B_fFWigqy-giVv-euumitos-A,7266
+unstructured_ingest/utils/data_prep.py,sha256=MfID_7SPZHeZztlNTSXIzilaWvv1mdfCcLlhqpGLYNg,7557
 unstructured_ingest/utils/dep_check.py,sha256=SXXcUna2H0RtxA6j1S2NGkvQa9JP2DujWhmyBa7776Y,2400
 unstructured_ingest/utils/google_filetype.py,sha256=YVspEkiiBrRUSGVeVbsavvLvTmizdy2e6TsjigXTSRU,468
 unstructured_ingest/utils/html.py,sha256=DGRDMqGbwH8RiF94Qh6NiqVkbbjZfe1h26dIehC-X7M,6340
@@ -433,7 +433,7 @@ unstructured_ingest/v2/processes/connectors/azure_ai_search.py,sha256=ngPDpU0oZ6
 unstructured_ingest/v2/processes/connectors/chroma.py,sha256=VHCnM56qNXuHzovJihrNfJnZbWLJShOe8j12PJFrbL0,7219
 unstructured_ingest/v2/processes/connectors/confluence.py,sha256=gSs4-AxL0gfeWdJfP7JfCrQSQNLoJRkvHquKK9RJvpQ,12043
 unstructured_ingest/v2/processes/connectors/couchbase.py,sha256=i7vuNKsUkN93JRVmg4--MO0ZgbjvhIqt46oYqk9zFSQ,12250
-unstructured_ingest/v2/processes/connectors/delta_table.py,sha256=xvLWTSFEC3gyGTwEISXxWmUoAfCgzdgZkETMMBOPHuI,7153
+unstructured_ingest/v2/processes/connectors/delta_table.py,sha256=bfEGiepJLOS9TxK-bMkjTTjHLHUc0q7qUzIYdwkLDMs,7104
 unstructured_ingest/v2/processes/connectors/discord.py,sha256=-e4-cBK4TnHkknK1qIb86AIVMy81lBgC288_iLpTzM8,5246
 unstructured_ingest/v2/processes/connectors/gitlab.py,sha256=ufE65Z8q_tC4oppGg5BsGXwSaL7RbEXcaagJQYsylNo,9984
 unstructured_ingest/v2/processes/connectors/google_drive.py,sha256=QzcHNelUbnubsDtanFIgDCRzmYTuP-GjJ_g9y8fButE,19623
@@ -442,8 +442,8 @@ unstructured_ingest/v2/processes/connectors/kdbai.py,sha256=VRDAiou_7oWOIAgQTdOG
 unstructured_ingest/v2/processes/connectors/local.py,sha256=FWPRjjUsnQjyZMChuZGuMU04AB5X0sFEOcAXhx1r9sk,7381
 unstructured_ingest/v2/processes/connectors/milvus.py,sha256=wmcu9NVy3gYlQGT25inN5w_QrhFoL8-hRq0pJFSNw8g,8866
 unstructured_ingest/v2/processes/connectors/mongodb.py,sha256=cL0QUQZF_s2brh3nNNeAywXVpaIiND4b5JTAFlYjLjw,14273
-unstructured_ingest/v2/processes/connectors/neo4j.py,sha256=vxf6Xuh-OMS09Y-mIF0PIwrFauqRtoI7vjeLBXsFwTk,18744
-unstructured_ingest/v2/processes/connectors/onedrive.py,sha256=EM9fq67RsiudZvZbi6nDXkS-i6W0xLvbkNvD0G-Ni5E,17779
+unstructured_ingest/v2/processes/connectors/neo4j.py,sha256=I-eDLAlThHKKFQfkZpQL8CLFBDy5krWgTQANLgMTwTk,18679
+unstructured_ingest/v2/processes/connectors/onedrive.py,sha256=5rg7t40gKxDHNcuJrJHmVzJ9uM7Ct4RBOvFsfwdGc5c,18002
 unstructured_ingest/v2/processes/connectors/outlook.py,sha256=KgNGM8hImRhy6_SpswRP2VwRD4VOrqqJoySgxf2oduI,9290
 unstructured_ingest/v2/processes/connectors/pinecone.py,sha256=O9lC4mZ9V_exg9apiCJSWHsgkuYDSEOlI6CaUS5ZB7c,13961
 unstructured_ingest/v2/processes/connectors/redisdb.py,sha256=p0AY4ukBNpwAemV4bWzpScvVbLTVlI3DzsCNUKiBI5M,6757
@@ -577,9 +577,9 @@ unstructured_ingest/v2/processes/connectors/zendesk/client.py,sha256=DDAYQB7catK
 unstructured_ingest/v2/processes/connectors/zendesk/zendesk.py,sha256=R8SXYkRhVUoWEHdGCt2CzcTxxuFundw_0GlGZ34YmbM,8987
 unstructured_ingest/v2/processes/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 unstructured_ingest/v2/processes/utils/blob_storage.py,sha256=EWvK4HRYubr9i1UyMhv5cU9u0UzVkCDC_BIm4Uxab7Y,964
-unstructured_ingest-0.5.17.dist-info/LICENSE.md,sha256=SxkKP_62uIAKb9mb1eH7FH4Kn2aYT09fgjKpJt5PyTk,11360
-unstructured_ingest-0.5.17.dist-info/METADATA,sha256=cfGRkKOAflmCnSh-KoaCVyBcWD2SN_onWWc0tVaZ8fc,8465
-unstructured_ingest-0.5.17.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-unstructured_ingest-0.5.17.dist-info/entry_points.txt,sha256=gUAAFnjFPnBgThJSEbw0N5ZjxtaKlT1s9e05_arQrNw,70
-unstructured_ingest-0.5.17.dist-info/top_level.txt,sha256=DMuDMHZRMdeay8v8Kdi855muIv92F0OkutvBCaBEW6M,25
-unstructured_ingest-0.5.17.dist-info/RECORD,,
+unstructured_ingest-0.5.18.dist-info/LICENSE.md,sha256=SxkKP_62uIAKb9mb1eH7FH4Kn2aYT09fgjKpJt5PyTk,11360
+unstructured_ingest-0.5.18.dist-info/METADATA,sha256=K47-NP1RfNwqRnvbZ8vO75ab5J5RSmb5nocwSXNwqko,8465
+unstructured_ingest-0.5.18.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+unstructured_ingest-0.5.18.dist-info/entry_points.txt,sha256=gUAAFnjFPnBgThJSEbw0N5ZjxtaKlT1s9e05_arQrNw,70
+unstructured_ingest-0.5.18.dist-info/top_level.txt,sha256=DMuDMHZRMdeay8v8Kdi855muIv92F0OkutvBCaBEW6M,25
+unstructured_ingest-0.5.18.dist-info/RECORD,,