unstructured-ingest 0.5.12__tar.gz → 0.5.14__tar.gz

{unstructured_ingest-0.5.12/unstructured_ingest.egg-info → unstructured_ingest-0.5.14}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: unstructured-ingest
-Version: 0.5.12
+Version: 0.5.14
 Summary: A library that prepares raw documents for downstream ML tasks.
 Home-page: https://github.com/Unstructured-IO/unstructured-ingest
 Author: Unstructured Technologies
@@ -22,13 +22,13 @@ Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Requires-Python: >=3.9.0,<3.14
 Description-Content-Type: text/markdown
 License-File: LICENSE.md
+Requires-Dist: pydantic>=2.7
+Requires-Dist: opentelemetry-sdk
+Requires-Dist: pandas
 Requires-Dist: tqdm
 Requires-Dist: dataclasses_json
 Requires-Dist: click
-Requires-Dist: opentelemetry-sdk
 Requires-Dist: python-dateutil
-Requires-Dist: pydantic>=2.7
-Requires-Dist: pandas
 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: adlfs; extra == "azure"
 Requires-Dist: fsspec; extra == "azure"
+Requires-Dist: adlfs; extra == "azure"
 Provides-Extra: azure-ai-search
 Requires-Dist: azure-search-documents; extra == "azure-ai-search"
 Provides-Extra: biomed
@@ -81,8 +81,8 @@ Requires-Dist: chromadb; extra == "chroma"
 Provides-Extra: clarifai
 Requires-Dist: clarifai; extra == "clarifai"
 Provides-Extra: confluence
-Requires-Dist: requests; extra == "confluence"
 Requires-Dist: atlassian-python-api; extra == "confluence"
+Requires-Dist: requests; extra == "confluence"
 Provides-Extra: couchbase
 Requires-Dist: couchbase; extra == "couchbase"
 Provides-Extra: delta-table
@@ -99,11 +99,11 @@ Provides-Extra: elasticsearch
 Requires-Dist: elasticsearch[async]; extra == "elasticsearch"
 Provides-Extra: gcs
 Requires-Dist: bs4; extra == "gcs"
-Requires-Dist: fsspec; extra == "gcs"
 Requires-Dist: gcsfs; extra == "gcs"
+Requires-Dist: fsspec; extra == "gcs"
 Provides-Extra: github
-Requires-Dist: pygithub>1.58.0; extra == "github"
 Requires-Dist: requests; extra == "github"
+Requires-Dist: pygithub>1.58.0; extra == "github"
 Provides-Extra: gitlab
 Requires-Dist: python-gitlab; extra == "gitlab"
 Provides-Extra: google-drive
@@ -128,9 +128,9 @@ Requires-Dist: neo4j-rust-ext; extra == "neo4j"
 Requires-Dist: cymple; extra == "neo4j"
 Requires-Dist: networkx; extra == "neo4j"
 Provides-Extra: notion
+Requires-Dist: backoff; extra == "notion"
 Requires-Dist: httpx; extra == "notion"
 Requires-Dist: notion-client; extra == "notion"
-Requires-Dist: backoff; extra == "notion"
 Requires-Dist: htmlBuilder; extra == "notion"
 Provides-Extra: onedrive
 Requires-Dist: bs4; extra == "onedrive"
@@ -160,8 +160,8 @@ Requires-Dist: Office365-REST-Python-Client; extra == "sharepoint"
 Provides-Extra: salesforce
 Requires-Dist: simple-salesforce; extra == "salesforce"
 Provides-Extra: sftp
-Requires-Dist: fsspec; extra == "sftp"
 Requires-Dist: paramiko; extra == "sftp"
+Requires-Dist: fsspec; extra == "sftp"
 Provides-Extra: slack
 Requires-Dist: slack_sdk[optional]; extra == "slack"
 Provides-Extra: snowflake
@@ -178,16 +178,16 @@ 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: httpx; extra == "vectara"
+Requires-Dist: aiofiles; extra == "vectara"
 Provides-Extra: vastdb
 Requires-Dist: ibis; extra == "vastdb"
 Requires-Dist: vastdb; extra == "vastdb"
 Requires-Dist: pyarrow; extra == "vastdb"
 Provides-Extra: zendesk
-Requires-Dist: httpx; extra == "zendesk"
 Requires-Dist: bs4; extra == "zendesk"
+Requires-Dist: httpx; extra == "zendesk"
 Requires-Dist: aiofiles; extra == "zendesk"
 Provides-Extra: embed-huggingface
 Requires-Dist: sentence-transformers; extra == "embed-huggingface"
@@ -204,8 +204,8 @@ Provides-Extra: openai
 Requires-Dist: tiktoken; extra == "openai"
 Requires-Dist: openai; extra == "openai"
 Provides-Extra: bedrock
-Requires-Dist: aioboto3; extra == "bedrock"
 Requires-Dist: boto3; extra == "bedrock"
+Requires-Dist: aioboto3; extra == "bedrock"
 Provides-Extra: togetherai
 Requires-Dist: together; extra == "togetherai"
 Dynamic: author

{unstructured_ingest-0.5.12 → unstructured_ingest-0.5.14}/test/integration/connectors/test_zendesk.py

@@ -11,7 +11,7 @@ from test.integration.connectors.utils.validation.source import (
 )
 from test.integration.utils import requires_env
 from unstructured_ingest.v2.errors import UserAuthError
-from unstructured_ingest.v2.processes.connectors.zendesk import (
+from unstructured_ingest.v2.processes.connectors.zendesk.zendesk import (
     CONNECTOR_TYPE,
     ZendeskAccessConfig,
     ZendeskConnectionConfig,

unstructured_ingest-0.5.14/unstructured_ingest/__version__.py

@@ -0,0 +1 @@
+__version__ = "0.5.14"  # pragma: no cover

{unstructured_ingest-0.5.12 → unstructured_ingest-0.5.14}/unstructured_ingest/v2/processes/connectors/__init__.py

@@ -64,6 +64,8 @@ from .slack import CONNECTOR_TYPE as SLACK_CONNECTOR_TYPE
 from .slack import slack_source_entry
 from .vectara import CONNECTOR_TYPE as VECTARA_CONNECTOR_TYPE
 from .vectara import vectara_destination_entry
+from .zendesk.zendesk import CONNECTOR_TYPE as ZENDESK_CONNECTOR_TYPE
+from .zendesk.zendesk import zendesk_source_entry
 
 add_source_entry(source_type=ASTRA_DB_CONNECTOR_TYPE, entry=astra_db_source_entry)
 add_destination_entry(destination_type=ASTRA_DB_CONNECTOR_TYPE, entry=astra_db_destination_entry)
@@ -119,3 +121,4 @@ add_source_entry(source_type=DISCORD_CONNECTOR_TYPE, entry=discord_source_entry)
 add_destination_entry(destination_type=REDIS_CONNECTOR_TYPE, entry=redis_destination_entry)
 
 add_source_entry(source_type=JIRA_CONNECTOR_TYPE, entry=jira_source_entry)
+add_source_entry(source_type=ZENDESK_CONNECTOR_TYPE, entry=zendesk_source_entry)

{unstructured_ingest-0.5.12 → unstructured_ingest-0.5.14}/unstructured_ingest/v2/processes/connectors/neo4j.py

@@ -154,7 +154,7 @@ class Neo4jUploadStager(UploadStager):
             self._add_entities(element, graph, element_node)
 
             if self._is_chunk(element):
-                for origin_element in format_and_truncate_orig_elements(element):
+                for origin_element in format_and_truncate_orig_elements(element, include_text=True):
                     origin_element_node = self._create_element_node(origin_element)
 
                     graph.add_edge(
@@ -327,13 +327,16 @@ class Neo4jUploader(Uploader):
     async def _create_vector_index(
         self, client: AsyncDriver, dimensions: int, similarity_function: SimilarityFunction
     ) -> None:
+        import neo4j.exceptions
+
         label = Label.CHUNK
         logger.info(
             f"Creating index on nodes labeled '{label.value}' if it does not already exist."
         )
         index_name = f"{label.value.lower()}_vector"
-        await client.execute_query(
-            f"""
+        try:
+            await client.execute_query(
+                f"""
             CREATE VECTOR INDEX {index_name} IF NOT EXISTS
             FOR (n:{label.value}) ON n.embedding
             OPTIONS {{indexConfig: {{
@@ -341,7 +344,12 @@ class Neo4jUploader(Uploader):
                 `vector.dimensions`: {dimensions}}}
             }}
             """
-        )
+            )
+        except neo4j.exceptions.ClientError as e:
+            if e.code == "Neo.ClientError.Schema.EquivalentSchemaRuleAlreadyExists":
+                logger.info(f"Index on nodes labeled '{label.value}' already exists.")
+            else:
+                raise
 
     async def _delete_old_data_if_exists(self, file_data: FileData, client: AsyncDriver) -> None:
         logger.info(f"Deleting old data for the record '{file_data.identifier}' (if present).")

{unstructured_ingest-0.5.12 → unstructured_ingest-0.5.14}/unstructured_ingest/v2/processes/connectors/utils.py

@@ -31,7 +31,9 @@ def conform_string_to_dict(value: Any) -> dict:
     raise ValidationError(f"Input could not be mapped to a valid dict: {value}")
 
 
-def format_and_truncate_orig_elements(element: dict) -> list[dict[str, Any]]:
+def format_and_truncate_orig_elements(
+    element: dict, include_text: bool = False
+) -> list[dict[str, Any]]:
     """
     This function is used to format and truncate the orig_elements field in the metadata.
     This is used to remove the text field and other larger fields from the orig_elements
@@ -42,7 +44,8 @@ def format_and_truncate_orig_elements(element: dict) -> list[dict[str, Any]]:
     orig_elements = []
     if raw_orig_elements is not None:
         for element in elements_from_base64_gzipped_json(raw_orig_elements):
-            element.pop("text", None)
+            if not include_text:
+                element.pop("text", None)
             for prop in (
                 "image_base64",
                 "text_as_html",

{unstructured_ingest-0.5.12 → unstructured_ingest-0.5.14/unstructured_ingest.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: unstructured-ingest
-Version: 0.5.12
+Version: 0.5.14
 Summary: A library that prepares raw documents for downstream ML tasks.
 Home-page: https://github.com/Unstructured-IO/unstructured-ingest
 Author: Unstructured Technologies
@@ -22,13 +22,13 @@ Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Requires-Python: >=3.9.0,<3.14
 Description-Content-Type: text/markdown
 License-File: LICENSE.md
+Requires-Dist: pydantic>=2.7
+Requires-Dist: opentelemetry-sdk
+Requires-Dist: pandas
 Requires-Dist: tqdm
 Requires-Dist: dataclasses_json
 Requires-Dist: click
-Requires-Dist: opentelemetry-sdk
 Requires-Dist: python-dateutil
-Requires-Dist: pydantic>=2.7
-Requires-Dist: pandas
 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: adlfs; extra == "azure"
 Requires-Dist: fsspec; extra == "azure"
+Requires-Dist: adlfs; extra == "azure"
 Provides-Extra: azure-ai-search
 Requires-Dist: azure-search-documents; extra == "azure-ai-search"
 Provides-Extra: biomed
@@ -81,8 +81,8 @@ Requires-Dist: chromadb; extra == "chroma"
 Provides-Extra: clarifai
 Requires-Dist: clarifai; extra == "clarifai"
 Provides-Extra: confluence
-Requires-Dist: requests; extra == "confluence"
 Requires-Dist: atlassian-python-api; extra == "confluence"
+Requires-Dist: requests; extra == "confluence"
 Provides-Extra: couchbase
 Requires-Dist: couchbase; extra == "couchbase"
 Provides-Extra: delta-table
@@ -99,11 +99,11 @@ Provides-Extra: elasticsearch
 Requires-Dist: elasticsearch[async]; extra == "elasticsearch"
 Provides-Extra: gcs
 Requires-Dist: bs4; extra == "gcs"
-Requires-Dist: fsspec; extra == "gcs"
 Requires-Dist: gcsfs; extra == "gcs"
+Requires-Dist: fsspec; extra == "gcs"
 Provides-Extra: github
-Requires-Dist: pygithub>1.58.0; extra == "github"
 Requires-Dist: requests; extra == "github"
+Requires-Dist: pygithub>1.58.0; extra == "github"
 Provides-Extra: gitlab
 Requires-Dist: python-gitlab; extra == "gitlab"
 Provides-Extra: google-drive
@@ -128,9 +128,9 @@ Requires-Dist: neo4j-rust-ext; extra == "neo4j"
 Requires-Dist: cymple; extra == "neo4j"
 Requires-Dist: networkx; extra == "neo4j"
 Provides-Extra: notion
+Requires-Dist: backoff; extra == "notion"
 Requires-Dist: httpx; extra == "notion"
 Requires-Dist: notion-client; extra == "notion"
-Requires-Dist: backoff; extra == "notion"
 Requires-Dist: htmlBuilder; extra == "notion"
 Provides-Extra: onedrive
 Requires-Dist: bs4; extra == "onedrive"
@@ -160,8 +160,8 @@ Requires-Dist: Office365-REST-Python-Client; extra == "sharepoint"
 Provides-Extra: salesforce
 Requires-Dist: simple-salesforce; extra == "salesforce"
 Provides-Extra: sftp
-Requires-Dist: fsspec; extra == "sftp"
 Requires-Dist: paramiko; extra == "sftp"
+Requires-Dist: fsspec; extra == "sftp"
 Provides-Extra: slack
 Requires-Dist: slack_sdk[optional]; extra == "slack"
 Provides-Extra: snowflake
@@ -178,16 +178,16 @@ 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: httpx; extra == "vectara"
+Requires-Dist: aiofiles; extra == "vectara"
 Provides-Extra: vastdb
 Requires-Dist: ibis; extra == "vastdb"
 Requires-Dist: vastdb; extra == "vastdb"
 Requires-Dist: pyarrow; extra == "vastdb"
 Provides-Extra: zendesk
-Requires-Dist: httpx; extra == "zendesk"
 Requires-Dist: bs4; extra == "zendesk"
+Requires-Dist: httpx; extra == "zendesk"
 Requires-Dist: aiofiles; extra == "zendesk"
 Provides-Extra: embed-huggingface
 Requires-Dist: sentence-transformers; extra == "embed-huggingface"
@@ -204,8 +204,8 @@ Provides-Extra: openai
 Requires-Dist: tiktoken; extra == "openai"
 Requires-Dist: openai; extra == "openai"
 Provides-Extra: bedrock
-Requires-Dist: aioboto3; extra == "bedrock"
 Requires-Dist: boto3; extra == "bedrock"
+Requires-Dist: aioboto3; extra == "bedrock"
 Provides-Extra: togetherai
 Requires-Dist: together; extra == "togetherai"
 Dynamic: author

unstructured_ingest-0.5.12/unstructured_ingest/__version__.py

@@ -1 +0,0 @@
-__version__ = "0.5.12"  # pragma: no cover

unstructured_ingest-0.5.12/unstructured_ingest/v2/processes/connectors/zendesk/__init__.py

@@ -1,31 +0,0 @@
-from unstructured_ingest.v2.processes.connector_registry import (
-    add_source_entry,
-)
-
-from .zendesk import (
-    CONNECTOR_TYPE,
-    ZendeskAccessConfig,
-    ZendeskClient,
-    ZendeskConnectionConfig,
-    ZendeskDownloader,
-    ZendeskDownloaderConfig,
-    ZendeskIndexer,
-    ZendeskIndexerConfig,
-    ZendeskTicket,
-    zendesk_source_entry,
-)
-
-__all__ = [
-    "add_source_entry",
-    "zendesk_source_entry",
-    "ZendeskAccessConfig",
-    "ZendeskClient",
-    "ZendeskConnectionConfig",
-    "ZendeskDownloader",
-    "ZendeskDownloaderConfig",
-    "ZendeskIndexer",
-    "ZendeskIndexerConfig",
-    "ZendeskTicket",
-]
-
-add_source_entry(source_type=CONNECTOR_TYPE, entry=zendesk_source_entry)