handle neo4j client error when initializing vector store (#421)

* handle neo4j client error when initializing vector store

* include text

* .

Author: vangheem

CHANGELOG.md

@@ -1,3 +1,10 @@
+## 0.5.13
+
+### Fixes 
+
+* **Handle schema conflict on neo4j**
+
+### Fixes
 
 ## 0.5.12
 

unstructured_ingest/__version__.py

@@ -1 +1 @@
-__version__ = "0.5.12"  # pragma: no cover
+__version__ = "0.5.13"  # pragma: no cover

unstructured_ingest/v2/processes/connectors/neo4j.py

@@ -154,7 +154,7 @@ def _create_lexical_graph(self, elements: list[dict], document_node: _Node) -> "
             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(
@@ -327,21 +327,29 @@ async def _create_uniqueness_constraints(self, client: AsyncDriver) -> None:
     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: {{
                 `vector.similarity_function`: '{similarity_function}',
                 `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/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",