cr

Pinecone can return multiple matches with different text keys in the
metadata, and the current behavior will throw a KeyError if one pops up
that isn't the expected text key. This PR only selects the matches with
the correct text key.

langchain/vectorstores/pinecone.py

@@ -1,6 +1,7 @@
 """Wrapper around Pinecone vector database."""
 from __future__ import annotations
 
+import logging
 import uuid
 from typing import Any, Callable, Iterable, List, Optional
 
@@ -8,6 +9,8 @@ from langchain.docstore.document import Document
 from langchain.embeddings.base import Embeddings
 from langchain.vectorstores.base import VectorStore
 
+logger = logging.getLogger(__name__)
+
 
 class Pinecone(VectorStore):
     """Wrapper around Pinecone vector database.
@@ -78,8 +81,13 @@ class Pinecone(VectorStore):
         results = self._index.query([query_obj], top_k=k, include_metadata=True)
         for res in results["matches"]:
             metadata = res["metadata"]
-            text = metadata.pop(self._text_key)
-            docs.append(Document(page_content=text, metadata=metadata))
+            if self._text_key in metadata:
+                text = metadata.pop(self._text_key)
+                docs.append(Document(page_content=text, metadata=metadata))
+            else:
+                logger.warning(
+                    f"Found document with no `{self._text_key}` key. Skipping."
+                )
         return docs
 
     @classmethod