fix(chroma): pydantic validation error when using retriever.invoke()

libs/partners/chroma/langchain_chroma/vectorstores.py

@@ -50,17 +50,33 @@ def _results_to_docs_and_scores(results: Any) -> list[tuple[Document, float]]:
             results["ids"][0],
             results["distances"][0],
         )
+        if result[0] is not None
     ]
 
 
 def _results_to_docs_and_vectors(results: Any) -> list[tuple[Document, np.ndarray]]:
+    """Convert ChromaDB results to documents and vectors, filtering out None content."""
     return [
-        (Document(page_content=result[0], metadata=result[1] or {}), result[2])
+        (
+            Document(page_content=result[0], metadata=result[1] or {}, id=result[3]),
+            result[2],
+        )
         for result in zip(
             results["documents"][0],
             results["metadatas"][0],
             results["embeddings"][0],
+            results["ids"][0],
         )
+        if result[0] is not None
+    ]
+
+
+def safe_results_to_docs_and_scores(results: Any) -> list[Document]:
+    """Convert results to documents and scores, filtering out None page_content."""
+    return [
+        Document(page_content=result[0], metadata=result[1] or {}, id=result[2])
+        for result in results
+        if result[0] is not None
     ]
 
 
@@ -1064,6 +1080,7 @@ def get_by_ids(self, ids: Sequence[str], /) -> list[Document]:
             for doc, meta, doc_id in zip(
                 results["documents"], results["metadatas"], results["ids"]
             )
+            if doc is not None  # Filter out documents with None page_content
         ]
 
     def update_document(self, document_id: str, document: Document) -> None:

libs/partners/chroma/tests/integration_tests/test_vectorstores.py

@@ -191,7 +191,10 @@ def test_chroma_with_metadatas_with_vectors() -> None:
     vec_1 = embeddings.embed_query(texts[0])
     output = docsearch.similarity_search_with_vectors("foo", k=1)
     docsearch.delete_collection()
-    assert output[0][0] == Document(page_content="foo", metadata={"page": "0"})
+    doc = output[0][0]
+    assert doc.page_content == "foo"
+    assert doc.metadata == {"page": "0"}
+    assert doc.id is not None
     assert (output[0][1] == vec_1).all()
 
 
@@ -805,3 +808,44 @@ def test_delete_where_clause(client: chromadb.ClientAPI) -> None:
     assert vectorstore._collection.count() == 1
     # Clean up
     vectorstore.delete_collection()
+
+
+def test_chroma_handles_none_page_content() -> None:
+    """Test that Chroma gracefully handles None page_content values."""
+    from langchain_chroma.vectorstores import _results_to_docs_and_scores
+
+    mock_results = {
+        "documents": [["valid content", None, "another valid content"]],
+        "metadatas": [[{"key": "value1"}, {"key": "value2"}, {"key": "value3"}]],
+        "ids": [["id1", "id2", "id3"]],
+        "distances": [[0.1, 0.2, 0.3]],
+    }
+
+    docs_and_scores = _results_to_docs_and_scores(mock_results)
+
+    assert len(docs_and_scores) == 2
+    assert docs_and_scores[0][0].page_content == "valid content"
+    assert docs_and_scores[1][0].page_content == "another valid content"
+    assert docs_and_scores[0][0].id == "id1"
+    assert docs_and_scores[1][0].id == "id3"
+
+
+def test_chroma_handles_none_page_content_with_vectors() -> None:
+    """Test that Chroma gracefully handles None page_content values with vectors."""
+    from langchain_chroma.vectorstores import _results_to_docs_and_vectors
+
+    mock_results = {
+        "documents": [["valid content", None, "another valid content"]],
+        "metadatas": [[{"key": "value1"}, {"key": "value2"}, {"key": "value3"}]],
+        "ids": [["id1", "id2", "id3"]],
+        "embeddings": [[[0.1, 0.2], [0.3, 0.4], [0.5, 0.6]]],
+    }
+    docs_and_vectors = _results_to_docs_and_vectors(mock_results)
+
+    assert len(docs_and_vectors) == 2
+    assert docs_and_vectors[0][0].page_content == "valid content"
+    assert docs_and_vectors[1][0].page_content == "another valid content"
+    assert docs_and_vectors[0][0].id == "id1"
+    assert docs_and_vectors[1][0].id == "id3"
+    assert docs_and_vectors[0][1] == [0.1, 0.2]
+    assert docs_and_vectors[1][1] == [0.5, 0.6]

pyproject.toml

@@ -2,7 +2,9 @@
 authors = []
 license = { text = "MIT" }
 requires-python = ">=3.9"
-dependencies = []
+dependencies = [
+    "pyproject-toml>=0.1.0",
+]
 name = "langchain-monorepo"
 version = "0.0.1"
 description = "LangChain mono-repo"
@@ -93,4 +95,4 @@ pydocstyle = { convention = "google" }
     "F841", # allow assignments to variables that are never read -- it's example code
 
 ]
-"!libs/langchain/langchain/model_laboratory.py" = ["D"]
+"!libs/langchain/langchain/model_laboratory.py" = ["D"]