Skip to content

The agent engineering platform.

feat(core): add `id` field to `Document` passed to filter for `InMemoryVectorStore` similarity search (#32688)

PieterKok-jaam33c7f230e058
diff --git a/libs/core/langchain_core/vectorstores/in_memory.py b/libs/core/langchain_core/vectorstores/in_memory.py
index d08a68c8fc..ab0899b04f 100644
--- a/libs/core/langchain_core/vectorstores/in_memory.py
+++ b/libs/core/langchain_core/vectorstores/in_memory.py
@@ -376,7 +376,11 @@ class InMemoryVectorStore(VectorStore):
docs = [
doc
for doc in docs
- if filter(Document(page_content=doc["text"], metadata=doc["metadata"]))
+ if filter(
+ Document(
+ id=doc["id"], page_content=doc["text"], metadata=doc["metadata"]
+ )
+ )
]
if not docs:
diff --git a/libs/core/tests/unit_tests/vectorstores/test_in_memory.py b/libs/core/tests/unit_tests/vectorstores/test_in_memory.py
index 5ebaf8633d..a935777594 100644
--- a/libs/core/tests/unit_tests/vectorstores/test_in_memory.py
+++ b/libs/core/tests/unit_tests/vectorstores/test_in_memory.py
@@ -117,6 +117,40 @@ async def test_inmemory_filter() -> None:
assert output == []
+async def test_inmemory_filter_by_document_id() -> None:
+ """Test filtering by document ID field."""
+ embedding = DeterministicFakeEmbedding(size=6)
+ store = InMemoryVectorStore(embedding=embedding)
+
+ # Add documents with specific IDs using add_documents
+ documents = [
+ Document(page_content="first document", id="doc_1"),
+ Document(page_content="second document", id="doc_2"),
+ Document(page_content="third document", id="doc_3"),
+ ]
+ store.add_documents(documents)
+
+ # Test filtering by specific document ID
+ output = store.similarity_search("document", filter=lambda doc: doc.id == "doc_2")
+ assert len(output) == 1
+ assert output[0].page_content == "second document"
+ assert output[0].id == "doc_2"
+
+ # Test async version
+ output = await store.asimilarity_search(
+ "document", filter=lambda doc: doc.id in ["doc_1", "doc_3"]
+ )
+ assert len(output) == 2
+ ids = {doc.id for doc in output}
+ assert ids == {"doc_1", "doc_3"}
+
+ # Test filtering with non-existent ID
+ output = store.similarity_search(
+ "document", filter=lambda doc: doc.id == "non_existent"
+ )
+ assert output == []
+
+
async def test_inmemory_upsert() -> None:
"""Test upsert documents."""
embedding = DeterministicFakeEmbedding(size=2)