how can i do a metadata filter in asimilarity_search() from AsyncPGVectorStore class?

[LucasdfRocha]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

from langchain_postgres.v2.async_vectorstore import AsyncPGVectorStore

async def get_top_4_docs(query: str, params: dict):
    vector_store = await AsyncPGVectorStore.create(
                engine=params['engine'],
                embedding_service=params['embeddings_model'],
                table_name=params['table_name'],
                schema_name=params['schema_name'],
                content_column=params['content_column'],
                embedding_column=params['embedding_column'],
                metadata_json_column=params['metadata_json_column'],
                id_column=params['id_column'],
            )

    docs = await vector_store.asimilarity_search(
                query=query,
                k=4,
                filter={"metadata->>foo_id": 3}
            )

    return docs

Description

Hi! I'm working on building an asynchronous retrieval function using AsyncPGVectorStore that fetches the top 4 most relevant chunks from my PostgreSQL vector store. I need to filter results based on the metadata field foo_id.

Here’s an example of the metadata stored in my documents:

{
  "loc": { "lines": { "from": 1, "to": 3 } },
  "source": "foo",
  "foo_id": 3,
  "contentType": "text/plain"
}

I encountered the following error:

Invalid field name: metadata->>'foo_id'. Expected a valid identifier.

Is there a supported way to filter documents by a specific metadata key, like foo_id, using AsyncPGVectorStore? If so, what is the correct filter syntax?

Any guidance would be greatly appreciated!

Thanks in advance!

System Info

• langchain==0.3.2
• langchain-anthropic==0.2.3
• langchain-community==0.3.1
• langchain-core==0.3.9
• langchain-google-vertexai==2.0.4
• langchain-text-splitters==0.3.0

---

plataform: mac
python version: Python 3.10.16