community[minor]: add mongodb byte store

[pprados]

The MongoDBStore can manage only documents.
It's not possible to use MongoDB for an CacheBackedEmbeddings.

With this new implementation, it's possible to use:

CacheBackedEmbeddings.from_bytes_store(
    underlying_embeddings=embeddings,
    document_embedding_cache=MongoDBByteStore(
      connection_string=db_uri,
      db_name=db_name,
      collection_name=collection_name,
  ),
)

and use MongoDB to cache the embeddings !

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Skipped Deployment

Name	Status	Preview	Comments	Updated (UTC)
langchain	⬜️ Ignored (Inspect)	Visit Preview		Jul 17, 2024 1:41pm

[eyurtsev]

@pprados could you add an integration test using langchain_standard_tests

Look at the following unit tests as examples:

langchain/libs/standard-tests/tests/unit_tests/test_in_memory_base_store.py

Line 16 in d311f22

return "foo", "bar", "buzz"

Essentially you only need to import the test suite and inherit from it and then provide a single fixture

class TestInMemoryStore(BaseStoreSyncTests):
    @pytest.fixture
    def three_values(self) -> Tuple[bytes, bytes, bytes]:  # <-- Provide 3 
        return b"foo", b"bar", b"buzz"

    @pytest.fixture
    def kv_store(self) -> Store:
       yield implementation using mongodb that has no keys stored in it

[pprados]

@eyurtsev
The new version use the generic test.