Embedded vectors returned as None despite setting return_embedding to True in MongoDBAtlasDocumentStore

[TechnoRahmon]

Description:

I'm encountering an issue with the MongoDBAtlasDocumentStore in the Haystack library. Despite setting return_embedding=True in the configuration, embedded vectors are returned as None when retrieving documents from the MongoDB collection.

Steps to Reproduce:

1. Set up a MongoDBAtlasDocumentStore with the following configuration:

from haystack.document_stores.mongodb_atlas import MongoDBAtlasDocumentStore

ds = MongoDBAtlasDocumentStore(
    mongo_connection_string="mongo_connection_string",
    database_name="squadDB",
    collection_name="dev",
    vector_search_index="vector_search_index",
    return_embedding=True,
    similarity='dotProduct'
)

2. Save documents to the MongoDB collection using ds.write_documents(squad_docs).

3. Retrieve documents from the collection using ds.get_all_documents().

4. Check the embedding of the retrieved documents, e.g., all_docs[0]['embedding'].

Expected Behavior:

The embedding field of the retrieved documents should contain the embedded vectors.

Actual Behavior:

The embedding field of the retrieved documents is None, despite setting return_embedding=True in the configuration.

Additional Information:

• Operating System: windows 11
• Python version: python 3.11
• farm-haystack version: 1.24.1

[anakin87]

Closing as duplicate of https://github.com/deepset-ai/haystack/issues/7037