`SupabasePgVectorDocumentStore` and `SupabasePgvectorEmbeddingRetriever`

[davidsbatista]

Implement SupabasePgVectorDocumentStore and SupabasePgvectorEmbeddingRetriever as a thin wrapper around the existing pgvector-haystack integration, pre-configured to work with Supabase's connection model.

Detailed design

Reuses PgvectorDocumentStore under the hood and handles Supabase-specific connection setup (URL + service role key/JWT).

from haystack_integrations.document_stores.supabase import SupabasePgVectorDocumentStore
from haystack_integrations.components.retrievers.supabase import SupabasePgvectorEmbeddingRetriever

document_store = SupabasePgVectorDocumentStore(
    supabase_url="https://<project>.supabase.co",
    supabase_key=Secret.from_env_var("SUPABASE_SERVICE_KEY"),
    table_name="haystack_documents",
    embedding_dimension=1536,
)

retriever = SupabasePgvectorEmbeddingRetriever(document_store=document_store, top_k=5)

Implementation notes

• Primary dependency: supabase-py + pgvector-haystack
• Auth via supabase_url and supabase_key (using Secret for secure handling)
• Supabase connection strings follow the pattern postgresql://postgres.<project>:<password>@aws-0-<region>.pooler.supabase.com:6543/postgres
• Support optional schema and table_name overrides
• Integration lives in integrations/supabase/

Checklist

• The code is documented with docstrings and was merged in the main branch
• There is a Github workflow running the tests for the integration nightly and at every PR

[pontuselmrin]

I might take a stab at addressing this!

[davidsbatista]

Thanks for the initiative @pontuselmrin - let me know if you have any questions - I'm also on discord in case you want to have call

[pontuselmrin]

@davidsbatista I shot you a message on discord. Let's try to have a call at one point!