quivr/backend/models/settings.py
2023-11-29 13:19:22 +01:00

60 lines
1.8 KiB
Python

from langchain.embeddings.openai import OpenAIEmbeddings
from models.databases.supabase.supabase import SupabaseDB
from pydantic import BaseSettings
from supabase.client import Client, create_client
from vectorstore.supabase import SupabaseVectorStore
class BrainRateLimiting(BaseSettings):
max_brain_per_user: int = 5
class BrainSettings(BaseSettings):
openai_api_key: str
supabase_url: str
supabase_service_key: str
resend_api_key: str = "null"
resend_email_address: str = "brain@mail.quivr.app"
class ContactsSettings(BaseSettings):
resend_contact_sales_from: str = "null"
resend_contact_sales_to: str = "null"
class ResendSettings(BaseSettings):
resend_api_key: str = "null"
def get_supabase_client() -> Client:
settings = BrainSettings() # pyright: ignore reportPrivateUsage=none
supabase_client: Client = create_client(
settings.supabase_url, settings.supabase_service_key
)
return supabase_client
def get_supabase_db() -> SupabaseDB:
supabase_client = get_supabase_client()
return SupabaseDB(supabase_client)
def get_embeddings() -> OpenAIEmbeddings:
settings = BrainSettings() # pyright: ignore reportPrivateUsage=none
embeddings = OpenAIEmbeddings(
openai_api_key=settings.openai_api_key
) # pyright: ignore reportPrivateUsage=none
return embeddings
def get_documents_vector_store() -> SupabaseVectorStore:
settings = BrainSettings() # pyright: ignore reportPrivateUsage=none
embeddings = get_embeddings()
supabase_client: Client = create_client(
settings.supabase_url, settings.supabase_service_key
)
documents_vector_store = SupabaseVectorStore(
supabase_client, embeddings, table_name="vectors"
)
return documents_vector_store