Multi-Rag / src /agents /nodes /retreiver.py
VashuTheGreat2's picture
Upload folder using huggingface_hub
644f326 verified
Raw History Blame Contribute Delete
1.98 kB
from src.app_config import get_app_config
from langchain_core.documents import Document
from ..models.state import AgentState
from ...core.exception import MyException
from ...core.logger import logger
from ...service.embedding.sentence_transformer import SentenceTransformerEmbedding, get_embedding_service
from ...service.vector_store.qdrant import QdrantVectorStore, get_qdrant_service
from ...service.retreival.reranker import flaskreranker
from langsmith import traceable
import sys
observe = traceable
app_config = get_app_config()
embedding_service: SentenceTransformerEmbedding = get_embedding_service()
vector_store_service: QdrantVectorStore = get_qdrant_service(collection_name=app_config.collection_name, embedding_service=embedding_service)
async def reranker_documents(query:str,documents:list[Document]):
"""this is the reranker documents"""
return await flaskreranker(query=query,documents=documents)
@observe(name="retriever_node", run_type="retriever")
async def retreived_node(state: AgentState):
"""Fetching relevant similar dcuments according to query"""
queries = state.get("queries")
with logger.span("Retriever node execution", queries=queries):
try:
logger.info("Querying vector store for documents", queries=queries)
documents = await vector_store_service.similarity_search(query=queries,k=15)
doc_count = len(documents) if documents else 0
if doc_count and queries:
logger.info("Reranking documents",queries=queries)
reranked = await reranker_documents(query=queries,documents=documents)
documents = reranked.get("reranked_content", documents)
logger.info("Documents retrieved successfully", count=doc_count)
return {"retreived_content": documents}
except Exception as e:
logger.error("Document retrieval failed", error=str(e))
raise MyException(e, sys)