app_p/utils/retriever.py

from langchain_huggingface import HuggingFaceEmbeddings
from langchain_mongodb.vectorstores import MongoDBAtlasVectorSearch
from langchain_mongodb.retrievers.hybrid_search import MongoDBAtlasHybridSearchRetriever
import os
from dotenv import load_dotenv

load_dotenv()

# ---- MongoDB credentials ----
# mongo_username = os.getenv('MONGO_USERNAME')
# mongo_password = os.getenv('MONGO_PASSWORD')
mongo_database = os.getenv('MONGO_DATABASE')
mongo_connection_str = os.getenv('MONGO_CONNECTION_STRING')
mongo_collection_name = os.getenv('MONGO_COLLECTION')

# ---- Common Configurations & Hybrid Retrieval Configuration ----
MODEL_KWARGS = {"device": "cpu"}
ENCODE_KWARGS = {"normalize_embeddings": True,
                 "batch_size": 32}
EMBEDDING_DIMENSIONS = 1024
MODEL_NAME = "BAAI/bge-m3"
FINAL_TOP_K = 50 # 30
HYBRID_FULLTEXT_PENALTY = 0 # 60
HYBRID_VECTOR_PENALTY = 0.8 # 60

# ---- Embedding model ----
embed_model = HuggingFaceEmbeddings(
    model_name=MODEL_NAME, 
    model_kwargs=MODEL_KWARGS, 
    encode_kwargs=ENCODE_KWARGS
)

# ---- Vectore Search ----
num_vector_candidates = max(20, 2 * FINAL_TOP_K)
num_text_candidates = max(20, 2 * FINAL_TOP_K)
vector_k = num_vector_candidates
vector_num_candidates_for_operator = vector_k * 10

# ---- Vectore Store ----
vector_store = MongoDBAtlasVectorSearch.from_connection_string(
    connection_string=mongo_connection_str,
    namespace=f"{mongo_database}.{mongo_collection_name}",
    embedding=embed_model,
    index_name="search_index_v1",
)

# ---- Retriever (Hybrid) ----

def get_retriever(**kwargs):
    """
    สร้าง Retriever โดยสามารถรับ filter พิเศษสำหรับ Vector Search ได้
    """
    # ดึง vector_search_filter ออกมาจาก kwargs
    vector_search_filter = kwargs.pop('vector_search_filter', None)
    
    # kwargs ที่เหลือ (ถ้ามี) จะถูกใช้เป็น pre_filter
    pre_filter = kwargs if kwargs else None

    retriever = MongoDBAtlasHybridSearchRetriever(
        vectorstore=vector_store,
        search_index_name='search_index_v1',
        embedding=embed_model,
        text_key= 'text',
        embedding_key='embedding',
        top_k=FINAL_TOP_K,
        vector_penalty=HYBRID_VECTOR_PENALTY,
        fulltext_penalty=HYBRID_FULLTEXT_PENALTY,
        vector_search_params={
            "k": vector_k,
            "numCandidates": vector_num_candidates_for_operator,
            # --- ส่ง filter ที่ถูกต้องเข้าไปในตำแหน่งที่ถูกต้อง ---
            "filter": vector_search_filter
        },
        text_search_params={
            "limit": max(20, 2 * FINAL_TOP_K)
        },
        pre_filter=pre_filter
    )
    return retriever

# def get_retriever(**kwargs):
#     retriever = MongoDBAtlasHybridSearchRetriever(
#         vectorstore=vector_store,
#         search_index_name='search_index_v1',
#         embedding=embed_model,
#         text_key= 'text', #'token',
#         embedding_key='embedding',
#         top_k=FINAL_TOP_K,
#         vector_penalty=HYBRID_VECTOR_PENALTY,
#         fulltext_penalty=HYBRID_FULLTEXT_PENALTY,
#         vector_search_params={
#             "k": vector_k,
#             "numCandidates": vector_num_candidates_for_operator
#         },
#         text_search_params={
#             "limit": num_text_candidates
#         },
#         pre_filter=kwargs
#     )
#     return retriever



# ---------- FILTER METAAAAA ----------
# ---------- FILTER METAAAAA ----------
# ---------- FILTER METAAAAA ----------
# ---------- FILTER METAAAAA ----------

# def get_retriever(**kwargs):
#     # ดึง filter ที่เราจะส่งมาจาก tool ออกมาจาก kwargs
#     # เราใช้ .pop() เพื่อเอามันออกมา จะได้ไม่ถูกส่งไปที่ pre_filter ซ้ำซ้อน
#     search_filter = kwargs.pop('filter', None)

#     retriever = MongoDBAtlasHybridSearchRetriever(
#         vectorstore=vector_store,
#         search_index_name='search_index_v1',
#         embedding=embed_model,
#         text_key= 'text',
#         embedding_key='embedding',
#         top_k=FINAL_TOP_K,
#         vector_penalty=HYBRID_VECTOR_PENALTY,
#         fulltext_penalty=HYBRID_FULLTEXT_PENALTY,
#         vector_search_params={
#             "k": vector_k,
#             "numCandidates": vector_num_candidates_for_operator,
#             "filter": search_filter
#         },
#         text_search_params={
#             "limit": num_text_candidates
#         },
#         pre_filter=kwargs 
#     )
#     return retriever