db_operations.py

import datetime
from qdrant_client import QdrantClient, models
from langchain_qdrant import Qdrant

class DatabaseOperations:

    def __init__(self):
        pass

    def save_user_history(client, collection_name, question, answer, embeddings, point_id, user_id, session_id):
    
        vector = embeddings.embed_documents([question])[0]
        client.upsert(
        collection_name=collection_name,
        points=[
            models.PointStruct(
                id=point_id,
                payload={"user_id": user_id,
                        "session_id": session_id,
                        "create_date": datetime.datetime.now().isoformat(),
                        "question": question,
                        "answer": answer},
                vector=vector,
            )
        ],
    )
        
    def save_user_history_demo(client, collection_name, question, answer, embeddings, point_id, manual, feedback):
    
        vector = embeddings.embed_documents([question])[0]
        client.upsert(
        collection_name=collection_name,
        points=[
            models.PointStruct(
                id=point_id,
                payload={"manual": manual,
                         "create_date": datetime.datetime.now().isoformat(),
                         "question": question,
                         "answer": answer,
                         "feedback": feedback},
                vector=vector,
            )
        ],
    )    
        
    def question_history_search(client, collection_name, car_id, question, embeddings, threshold=0.9):

        CAR_ID = car_id

        vector = embeddings.embed_documents([question])[0]

        search_result = client.search(collection_name=collection_name,
                query_vector=vector,
                query_filter=models.Filter(
                    must=[
                        models.FieldCondition(key="car_id", match=models.MatchValue(value=CAR_ID)),
                        models.FieldCondition(key="source_name", match=models.MatchValue(value="User Question"))
                    ]
                ),
                score_threshold=threshold,
                limit=1)
        return search_result
    
    def user_history_scroll(client, collection_name, user_id, key="user_id"):

        history = client.scroll(collection_name=collection_name,
                scroll_filter=models.Filter(
                    must=[
                        models.FieldCondition(key=key, match=models.MatchValue(value=user_id))
                    ]
                ))
        return history

    def save_question_history(client, collection_name, question, answer, embeddings, 
                          point_id, car_id, source_name, model_year):
    
        vector = embeddings.embed_documents([question])[0]
        client.upsert(
        collection_name=collection_name,
        points=[
            models.PointStruct(
                id=point_id,
                payload={"source_name": source_name,
                        "page_content": question,
                        "car_id": car_id,
                        "model_year": model_year,
                        "create_date": datetime.datetime.now().isoformat(),
                        "answer": answer},
                vector=vector,
            )
        ],
    )