from metadata_filtering import build_filters_from_query
from llama_index.llms.openai import OpenAI
import os
from llama_index.vector_stores.qdrant import QdrantVectorStore
from qdrant_client import QdrantClient
from llama_index.core import VectorStoreIndex
from qdrant_client.models import PayloadSchemaType
from llama_index.embeddings.huggingface import HuggingFaceEmbedding
from perplexity import  get_response_from_perplexity
from reranker import  reranker
#from Evaluation import evaluation


def query_knowledge_base(index, user_query, perplexity_key, cohere_key):

  qdrant_key = os.getenv("Qdrant_key")

  llm = OpenAI(model="gpt-4o-mini")
  perplexity_context = get_response_from_perplexity(user_query, perplexity_key)
  cohere_rerank3 = reranker(cohere_key)
  filters = build_filters_from_query(user_query)

  query_engine = index.as_query_engine(
      similarity_top_k=25,
      node_postprocessors=[cohere_rerank3],
      filters = filters
  )

  response = query_engine.query(user_query)

  retrieved_context = "\n\n".join(
    node.node.get_content()
    for node in response.source_nodes
  )

  final_prompt = f"""
  
  You are an AI Tutor specialized in Artificial Intelligence, Machine Learning, Deep Learning, Large Language Models, and related subfields.
  You must answer ONLY questions related to AI/ML and its subfields.
  If the user asks any question outside these topics, politely respond:
  "I can answer only AI-related questions."

  
  You are answering a question using two sources:
  1. Retrieved knowledge base context
  2. External context

  Use retrieved context as primary truth.
  Use external context only to supplement or clarify.

  Retrieved context:
  {retrieved_context}

  External context:
  {perplexity_context}

  Question:
  {user_query}
  """

  final_answer = llm.complete(final_prompt)

  return final_answer



def connect_tovector_store():

  qdrant_key = os.getenv("Qdrant_key")  
  qdrant_client = QdrantClient(
      url="https://afc34f29-812e-40ea-b515-a8cc6ae9ed37.us-east4-0.gcp.cloud.qdrant.io:6333",
      api_key=qdrant_key,
      prefer_grpc=False
  )
  vector_store = QdrantVectorStore(
      client=qdrant_client,
      collection_name="ai_tutor_knowledge",
  )

  embed_model = HuggingFaceEmbedding(
      model_name="sentence-transformers/all-MiniLM-L6-v2"
  )

  index = VectorStoreIndex.from_vector_store(vector_store, embed_model=embed_model)

  qdrant_client.create_payload_index(
      collection_name="ai_tutor_knowledge",
      field_name="excerpt_keywords",
      field_schema=PayloadSchemaType.TEXT
  )

  return index