"""
app/routers/chat.py — RAG Endpunkt (WP-05)

Version:
    0.1.1 (Debug-Logging enabled)
"""

from fastapi import APIRouter, HTTPException, Depends
from typing import List
import time
import uuid
import logging

from app.models.dto import ChatRequest, ChatResponse, QueryRequest, QueryHit
from app.services.llm_service import LLMService
# Annahme: Der Retriever aus WP-04 liegt hier.
from app.core.retriever import Retriever 

router = APIRouter()
logger = logging.getLogger(__name__)

def get_llm_service():
    return LLMService()

def get_retriever():
    return Retriever()

def _build_context_from_hits(hits: List[QueryHit]) -> str:
    """
    Formatiert die Suchtreffer zu einem String für den Prompt.
    """
    context_parts = []
    for i, hit in enumerate(hits, 1):
        source = hit.source or {}
        # Wir probieren alle möglichen Felder, in denen Text stecken könnte
        content = (
            source.get("text") or 
            source.get("content") or 
            source.get("page_content") or 
            source.get("chunk_text") or # Oft verwendet
            "[[LEERER INHALT - PAYLOAD PRÜFEN]]"
        )
        
        title = hit.note_id or "Unknown Note"
        
        # Debug Log für jeden Hit
        logger.info(f"Building Context [{i}]: ID={hit.node_id} Content-Length={len(str(content))}")
        
        entry = (
            f"SOURCE [{i}]: {title} (Score: {hit.total_score:.2f})\n"
            f"CONTENT: {content}\n"
        )
        context_parts.append(entry)
    
    return "\n---\n".join(context_parts)

@router.post("/", response_model=ChatResponse)
async def chat_endpoint(
    request: ChatRequest,
    llm: LLMService = Depends(get_llm_service),
    retriever: Retriever = Depends(get_retriever)
):
    start_time = time.time()
    query_id = str(uuid.uuid4())
    
    logger.info(f"Chat request [{query_id}]: {request.message}")

    try:
        # 1. Retrieval
        query_req = QueryRequest(
            query=request.message,
            mode="hybrid",
            top_k=request.top_k,
            explain=request.explain
        )
        
        retrieve_result = await retriever.search(query_req)
        hits = retrieve_result.results

        # 2. Kontext bauen
        if not hits:
            logger.info(f"[{query_id}] No hits found for context.")
            context_str = "Keine relevanten Notizen gefunden."
        else:
            context_str = _build_context_from_hits(hits)
            
        # WICHTIG: Wir loggen den ersten Teil des Kontextes, um zu sehen, was das LLM bekommt
        logger.info(f"--- LLM CONTEXT PREVIEW ---\n{context_str[:500]}\n--- END PREVIEW ---")

        # 3. LLM Generation
        answer_text = await llm.generate_rag_response(
            query=request.message,
            context_str=context_str
        )

        # 4. Response
        duration_ms = int((time.time() - start_time) * 1000)
        
        return ChatResponse(
            query_id=retrieve_result.query_id,
            answer=answer_text,
            sources=hits,
            latency_ms=duration_ms
        )

    except Exception as e:
        logger.error(f"Error in chat endpoint: {e}", exc_info=True)
        raise HTTPException(status_code=500, detail=str(e))