# app.py
from fastapi import FastAPI
from pydantic import BaseModel
from fastapi.middleware.cors import CORSMiddleware
#from bot_instance import gemini_bot, llama_bot  # singleton ErrorBot instance
from typing import List, Optional,Any

import os
from dotenv import load_dotenv
from util import ErrorBot

# from mongo_to_qdrant_ingestor import MongoToQdrantIngestor
# from qdrant_instance import qdrant
# from embedding_model_instance import embedding_model
# from json_to_qdrant_ingestor import JsonToQdrantIngestor

app = FastAPI(title="ErrorBot API")

# ✅ Allow all origins (adjust in production)
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

# ---------------- Request Models ----------------
class MessageItem(BaseModel):
    role: str  # "user" or "bot"
    content: str

class ChatRequest(BaseModel):
    message: str
    history: Optional[List[MessageItem]] = []  # optional conversation history
    lastContext: List[Any] = None

# ---------------- Endpoints ----------------
@app.get("/")
def root():
    return {"status": "ok"}

# @app.post("/chat")
# def chat(request: ChatRequest):
#     """
#     Main chat endpoint:
#     - Accepts a message and optional conversation history
#     - Uses ErrorBot with RAG + LLM
#     """
#     history_list = [
#         {"role": msg.role, "content": msg.content} for msg in request.history
#     ]

#     # Ask bot with history
#     answer = bot.ask(request.message, history=history_list)

#     return {"reply": answer}

load_dotenv()
#GOOGLE_API_KEY = os.environ.get("GOOGLE_API_KEY")

#GROQ_API_KEY = os.getenv("GROQ_API_KEY")

#EMBEDDING_MODEL = "BAAI/bge-large-en-v1.5"
#EMBEDDING_MODEL = "all-MiniLM-L6-v2"
#EMBEDDING_MODEL = "BAAI/bge-m3"

@app.post("/gemini/chat")
def gemini_chat(request: ChatRequest):
    history_list = [{"role": msg.role, "content": msg.content} for msg in request.history]
    gemini_bot = ErrorBot(
        llm_model_name="gemini-2.5-flash",
        llm_provider="gemini",
        last_context = request.lastContext
    )
    #print("In App.py")
    #print(request.lastContext)
    answer, last_context = gemini_bot.ask(request.message, history=history_list)
    #print(answer)
    #print(last_context)
    return {"reply": answer, "last_context": last_context}

@app.post("/llama/chat")
def llama_chat(request: ChatRequest):
    history_list = [{"role": msg.role, "content": msg.content} for msg in request.history]
    llama_bot = ErrorBot(
        llm_model_name="llama-3.3-70b-versatile",
        llm_provider="groq",
        last_context = request.lastContext

    )
    answer, last_context = llama_bot.ask(request.message, history=history_list)
    #print(answer)
    #print(last_context)
    return {"reply": answer, "last_context": last_context}

# @app.post("/ingest/mongodb")
# def ingest_mongodb():
#     """
#     Ingest documents from MongoDB into the bot's knowledge base.
#     """

#     ingestor = MongoToQdrantIngestor(qdrant, embedding_model, collection_name="technical_errors")


#     # def build_content(doc, entity_type):
#     #     """Simple example function to build textual content."""
#     #     return f"{entity_type}: {doc.get('title', '')} {doc.get('description', '')}"

#     def build_content(doc: dict, entity_type: str) -> str:
#         """Convert MongoDB document into natural text for embeddings."""
#         parts = [f"{entity_type} ID: {doc.get('id', str(doc.get('_id', '')))}"]
#         for k, v in doc.items():
#             if k in ["_id"]:  # skip ObjectId
#                 continue
#             if isinstance(v, list):
#                 parts.append(f"{k}: {', '.join(map(str, v))}")
#             elif isinstance(v, dict):
#                 nested = "; ".join([f"{nk}: {nv}" for nk, nv in v.items() if nv])
#                 parts.append(f"{k}: {nested}")
#             else:
#                 if v:
#                     parts.append(f"{k}: {v}")
#         return "\n".join(parts)

#     ingestor.ingest_from_mongodb(
     
#         build_content_fn=build_content,
#         batch_size=500,
#     )
#     return {"status": "ingestion started"}


# @app.post("/ingest/json")
# def ingest_json():
#     """
#     Ingest documents from JSON into the bot's knowledge base.
#     """

#     json_sources = {
#         "ProblemReport": "./json/problemReports.json",
#         "Correction": "./json/corrections.json",
#         "FaultAnalysis": "./json/faultanalysis.json",
#     }
#     ingestor = JsonToQdrantIngestor(qdrant, embedding_model, collection_name="technical_errors")


#     def build_content(doc, entity_type):
#         """Simple example function to build textual content."""
#         return f"{entity_type}: {doc.get('title', '')} {doc.get('description', '')}"

#     ingestor.ingest_from_json(
#         json_sources,
#         build_content_fn=build_content,
#         batch_size=500,
#     )
#     return {"status": "ingestion started"}