backend/api_backend.py

from fastapi import FastAPI, UploadFile, File, HTTPException
from fastapi.middleware.cors import CORSMiddleware
from pydantic import BaseModel
from typing import List, Optional
import os

# Import your agent directly (same repo)
from nivra_agent import nivra_chat

app = FastAPI(
    title="Nivra AI Healthcare Assistant API",
    description="🩺 India-first AI Healthcare Assistant with ClinicalBERT + Groq",
    version="1.0.0"
)

# CORS for Flutter app (production-ready)
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],  # Lock this to your Flutter app domain in production
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

class SymptomInput(BaseModel):
    symptoms: List[str] = []
    language: str = "en"
    age: Optional[int] = None
    gender: Optional[str] = None

class DiagnosisResponse(BaseModel):
    diagnosis: str
    confidence: float = 0.85
    recommendations: str = ""
    urgency: str = "low"
    audio_url: Optional[str] = None
    success: bool = True

@app.post("/diagnose/text", response_model=DiagnosisResponse)
async def diagnose_text_symptoms(input: SymptomInput):
    """
    Main App endpoint - Text-based symptom diagnosis
    Calls Nivra AI Agent for diagnosis via text
    """
    try:
        # Format prompt for your agent
        symptoms_text = "Patient age: {} {}, symptoms: {}".format(
            input.age or "unknown", 
            input.gender or "unknown", 
            ", ".join(input.symptoms)
        )
        
        # Call YOUR existing nivra_chat agent directly (no HTTP calls!)
        diagnosis = nivra_chat(symptoms_text)
        
        # Parse urgency from diagnosis (simple keyword matching)
        urgency = "low"
        if any(word in diagnosis.lower() for word in ["critical", "emergency", "severe"]):
            urgency = "critical"
        elif any(word in diagnosis.lower() for word in ["consult doctor", "see specialist"]):
            urgency = "medium"
        
        return DiagnosisResponse(
            diagnosis=diagnosis,
            confidence=0.85,
            recommendations="Follow the guidance above. Consult a doctor if symptoms worsen.",
            urgency=urgency,
            audio_url=f"https://huggingface.co/spaces/nivra/tts/{input.language}",  # TTS endpoint
            success=True
        )
    except Exception as e:
        raise HTTPException(
            status_code=500, 
            detail=f"Diagnosis failed: {str(e)}"
        )

@app.post("/diagnose/image")
async def diagnose_image_symptoms(
    file: UploadFile = File(...),
    age: Optional[int] = None,
    gender: Optional[str] = None
):
    """
    Image-based diagnosis endpoint
    Uses your image_symptom_tool.py
    """
    try:
        # Save uploaded image temporarily
        image_path = f"/tmp/{file.filename}"
        with open(image_path, "wb") as f:
            f.write(await file.read())
        
        # Call your agent with image context
        prompt = f"Patient image analysis: {image_path}"
        if age or gender:
            prompt += f"\nPatient: {age}yo {gender}"
        
        diagnosis = nivra_chat(prompt)
        
        return {
            "diagnosis": diagnosis,
            "type": "image_analysis",
            "success": True
        }
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

@app.get("/")
async def root():
    """Root endpoint - API info"""
    return {
        "message": "🩺 Nivra AI Healthcare API",
        "version": "1.0.0",
        "endpoints": {
            "text_diagnosis": "/diagnose/text",
            "image_diagnosis": "/diagnose/image",
            "health_check": "/health",
            "docs": "/docs"
        }
    }

@app.get("/health")
async def health_check():
    """Health check for monitoring"""
    return {
        "status": "healthy",
        "agent": "nivra_chat loaded",
        "models": ["ClinicalBERT", "Groq LLM", "Indic Parler-TTS"]
    }

# Environment info (useful for debugging on HF Spaces)
@app.get("/info")
async def system_info():
    """System information"""
    return {
        "space_author": os.getenv("SPACE_AUTHOR_NAME", "unknown"),
        "space_repo": os.getenv("SPACE_REPO_NAME", "unknown"),
        "space_id": os.getenv("SPACE_ID", "unknown"),
        "host": os.getenv("SPACE_HOST", "localhost")
    }