checklist/backend/app/main.py

# ============= LOGO CONFIGURABLE =============

from fastapi import FastAPI, File, UploadFile, Form, Depends, HTTPException, status
from fastapi.middleware.cors import CORSMiddleware
from fastapi.security import HTTPBearer, HTTPAuthorizationCredentials
from sqlalchemy.orm import Session, joinedload
from sqlalchemy import func, case, or_
from typing import List, Optional
from io import BytesIO
import os
import boto3
from botocore.client import Config
import uuid
from app.core import config as app_config
from app.core.database import engine, get_db, Base
from app.core.security import verify_password, get_password_hash, create_access_token, decode_access_token
from app import models, schemas
import shutil
from datetime import datetime, timedelta
import sys
import requests

# Función para enviar notificaciones al webhook
def send_answer_notification(answer, question, mechanic, db):
    """Envía notificación al webhook cuando se responde una pregunta marcada"""
    try:
        if not app_config.settings.NOTIFICACION_ENDPOINT:
            print("No hay endpoint de notificación configurado")
            return
        
        # Obtener datos de la inspección
        inspection = db.query(models.Inspection).filter(
            models.Inspection.id == answer.inspection_id
        ).first()
        
        if not inspection:
            return
        
        # Preparar datos para enviar
        notification_data = {
            "tipo": "respuesta_pregunta",
            "pregunta": {
                "id": question.id,
                "texto": question.text,
                "seccion": question.section
            },
            "respuesta": {
                "id": answer.id,
                "valor": answer.answer_value,
                "estado": answer.status,
                "comentario": answer.comment,
                "puntos": answer.points_earned
            },
            "inspeccion": {
                "id": inspection.id,
                "vehiculo_placa": inspection.vehicle_plate,
                "vehiculo_marca": inspection.vehicle_brand,
                "vehiculo_modelo": inspection.vehicle_model,
                "pedido": inspection.order_number,
                "or_number": inspection.or_number
            },
            "mecanico": {
                "id": mechanic.id,
                "nombre": mechanic.full_name,
                "email": mechanic.email
            },
            "timestamp": datetime.utcnow().isoformat()
        }
        
        # Enviar al webhook
        response = requests.post(
            app_config.settings.NOTIFICACION_ENDPOINT,
            json=notification_data,
            timeout=5
        )
        
        if response.status_code == 200:
            print(f"✅ Notificación enviada para pregunta {question.id}")
        else:
            print(f"⚠️ Error al enviar notificación: {response.status_code}")
            
    except Exception as e:
        print(f"❌ Error enviando notificación: {e}")
        # No lanzamos excepción para no interrumpir el flujo normal


def send_completed_inspection_to_n8n(inspection, db):
    """Envía la inspección completa con todas las respuestas e imágenes a n8n"""
    try:
        if not app_config.settings.NOTIFICACION_ENDPOINT:
            print("No hay endpoint de notificación configurado")
            return
        
        print(f"\n🚀 Enviando inspección #{inspection.id} a n8n...")
        
        # Obtener datos del mecánico
        mechanic = db.query(models.User).filter(models.User.id == inspection.mechanic_id).first()
        
        # Obtener checklist
        checklist = db.query(models.Checklist).filter(models.Checklist.id == inspection.checklist_id).first()
        
        # Obtener todas las respuestas con sus imágenes - SOLO de preguntas NO eliminadas
        answers = db.query(models.Answer).options(
            joinedload(models.Answer.media_files),
            joinedload(models.Answer.question)
        ).join(models.Question).filter(
            models.Answer.inspection_id == inspection.id,
            models.Question.is_deleted == False  # Excluir preguntas eliminadas
        ).all()
        
        # Preparar respuestas con imágenes
        respuestas_data = []
        for answer in answers:
            # Obtener URLs de imágenes
            imagenes = []
            for media in answer.media_files:
                if media.file_type == "image":
                    # Extraer filename del file_path (última parte de la URL)
                    filename = media.file_path.split('/')[-1] if media.file_path else "imagen.jpg"
                    
                    imagenes.append({
                        "id": media.id,
                        "url": media.file_path,
                        "filename": filename
                    })
            
            respuestas_data.append({
                "id": answer.id,
                "pregunta": {
                    "id": answer.question.id,
                    "texto": answer.question.text,
                    "seccion": answer.question.section,
                    "orden": answer.question.order,
                    "tipo": answer.question.type
                },
                "respuesta": answer.answer_value,
                "estado": answer.status,
                "comentario": answer.comment,
                "puntos_obtenidos": answer.points_earned,
                "es_critico": answer.is_flagged,
                "imagenes": imagenes,
                "ai_analysis": answer.ai_analysis,
                "chat_history": answer.chat_history  # Incluir historial de chat si existe
            })
        
        # Preparar datos completos de la inspección
        inspeccion_data = {
            "tipo": "inspeccion_completada",
            "inspeccion": {
                "id": inspection.id,
                "estado": inspection.status,
                "or_number": inspection.or_number,
                "work_order_number": inspection.work_order_number,
                "vehiculo": {
                    "placa": inspection.vehicle_plate,
                    "marca": inspection.vehicle_brand,
                    "modelo": inspection.vehicle_model,
                    "kilometraje": inspection.vehicle_km
                },
                "pedido": inspection.order_number,
                "mecanico": {
                    "id": mechanic.id if mechanic else None,
                    "nombre": mechanic.full_name if mechanic else None,
                    "email": mechanic.email if mechanic else None,
                    "codigo_operario": inspection.mechanic_employee_code
                },
                "checklist": {
                    "id": checklist.id if checklist else None,
                    "nombre": checklist.name if checklist else None
                },
                "puntuacion": {
                    "obtenida": inspection.score,
                    "maxima": inspection.max_score,
                    "porcentaje": round(inspection.percentage, 2),
                    "items_criticos": inspection.flagged_items_count
                },
                "fechas": {
                    "inicio": inspection.started_at.isoformat() if inspection.started_at else None,
                    "completado": inspection.completed_at.isoformat() if inspection.completed_at else None
                },
                "pdf_url": inspection.pdf_url,
                "firma": inspection.signature_data
            },
            "respuestas": respuestas_data,
            "timestamp": datetime.utcnow().isoformat()
        }
        
        # Enviar al webhook de n8n
        print(f"📤 Enviando {len(respuestas_data)} respuestas con imágenes a n8n...")
        response = requests.post(
            app_config.settings.NOTIFICACION_ENDPOINT,
            json=inspeccion_data,
            timeout=30  # Timeout más largo para inspecciones completas
        )
        
        if response.status_code == 200:
            print(f"✅ Inspección #{inspection.id} enviada exitosamente a n8n")
            print(f"   - {len(respuestas_data)} respuestas")
            print(f"   - {sum(len(r['imagenes']) for r in respuestas_data)} imágenes")
        else:
            print(f"⚠️ Error al enviar inspección a n8n: {response.status_code}")
            print(f"   Response: {response.text[:200]}")
            
    except Exception as e:
        print(f"❌ Error enviando inspección a n8n: {e}")
        import traceback
        traceback.print_exc()
        # No lanzamos excepción para no interrumpir el flujo normal


BACKEND_VERSION = "1.0.92"
app = FastAPI(title="Checklist Inteligente API", version=BACKEND_VERSION)

# S3/MinIO configuration
S3_ENDPOINT = app_config.MINIO_ENDPOINT
S3_ACCESS_KEY = app_config.MINIO_ACCESS_KEY
S3_SECRET_KEY = app_config.MINIO_SECRET_KEY
S3_IMAGE_BUCKET = app_config.MINIO_IMAGE_BUCKET
S3_PDF_BUCKET = app_config.MINIO_PDF_BUCKET

s3_client = boto3.client(
    's3',
    endpoint_url=S3_ENDPOINT,
    aws_access_key_id=S3_ACCESS_KEY,
    aws_secret_access_key=S3_SECRET_KEY,
    config=Config(signature_version='s3v4'),
    region_name='us-east-1'
)

# Crear tablas
Base.metadata.create_all(bind=engine)

# Información visual al iniciar el backend
print("\n================ BACKEND STARTUP INFO ================")
print(f"Backend version: {BACKEND_VERSION}")
print(f"Database URL: {app_config.settings.DATABASE_URL}")
print(f"Environment: {app_config.settings.ENVIRONMENT}")
print(f"MinIO endpoint: {app_config.MINIO_ENDPOINT}")
print("====================================================\n", flush=True)

# CORS
app.add_middleware(
    CORSMiddleware,
    allow_origins=["http://localhost:5173", "http://localhost:3000"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

security = HTTPBearer()

# Dependency para obtener usuario actual
def get_current_user(
    credentials: HTTPAuthorizationCredentials = Depends(security),
    db: Session = Depends(get_db)
):
    token = credentials.credentials
    
    # Verificar si es un API token (comienza con "syntria_")
    if token.startswith("syntria_"):
        api_token = db.query(models.APIToken).filter(
            models.APIToken.token == token,
            models.APIToken.is_active == True
        ).first()
        
        if not api_token:
            raise HTTPException(
                status_code=status.HTTP_401_UNAUTHORIZED,
                detail="API Token inválido o inactivo"
            )
        
        # Actualizar último uso
        api_token.last_used_at = datetime.utcnow()
        db.commit()
        
        # Obtener usuario
        user = db.query(models.User).filter(models.User.id == api_token.user_id).first()
        if not user or not user.is_active:
            raise HTTPException(
                status_code=status.HTTP_401_UNAUTHORIZED,
                detail="Usuario inválido o inactivo"
            )
        
        return user
    
    # Si no es API token, es JWT token
    payload = decode_access_token(token)
    print(f"Token payload: {payload}")  # Debug
    if payload is None:
        raise HTTPException(
            status_code=status.HTTP_401_UNAUTHORIZED,
            detail="Token inválido o expirado"
        )
    
    user_id = int(payload.get("sub"))
    print(f"Looking for user ID: {user_id}")  # Debug
    user = db.query(models.User).filter(models.User.id == user_id).first()
    if user is None:
        print(f"User not found with ID: {user_id}")  # Debug
        raise HTTPException(status_code=404, detail="Usuario no encontrado")
    
    return user


@app.post("/api/config/logo", response_model=dict)
async def upload_logo(
    file: UploadFile = File(...),
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Sube un logo y lo guarda en MinIO, actualiza la configuración."""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden cambiar el logo")

    # Subir imagen a MinIO
    file_extension = file.filename.split(".")[-1]
    now = datetime.now()
    folder = f"logo"
    file_name = f"logo_{now.strftime('%Y%m%d_%H%M%S')}.{file_extension}"
    s3_key = f"{folder}/{file_name}"
    s3_client.upload_fileobj(file.file, S3_IMAGE_BUCKET, s3_key, ExtraArgs={"ContentType": file.content_type})
    logo_url = f"{S3_ENDPOINT}/{S3_IMAGE_BUCKET}/{s3_key}"

    # Guardar en configuración (puedes tener una tabla Config o usar AIConfiguration)
    config = db.query(models.AIConfiguration).filter(models.AIConfiguration.is_active == True).first()
    if config:
        config.logo_url = logo_url
        db.commit()
        db.refresh(config)
    # Si no hay config, solo retorna la url
    return {"logo_url": logo_url}

@app.get("/api/config/logo", response_model=dict)
def get_logo_url(
    db: Session = Depends(get_db)
):
    config = db.query(models.AIConfiguration).filter(models.AIConfiguration.is_active == True).first()
    if config and getattr(config, "logo_url", None):
        return {"logo_url": config.logo_url}
    # Default logo (puedes poner una url por defecto)
    return {"logo_url": f"{S3_ENDPOINT}/{S3_IMAGE_BUCKET}/logo/default_logo.png"}


# ============= AUTH ENDPOINTS =============
@app.post("/api/auth/register", response_model=schemas.User)
def register(user: schemas.UserCreate, db: Session = Depends(get_db)):
    # Verificar si usuario existe
    db_user = db.query(models.User).filter(models.User.username == user.username).first()
    if db_user:
        raise HTTPException(status_code=400, detail="Usuario ya existe")
    
    # Crear usuario
    hashed_password = get_password_hash(user.password)
    db_user = models.User(
        username=user.username,
        email=user.email,
        full_name=user.full_name,
        role=user.role,
        password_hash=hashed_password
    )
    db.add(db_user)
    db.commit()
    db.refresh(db_user)
    return db_user


@app.post("/api/auth/login", response_model=schemas.Token)
def login(user_login: schemas.UserLogin, db: Session = Depends(get_db)):
    user = db.query(models.User).filter(models.User.username == user_login.username).first()
    if not user or not verify_password(user_login.password, user.password_hash):
        raise HTTPException(
            status_code=status.HTTP_401_UNAUTHORIZED,
            detail="Usuario o contraseña incorrectos"
        )
    
    access_token = create_access_token(data={"sub": str(user.id), "role": user.role})
    return {
        "access_token": access_token,
        "token_type": "bearer",
        "user": user
    }


@app.get("/api/auth/me", response_model=schemas.User)
def get_me(current_user: models.User = Depends(get_current_user)):
    return current_user


# ============= USER ENDPOINTS =============
@app.get("/api/users", response_model=List[schemas.User])
def get_users(
    skip: int = 0,
    limit: int = 100,
    active_only: bool = False,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Solo admin puede ver todos los usuarios
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No tienes permisos para ver usuarios")
    
    query = db.query(models.User)
    if active_only:
        query = query.filter(models.User.is_active == True)
    
    return query.offset(skip).limit(limit).all()


@app.get("/api/users/{user_id}", response_model=schemas.User)
def get_user(
    user_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Solo admin puede ver otros usuarios
    if current_user.role != "admin" and current_user.id != user_id:
        raise HTTPException(status_code=403, detail="No tienes permisos para ver este usuario")
    
    user = db.query(models.User).filter(models.User.id == user_id).first()
    if not user:
        raise HTTPException(status_code=404, detail="Usuario no encontrado")
    
    return user


@app.post("/api/users", response_model=schemas.User)
def create_user(
    user: schemas.UserCreate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Solo admin puede crear usuarios
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No tienes permisos para crear usuarios")
    
    # Verificar si usuario existe
    db_user = db.query(models.User).filter(models.User.username == user.username).first()
    if db_user:
        raise HTTPException(status_code=400, detail="Usuario ya existe")
    
    # Verificar si email existe
    if user.email:
        db_email = db.query(models.User).filter(models.User.email == user.email).first()
        if db_email:
            raise HTTPException(status_code=400, detail="Email ya está en uso")
    
    # Crear usuario
    hashed_password = get_password_hash(user.password)
    db_user = models.User(
        username=user.username,
        email=user.email,
        full_name=user.full_name,
        employee_code=user.employee_code,
        role=user.role,
        password_hash=hashed_password,
        is_active=True
    )
    db.add(db_user)
    db.commit()
    db.refresh(db_user)
    return db_user


@app.put("/api/users/{user_id}", response_model=schemas.User)
def update_user(
    user_id: int,
    user_update: schemas.UserUpdate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Solo admin puede actualizar otros usuarios
    if current_user.role != "admin" and current_user.id != user_id:
        raise HTTPException(status_code=403, detail="No tienes permisos para actualizar este usuario")
    
    db_user = db.query(models.User).filter(models.User.id == user_id).first()
    if not db_user:
        raise HTTPException(status_code=404, detail="Usuario no encontrado")
    
    # Actualizar campos
    if user_update.username is not None:
        # Verificar si username está en uso
        existing = db.query(models.User).filter(
            models.User.username == user_update.username,
            models.User.id != user_id
        ).first()
        if existing:
            raise HTTPException(status_code=400, detail="Nombre de usuario ya está en uso")
        db_user.username = user_update.username
    
    if user_update.email is not None:
        # Verificar si email está en uso
        existing = db.query(models.User).filter(
            models.User.email == user_update.email,
            models.User.id != user_id
        ).first()
        if existing:
            raise HTTPException(status_code=400, detail="Email ya está en uso")
        db_user.email = user_update.email
    
    if user_update.full_name is not None:
        db_user.full_name = user_update.full_name
    
    if user_update.employee_code is not None:
        db_user.employee_code = user_update.employee_code
    
    # Solo admin puede cambiar roles
    if user_update.role is not None:
        if current_user.role != "admin":
            raise HTTPException(status_code=403, detail="No tienes permisos para cambiar roles")
        db_user.role = user_update.role
    
    db.commit()
    db.refresh(db_user)
    return db_user


@app.patch("/api/users/{user_id}/deactivate")
def deactivate_user(
    user_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Solo admin puede inactivar usuarios
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No tienes permisos para inactivar usuarios")
    
    # No permitir auto-inactivación
    if current_user.id == user_id:
        raise HTTPException(status_code=400, detail="No puedes inactivar tu propio usuario")
    
    db_user = db.query(models.User).filter(models.User.id == user_id).first()
    if not db_user:
        raise HTTPException(status_code=404, detail="Usuario no encontrado")
    
    db_user.is_active = False
    db.commit()
    
    return {"message": "Usuario inactivado correctamente", "user_id": user_id}


@app.patch("/api/users/{user_id}/activate")
def activate_user(
    user_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Solo admin puede activar usuarios
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No tienes permisos para activar usuarios")
    
    db_user = db.query(models.User).filter(models.User.id == user_id).first()
    if not db_user:
        raise HTTPException(status_code=404, detail="Usuario no encontrado")
    
    db_user.is_active = True
    db.commit()
    
    return {"message": "Usuario activado correctamente", "user_id": user_id}


@app.patch("/api/users/{user_id}/password")
def change_user_password(
    user_id: int,
    password_update: schemas.AdminPasswordUpdate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Solo admin puede cambiar contraseñas de otros usuarios
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No tienes permisos para cambiar contraseñas")
    
    db_user = db.query(models.User).filter(models.User.id == user_id).first()
    if not db_user:
        raise HTTPException(status_code=404, detail="Usuario no encontrado")
    
    # Cambiar contraseña
    db_user.password_hash = get_password_hash(password_update.new_password)
    db.commit()
    
    return {"message": "Contraseña actualizada correctamente"}


@app.patch("/api/users/me/password")
def change_my_password(
    password_update: schemas.UserPasswordUpdate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Verificar contraseña actual
    if not verify_password(password_update.current_password, current_user.password_hash):
        raise HTTPException(status_code=400, detail="Contraseña actual incorrecta")
    
    # Cambiar contraseña
    current_user.password_hash = get_password_hash(password_update.new_password)
    db.commit()
    
    return {"message": "Contraseña actualizada correctamente"}


@app.put("/api/users/me", response_model=schemas.User)
def update_my_profile(
    user_update: schemas.UserUpdate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Actualizar email
    if user_update.email is not None:
        # Verificar si email está en uso
        existing = db.query(models.User).filter(
            models.User.email == user_update.email,
            models.User.id != current_user.id
        ).first()
        if existing:
            raise HTTPException(status_code=400, detail="Email ya está en uso")
        current_user.email = user_update.email
    
    # Actualizar nombre
    if user_update.full_name is not None:
        current_user.full_name = user_update.full_name
    
    # No permitir cambio de rol desde perfil
    db.commit()
    db.refresh(current_user)
    return current_user


# ============= API TOKENS ENDPOINTS =============
@app.get("/api/users/me/tokens", response_model=List[schemas.APIToken])
def get_my_tokens(
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Obtener todos mis API tokens"""
    tokens = db.query(models.APIToken).filter(
        models.APIToken.user_id == current_user.id
    ).all()
    return tokens


@app.post("/api/users/me/tokens", response_model=schemas.APITokenWithValue)
def create_my_token(
    token_create: schemas.APITokenCreate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Generar un nuevo API token"""
    from app.core.security import generate_api_token
    
    # Generar token único
    token_value = generate_api_token()
    
    # Crear registro
    api_token = models.APIToken(
        user_id=current_user.id,
        token=token_value,
        description=token_create.description,
        is_active=True
    )
    
    db.add(api_token)
    db.commit()
    db.refresh(api_token)
    
    # Retornar con el token completo (solo esta vez)
    return schemas.APITokenWithValue(
        id=api_token.id,
        token=api_token.token,
        description=api_token.description,
        is_active=api_token.is_active,
        last_used_at=api_token.last_used_at,
        created_at=api_token.created_at
    )


@app.delete("/api/users/me/tokens/{token_id}")
def delete_my_token(
    token_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Revocar uno de mis API tokens"""
    api_token = db.query(models.APIToken).filter(
        models.APIToken.id == token_id,
        models.APIToken.user_id == current_user.id
    ).first()
    
    if not api_token:
        raise HTTPException(status_code=404, detail="Token no encontrado")
    
    api_token.is_active = False
    db.commit()
    
    return {"message": "Token revocado correctamente", "token_id": token_id}


@app.get("/api/users/{user_id}/tokens", response_model=List[schemas.APIToken])
def get_user_tokens(
    user_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Obtener tokens de un usuario (solo admin)"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No tienes permisos")
    
    tokens = db.query(models.APIToken).filter(
        models.APIToken.user_id == user_id
    ).all()
    return tokens


@app.delete("/api/users/{user_id}/tokens/{token_id}")
def delete_user_token(
    user_id: int,
    token_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Revocar token de un usuario (solo admin)"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No tienes permisos")
    
    api_token = db.query(models.APIToken).filter(
        models.APIToken.id == token_id,
        models.APIToken.user_id == user_id
    ).first()
    
    if not api_token:
        raise HTTPException(status_code=404, detail="Token no encontrado")
    
    api_token.is_active = False
    db.commit()
    
    return {"message": "Token revocado correctamente", "token_id": token_id}


# ============= CHECKLIST ENDPOINTS =============
@app.get("/api/checklists", response_model=List[schemas.Checklist])
def get_checklists(
    skip: int = 0,
    limit: int = 100,
    active_only: bool = False,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    query = db.query(models.Checklist)
    if active_only:
        query = query.filter(models.Checklist.is_active == True)
    
    # Si es mecánico, solo ver checklists con permiso
    if current_user.role == "mechanic":
        # Obtener IDs de checklists con permiso o sin permisos (acceso global)
        permitted_checklist_ids = db.query(models.ChecklistPermission.checklist_id).filter(
            models.ChecklistPermission.mechanic_id == current_user.id
        ).distinct().all()
        permitted_ids = [id[0] for id in permitted_checklist_ids]
        
        # Checklists sin permisos = acceso global
        checklists_without_permissions = db.query(models.Checklist.id).outerjoin(
            models.ChecklistPermission
        ).group_by(models.Checklist.id).having(
            func.count(models.ChecklistPermission.id) == 0
        ).all()
        global_ids = [id[0] for id in checklists_without_permissions]
        
        all_allowed_ids = list(set(permitted_ids + global_ids))
        if all_allowed_ids:
            query = query.filter(models.Checklist.id.in_(all_allowed_ids))
        else:
            # Si no hay permisos, devolver lista vacía
            return []
    
    checklists = query.offset(skip).limit(limit).all()
    
    # Agregar allowed_mechanics a cada checklist
    for checklist in checklists:
        permissions = db.query(models.ChecklistPermission.mechanic_id).filter(
            models.ChecklistPermission.checklist_id == checklist.id
        ).all()
        checklist.allowed_mechanics = [p[0] for p in permissions]
    
    return checklists


@app.get("/api/checklists/{checklist_id}", response_model=schemas.ChecklistWithQuestions)
def get_checklist(checklist_id: int, db: Session = Depends(get_db)):
    checklist = db.query(models.Checklist).filter(models.Checklist.id == checklist_id).first()
    
    if not checklist:
        raise HTTPException(status_code=404, detail="Checklist no encontrado")
    
    # Cargar solo preguntas NO eliminadas
    checklist.questions = db.query(models.Question).filter(
        models.Question.checklist_id == checklist_id,
        models.Question.is_deleted == False
    ).order_by(models.Question.order).all()
    
    # Agregar allowed_mechanics
    permissions = db.query(models.ChecklistPermission.mechanic_id).filter(
        models.ChecklistPermission.checklist_id == checklist.id
    ).all()
    checklist.allowed_mechanics = [p[0] for p in permissions]
    
    return checklist


@app.post("/api/checklists", response_model=schemas.Checklist)
def create_checklist(
    checklist: schemas.ChecklistCreate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No autorizado")
    
    # Extraer mechanic_ids antes de crear el checklist
    checklist_data = checklist.dict(exclude={'mechanic_ids'})
    mechanic_ids = checklist.mechanic_ids or []
    
    db_checklist = models.Checklist(**checklist_data, created_by=current_user.id)
    db.add(db_checklist)
    db.flush()  # Para obtener el ID
    
    # Crear permisos para mecánicos seleccionados
    for mechanic_id in mechanic_ids:
        permission = models.ChecklistPermission(
            checklist_id=db_checklist.id,
            mechanic_id=mechanic_id
        )
        db.add(permission)
    
    db.commit()
    db.refresh(db_checklist)
    
    # Agregar allowed_mechanics a la respuesta
    db_checklist.allowed_mechanics = mechanic_ids
    
    return db_checklist


@app.put("/api/checklists/{checklist_id}", response_model=schemas.Checklist)
def update_checklist(
    checklist_id: int,
    checklist: schemas.ChecklistUpdate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No autorizado")
    
    db_checklist = db.query(models.Checklist).filter(models.Checklist.id == checklist_id).first()
    if not db_checklist:
        raise HTTPException(status_code=404, detail="Checklist no encontrado")
    
    # Extraer mechanic_ids si se envía
    update_data = checklist.dict(exclude_unset=True, exclude={'mechanic_ids'})
    mechanic_ids = checklist.mechanic_ids
    
    # Actualizar campos del checklist
    for key, value in update_data.items():
        setattr(db_checklist, key, value)
    
    # Si se proporcionan mechanic_ids, actualizar permisos
    if mechanic_ids is not None:
        # Eliminar permisos existentes
        db.query(models.ChecklistPermission).filter(
            models.ChecklistPermission.checklist_id == checklist_id
        ).delete()
        
        # Crear nuevos permisos
        for mechanic_id in mechanic_ids:
            permission = models.ChecklistPermission(
                checklist_id=checklist_id,
                mechanic_id=mechanic_id
            )
            db.add(permission)
    
    db.commit()
    db.refresh(db_checklist)
    
    # Agregar allowed_mechanics a la respuesta
    permissions = db.query(models.ChecklistPermission.mechanic_id).filter(
        models.ChecklistPermission.checklist_id == checklist_id
    ).all()
    db_checklist.allowed_mechanics = [p[0] for p in permissions]
    
    return db_checklist


@app.post("/api/checklists/{checklist_id}/upload-logo")
async def upload_checklist_logo(
    checklist_id: int,
    file: UploadFile = File(...),
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Subir logo para un checklist (solo admin)"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden subir logos")
    
    # Verificar que el checklist existe
    checklist = db.query(models.Checklist).filter(models.Checklist.id == checklist_id).first()
    if not checklist:
        raise HTTPException(status_code=404, detail="Checklist no encontrado")
    
    # Validar que es una imagen
    if not file.content_type.startswith('image/'):
        raise HTTPException(status_code=400, detail="El archivo debe ser una imagen")
    
    # Subir a S3/MinIO
    file_extension = file.filename.split(".")[-1]
    now = datetime.now()
    folder = f"checklist-logos/{now.year}/{now.month:02d}"
    file_name = f"checklist_{checklist_id}_{uuid.uuid4().hex}.{file_extension}"
    s3_key = f"{folder}/{file_name}"
    
    file_content = await file.read()
    s3_client.upload_fileobj(
        BytesIO(file_content),
        S3_IMAGE_BUCKET,
        s3_key,
        ExtraArgs={"ContentType": file.content_type}
    )
    
    logo_url = f"{S3_ENDPOINT}/{S3_IMAGE_BUCKET}/{s3_key}"
    
    # Actualizar checklist
    checklist.logo_url = logo_url
    db.commit()
    db.refresh(checklist)
    
    return {"logo_url": logo_url, "message": "Logo subido exitosamente"}


@app.delete("/api/checklists/{checklist_id}/logo")
def delete_checklist_logo(
    checklist_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Eliminar logo de un checklist (solo admin)"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden eliminar logos")
    
    checklist = db.query(models.Checklist).filter(models.Checklist.id == checklist_id).first()
    if not checklist:
        raise HTTPException(status_code=404, detail="Checklist no encontrado")
    
    checklist.logo_url = None
    db.commit()
    
    return {"message": "Logo eliminado exitosamente"}


# ============= QUESTION ENDPOINTS =============
@app.post("/api/questions", response_model=schemas.Question)
def create_question(
    question: schemas.QuestionCreate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No autorizado")
    
    db_question = models.Question(**question.dict())
    db.add(db_question)
    db.commit()
    db.refresh(db_question)
    
    # Recalcular max_score del checklist DESPUÉS de persistir
    recalculate_checklist_max_score(question.checklist_id, db)
    db.commit()
    
    # Registrar auditoría
    audit_log = models.QuestionAuditLog(
        question_id=db_question.id,
        checklist_id=question.checklist_id,
        user_id=current_user.id,
        action="created",
        new_value=f"Pregunta creada: {question.text}",
        comment=f"Sección: {question.section}, Tipo: {question.type}, Puntos: {question.points}"
    )
    db.add(audit_log)
    db.commit()
    
    return db_question


# Helper function para recalcular max_score de un checklist
def recalculate_checklist_max_score(checklist_id: int, db: Session):
    """Recalcula el max_score sumando los puntos de todas las preguntas NO eliminadas"""
    total_score = db.query(func.sum(models.Question.points)).filter(
        models.Question.checklist_id == checklist_id,
        models.Question.is_deleted == False
    ).scalar() or 0
    
    checklist = db.query(models.Checklist).filter(models.Checklist.id == checklist_id).first()
    if checklist:
        checklist.max_score = total_score
        print(f"✅ Checklist #{checklist_id} max_score recalculado: {total_score}")
    
    return total_score


@app.put("/api/questions/{question_id}", response_model=schemas.Question)
def update_question(
    question_id: int,
    question: schemas.QuestionUpdate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No autorizado")
    
    db_question = db.query(models.Question).filter(models.Question.id == question_id).first()
    if not db_question:
        raise HTTPException(status_code=404, detail="Pregunta no encontrada")
    
    # Guardar valores anteriores para auditoría
    import json
    changes = []
    
    for key, value in question.dict(exclude_unset=True).items():
        old_value = getattr(db_question, key)
        if old_value != value:
            # Convertir a string para comparación y almacenamiento
            old_str = json.dumps(old_value, ensure_ascii=False) if isinstance(old_value, (dict, list)) else str(old_value)
            new_str = json.dumps(value, ensure_ascii=False) if isinstance(value, (dict, list)) else str(value)
            
            changes.append({
                'field': key,
                'old': old_str,
                'new': new_str
            })
            setattr(db_question, key, value)
    
    # Si cambiaron los puntos, hacer flush y recalcular
    points_changed = any(change['field'] == 'points' for change in changes)
    
    db.commit()
    db.refresh(db_question)
    
    # Registrar auditoría para cada campo cambiado
    for change in changes:
        audit_log = models.QuestionAuditLog(
            question_id=question_id,
            checklist_id=db_question.checklist_id,
            user_id=current_user.id,
            action="updated",
            field_name=change['field'],
            old_value=change['old'],
            new_value=change['new'],
            comment=f"Campo '{change['field']}' modificado"
        )
        db.add(audit_log)
    
    if changes:
        db.commit()
    
    # Si cambiaron los puntos, recalcular DESPUÉS del commit
    if points_changed:
        recalculate_checklist_max_score(db_question.checklist_id, db)
        db.commit()
    
    return db_question


@app.patch("/api/checklists/{checklist_id}/questions/reorder")
def reorder_questions(
    checklist_id: int,
    reorder_data: List[schemas.QuestionReorder],
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Reordenar preguntas de un checklist"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No autorizado")
    
    # Verificar que el checklist existe
    checklist = db.query(models.Checklist).filter(models.Checklist.id == checklist_id).first()
    if not checklist:
        raise HTTPException(status_code=404, detail="Checklist no encontrado")
    
    # Actualizar el orden de cada pregunta
    for item in reorder_data:
        question = db.query(models.Question).filter(
            models.Question.id == item.question_id,
            models.Question.checklist_id == checklist_id
        ).first()
        
        if question:
            old_order = question.order
            question.order = item.new_order
            question.updated_at = datetime.utcnow()
            
            # Registrar auditoría
            audit_log = models.QuestionAuditLog(
                question_id=question.id,
                checklist_id=checklist_id,
                user_id=current_user.id,
                action="updated",
                field_name="order",
                old_value=str(old_order),
                new_value=str(item.new_order),
                comment="Orden de pregunta actualizado"
            )
            db.add(audit_log)
    
    db.commit()
    
    return {"message": "Orden de preguntas actualizado exitosamente", "updated_count": len(reorder_data)}


@app.delete("/api/questions/{question_id}")
def delete_question(
    question_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No autorizado")
    
    db_question = db.query(models.Question).filter(models.Question.id == question_id).first()
    if not db_question:
        raise HTTPException(status_code=404, detail="Pregunta no encontrada")
    
    if db_question.is_deleted:
        raise HTTPException(status_code=400, detail="La pregunta ya está eliminada")
    
    # Registrar auditoría antes de eliminar
    audit_log = models.QuestionAuditLog(
        question_id=question_id,
        checklist_id=db_question.checklist_id,
        user_id=current_user.id,
        action="deleted",
        old_value=f"Pregunta eliminada: {db_question.text}",
        comment=f"Sección: {db_question.section}, Tipo: {db_question.type}, Puntos: {db_question.points}"
    )
    db.add(audit_log)
    
    # SOFT DELETE: marcar como eliminada
    db_question.is_deleted = True
    db_question.updated_at = datetime.utcnow()
    
    # También marcar como eliminadas todas las subpreguntas (en cascada)
    subquestions = db.query(models.Question).filter(
        models.Question.parent_question_id == question_id,
        models.Question.is_deleted == False
    ).all()
    
    subquestion_count = 0
    for subq in subquestions:
        subq.is_deleted = True
        subq.updated_at = datetime.utcnow()
        subquestion_count += 1
        
        # Registrar auditoría de subpregunta
        sub_audit_log = models.QuestionAuditLog(
            question_id=subq.id,
            checklist_id=subq.checklist_id,
            user_id=current_user.id,
            action="deleted",
            old_value=f"Subpregunta eliminada en cascada: {subq.text}",
            comment=f"Eliminada junto con pregunta padre #{question_id}"
        )
        db.add(sub_audit_log)
    
    db.commit()
    
    # Recalcular max_score del checklist DESPUÉS del commit
    recalculate_checklist_max_score(db_question.checklist_id, db)
    db.commit()
    
    message = "Pregunta eliminada exitosamente"
    if subquestion_count > 0:
        message += f" junto con {subquestion_count} subpregunta(s)"
    
    return {
        "message": message,
        "id": question_id,
        "subquestions_deleted": subquestion_count,
        "note": "Las respuestas históricas se mantienen intactas. Las preguntas no aparecerán en nuevas inspecciones."
    }


@app.get("/api/questions/{question_id}/audit", response_model=List[schemas.QuestionAuditLog])
def get_question_audit_history(
    question_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Obtener historial de cambios de una pregunta"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden ver el historial")
    
    audit_logs = db.query(models.QuestionAuditLog).filter(
        models.QuestionAuditLog.question_id == question_id
    ).order_by(models.QuestionAuditLog.created_at.desc()).all()
    
    return audit_logs


@app.get("/api/checklists/{checklist_id}/questions/audit", response_model=List[schemas.QuestionAuditLog])
def get_checklist_questions_audit_history(
    checklist_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Obtener historial de cambios de todas las preguntas de un checklist"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden ver el historial")
    
    audit_logs = db.query(models.QuestionAuditLog).filter(
        models.QuestionAuditLog.checklist_id == checklist_id
    ).order_by(models.QuestionAuditLog.created_at.desc()).all()
    
    return audit_logs


# ============= INSPECTION ENDPOINTS =============
@app.get("/api/inspections", response_model=List[schemas.Inspection])
def get_inspections(
    skip: int = 0,
    limit: int = 100,
    vehicle_plate: str = None,
    status: str = None,
    show_inactive: bool = False,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    query = db.query(models.Inspection)
    
    # Por defecto, solo mostrar inspecciones activas
    if not show_inactive:
        query = query.filter(models.Inspection.is_active == True)
    
    # Mecánicos solo ven sus inspecciones
    if current_user.role == "mechanic":
        query = query.filter(models.Inspection.mechanic_id == current_user.id)
    
    if vehicle_plate:
        query = query.filter(models.Inspection.vehicle_plate.contains(vehicle_plate))
    
    if status:
        query = query.filter(models.Inspection.status == status)
    
    return query.order_by(models.Inspection.created_at.desc()).offset(skip).limit(limit).all()


@app.get("/api/inspections/{inspection_id}", response_model=schemas.InspectionDetail)
def get_inspection(
    inspection_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    inspection = db.query(models.Inspection).options(
        joinedload(models.Inspection.checklist),
        joinedload(models.Inspection.mechanic),
        joinedload(models.Inspection.answers).joinedload(models.Answer.question),
        joinedload(models.Inspection.answers).joinedload(models.Answer.media_files)
    ).filter(models.Inspection.id == inspection_id).first()
    
    if not inspection:
        raise HTTPException(status_code=404, detail="Inspección no encontrada")
    
    # Cargar solo preguntas NO eliminadas del checklist
    if inspection.checklist:
        inspection.checklist.questions = db.query(models.Question).filter(
            models.Question.checklist_id == inspection.checklist.id,
            models.Question.is_deleted == False
        ).order_by(models.Question.order).all()
    
    return inspection


@app.post("/api/inspections", response_model=schemas.Inspection)
def create_inspection(
    inspection: schemas.InspectionCreate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Obtener max_score del checklist
    checklist = db.query(models.Checklist).filter(
        models.Checklist.id == inspection.checklist_id
    ).first()
    
    if not checklist:
        raise HTTPException(status_code=404, detail="Checklist no encontrado")
    
    # Crear inspección con el employee_code del mecánico actual
    inspection_data = inspection.dict()
    inspection_data['mechanic_employee_code'] = current_user.employee_code  # Agregar código de operario automáticamente
    
    db_inspection = models.Inspection(
        **inspection_data,
        mechanic_id=current_user.id,
        max_score=checklist.max_score
    )
    db.add(db_inspection)
    db.commit()
    db.refresh(db_inspection)
    return db_inspection


@app.put("/api/inspections/{inspection_id}", response_model=schemas.Inspection)
def update_inspection(
    inspection_id: int,
    inspection: schemas.InspectionUpdate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    db_inspection = db.query(models.Inspection).filter(
        models.Inspection.id == inspection_id
    ).first()
    
    if not db_inspection:
        raise HTTPException(status_code=404, detail="Inspección no encontrada")
    
    for key, value in inspection.dict(exclude_unset=True).items():
        setattr(db_inspection, key, value)
    
    db.commit()
    db.refresh(db_inspection)
    return db_inspection


def generate_inspection_pdf(inspection_id: int, db: Session) -> str:
    """
    Genera el PDF de una inspección y lo sube a S3.
    Retorna la URL del PDF generado.
    """
    from reportlab.lib.pagesizes import A4
    from reportlab.lib import colors
    from reportlab.lib.units import inch, mm
    from reportlab.platypus import SimpleDocTemplate, Table, TableStyle, Paragraph, Spacer, Image as RLImage, PageBreak, KeepTogether
    from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
    from reportlab.lib.enums import TA_CENTER, TA_LEFT, TA_RIGHT, TA_JUSTIFY
    from io import BytesIO
    import requests
    
    inspection = db.query(models.Inspection).filter(models.Inspection.id == inspection_id).first()
    if not inspection:
        raise HTTPException(status_code=404, detail="Inspección no encontrada")
    
    buffer = BytesIO()
    doc = SimpleDocTemplate(
        buffer, 
        pagesize=A4,
        rightMargin=15*mm,
        leftMargin=15*mm,
        topMargin=20*mm,
        bottomMargin=20*mm
    )
    
    elements = []
    styles = getSampleStyleSheet()
    
    # Estilos personalizados
    title_style = ParagraphStyle(
        'CustomTitle',
        parent=styles['Heading1'],
        fontSize=24,
        textColor=colors.HexColor('#1e3a8a'),
        spaceAfter=6,
        alignment=TA_CENTER,
        fontName='Helvetica-Bold'
    )
    
    subtitle_style = ParagraphStyle(
        'CustomSubtitle',
        parent=styles['Normal'],
        fontSize=11,
        textColor=colors.HexColor('#475569'),
        spaceAfter=20,
        alignment=TA_CENTER
    )
    
    section_header_style = ParagraphStyle(
        'SectionHeader',
        parent=styles['Heading2'],
        fontSize=14,
        textColor=colors.HexColor('#1e40af'),
        spaceBefore=16,
        spaceAfter=10,
        fontName='Helvetica-Bold',
        borderWidth=0,
        borderColor=colors.HexColor('#3b82f6'),
        borderPadding=6,
        backColor=colors.HexColor('#eff6ff')
    )
    
    info_style = ParagraphStyle(
        'InfoStyle',
        parent=styles['Normal'],
        fontSize=10,
        textColor=colors.HexColor('#334155'),
        spaceAfter=4
    )
    
    small_style = ParagraphStyle(
        'SmallStyle',
        parent=styles['Normal'],
        fontSize=8,
        textColor=colors.HexColor('#64748b')
    )
    
    # Estilos mejorados para preguntas y respuestas
    question_style = ParagraphStyle(
        'QuestionStyle',
        parent=styles['Normal'],
        fontSize=11,
        textColor=colors.HexColor('#1f2937'),
        spaceAfter=3,
        fontName='Helvetica-Bold'
    )
    
    answer_style = ParagraphStyle(
        'AnswerStyle',
        parent=styles['Normal'],
        fontSize=10,
        textColor=colors.HexColor('#374151'),
        spaceAfter=4
    )
    
    comment_style = ParagraphStyle(
        'CommentStyle',
        parent=styles['Normal'],
        fontSize=9,
        textColor=colors.HexColor('#6b7280'),
        spaceAfter=6,
        leftIndent=10,
        rightIndent=10
    )
    
    # Obtener datos
    mechanic = db.query(models.User).filter(models.User.id == inspection.mechanic_id).first()
    checklist = db.query(models.Checklist).filter(models.Checklist.id == inspection.checklist_id).first()
    
    print(f"🔍 DEBUG: Checklist ID: {inspection.checklist_id}")
    print(f"🔍 DEBUG: Checklist obtenido: {checklist}")
    if checklist:
        print(f"🔍 DEBUG: Checklist.logo_url = {getattr(checklist, 'logo_url', 'NO EXISTE')}")
    
    # Obtener logo principal de configuración (empresa)
    config = db.query(models.AIConfiguration).filter(models.AIConfiguration.is_active == True).first()
    company_logo_url = None
    if config and getattr(config, "logo_url", None):
        company_logo_url = config.logo_url
        print(f"📸 Logo de la empresa: {company_logo_url}")
    else:
        print("ℹ️ No hay logo de empresa configurado")
    
    # Obtener logo del checklist (NO usar fallback)
    checklist_logo_url = None
    if checklist and getattr(checklist, "logo_url", None):
        checklist_logo_url = checklist.logo_url
        print(f"📋 Logo del checklist: {checklist_logo_url}")
    else:
        print(f"ℹ️ Checklist sin logo propio")
    
    print(f"🎯 RESULTADO: company_logo={company_logo_url}, checklist_logo={checklist_logo_url}")
    
    # ===== PORTADA =====
    elements.append(Spacer(1, 10*mm))
    
    # Función helper para cargar y dimensionar logos (optimizada)
    def load_logo(logo_url, max_width_mm=45, max_height_mm=35):
        """Carga un logo desde URL y retorna objeto Image con dimensiones ajustadas"""
        if not logo_url:
            return None
        try:
            # Reducir timeout para respuestas más rápidas
            logo_resp = requests.get(logo_url, timeout=5)
            
            if logo_resp.status_code == 200:
                logo_bytes = BytesIO(logo_resp.content)
                logo_img = RLImage(logo_bytes)
                
                # Ajustar tamaño manteniendo aspect ratio
                aspect = logo_img.imageHeight / float(logo_img.imageWidth)
                logo_width = max_width_mm * mm
                logo_height = logo_width * aspect
                
                # Si la altura excede el máximo, ajustar por altura
                if logo_height > max_height_mm * mm:
                    logo_height = max_height_mm * mm
                    logo_width = logo_height / aspect
                
                logo_img.drawWidth = logo_width
                logo_img.drawHeight = logo_height
                
                return logo_img
            else:
                print(f"❌ Error HTTP cargando logo: {logo_resp.status_code}")
        except Exception as e:
            print(f"⚠️ Error cargando logo: {str(e)[:100]}")
        return None
    
    # Cargar ambos logos en paralelo usando ThreadPoolExecutor
    from concurrent.futures import ThreadPoolExecutor, as_completed
    
    company_logo = None
    checklist_logo = None
    
    with ThreadPoolExecutor(max_workers=2) as executor:
        futures = {}
        if company_logo_url:
            futures[executor.submit(load_logo, company_logo_url, 50, 35)] = 'company'
        if checklist_logo_url:
            futures[executor.submit(load_logo, checklist_logo_url, 50, 35)] = 'checklist'
        
        for future in as_completed(futures):
            logo_type = futures[future]
            try:
                result = future.result()
                if logo_type == 'company':
                    company_logo = result
                    if result:
                        print(f"✅ Logo empresa cargado")
                elif logo_type == 'checklist':
                    checklist_logo = result
                    if result:
                        print(f"✅ Logo checklist cargado")
            except Exception as e:
                print(f"❌ Error procesando logo {logo_type}: {e}")
    
    # Crear tabla con logos en los extremos (ancho total disponible ~180mm)
    logo_row = []
    
    # Logo empresa (izquierda)
    if company_logo:
        logo_row.append(company_logo)
    else:
        logo_row.append(Paragraph("", styles['Normal']))  # Espacio vacío
    
    # Espaciador central flexible
    logo_row.append(Paragraph("", styles['Normal']))
    
    # Logo checklist (derecha)
    if checklist_logo:
        logo_row.append(checklist_logo)
    else:
        logo_row.append(Paragraph("", styles['Normal']))  # Espacio vacío
    
    # Crear tabla con logos - columnas ajustadas para maximizar separación
    # Columna 1: 55mm (logo empresa), Columna 2: 70mm (espacio), Columna 3: 55mm (logo checklist)
    logo_table = Table([logo_row], colWidths=[55*mm, 70*mm, 55*mm])
    logo_table.setStyle(TableStyle([
        ('ALIGN', (0, 0), (0, 0), 'LEFT'),      # Logo empresa a la izquierda
        ('ALIGN', (1, 0), (1, 0), 'CENTER'),    # Centro vacío
        ('ALIGN', (2, 0), (2, 0), 'RIGHT'),     # Logo checklist a la derecha
        ('VALIGN', (0, 0), (-1, -1), 'MIDDLE'), # Alineación vertical al centro
        # DEBUG: Agregar bordes para ver la distribución
        # ('GRID', (0, 0), (-1, -1), 0.5, colors.red),
    ]))
    elements.append(logo_table)
    elements.append(Spacer(1, 5*mm))
    
    # Título con diseño moderno
    elements.append(Paragraph("📋 INFORME DE INSPECCIÓN VEHICULAR", title_style))
    elements.append(Paragraph(f"N° {inspection.id}", subtitle_style))
    elements.append(Spacer(1, 10*mm))
    
    # Estilo para etiquetas de información
    label_style = ParagraphStyle(
        'LabelStyle',
        parent=styles['Normal'],
        fontSize=9,
        textColor=colors.HexColor('#64748b'),
        spaceAfter=2
    )
    
    value_style = ParagraphStyle(
        'ValueStyle',
        parent=styles['Normal'],
        fontSize=11,
        textColor=colors.HexColor('#1e293b'),
        fontName='Helvetica-Bold'
    )
    
    # Cuadro de información del vehículo con diseño moderno
    vehicle_header = Table(
        [[Paragraph("<b>🚗 INFORMACIÓN DEL VEHÍCULO</b>", ParagraphStyle('veh_header', parent=info_style, fontSize=12, textColor=colors.white))]],
        colWidths=[85*mm]
    )
    vehicle_header.setStyle(TableStyle([
        ('BACKGROUND', (0, 0), (-1, -1), colors.HexColor('#2563eb')),
        ('PADDING', (0, 0), (-1, -1), 10),
        ('ALIGN', (0, 0), (-1, -1), 'LEFT'),
        ('ROUNDEDCORNERS', [6, 6, 0, 0]),
    ]))
    
    vehicle_content = Table([
        [Paragraph("Placa", label_style), Paragraph(f"{inspection.vehicle_plate}", value_style)],
        [Paragraph("Marca", label_style), Paragraph(f"{inspection.vehicle_brand or 'N/A'}", value_style)],
        [Paragraph("Modelo", label_style), Paragraph(f"{inspection.vehicle_model or 'N/A'}", value_style)],
        [Paragraph("Kilometraje", label_style), Paragraph(f"{inspection.vehicle_km or 'N/A'} km", value_style)]
    ], colWidths=[25*mm, 60*mm])
    vehicle_content.setStyle(TableStyle([
        ('PADDING', (0, 0), (-1, -1), 10),
        ('BACKGROUND', (0, 0), (-1, -1), colors.white),
        ('VALIGN', (0, 0), (-1, -1), 'TOP'),
        ('LINEBELOW', (0, 0), (-1, -2), 0.5, colors.HexColor('#e2e8f0')),
    ]))
    
    vehicle_table = Table(
        [[vehicle_header], [vehicle_content]],
        colWidths=[85*mm]
    )
    vehicle_table.setStyle(TableStyle([
        ('BOX', (0, 0), (-1, -1), 1.5, colors.HexColor('#2563eb')),
        ('VALIGN', (0, 0), (-1, -1), 'TOP'),
        ('ROUNDEDCORNERS', [6, 6, 6, 6]),
    ]))
    
    # Cuadro de información del cliente e inspección (sin nombre de mecánico por privacidad)
    client_header = Table(
        [[Paragraph("<b>📄 INFORMACIÓN DE LA INSPECCIÓN</b>", ParagraphStyle('client_header', parent=info_style, fontSize=12, textColor=colors.white))]],
        colWidths=[85*mm]
    )
    client_header.setStyle(TableStyle([
        ('BACKGROUND', (0, 0), (-1, -1), colors.HexColor('#16a34a')),
        ('PADDING', (0, 0), (-1, -1), 10),
        ('ALIGN', (0, 0), (-1, -1), 'LEFT'),
        ('ROUNDEDCORNERS', [6, 6, 0, 0]),
    ]))
    
    client_content = Table([
        [Paragraph("Nº Pedido", label_style), Paragraph(f"{inspection.order_number or 'N/A'}", value_style)],
        [Paragraph("OR N°", label_style), Paragraph(f"{inspection.or_number or 'N/A'}", value_style)],
        [Paragraph("Cód. Operario", label_style), Paragraph(f"{inspection.mechanic_employee_code or 'N/A'}", value_style)],
        [Paragraph("Fecha", label_style), Paragraph(f"{inspection.started_at.strftime('%d/%m/%Y %H:%M') if inspection.started_at else 'N/A'}", value_style)]
    ], colWidths=[25*mm, 60*mm])
    client_content.setStyle(TableStyle([
        ('PADDING', (0, 0), (-1, -1), 10),
        ('BACKGROUND', (0, 0), (-1, -1), colors.white),
        ('VALIGN', (0, 0), (-1, -1), 'TOP'),
        ('LINEBELOW', (0, 0), (-1, -2), 0.5, colors.HexColor('#e2e8f0')),
    ]))
    
    inspection_info_table = Table(
        [[client_header], [client_content]],
        colWidths=[85*mm]
    )
    inspection_info_table.setStyle(TableStyle([
        ('BOX', (0, 0), (-1, -1), 1.5, colors.HexColor('#16a34a')),
        ('VALIGN', (0, 0), (-1, -1), 'TOP'),
        ('ROUNDEDCORNERS', [6, 6, 6, 6]),
    ]))
    
    # Tabla con ambos cuadros lado a lado
    info_table = Table([[vehicle_table, inspection_info_table]], colWidths=[90*mm, 90*mm])
    info_table.setStyle(TableStyle([
        ('ALIGN', (0, 0), (-1, -1), 'CENTER'),
        ('VALIGN', (0, 0), (-1, -1), 'TOP'),
    ]))
    elements.append(info_table)
    elements.append(Spacer(1, 8*mm))
    
    # Resumen de puntuación con diseño mejorado
    percentage = inspection.percentage
    score_color = colors.HexColor('#22c55e') if percentage >= 80 else colors.HexColor('#eab308') if percentage >= 60 else colors.HexColor('#ef4444')
    score_label = "EXCELENTE" if percentage >= 80 else "ACEPTABLE" if percentage >= 60 else "DEFICIENTE"
    
    # Título de resumen
    score_title = Table(
        [[Paragraph("<b>📊 RESUMEN DE EVALUACIÓN</b>", ParagraphStyle('score_title', parent=info_style, fontSize=14, textColor=colors.HexColor('#1e293b'), alignment=TA_CENTER))]],
        colWidths=[180*mm]
    )
    score_title.setStyle(TableStyle([
        ('PADDING', (0, 0), (-1, -1), 8),
        ('ALIGN', (0, 0), (-1, -1), 'CENTER'),
    ]))
    elements.append(score_title)
    elements.append(Spacer(1, 3*mm))
    
    # Cuadro de métricas con diseño moderno
    metric_label = ParagraphStyle('metric_label', parent=small_style, fontSize=9, textColor=colors.HexColor('#64748b'), alignment=TA_CENTER)
    metric_value = ParagraphStyle('metric_value', parent=info_style, fontSize=16, fontName='Helvetica-Bold', alignment=TA_CENTER)
    
    metrics_data = [
        [Paragraph("Puntuación", metric_label), Paragraph("Porcentaje", metric_label), Paragraph("Estado", metric_label), Paragraph("Ítems Críticos", metric_label)],
        [
            Paragraph(f"<b>{inspection.score}</b> / {inspection.max_score}", metric_value),
            Paragraph(f"<b>{percentage:.1f}%</b>", metric_value),
            Paragraph(f"<b>{score_label}</b>", ParagraphStyle('status_value', parent=metric_value, textColor=score_color)),
            Paragraph(f"<b>{inspection.flagged_items_count}</b>", metric_value)
        ]
    ]
    
    score_table = Table(metrics_data, colWidths=[45*mm, 45*mm, 45*mm, 45*mm])
    score_table.setStyle(TableStyle([
        ('BACKGROUND', (0, 0), (-1, 0), colors.HexColor('#f8fafc')),
        ('BACKGROUND', (0, 1), (-1, -1), colors.white),
        ('ALIGN', (0, 0), (-1, -1), 'CENTER'),
        ('VALIGN', (0, 0), (-1, -1), 'MIDDLE'),
        ('PADDING', (0, 0), (-1, -1), 12),
        ('BOX', (0, 0), (-1, -1), 2, score_color),
        ('LINEABOVE', (0, 1), (-1, 1), 1.5, score_color),
        ('GRID', (0, 0), (-1, -1), 0.5, colors.HexColor('#e2e8f0')),
        ('ROUNDEDCORNERS', [8, 8, 8, 8]),
    ]))
    elements.append(score_table)
    elements.append(PageBreak())
    
    # ===== DETALLE DE RESPUESTAS =====
    elements.append(Paragraph("📝 DETALLE DE LA INSPECCIÓN", section_header_style))
    elements.append(Spacer(1, 5*mm))
    
    # Obtener respuestas agrupadas por sección - SOLO de preguntas NO eliminadas
    answers = db.query(models.Answer).options(
        joinedload(models.Answer.media_files),
        joinedload(models.Answer.question)
    ).join(models.Question).filter(
        models.Answer.inspection_id == inspection_id,
        models.Question.is_deleted == False  # Excluir preguntas eliminadas
    ).order_by(
        models.Question.section,
        models.Question.order
    ).all()
    
    # Función helper para convertir valores técnicos a etiquetas legibles
    def get_readable_answer(answer_value, question_options):
        """
        Convierte el valor técnico de la respuesta a su etiqueta legible.
        Ej: 'option1' -> 'Bueno', 'pass' -> 'Pasa'
        """
        if not answer_value or not question_options:
            return answer_value or 'Sin respuesta'
        
        config = question_options
        question_type = config.get('type', '')
        
        # Para tipos con choices (boolean, single_choice, multiple_choice)
        if question_type in ['boolean', 'single_choice', 'multiple_choice'] and config.get('choices'):
            # Si es multiple_choice, puede tener varios valores separados por coma
            if question_type == 'multiple_choice' and ',' in answer_value:
                values = answer_value.split(',')
                labels = []
                for val in values:
                    val = val.strip()
                    choice = next((c for c in config['choices'] if c.get('value') == val), None)
                    if choice:
                        labels.append(choice.get('label', val))
                    else:
                        labels.append(val)
                return ', '.join(labels)
            else:
                # Buscar la etiqueta correspondiente al valor
                choice = next((c for c in config['choices'] if c.get('value') == answer_value), None)
                if choice:
                    return choice.get('label', answer_value)
        
        # Para tipos scale, text, number, date, time - devolver el valor tal cual
        return answer_value
    
    current_section = None
    for ans in answers:
        question = ans.question
        
        # Nueva sección
        if question.section != current_section:
            if current_section is not None:
                elements.append(Spacer(1, 5*mm))
            current_section = question.section
            elements.append(Paragraph(f"▶ {question.section or 'General'}", section_header_style))
            elements.append(Spacer(1, 3*mm))
        
        # Estado visual
        status_colors = {
            'ok': colors.HexColor('#22c55e'),
            'warning': colors.HexColor('#eab308'),
            'critical': colors.HexColor('#ef4444')
        }
        status_icons = {
            'ok': '✓',
            'warning': '⚠',
            'critical': '✕'
        }
        status_color = status_colors.get(ans.status, colors.HexColor('#64748b'))
        status_icon = status_icons.get(ans.status, '●')
        
        # Tabla de pregunta/respuesta
        question_data = []
        
        # Fila 1: Pregunta con estilo mejorado
        question_data.append([
            Paragraph(f"<b>{status_icon} {question.text}</b>", question_style),
        ])
        
        # Fila 2: Respuesta y estado - Convertir valor técnico a etiqueta legible
        answer_text = get_readable_answer(ans.answer_value, question.options)
        question_data.append([
            Table([
                [
                    Paragraph(f"<b>Respuesta:</b> {answer_text}", answer_style),
                    Paragraph(f"<b>Estado:</b> {ans.status.upper()}", ParagraphStyle('status', parent=answer_style, textColor=status_color, fontName='Helvetica-Bold'))
                ]
            ], colWidths=[120*mm, 50*mm])
        ])
        
        # Fila 3: Comentario mejorado (si existe)
        if ans.comment:
            comment_text = ans.comment
            
            # Limpiar prefijo de análisis automático/IA si existe (con cualquier porcentaje)
            import re
            # Patrón para detectar "Análisis Automático (XX% confianza): " o "Análisis IA (XX% confianza): "
            comment_text = re.sub(r'^(Análisis Automático|Análisis IA)\s*\(\d+%\s*confianza\):\s*', '', comment_text)
            # También remover variantes sin emoji
            comment_text = re.sub(r'^🤖\s*(Análisis Automático|Análisis IA)\s*\(\d+%\s*confianza\):\s*', '', comment_text)
            
            # Separar análisis y recomendaciones con salto de línea
            if "Recomendaciones:" in comment_text or "Recomendación:" in comment_text:
                comment_text = comment_text.replace("Recomendaciones:", "<br/><br/><b>Recomendaciones:</b>")
                comment_text = comment_text.replace("Recomendación:", "<br/><br/><b>Recomendación:</b>")
            
            question_data.append([
                Paragraph(f"<b>Comentario:</b> {comment_text}", comment_style)
            ])
        
        # Fila 4: Imágenes (si existen)
        if ans.media_files:
            media_imgs = []
            for media in ans.media_files:
                if media.file_type == "image":
                    try:
                        img_resp = requests.get(media.file_path, timeout=10)
                        if img_resp.status_code == 200:
                            img_bytes = BytesIO(img_resp.content)
                            rl_img = RLImage(img_bytes, width=25*mm, height=25*mm)
                            media_imgs.append(rl_img)
                    except Exception as e:
                        print(f"Error cargando imagen {media.file_path}: {e}")
            
            if media_imgs:
                # Crear tabla de miniaturas (máximo 6 por fila)
                img_rows = []
                for i in range(0, len(media_imgs), 6):
                    img_rows.append(media_imgs[i:i+6])
                
                img_table = Table(img_rows)
                img_table.setStyle(TableStyle([
                    ('ALIGN', (0, 0), (-1, -1), 'LEFT'),
                    ('VALIGN', (0, 0), (-1, -1), 'MIDDLE'),
                    ('PADDING', (0, 0), (-1, -1), 2),
                ]))
                question_data.append([img_table])
        
        # Tabla de la pregunta completa
        q_table = Table(question_data, colWidths=[180*mm])
        q_table.setStyle(TableStyle([
            ('BACKGROUND', (0, 0), (-1, 0), colors.HexColor('#f1f5f9')),
            ('PADDING', (0, 0), (-1, -1), 6),
            ('BOX', (0, 0), (-1, -1), 0.5, colors.HexColor('#cbd5e1')),
            ('LEFTPADDING', (0, 0), (-1, -1), 8),
            ('VALIGN', (0, 0), (-1, -1), 'TOP'),
        ]))
        
        elements.append(KeepTogether(q_table))
        elements.append(Spacer(1, 3*mm))
    
    # ===== FIRMA =====
    if inspection.signature_data:
        elements.append(PageBreak())
        elements.append(Spacer(1, 10*mm))
        elements.append(Paragraph("✍️ FIRMA DEL OPERARIO", section_header_style))
        elements.append(Spacer(1, 5*mm))
        
        try:
            # Decodificar firma base64
            import base64
            signature_bytes = base64.b64decode(inspection.signature_data.split(',')[1] if ',' in inspection.signature_data else inspection.signature_data)
            signature_img_buffer = BytesIO(signature_bytes)
            signature_img = RLImage(signature_img_buffer, width=80*mm, height=40*mm)
            
            # Tabla con la firma y datos
            signature_data = [
                [signature_img],
                [Paragraph(f"<b>Operario:</b> {inspection.mechanic_employee_code or 'N/A'}", info_style)],
                [Paragraph(f"<b>Fecha de finalización:</b> {inspection.completed_at.strftime('%d/%m/%Y %H:%M') if inspection.completed_at else 'N/A'}", info_style)]
            ]
            
            signature_table = Table(signature_data, colWidths=[180*mm])
            signature_table.setStyle(TableStyle([
                ('ALIGN', (0, 0), (-1, -1), 'CENTER'),
                ('VALIGN', (0, 0), (-1, -1), 'MIDDLE'),
                ('LINEABOVE', (0, 0), (0, 0), 1, colors.HexColor('#cbd5e1')),
                ('PADDING', (0, 0), (-1, -1), 8),
            ]))
            
            elements.append(signature_table)
            print(f"✅ Firma agregada al PDF")
        except Exception as e:
            print(f"⚠️ Error agregando firma al PDF: {e}")
            elements.append(Paragraph(
                f"<i>Error al cargar la firma</i>",
                ParagraphStyle('error', parent=small_style, alignment=TA_CENTER, textColor=colors.HexColor('#ef4444'))
            ))
    
    # ===== FOOTER =====
    elements.append(Spacer(1, 10*mm))
    elements.append(Paragraph(
        f"Documento generado automáticamente por Checklist Inteligente el {datetime.now().strftime('%d/%m/%Y a las %H:%M')}",
        ParagraphStyle('footer', parent=small_style, alignment=TA_CENTER, textColor=colors.HexColor('#94a3b8'))
    ))
    
    # Generar PDF
    try:
        doc.build(elements)
    except Exception as e:
        print(f"❌ Error al generar PDF: {e}")
        import traceback
        traceback.print_exc()
        raise HTTPException(status_code=500, detail=f"Error al generar PDF: {str(e)}")
    
    # Subir a S3
    buffer.seek(0)
    now = datetime.now()
    folder = f"{now.year}/{now.month:02d}"
    filename = f"inspeccion_{inspection_id}_{inspection.vehicle_plate or 'sin-patente'}.pdf"
    s3_key = f"{folder}/{filename}"
    buffer.seek(0)
    s3_client.upload_fileobj(buffer, S3_PDF_BUCKET, s3_key, ExtraArgs={"ContentType": "application/pdf"})
    pdf_url = f"{S3_ENDPOINT}/{S3_PDF_BUCKET}/{s3_key}"
    
    print(f"✅ PDF generado y subido a S3: {pdf_url}")
    return pdf_url


@app.post("/api/inspections/{inspection_id}/complete", response_model=schemas.Inspection)
def complete_inspection(
    inspection_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    inspection = db.query(models.Inspection).filter(
        models.Inspection.id == inspection_id
    ).first()
    
    if not inspection:
        raise HTTPException(status_code=404, detail="Inspección no encontrada")
    
    # Calcular score - SOLO de preguntas NO eliminadas
    answers = db.query(models.Answer).join(models.Question).filter(
        models.Answer.inspection_id == inspection_id,
        models.Question.is_deleted == False  # Excluir preguntas eliminadas
    ).all()
    total_score = sum(a.points_earned for a in answers)
    flagged_count = sum(1 for a in answers if a.is_flagged)
    
    inspection.score = total_score
    inspection.percentage = (total_score / inspection.max_score * 100) if inspection.max_score > 0 else 0
    inspection.flagged_items_count = flagged_count
    inspection.status = "completed"
    inspection.completed_at = datetime.utcnow()

    # Generar PDF usando función reutilizable
    pdf_url = generate_inspection_pdf(inspection_id, db)
    inspection.pdf_url = pdf_url
    db.commit()
    db.refresh(inspection)
    
    # Enviar inspección completa a n8n con todas las respuestas e imágenes
    send_completed_inspection_to_n8n(inspection, db)
    
    return inspection


@app.patch("/api/inspections/{inspection_id}/deactivate")
def deactivate_inspection(
    inspection_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Solo admin puede inactivar
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="No tienes permisos para inactivar inspecciones")
    
    inspection = db.query(models.Inspection).filter(
        models.Inspection.id == inspection_id
    ).first()
    
    if not inspection:
        raise HTTPException(status_code=404, detail="Inspección no encontrada")
    
    inspection.is_active = False
    inspection.status = "inactive"
    
    db.commit()
    db.refresh(inspection)
    
    return {"message": "Inspección inactivada correctamente", "inspection_id": inspection_id}


# ============= ANSWER ENDPOINTS =============
@app.post("/api/answers", response_model=schemas.Answer)
def create_answer(
    answer: schemas.AnswerCreate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Obtener la pregunta para saber los puntos
    question = db.query(models.Question).filter(models.Question.id == answer.question_id).first()
    
    if not question:
        raise HTTPException(status_code=404, detail="Pregunta no encontrada")
    
    # Sistema simplificado: 1 punto por pregunta correcta
    points_earned = 0
    if answer.status == "ok":
        points_earned = 1
    elif answer.status == "warning":
        points_earned = 0.5
    
    # Buscar si ya existe una respuesta para esta inspección y pregunta
    existing_answer = db.query(models.Answer).filter(
        models.Answer.inspection_id == answer.inspection_id,
        models.Answer.question_id == answer.question_id
    ).first()
    if existing_answer:
        # Actualizar la respuesta existente
        # Si status es pass/fail, no poner valor por defecto en answer_value
        if answer.status in ["pass", "fail"] and not answer.answer_value:
            existing_answer.answer_value = None
        else:
            existing_answer.answer_value = answer.answer_value
        existing_answer.status = answer.status
        existing_answer.comment = getattr(answer, "comment", None)
        existing_answer.ai_analysis = getattr(answer, "ai_analysis", None)
        existing_answer.is_flagged = getattr(answer, "is_flagged", False)
        existing_answer.points_earned = points_earned
        existing_answer.updated_at = datetime.utcnow()
        db.commit()
        db.refresh(existing_answer)
        
        
        # Solo enviar si tiene valor real (no vacío ni None)
        if question.send_notification and answer.answer_value:
            print(f"✅ Enviando notificación para pregunta #{question.id}")
            send_answer_notification(existing_answer, question, current_user, db)
        else:
            if not question.send_notification:
                print(f"❌ NO se envía notificación (send_notification=False) para pregunta #{question.id}")
            else:
                print(f"⏭️  NO se envía notificación (respuesta vacía) para pregunta #{question.id}")
        
        return existing_answer
    else:
        # Si status es pass/fail y no hay valor, no poner valor por defecto en answer_value
        answer_data = answer.dict()
        if answer.status in ["pass", "fail"] and not answer.answer_value:
            answer_data["answer_value"] = None
        db_answer = models.Answer(
            **answer_data,
            points_earned=points_earned
        )
        db.add(db_answer)
        db.commit()
        db.refresh(db_answer)
        
    
        
        # Solo enviar si tiene valor real (no vacío ni None)
        if question.send_notification and answer.answer_value:
            print(f"✅ Enviando notificación para pregunta #{question.id}")
            send_answer_notification(db_answer, question, current_user, db)
        else:
            if not question.send_notification:
                print(f"❌ NO se envía notificación (send_notification=False) para pregunta #{question.id}")
            else:
                print(f"⏭️  NO se envía notificación (respuesta vacía) para pregunta #{question.id}")
        
        return db_answer


@app.put("/api/answers/{answer_id}", response_model=schemas.Answer)
def update_answer(
    answer_id: int,
    answer: schemas.AnswerUpdate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    db_answer = db.query(models.Answer).filter(models.Answer.id == answer_id).first()
    
    if not db_answer:
        raise HTTPException(status_code=404, detail="Respuesta no encontrada")
    
    # Obtener la inspección para verificar si está completada
    inspection = db.query(models.Inspection).filter(
        models.Inspection.id == db_answer.inspection_id
    ).first()
    
    if not inspection:
        raise HTTPException(status_code=404, detail="Inspección no encontrada")
    
    # Recalcular puntos si cambió el status
    if answer.status and answer.status != db_answer.status:
        question = db.query(models.Question).filter(
            models.Question.id == db_answer.question_id
        ).first()
        
        if answer.status == "ok":
            db_answer.points_earned = question.points
        elif answer.status == "warning":
            db_answer.points_earned = int(question.points * 0.5)
        else:
            db_answer.points_earned = 0
    
    for key, value in answer.dict(exclude_unset=True).items():
        setattr(db_answer, key, value)
    
    db.commit()
    db.refresh(db_answer)
    
    # Si la inspección está completada, regenerar PDF con los cambios
    if inspection.status == "completed":
        print(f"🔄 Regenerando PDF para inspección completada #{inspection.id}")
        
        # Recalcular score de la inspección - SOLO de preguntas NO eliminadas
        answers = db.query(models.Answer).join(models.Question).filter(
            models.Answer.inspection_id == inspection.id,
            models.Question.is_deleted == False  # Excluir preguntas eliminadas
        ).all()
        
        inspection.score = sum(a.points_earned for a in answers)
        inspection.percentage = (inspection.score / inspection.max_score * 100) if inspection.max_score > 0 else 0
        inspection.flagged_items_count = sum(1 for a in answers if a.is_flagged)
        
        # Regenerar PDF
        try:
            pdf_url = generate_inspection_pdf(inspection.id, db)
            inspection.pdf_url = pdf_url
            db.commit()
            print(f"✅ PDF regenerado exitosamente: {pdf_url}")
        except Exception as e:
            print(f"❌ Error regenerando PDF: {e}")
            import traceback
            traceback.print_exc()
            # No lanzamos excepción para no interrumpir la actualización de la respuesta
    
    return db_answer


# ============= AUDIT LOG ENDPOINTS =============
@app.get("/api/inspections/{inspection_id}/audit-log", response_model=List[schemas.AuditLog])
def get_inspection_audit_log(
    inspection_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Obtener el historial de cambios de una inspección"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden ver el historial")
    
    logs = db.query(models.InspectionAuditLog).filter(
        models.InspectionAuditLog.inspection_id == inspection_id
    ).order_by(models.InspectionAuditLog.created_at.desc()).all()
    
    # Agregar nombre de usuario a cada log
    result = []
    for log in logs:
        log_dict = {
            "id": log.id,
            "inspection_id": log.inspection_id,
            "answer_id": log.answer_id,
            "user_id": log.user_id,
            "action": log.action,
            "entity_type": log.entity_type,
            "field_name": log.field_name,
            "old_value": log.old_value,
            "new_value": log.new_value,
            "comment": log.comment,
            "created_at": log.created_at,
            "user_name": None
        }
        
        user = db.query(models.User).filter(models.User.id == log.user_id).first()
        if user:
            log_dict["user_name"] = user.full_name or user.username
        
        result.append(schemas.AuditLog(**log_dict))
    
    return result


@app.put("/api/answers/{answer_id}/admin-edit", response_model=schemas.Answer)
def admin_edit_answer(
    answer_id: int,
    answer_edit: schemas.AnswerEdit,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Editar una respuesta (solo admin) con registro de auditoría"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden editar respuestas")
    
    db_answer = db.query(models.Answer).filter(models.Answer.id == answer_id).first()
    
    if not db_answer:
        raise HTTPException(status_code=404, detail="Respuesta no encontrada")
    
    # Registrar cambios en el log de auditoría
    changes = []
    
    if answer_edit.answer_value is not None and answer_edit.answer_value != db_answer.answer_value:
        changes.append({
            "field_name": "answer_value",
            "old_value": db_answer.answer_value,
            "new_value": answer_edit.answer_value
        })
        db_answer.answer_value = answer_edit.answer_value
    
    if answer_edit.status is not None and answer_edit.status != db_answer.status:
        changes.append({
            "field_name": "status",
            "old_value": db_answer.status,
            "new_value": answer_edit.status
        })
        
        # Recalcular puntos
        question = db.query(models.Question).filter(
            models.Question.id == db_answer.question_id
        ).first()
        
        old_points = db_answer.points_earned
        if answer_edit.status == "ok":
            db_answer.points_earned = question.points
        elif answer_edit.status == "warning":
            db_answer.points_earned = int(question.points * 0.5)
        else:
            db_answer.points_earned = 0
        
        if old_points != db_answer.points_earned:
            changes.append({
                "field_name": "points_earned",
                "old_value": str(old_points),
                "new_value": str(db_answer.points_earned)
            })
        
        db_answer.status = answer_edit.status
    
    if answer_edit.comment is not None and answer_edit.comment != db_answer.comment:
        changes.append({
            "field_name": "comment",
            "old_value": db_answer.comment or "",
            "new_value": answer_edit.comment
        })
        db_answer.comment = answer_edit.comment
    
    if answer_edit.is_flagged is not None and answer_edit.is_flagged != db_answer.is_flagged:
        changes.append({
            "field_name": "is_flagged",
            "old_value": str(db_answer.is_flagged),
            "new_value": str(answer_edit.is_flagged)
        })
        db_answer.is_flagged = answer_edit.is_flagged
    
    # Crear registros de auditoría para cada cambio
    for change in changes:
        audit_log = models.InspectionAuditLog(
            inspection_id=db_answer.inspection_id,
            answer_id=answer_id,
            user_id=current_user.id,
            action="updated",
            entity_type="answer",
            field_name=change["field_name"],
            old_value=change["old_value"],
            new_value=change["new_value"],
            comment=answer_edit.edit_comment or "Editado por administrador"
        )
        db.add(audit_log)
    
    db_answer.updated_at = datetime.utcnow()
    db.commit()
    db.refresh(db_answer)
    
    # Si la inspección está completada, regenerar PDF con los cambios
    inspection = db.query(models.Inspection).filter(
        models.Inspection.id == db_answer.inspection_id
    ).first()
    
    if inspection and inspection.status == "completed":
        print(f"🔄 Regenerando PDF para inspección completada #{inspection.id} (admin-edit)")
        
        # Recalcular score de la inspección - SOLO de preguntas NO eliminadas
        answers = db.query(models.Answer).join(models.Question).filter(
            models.Answer.inspection_id == inspection.id,
            models.Question.is_deleted == False  # Excluir preguntas eliminadas
        ).all()
        
        inspection.score = sum(a.points_earned for a in answers)
        inspection.percentage = (inspection.score / inspection.max_score * 100) if inspection.max_score > 0 else 0
        inspection.flagged_items_count = sum(1 for a in answers if a.is_flagged)
        
        # Regenerar PDF
        try:
            pdf_url = generate_inspection_pdf(inspection.id, db)
            inspection.pdf_url = pdf_url
            db.commit()
            print(f"✅ PDF regenerado exitosamente: {pdf_url}")
        except Exception as e:
            print(f"❌ Error regenerando PDF: {e}")
            import traceback
            traceback.print_exc()
            # No lanzamos excepción para no interrumpir la actualización de la respuesta
    
    return db_answer


# ============= MEDIA FILE ENDPOINTS =============
@app.post("/api/answers/{answer_id}/upload", response_model=schemas.MediaFile)
async def upload_photo(
    answer_id: int,
    file: UploadFile = File(...),
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    # Verificar que la respuesta existe
    answer = db.query(models.Answer).filter(models.Answer.id == answer_id).first()
    if not answer:
        raise HTTPException(status_code=404, detail="Respuesta no encontrada")
    
    # Subir imagen a S3/MinIO
    file_extension = file.filename.split(".")[-1]
    now = datetime.now()
    folder = f"{now.year}/{now.month:02d}"
    file_name = f"answer_{answer_id}_{uuid.uuid4().hex}.{file_extension}"
    s3_key = f"{folder}/{file_name}"
    s3_client.upload_fileobj(file.file, S3_IMAGE_BUCKET, s3_key, ExtraArgs={"ContentType": file.content_type})
    # Generar URL pública (ajusta si usas presigned)
    image_url = f"{S3_ENDPOINT}/{S3_IMAGE_BUCKET}/{s3_key}"
    # Crear registro en BD
    media_file = models.MediaFile(
        answer_id=answer_id,
        file_path=image_url,
        file_type="image"
    )
    db.add(media_file)
    db.commit()
    db.refresh(media_file)
    return media_file


# ============= AI ANALYSIS =============
@app.get("/api/ai/models", response_model=List[schemas.AIModelInfo])
def get_available_ai_models(current_user: models.User = Depends(get_current_user)):
    """Obtener lista de modelos de IA disponibles"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden ver modelos de IA")
    
    models_list = [
        # OpenAI Models
        {
            "id": "gpt-4o",
            "name": "GPT-4o (Recomendado)",
            "provider": "openai",
            "description": "Modelo multimodal más avanzado de OpenAI, rápido y preciso para análisis de imágenes"
        },
        {
            "id": "gpt-4o-mini",
            "name": "GPT-4o Mini",
            "provider": "openai",
            "description": "Versión compacta y económica de GPT-4o, ideal para análisis rápidos"
        },
        {
            "id": "gpt-4-turbo",
            "name": "GPT-4 Turbo",
            "provider": "openai",
            "description": "Modelo potente con capacidades de visión y contexto amplio"
        },
        {
            "id": "gpt-4-vision-preview",
            "name": "GPT-4 Vision (Preview)",
            "provider": "openai",
            "description": "Modelo específico para análisis de imágenes (versión previa)"
        },
        # Gemini Models - Actualizados a versiones 2.0, 2.5 y 3.0
        {
            "id": "gemini-3-pro-preview",
            "name": "Gemini 3 Pro Preview (Último)",
            "provider": "gemini",
            "description": "Modelo de próxima generación en preview, máxima capacidad de análisis"
        },
        {
            "id": "gemini-2.5-pro",
            "name": "Gemini 2.5 Pro (Recomendado)",
            "provider": "gemini",
            "description": "Último modelo estable con excelente análisis visual y razonamiento avanzado"
        },
        {
            "id": "gemini-2.5-flash",
            "name": "Gemini 2.5 Flash",
            "provider": "gemini",
            "description": "Versión rápida del 2.5, ideal para inspecciones en tiempo real"
        },
        {
            "id": "gemini-2.0-flash",
            "name": "Gemini 2.0 Flash",
            "provider": "gemini",
            "description": "Modelo rápido y eficiente de la generación 2.0"
        },
        {
            "id": "gemini-1.5-pro-latest",
            "name": "Gemini 1.5 Pro Latest",
            "provider": "gemini",
            "description": "Versión estable 1.5 con contexto de 2M tokens"
        },
        {
            "id": "gemini-1.5-flash-latest",
            "name": "Gemini 1.5 Flash Latest",
            "provider": "gemini",
            "description": "Modelo 1.5 rápido para análisis básicos"
        }
    ]
    
    return models_list


@app.get("/api/ai/configuration", response_model=schemas.AIConfiguration)
def get_ai_configuration(
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Obtener configuración de IA actual"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden ver configuración de IA")
    
    config = db.query(models.AIConfiguration).filter(
        models.AIConfiguration.is_active == True
    ).first()
    
    if not config:
        raise HTTPException(status_code=404, detail="No hay configuración de IA activa")
    
    return config


@app.post("/api/ai/configuration", response_model=schemas.AIConfiguration)
def create_ai_configuration(
    config: schemas.AIConfigurationCreate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Crear o actualizar configuración de IA"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden configurar IA")
    
    # Desactivar configuraciones anteriores
    db.query(models.AIConfiguration).update({"is_active": False})
    
    # Determinar modelo por defecto según el proveedor si no se especifica
    model_name = config.model_name
    if not model_name:
        if config.provider == "openai":
            model_name = "gpt-4o"
        elif config.provider == "gemini":
            model_name = "gemini-2.5-pro"
        else:
            model_name = "default"
    
    # Crear nueva configuración
    new_config = models.AIConfiguration(
        provider=config.provider,
        api_key=config.api_key,
        model_name=model_name,
        is_active=True
    )
    
    db.add(new_config)
    db.commit()
    db.refresh(new_config)
    
    return new_config


@app.put("/api/ai/configuration/{config_id}", response_model=schemas.AIConfiguration)
def update_ai_configuration(
    config_id: int,
    config_update: schemas.AIConfigurationUpdate,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Actualizar configuración de IA existente"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden actualizar configuración de IA")
    
    config = db.query(models.AIConfiguration).filter(
        models.AIConfiguration.id == config_id
    ).first()
    
    if not config:
        raise HTTPException(status_code=404, detail="Configuración no encontrada")
    
    # Actualizar campos
    for key, value in config_update.dict(exclude_unset=True).items():
        setattr(config, key, value)
    
    db.commit()
    db.refresh(config)
    
    return config


@app.delete("/api/ai/configuration/{config_id}")
def delete_ai_configuration(
    config_id: int,
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """Eliminar configuración de IA"""
    if current_user.role != "admin":
        raise HTTPException(status_code=403, detail="Solo administradores pueden eliminar configuración de IA")
    
    config = db.query(models.AIConfiguration).filter(
        models.AIConfiguration.id == config_id
    ).first()
    
    if not config:
        raise HTTPException(status_code=404, detail="Configuración no encontrada")
    
    db.delete(config)
    db.commit()
    
    return {"message": "Configuración eliminada correctamente"}


@app.post("/api/analyze-image")
async def analyze_image(
    file: UploadFile = File(...),
    question_id: int = Form(None),
    inspection_id: int = Form(None),
    custom_prompt: str = Form(None),
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """
    Analiza una imagen usando IA para sugerir respuestas
    Usa la configuración de IA activa (OpenAI o Gemini)
    Incluye contexto del vehículo si se proporciona inspection_id
    """
    print("\n" + "="*80)
    print("🔍 ANALYZE IMAGE - DEBUG")
    print("="*80)
    print(f"📥 Parámetros recibidos:")
    print(f"   - file: {file.filename}")
    print(f"   - question_id: {question_id}")
    print(f"   - inspection_id: {inspection_id}")
    print(f"   - custom_prompt (del Form): {custom_prompt[:100] if custom_prompt else 'NO RECIBIDO'}")
    
    # Obtener configuración de IA activa
    ai_config = db.query(models.AIConfiguration).filter(
        models.AIConfiguration.is_active == True
    ).first()
    
    if not ai_config:
        return {
            "status": "disabled",
            "message": "No hay configuración de IA activa. Configure en Settings."
        }
    
    # Guardar archivo temporalmente y procesar según tipo
    import base64
    from pypdf import PdfReader
    
    contents = await file.read()
    file_type = file.content_type
    
    print(f"📄 Tipo de archivo: {file_type}")
    
    # Detectar si es PDF
    is_pdf = file_type == 'application/pdf' or file.filename.lower().endswith('.pdf')
    
    if is_pdf:
        print("📕 Detectado PDF - extrayendo texto...")
        try:
            from io import BytesIO
            pdf_file = BytesIO(contents)
            pdf_reader = PdfReader(pdf_file)
            
            # Extraer texto de todas las páginas
            pdf_text = ""
            for page_num, page in enumerate(pdf_reader.pages):
                pdf_text += f"\n--- Página {page_num + 1} ---\n"
                pdf_text += page.extract_text()
            
            print(f"✅ Texto extraído: {len(pdf_text)} caracteres")
            
            if not pdf_text.strip():
                return {
                    "status": "error",
                    "message": "No se pudo extraer texto del PDF. Puede ser un PDF escaneado sin OCR."
                }
                
            # Para PDFs usamos análisis de texto, no de imagen
            image_b64 = None
            
        except Exception as e:
            print(f"❌ Error al procesar PDF: {str(e)}")
            return {
                "status": "error",
                "message": f"Error al procesar PDF: {str(e)}"
            }
    else:
        # Es una imagen
        image_b64 = base64.b64encode(contents).decode('utf-8')
        pdf_text = None
    
    # Obtener contexto de la pregunta si se proporciona
    question_obj = None
    if question_id:
        question_obj = db.query(models.Question).filter(models.Question.id == question_id).first()
        print(f"📋 Pregunta encontrada:")
        print(f"   - ID: {question_obj.id}")
        print(f"   - Texto: {question_obj.text}")
        print(f"   - ai_prompt en DB: {question_obj.ai_prompt[:100] if question_obj.ai_prompt else 'NO TIENE'}")
    
    # Si no se proporciona custom_prompt en el Form, usar el de la pregunta
    if not custom_prompt and question_obj and question_obj.ai_prompt:
        custom_prompt = question_obj.ai_prompt
        print(f"✅ Usando ai_prompt de la pregunta de la DB")
    elif custom_prompt:
        print(f"✅ Usando custom_prompt del Form")
    else:
        print(f"⚠️ NO HAY custom_prompt (ni del Form ni de la DB)")
    
    print(f"📝 Custom prompt FINAL a usar: {custom_prompt[:150] if custom_prompt else 'NINGUNO'}...")
    
    # Obtener contexto del vehículo si se proporciona inspection_id
    vehicle_context = ""
    if inspection_id:
        inspection = db.query(models.Inspection).filter(models.Inspection.id == inspection_id).first()
        if inspection:
            print(f"🚗 Contexto del vehículo agregado: {inspection.vehicle_brand} {inspection.vehicle_model}")
            vehicle_context = f"""
INFORMACIÓN DEL VEHÍCULO INSPECCIONADO:
- Marca: {inspection.vehicle_brand}
- Modelo: {inspection.vehicle_model}
- Placa: {inspection.vehicle_plate}
- Kilometraje: {inspection.vehicle_km} km
- Nº Pedido: {inspection.order_number}
- OR/Orden: {inspection.or_number}
"""
        else:
            print(f"⚠️ inspection_id {inspection_id} no encontrado en DB")
    else:
        print(f"⚠️ NO se proporcionó inspection_id, sin contexto de vehículo")
    
    try:
        # Construir prompt dinámico basado en la pregunta específica
        if question_obj:
            # Usar prompt personalizado si está disponible
            if custom_prompt:
                # Prompt personalizado - DIRECTO Y SIMPLE
                system_prompt = f"""Eres un mecánico experto realizando una inspección vehicular.

{vehicle_context}

TAREA ESPECÍFICA:
{custom_prompt}

Responde SOLO en formato JSON válido (sin markdown, sin ```json):
{{
  "status": "ok",
  "observations": "Describe lo que observas en la imagen en relación a la tarea solicitada",
  "recommendation": "Acción sugerida basada en lo observado",
  "confidence": 0.85,
  "context_match": true
}}

VALORES DE STATUS:
- "ok": Cumple con lo esperado según la tarea
- "minor": Presenta observaciones menores o advertencias
- "critical": Presenta problemas graves o no cumple con lo esperado

VALOR DE CONTEXT_MATCH:
- true: La imagen SÍ corresponde al contexto de la pregunta/tarea
- false: La imagen NO corresponde (ej: pregunta sobre luces pero muestra motor)

IMPORTANTE: 
- Si la imagen NO corresponde al contexto de la pregunta, establece context_match=false y en observations indica qué se esperaba ver vs qué se muestra
- Si la tarea requiere verificar funcionamiento (algo encendido, prendido, activo) pero la imagen muestra el componente apagado o en reposo, usa status "critical" y context_match=false, indica en "recommendation" que se necesita una foto con el componente funcionando o un video."""
                
                user_message = f"Pregunta de inspección: {question_obj.text}\n\nAnaliza esta imagen según la tarea especificada."
            else:
                # Prompt altamente específico para la pregunta
                question_text = question_obj.text
                question_type = question_obj.type
                section = question_obj.section
                
                system_prompt = f"""Eres un mecánico experto realizando una inspección vehicular.

{vehicle_context}

PREGUNTA ESPECÍFICA A RESPONDER: "{question_text}"
Sección: {section}

Analiza la imagen ÚNICAMENTE para responder esta pregunta específica. 
Sé directo y enfócate solo en lo que la pregunta solicita.
Considera el kilometraje y características del vehículo para contextualizar tu análisis.

VALIDACIÓN DE IMAGEN:
- Si la imagen NO corresponde al contexto de la pregunta, establece context_match=false y explica en observations qué se esperaba vs qué se muestra
- Si la imagen es borrosa o no permite análisis, establece context_match=false e indica en recommendation que tomen otra foto más clara

Responde SOLO en formato JSON válido (sin markdown, sin ```json):
{{
  "status": "ok",
  "observations": "Respuesta técnica específica a: {question_text}",
  "recommendation": "Acción técnica recomendada o mensaje si la foto no es apropiada",
  "confidence": 0.85,
  "context_match": true
}}

NOTA IMPORTANTE sobre el campo "status":
- Usa "ok" si el componente está en buen estado y pasa la inspección
- Usa "minor" si hay problemas leves que requieren atención pero no son críticos
- Usa "critical" si hay problemas graves que requieren reparación inmediata

VALOR DE CONTEXT_MATCH:
- true: La imagen SÍ corresponde y es apropiada para responder la pregunta
- false: La imagen NO corresponde al contexto de la pregunta (ej: pregunta sobre luces pero imagen muestra motor)

RECUERDA: 
- Responde SOLO lo que la pregunta pide
- No des información genérica del vehículo
- Sé específico y técnico"""

            if vehicle_context:
                user_message = f"Inspecciona esta imagen del vehículo y responde específicamente: {question_obj.text}. En tus observaciones, menciona si el estado es apropiado para el kilometraje y marca/modelo del vehículo."
            else:
                user_message = f"Inspecciona la imagen y responde específicamente: {question_obj.text}"
        else:
            # Fallback para análisis general
            system_prompt = f"""Eres un experto mecánico automotriz. 

{vehicle_context}

Analiza la imagen y proporciona:
1. Estado del componente (bueno/regular/malo)
2. Nivel de criticidad (ok/minor/critical)
3. Observaciones técnicas breves
4. Recomendación de acción
5. Si la imagen corresponde al contexto automotriz

Responde SOLO en formato JSON válido (sin markdown, sin ```json):
{{
  "status": "ok",
  "observations": "descripción técnica del componente",
  "recommendation": "acción sugerida",
  "confidence": 0.85,
  "context_match": true
}}

NOTA: 
- "status" debe ser "ok" (bueno), "minor" (problemas leves) o "critical" (problemas graves)
- "context_match" debe ser true si la imagen muestra un componente vehicular relevante, false si no corresponde."""
            user_message = "Analiza este componente del vehículo para la inspección general."
        
        # Ajustar prompt si es PDF en lugar de imagen
        if is_pdf:
            system_prompt = system_prompt.replace("Analiza la imagen", "Analiza el documento PDF")
            system_prompt = system_prompt.replace("la imagen", "el documento")
            system_prompt = system_prompt.replace("context_match", "document_relevance")
            user_message = user_message.replace("imagen", "documento PDF")

        print(f"\n🤖 PROMPT ENVIADO AL AI:")
        print(f"Provider: {ai_config.provider}")
        print(f"Model: {ai_config.model_name}")
        print(f"System prompt (primeros 200 chars): {system_prompt[:200]}...")
        print(f"User message: {user_message}")
        print("="*80 + "\n")

        if ai_config.provider == "openai":
            import openai
            openai.api_key = ai_config.api_key
            
            # Construir mensaje según si es PDF o imagen
            if is_pdf:
                # Para PDF, solo texto
                messages_content = [
                    {"role": "system", "content": system_prompt},
                    {
                        "role": "user",
                        "content": f"{user_message}\n\n--- CONTENIDO DEL DOCUMENTO PDF ---\n{pdf_text[:4000]}"  # Limitar a 4000 chars
                    }
                ]
            else:
                # Para imagen, usar vision
                messages_content = [
                    {"role": "system", "content": system_prompt},
                    {
                        "role": "user",
                        "content": [
                            {
                                "type": "text",
                                "text": user_message
                            },
                            {
                                "type": "image_url",
                                "image_url": {"url": f"data:image/jpeg;base64,{image_b64}"}
                            }
                        ]
                    }
                ]
            
            response = openai.ChatCompletion.create(
                model=ai_config.model_name,
                messages=messages_content,
                max_tokens=500
            )
            
            ai_response = response.choices[0].message.content
            
        elif ai_config.provider == "gemini":
            import google.generativeai as genai
            from PIL import Image
            from io import BytesIO
            
            genai.configure(api_key=ai_config.api_key)
            model = genai.GenerativeModel(ai_config.model_name)
            
            prompt = f"{system_prompt}\n\n{user_message}"
            
            if is_pdf:
                # Para PDF, solo texto
                prompt_with_content = f"{prompt}\n\n--- CONTENIDO DEL DOCUMENTO PDF ---\n{pdf_text[:4000]}"
                response = model.generate_content(prompt_with_content)
            else:
                # Para imagen, incluir imagen
                image = Image.open(BytesIO(contents))
                response = model.generate_content([prompt, image])
                
            ai_response = response.text
            
        else:
            return {
                "success": False,
                "error": f"Provider {ai_config.provider} no soportado"
            }
        
        # Intentar parsear como JSON, si falla, usar texto plano
        try:
            import json
            import re
            
            # Limpiar markdown code blocks si existen
            cleaned_response = ai_response.strip()
            
            # Remover ```json ... ``` si existe
            if cleaned_response.startswith('```'):
                # Extraer contenido entre ``` markers
                match = re.search(r'```(?:json)?\s*\n?(.*?)\n?```', cleaned_response, re.DOTALL)
                if match:
                    cleaned_response = match.group(1).strip()
            
            analysis = json.loads(cleaned_response)
        except:
            # Si no es JSON válido, crear estructura básica
            analysis = {
                "status": "ok",
                "observations": ai_response,
                "recommendation": "Revisar manualmente",
                "confidence": 0.7
            }
        
        return {
            "success": True,
            "analysis": analysis,
            "raw_response": ai_response,
            "model": ai_config.model_name,
            "provider": ai_config.provider
        }
        
    except Exception as e:
        print(f"Error en análisis AI: {e}")
        import traceback
        traceback.print_exc()
        return {
            "success": False,
            "error": str(e),
            "message": "Error analyzing image with AI. Please check AI configuration in Settings."
        }
    
    try:
        import openai
        openai.api_key = settings.OPENAI_API_KEY
        
        # Prompt especializado para inspección vehicular
        system_prompt = """Eres un experto mecánico automotriz. Analiza la imagen y proporciona:
1. Estado del componente (bueno/regular/malo)
2. Nivel de criticidad (ok/minor/critical)
3. Observaciones técnicas breves
4. Recomendación de acción

Responde en formato JSON:
{
  "status": "ok|minor|critical",
  "observations": "descripción técnica",
  "recommendation": "acción sugerida",
  "confidence": 0.0-1.0
}"""

        response = openai.ChatCompletion.create(
            model="gpt-4-vision-preview" if "gpt-4" in str(settings.OPENAI_API_KEY) else "gpt-4o",
            messages=[
                {
                    "role": "system",
                    "content": system_prompt
                },
                {
                    "role": "user",
                    "content": [
                        {
                            "type": "text",
                            "text": f"Analiza este componente del vehículo.\n{question_context}"
                        },
                        {
                            "type": "image_url",
                            "image_url": {
                                "url": f"data:image/jpeg;base64,{image_b64}"
                            }
                        }
                    ]
                }
            ],
            max_tokens=500
        )
        
        ai_response = response.choices[0].message.content
        
        # Intentar parsear como JSON, si falla, usar texto plano
        try:
            import json
            analysis = json.loads(ai_response)
        except:
            # Si no es JSON válido, crear estructura básica
            analysis = {
                "status": "ok",
                "observations": ai_response,
                "recommendation": "Revisar manualmente",
                "confidence": 0.7
            }
        
        return {
            "success": True,
            "analysis": analysis,
            "raw_response": ai_response,
            "model": "gpt-4-vision"
        }
        
    except Exception as e:
        print(f"Error en análisis AI: {e}")
        return {
            "success": False,
            "error": str(e),
            "message": "Error analyzing image with AI"
        }


@app.post("/api/ai/chat-assistant")
async def chat_with_ai_assistant(
    question_id: int = Form(...),
    inspection_id: int = Form(...),
    user_message: str = Form(""),
    chat_history: str = Form("[]"),
    context_photos: str = Form("[]"),
    vehicle_info: str = Form("{}"),
    assistant_prompt: str = Form(""),
    assistant_instructions: str = Form(""),
    response_length: str = Form("medium"),
    files: List[UploadFile] = File(default=[]),
    db: Session = Depends(get_db),
    current_user: models.User = Depends(get_current_user)
):
    """
    Chat conversacional con IA usando contexto de fotos anteriores
    El asistente tiene acceso a fotos de preguntas previas para dar mejor contexto
    Ahora soporta archivos adjuntos (imágenes y PDFs)
    """
    print("\n" + "="*80)
    print("🤖 AI CHAT ASSISTANT")
    print("="*80)
    
    # Parsear JSON strings
    import json
    chat_history_list = json.loads(chat_history)
    context_photos_list = json.loads(context_photos)
    vehicle_info_dict = json.loads(vehicle_info)
    
    print(f"📋 Question ID: {question_id}")
    print(f"🚗 Inspection ID: {inspection_id}")
    print(f"💬 User message: {user_message}")
    print(f"📎 Attached files: {len(files)}")
    print(f"📸 Context photos: {len(context_photos_list)} fotos")
    print(f"💭 Chat history: {len(chat_history_list)} mensajes previos")
    
    # Procesar archivos adjuntos
    attached_files_data = []
    if files:
        import base64
        from pypdf import PdfReader
        from io import BytesIO
        
        for file in files:
            file_content = await file.read()
            file_type = file.content_type
            
            file_info = {
                'filename': file.filename,
                'type': file_type,
                'size': len(file_content)
            }
            
            # Si es PDF, extraer texto
            if file_type == 'application/pdf' or file.filename.lower().endswith('.pdf'):
                try:
                    pdf_file = BytesIO(file_content)
                    pdf_reader = PdfReader(pdf_file)
                    pdf_text = ""
                    for page in pdf_reader.pages:
                        pdf_text += page.extract_text()
                    file_info['content_type'] = 'pdf'
                    file_info['text'] = pdf_text[:2000]  # Limitar texto
                    print(f"📄 PDF procesado: {file.filename} - {len(pdf_text)} caracteres")
                except Exception as e:
                    print(f"❌ Error procesando PDF {file.filename}: {str(e)}")
                    file_info['error'] = str(e)
            
            # Si es imagen, convertir a base64
            elif file_type.startswith('image/'):
                file_info['content_type'] = 'image'
                file_info['base64'] = base64.b64encode(file_content).decode('utf-8')
                print(f"🖼️ Imagen procesada: {file.filename}")
            
            attached_files_data.append(file_info)
    
    # Obtener configuración de IA
    ai_config = db.query(models.AIConfiguration).filter(
        models.AIConfiguration.is_active == True
    ).first()
    
    if not ai_config:
        return {
            "success": False,
            "response": "No hay configuración de IA activa. Por favor configura en Settings.",
            "confidence": 0
        }
    
    try:
        # Construir el contexto del vehículo
        vehicle_context = f"""
INFORMACIÓN DEL VEHÍCULO:
- Marca: {vehicle_info_dict.get('brand', 'N/A')}
- Modelo: {vehicle_info_dict.get('model', 'N/A')}
- Placa: {vehicle_info_dict.get('plate', 'N/A')}
- Kilometraje: {vehicle_info_dict.get('km', 'N/A')} km
"""
        
        # Construir el contexto de las fotos anteriores
        photos_context = ""
        if context_photos_list:
            photos_context = f"\n\nFOTOS ANALIZADAS PREVIAMENTE ({len(context_photos_list)} imágenes):\n"
            for idx, photo in enumerate(context_photos_list[:10], 1):  # Limitar a 10 fotos
                ai_analysis = photo.get('aiAnalysis', [])
                if ai_analysis and len(ai_analysis) > 0:
                    analysis_text = ai_analysis[0].get('analysis', {})
                    obs = analysis_text.get('observations', 'Sin análisis')
                    status = analysis_text.get('status', 'unknown')
                    photos_context += f"\n{idx}. Pregunta ID {photo.get('questionId')}: Status={status}\n   Observaciones: {obs[:200]}...\n"
        
        # Definir la longitud de respuesta
        max_tokens_map = {
            'short': 200,
            'medium': 400,
            'long': 800
        }
        max_tokens = max_tokens_map.get(response_length, 400)
        
        # Construir contexto de archivos adjuntos
        attached_context = ""
        if attached_files_data:
            attached_context = f"\n\nARCHIVOS ADJUNTOS EN ESTE MENSAJE ({len(attached_files_data)} archivos):\n"
            for idx, file_info in enumerate(attached_files_data, 1):
                if file_info.get('content_type') == 'pdf':
                    attached_context += f"\n{idx}. PDF: {file_info['filename']}\n"
                    if 'text' in file_info:
                        attached_context += f"   Contenido: {file_info['text'][:500]}...\n"
                elif file_info.get('content_type') == 'image':
                    attached_context += f"\n{idx}. Imagen: {file_info['filename']}\n"
        
        # Construir el system prompt
        base_prompt = assistant_prompt or "Eres un experto mecánico automotriz que ayuda a diagnosticar problemas."
        
        system_prompt = f"""{base_prompt}

{vehicle_context}

{photos_context}

{attached_context}

INSTRUCCIONES ADICIONALES:
{assistant_instructions if assistant_instructions else "Sé técnico, claro y directo en tus respuestas."}

FORMATO DE RESPUESTA:
- Sé {response_length} en tus respuestas
- Usa lenguaje técnico pero comprensible
- Si ves algo preocupante en las fotos analizadas, menciónalo
- Proporciona recomendaciones específicas cuando sea relevante
- Si no tienes suficiente información, pide más detalles
"""
        
        # Construir el historial de mensajes para la IA
        messages = [{"role": "system", "content": system_prompt}]
        
        # Agregar historial previo (últimos 10 mensajes para no saturar)
        for msg in chat_history_list[-10:]:
            messages.append({
                "role": msg.get('role'),
                "content": msg.get('content')
            })
        
        # Agregar el mensaje actual del usuario con imágenes si hay
        has_images = any(f.get('content_type') == 'image' for f in attached_files_data)
        
        if has_images:
            # Formato multimodal para OpenAI/Gemini
            user_content = []
            if user_message:
                user_content.append({"type": "text", "text": user_message})
            
            # Agregar imágenes
            for file_info in attached_files_data:
                if file_info.get('content_type') == 'image':
                    user_content.append({
                        "type": "image_url",
                        "image_url": {"url": f"data:image/jpeg;base64,{file_info['base64']}"}
                    })
            
            messages.append({
                "role": "user",
                "content": user_content
            })
        else:
            # Solo texto
            messages.append({
                "role": "user",
                "content": user_message
            })
        
        print(f"🔧 Enviando a {ai_config.provider} con {len(messages)} mensajes")
        
        # Llamar a la IA según el proveedor
        if ai_config.provider == 'openai':
            from openai import OpenAI
            client = OpenAI(api_key=ai_config.api_key)
            
            response = client.chat.completions.create(
                model=ai_config.model_name or "gpt-4",
                messages=messages,
                max_tokens=max_tokens,
                temperature=0.7
            )
            
            ai_response = response.choices[0].message.content
            confidence = 0.85  # OpenAI no devuelve confidence directo
            
        elif ai_config.provider == 'gemini':
            import google.generativeai as genai
            genai.configure(api_key=ai_config.api_key)
            
            model = genai.GenerativeModel(ai_config.model_name or 'gemini-pro')
            
            # Gemini maneja el chat diferente
            # Convertir mensajes al formato de Gemini
            chat_content = ""
            for msg in messages[1:]:  # Skip system message
                role_label = "Usuario" if msg['role'] == 'user' else "Asistente"
                chat_content += f"\n{role_label}: {msg['content']}\n"
            
            full_prompt = f"{system_prompt}\n\nCONVERSACIÓN:\n{chat_content}\n\nAsistente:"
            
            response = model.generate_content(full_prompt)
            ai_response = response.text
            confidence = 0.80
        
        else:
            raise ValueError(f"Proveedor no soportado: {ai_config.provider}")
        
        print(f"✅ Respuesta generada: {len(ai_response)} caracteres")
        
        return {
            "success": True,
            "response": ai_response,
            "confidence": confidence,
            "provider": ai_config.provider,
            "model": ai_config.model_name,
            "attached_files": [{'filename': f['filename'], 'type': f['type']} for f in attached_files_data]
        }
        
    except Exception as e:
        print(f"❌ Error en chat IA: {e}")
        import traceback
        traceback.print_exc()
        
        return {
            "success": False,
            "response": f"Error al comunicarse con el asistente: {str(e)}",
            "confidence": 0
        }


# ============= REPORTS =============
@app.get("/api/reports/dashboard", response_model=schemas.DashboardData)
def get_dashboard_data(
    start_date: Optional[str] = None,
    end_date: Optional[str] = None,
    mechanic_id: Optional[int] = None,
    current_user: models.User = Depends(get_current_user),
    db: Session = Depends(get_db)
):
    """Obtener datos del dashboard de informes"""
    if current_user.role not in ["admin", "asesor"]:
        raise HTTPException(status_code=403, detail="No tienes permisos para acceder a reportes")
    
    # Construir query base
    query = db.query(models.Inspection)
    
    # Aplicar filtros de fecha
    if start_date:
        # Parsear fecha y establecer al inicio del día en UTC-3
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        start = datetime.fromisoformat(start_date).replace(hour=0, minute=0, second=0, microsecond=0)
        if start.tzinfo is None:
            start = start.replace(tzinfo=local_tz)
        query = query.filter(models.Inspection.started_at >= start)
    if end_date:
        # Parsear fecha y establecer al final del día en UTC-3
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        end = datetime.fromisoformat(end_date).replace(hour=23, minute=59, second=59, microsecond=999999)
        if end.tzinfo is None:
            end = end.replace(tzinfo=local_tz)
        query = query.filter(models.Inspection.started_at <= end)
    
    # Filtro por mecánico
    if mechanic_id:
        query = query.filter(models.Inspection.mechanic_id == mechanic_id)
    
    # Solo inspecciones activas
    query = query.filter(models.Inspection.is_active == True)
    
    # ESTADÍSTICAS GENERALES
    total = query.count()
    completed = query.filter(models.Inspection.status == "completed").count()
    pending = total - completed
    
    # Score promedio
    avg_score_result = query.filter(
        models.Inspection.score.isnot(None),
        models.Inspection.max_score.isnot(None),
        models.Inspection.max_score > 0
    ).with_entities(
        func.avg(models.Inspection.score * 100.0 / models.Inspection.max_score)
    ).scalar()
    avg_score = round(avg_score_result, 2) if avg_score_result else 0.0
    
    # Items señalados
    flagged_items = db.query(func.count(models.Answer.id))\
        .filter(models.Answer.is_flagged == True)\
        .join(models.Inspection)\
        .filter(models.Inspection.is_active == True)
    
    if start_date:
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        start = datetime.fromisoformat(start_date).replace(hour=0, minute=0, second=0, microsecond=0)
        if start.tzinfo is None:
            start = start.replace(tzinfo=local_tz)
        flagged_items = flagged_items.filter(models.Inspection.started_at >= start)
    if end_date:
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        end = datetime.fromisoformat(end_date).replace(hour=23, minute=59, second=59, microsecond=999999)
        if end.tzinfo is None:
            end = end.replace(tzinfo=local_tz)
        flagged_items = flagged_items.filter(models.Inspection.started_at <= end)
    if mechanic_id:
        flagged_items = flagged_items.filter(models.Inspection.mechanic_id == mechanic_id)
    
    total_flagged = flagged_items.scalar() or 0
    
    stats = schemas.InspectionStats(
        total_inspections=total,
        completed_inspections=completed,
        pending_inspections=pending,
        completion_rate=round((completed / total * 100) if total > 0 else 0, 2),
        avg_score=avg_score,
        total_flagged_items=total_flagged
    )
    
    # RANKING DE MECÁNICOS
    mechanic_stats = db.query(
        models.User.id,
        models.User.full_name,
        func.count(models.Inspection.id).label('total'),
        func.avg(
            case(
                (models.Inspection.max_score > 0, models.Inspection.score * 100.0 / models.Inspection.max_score),
                else_=None
            )
        ).label('avg_score'),
        func.count(case((models.Inspection.status == 'completed', 1))).label('completed')
    ).join(models.Inspection, models.Inspection.mechanic_id == models.User.id)\
     .filter(models.User.role.in_(['mechanic', 'mecanico']))\
     .filter(models.User.is_active == True)\
     .filter(models.Inspection.is_active == True)
    
    if start_date:
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        start = datetime.fromisoformat(start_date).replace(hour=0, minute=0, second=0, microsecond=0)
        if start.tzinfo is None:
            start = start.replace(tzinfo=local_tz)
        mechanic_stats = mechanic_stats.filter(models.Inspection.started_at >= start)
    if end_date:
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        end = datetime.fromisoformat(end_date).replace(hour=23, minute=59, second=59, microsecond=999999)
        if end.tzinfo is None:
            end = end.replace(tzinfo=local_tz)
        mechanic_stats = mechanic_stats.filter(models.Inspection.started_at <= end)
    
    mechanic_stats = mechanic_stats.group_by(models.User.id, models.User.full_name)\
                                   .order_by(func.count(models.Inspection.id).desc())\
                                   .all()
    
    mechanic_ranking = [
        schemas.MechanicRanking(
            mechanic_id=m.id,
            mechanic_name=m.full_name or "Sin nombre",
            total_inspections=m.total,
            avg_score=round(m.avg_score, 2) if m.avg_score else 0.0,
            completion_rate=round((m.completed / m.total * 100) if m.total > 0 else 0, 2)
        )
        for m in mechanic_stats if m.full_name
    ]
    
    # ESTADÍSTICAS POR CHECKLIST
    checklist_stats_query = db.query(
        models.Checklist.id,
        models.Checklist.name,
        func.count(models.Inspection.id).label('total'),
        func.avg(
            case(
                (models.Inspection.max_score > 0, models.Inspection.score * 100.0 / models.Inspection.max_score),
                else_=None
            )
        ).label('avg_score')
    ).join(models.Inspection)\
     .filter(models.Inspection.is_active == True)\
     .filter(models.Checklist.is_active == True)
    
    if start_date:
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        start = datetime.fromisoformat(start_date).replace(hour=0, minute=0, second=0, microsecond=0)
        if start.tzinfo is None:
            start = start.replace(tzinfo=local_tz)
        checklist_stats_query = checklist_stats_query.filter(models.Inspection.started_at >= start)
    if end_date:
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        end = datetime.fromisoformat(end_date).replace(hour=23, minute=59, second=59, microsecond=999999)
        if end.tzinfo is None:
            end = end.replace(tzinfo=local_tz)
        checklist_stats_query = checklist_stats_query.filter(models.Inspection.started_at <= end)
    if mechanic_id:
        checklist_stats_query = checklist_stats_query.filter(models.Inspection.mechanic_id == mechanic_id)
    
    checklist_stats_query = checklist_stats_query.group_by(models.Checklist.id, models.Checklist.name)
    checklist_stats_data = checklist_stats_query.all()
    
    checklist_stats = [
        schemas.ChecklistStats(
            checklist_id=c.id,
            checklist_name=c.name or "Sin nombre",
            total_inspections=c.total,
            avg_score=round(c.avg_score, 2) if c.avg_score else 0.0
        )
        for c in checklist_stats_data if c.name
    ]
    
    # INSPECCIONES POR FECHA (últimos 30 días)
    end_date_obj = datetime.fromisoformat(end_date) if end_date else datetime.now()
    start_date_obj = datetime.fromisoformat(start_date) if start_date else end_date_obj - timedelta(days=30)
    
    inspections_by_date_query = db.query(
        func.date(models.Inspection.started_at).label('date'),
        func.count(models.Inspection.id).label('count')
    ).filter(
        models.Inspection.started_at.between(start_date_obj, end_date_obj),
        models.Inspection.is_active == True
    )
    
    if mechanic_id:
        inspections_by_date_query = inspections_by_date_query.filter(
            models.Inspection.mechanic_id == mechanic_id
        )
    
    inspections_by_date_data = inspections_by_date_query.group_by(
        func.date(models.Inspection.started_at)
    ).all()
    
    inspections_by_date = {
        str(d.date): d.count for d in inspections_by_date_data
    }
    
    # RATIO PASS/FAIL
    pass_fail_data = db.query(
        models.Answer.answer_value,
        func.count(models.Answer.id).label('count')
    ).join(models.Inspection)\
     .filter(models.Inspection.is_active == True)\
     .filter(models.Answer.answer_value.in_(['pass', 'fail', 'good', 'bad', 'regular']))
    
    if start_date:
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        start = datetime.fromisoformat(start_date).replace(hour=0, minute=0, second=0, microsecond=0)
        if start.tzinfo is None:
            start = start.replace(tzinfo=local_tz)
        pass_fail_data = pass_fail_data.filter(models.Inspection.started_at >= start)
    if end_date:
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        end = datetime.fromisoformat(end_date).replace(hour=23, minute=59, second=59, microsecond=999999)
        if end.tzinfo is None:
            end = end.replace(tzinfo=local_tz)
        pass_fail_data = pass_fail_data.filter(models.Inspection.started_at <= end)
    if mechanic_id:
        pass_fail_data = pass_fail_data.filter(models.Inspection.mechanic_id == mechanic_id)
    
    pass_fail_data = pass_fail_data.group_by(models.Answer.answer_value).all()
    
    pass_fail_ratio = {d.answer_value: d.count for d in pass_fail_data}
    
    return schemas.DashboardData(
        stats=stats,
        mechanic_ranking=mechanic_ranking,
        checklist_stats=checklist_stats,
        inspections_by_date=inspections_by_date,
        pass_fail_ratio=pass_fail_ratio
    )


@app.get("/api/reports/inspections")
def get_inspections_report(
    start_date: Optional[str] = None,
    end_date: Optional[str] = None,
    mechanic_id: Optional[int] = None,
    checklist_id: Optional[int] = None,
    status: Optional[str] = None,
    limit: int = 100,
    current_user: models.User = Depends(get_current_user),
    db: Session = Depends(get_db)
):
    """Obtener lista de inspecciones con filtros"""
    if current_user.role not in ["admin", "asesor"]:
        raise HTTPException(status_code=403, detail="No tienes permisos para acceder a reportes")
    
    # Query base con select_from explícito
    query = db.query(
        models.Inspection.id,
        models.Inspection.vehicle_plate,
        models.Inspection.checklist_id,
        models.Checklist.name.label('checklist_name'),
        models.User.full_name.label('mechanic_name'),
        models.Inspection.status,
        models.Inspection.score,
        models.Inspection.max_score,
        models.Inspection.started_at,
        models.Inspection.completed_at,
        func.coalesce(
            func.count(case((models.Answer.is_flagged == True, 1))),
            0
        ).label('flagged_items')
    ).select_from(models.Inspection)\
     .join(models.Checklist, models.Inspection.checklist_id == models.Checklist.id)\
     .join(models.User, models.Inspection.mechanic_id == models.User.id)\
     .outerjoin(models.Answer, models.Answer.inspection_id == models.Inspection.id)\
     .outerjoin(models.Question, models.Answer.question_id == models.Question.id)\
     .filter(
         models.Inspection.is_active == True,
         or_(models.Question.is_deleted == False, models.Question.id == None)  # Solo contar answers de preguntas no eliminadas o si no hay answer
     )
    
    # Aplicar filtros
    if start_date:
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        start = datetime.fromisoformat(start_date).replace(hour=0, minute=0, second=0, microsecond=0)
        if start.tzinfo is None:
            start = start.replace(tzinfo=local_tz)
        query = query.filter(models.Inspection.started_at >= start)
    if end_date:
        from datetime import timezone
        local_tz = timezone(timedelta(hours=-3))
        end = datetime.fromisoformat(end_date).replace(hour=23, minute=59, second=59, microsecond=999999)
        if end.tzinfo is None:
            end = end.replace(tzinfo=local_tz)
        query = query.filter(models.Inspection.started_at <= end)
    if mechanic_id:
        query = query.filter(models.Inspection.mechanic_id == mechanic_id)
    if checklist_id:
        query = query.filter(models.Inspection.checklist_id == checklist_id)
    if status:
        query = query.filter(models.Inspection.status == status)
    
    # Group by y order
    query = query.group_by(
        models.Inspection.id,
        models.Checklist.name,
        models.User.full_name
    ).order_by(models.Inspection.started_at.desc())\
     .limit(limit)
    
    results = query.all()
    
    return [
        {
            "id": r.id,
            "vehicle_plate": r.vehicle_plate,
            "checklist_id": r.checklist_id,
            "checklist_name": r.checklist_name or "Sin nombre",
            "mechanic_name": r.mechanic_name or "Sin nombre",
            "status": r.status,
            "score": r.score,
            "max_score": r.max_score,
            "flagged_items": r.flagged_items,
            "started_at": r.started_at.isoformat() if r.started_at else None,
            "completed_at": r.completed_at.isoformat() if r.completed_at else None
        }
        for r in results
    ]


@app.get("/api/inspections/{inspection_id}/pdf")
def export_inspection_to_pdf(
    inspection_id: int,
    current_user: models.User = Depends(get_current_user),
    db: Session = Depends(get_db)
):
    """Descargar el PDF guardado en MinIO para la inspección"""
    from fastapi.responses import StreamingResponse
    import requests
    # Obtener inspección
    inspection = db.query(models.Inspection).filter(
        models.Inspection.id == inspection_id
    ).first()
    if not inspection:
        raise HTTPException(status_code=404, detail="Inspección no encontrada")
    if current_user.role not in ["admin", "asesor"] and inspection.mechanic_id != current_user.id:
        raise HTTPException(status_code=403, detail="No tienes permisos para ver esta inspección")
    # Si existe pdf_url, descargar desde MinIO y devolverlo
    if inspection.pdf_url:
        try:
            pdf_resp = requests.get(inspection.pdf_url, stream=True)
            if pdf_resp.status_code == 200:
                filename = inspection.pdf_url.split("/")[-1]
                return StreamingResponse(pdf_resp.raw, media_type="application/pdf", headers={
                    "Content-Disposition": f"attachment; filename={filename}"
                })
            else:
                raise HTTPException(status_code=404, detail="No se pudo descargar el PDF desde MinIO")
        except Exception as e:
            raise HTTPException(status_code=500, detail=f"Error al descargar PDF: {e}")
    else:
        raise HTTPException(status_code=404, detail="La inspección no tiene PDF generado")


# ============= HEALTH CHECK =============
@app.get("/")
def root():
    return {"message": "Checklist Inteligente API", "version": "1.0.0", "status": "running"}


@app.get("/health")
def health_check():
    return {"status": "healthy"}