"""DAG diario – Metrocuadrado CO (Colombia) Scraper RSC-based que itera categorías × ciudades × barrios. """ from datetime import datetime, timedelta import paramiko from airflow import DAG from airflow.operators.python import PythonOperator VPS_HOST = '152.53.242.31' VPS_USER = 'root' VPS_KEY = '/opt/airflow/config/.ssh/id_rsa' SCRAPER_CMD = ( 'METRO_DAILY=1 ' 'timeout 18000 ' '/home/jpuma/proyectos/bot-sunat/.venv/bin/python ' '/opt/scrapers/metrocuadrado_co/scraper.py 2>&1' ) def run_scraper(**ctx): client = paramiko.SSHClient() client.set_missing_host_key_policy(paramiko.AutoAddPolicy()) client.connect(VPS_HOST, username=VPS_USER, key_filename=VPS_KEY, timeout=30) _, stdout, stderr = client.exec_command(SCRAPER_CMD, timeout=18000) output = stdout.read().decode('utf-8', errors='replace') err = stderr.read().decode('utf-8', errors='replace') client.close() print(output[-4000:] if len(output) > 4000 else output) if err: print('STDERR:', err[-500:]) if 'Finalizado:' not in output: raise RuntimeError('Scraper no imprimió "Finalizado:" — posible error') return output.split('Finalizado:')[-1].strip() default_args = { 'owner': 'jpuma', 'retries': 0, 'execution_timeout': timedelta(hours=6), } with DAG( dag_id='metrocuadrado_co_daily', default_args=default_args, description='Scraper diario Metrocuadrado CO (RSC Next.js)', schedule_interval='0 5 * * *', start_date=datetime(2026, 8, 1), catchup=False, max_active_runs=1, tags=['colombia', 'daily', 'inmuebles'], ) as dag: PythonOperator( task_id='scrape_metrocuadrado_co', python_callable=run_scraper, )