Files
portales-hispanohablantes/dags/portal_inmobiliario_cl_daily.py
T

75 lines
2.3 KiB
Python

from datetime import datetime, timedelta
import paramiko
from airflow import DAG
from airflow.operators.python import PythonOperator
# ─── Config SSH / scraper ───────────────────────────────────────────────────────
SSH_HOST = '152.53.242.31'
SSH_PORT = 22
SSH_USER = 'root'
SSH_PASS = 'XpvvdHnNr0Bq9qU'
PYTHON_BIN = '/home/jpuma/proyectos/bot-sunat/.venv/bin/python'
SCRAPER_PATH = '/opt/scrapers/portal_inmobiliario_cl/scraper.py'
default_args = {
'owner': 'jpuma',
'retries': 2,
'retry_delay': timedelta(minutes=15),
'email': ['jpuma@redneurocom.com'],
'email_on_failure': True,
'email_on_retry': False,
'execution_timeout': timedelta(hours=6),
}
def run_scraper(**ctx):
client = paramiko.SSHClient()
client.set_missing_host_key_policy(paramiko.AutoAddPolicy())
try:
client.connect(
hostname=SSH_HOST,
port=SSH_PORT,
username=SSH_USER,
password=SSH_PASS,
timeout=30,
)
client.get_transport().set_keepalive(30)
cmd = '{} {}'.format(PYTHON_BIN, SCRAPER_PATH)
_, stdout, stderr = client.exec_command(cmd, timeout=21600) # 6 h max
out = stdout.read().decode('utf-8', errors='replace')
err = stderr.read().decode('utf-8', errors='replace')
print("=== STDOUT ===\n{}".format(out))
if err.strip():
print("=== STDERR ===\n{}".format(err))
if 'Finalizado:' not in out:
raise RuntimeError(
"El scraper no imprimio 'Finalizado:'. "
"STDOUT: {!r} | STDERR: {!r}".format(
out[-2000:], err[-1000:]
)
)
finally:
client.close()
with DAG(
dag_id='portal_inmobiliario_cl_daily',
default_args=default_args,
description='Scraper diario de PortalInmobiliario.com (Chile - MercadoLibre)',
schedule_interval='30 7 * * *',
start_date=datetime(2024, 1, 1),
catchup=False,
max_active_runs=1,
tags=['scraping', 'portal-inmobiliario', 'cl', 'inmuebles'],
) as dag:
scrape_task = PythonOperator(
task_id='run_portal_inmobiliario_scraper',
python_callable=run_scraper,
)