feat: DAG backfill Yapo CL

2026-08-02 18:05:35 +02:00
parent b6648d3234
commit 91bf501e3b
1 changed files with 67 additions and 0 deletions
@@ -0,0 +1,67 @@
+"""
+DAG: yapo_cl_backfill
+Backfill manual del catalogo completo de Yapo.cl Chile (7 subcategorias).
+Reanudable: checkpoint en /opt/scrapers/yapo_cl/backfill_state.json
+Para reiniciar desde cero: borrar ese archivo antes de disparar el DAG.
+"""
+from datetime import datetime, timedelta
+import paramiko
+from airflow import DAG
+from airflow.operators.python import PythonOperator
+
+VPS_HOST    = '152.53.242.31'
+VPS_PORT    = 22
+VPS_USER    = 'root'
+VPS_PASS    = 'XpvvdHnNr0Bq9qU'
+VENV_PYTHON = '/home/jpuma/proyectos/bot-sunat/.venv/bin/python'
+BACKFILL_PY = '/opt/scrapers/yapo_cl/backfill.py'
+
+default_args = {
+    'owner':             'jpuma',
+    'retries':           0,
+    'email':             ['jpuma@redneurocom.com'],
+    'email_on_failure':  True,
+    'email_on_retry':    False,
+    'execution_timeout': timedelta(hours=12),
+}
+
+
+def run_backfill(**ctx):
+    ssh = paramiko.SSHClient()
+    ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
+    ssh.connect(VPS_HOST, port=VPS_PORT, username=VPS_USER, password=VPS_PASS, timeout=30)
+    ssh.get_transport().set_keepalive(30)
+
+    cmd = f'BF_MAX_PAGES=300 BF_STREAK=6 {VENV_PYTHON} {BACKFILL_PY} 2>&1'
+    _, stdout, _ = ssh.exec_command(cmd, timeout=43200)
+    output    = stdout.read().decode('utf-8', errors='replace')
+    exit_code = stdout.channel.recv_exit_status()
+    ssh.close()
+
+    tail = output[-5000:] if len(output) > 5000 else output
+    print('[yapo_cl_backfill] OUTPUT:\n' + tail)
+
+    if 'Finalizado:' not in output:
+        raise RuntimeError(
+            f'Backfill no finalizó correctamente (exit={exit_code}). '
+            f'Ultimas lineas: {output[-600:]}'
+        )
+    if exit_code != 0:
+        raise RuntimeError(f'Backfill terminó con exit code {exit_code}')
+
+
+with DAG(
+    dag_id            = 'yapo_cl_backfill',
+    default_args      = default_args,
+    description       = 'Backfill catalogo completo Yapo.cl Chile — disparo manual, reanudable',
+    schedule_interval = None,
+    start_date        = datetime(2026, 7, 1),
+    catchup           = False,
+    max_active_runs   = 1,
+    tags              = ['backfill', 'yapo', 'cl', 'inmuebles'],
+) as dag:
+
+    PythonOperator(
+        task_id         = 'run_yapo_backfill',
+        python_callable = run_backfill,
+    )