feat: add DAG encuentra24_ca_daily (schedule 08:30 UTC)

2026-08-02 04:22:56 +02:00
parent cec3e88962
commit 29e52fcadb
1 changed files with 79 additions and 0 deletions
@@ -0,0 +1,79 @@
+"""
+DAG: encuentra24_ca_daily
+Portal: Encuentra24.com – Bienes Raices Centroamerica (CR, GT, HN, NI, SV)
+Schedule: 08:30 UTC diario
+"""
+
+import paramiko
+from datetime import datetime, timedelta
+from airflow import DAG
+from airflow.operators.python import PythonOperator
+
+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/encuentra24_ca/scraper.py'
+CMD_TIMEOUT  = 21600  # 6 horas
+
+default_args = {
+    'owner':             'jpuma',
+    'depends_on_past':   False,
+    '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(SSH_HOST, port=SSH_PORT, username=SSH_USER,
+                       password=SSH_PASS, timeout=30)
+        client.get_transport().set_keepalive(30)
+
+        cmd = f'{PYTHON_BIN} {SCRAPER_PATH}'
+        print(f'Ejecutando en VPS: {cmd}')
+
+        _, stdout, stderr = client.exec_command(cmd, timeout=CMD_TIMEOUT)
+        output = stdout.read().decode('utf-8', errors='replace')
+        errors = stderr.read().decode('utf-8', errors='replace')
+
+        if output:
+            print(f'[STDOUT]\n{output}')
+        if errors:
+            print(f'[STDERR]\n{errors}')
+
+        log_text = output + errors
+        if 'Finalizado:' not in log_text:
+            raise RuntimeError(
+                'Scraper encuentra24_ca no completó correctamente. '
+                f'Últimas líneas:\n{log_text[-3000:]}'
+            )
+
+        print('Scraper Encuentra24 CA finalizado correctamente.')
+    finally:
+        client.close()
+
+
+with DAG(
+    dag_id='encuentra24_ca_daily',
+    default_args=default_args,
+    description='Scraper diario de bienes raices en Encuentra24 CA (CR/GT/HN/NI/SV)',
+    schedule_interval='30 8 * * *',
+    start_date=datetime(2024, 1, 1),
+    catchup=False,
+    max_active_runs=1,
+    tags=['scraping', 'encuentra24', 'ca', 'inmuebles'],
+) as dag:
+
+    PythonOperator(
+        task_id='run_encuentra24_ca_scraper',
+        python_callable=run_scraper,
+        provide_context=True,
+    )