add workflow 市场News-Agents调度,dev

This commit is contained in:
root 2025-12-19 15:18:32 +08:00
parent c39851a33d
commit d202e3fb81
1 changed files with 119 additions and 0 deletions

View File

@ -0,0 +1,119 @@
#!/usr/bin/python
# -*- encoding=utf-8 -*-
from airflow import DAG
from datetime import datetime, timedelta
from airflow.contrib.hooks.ssh_hook import SSHHook
from airflow.contrib.operators.ssh_operator import SSHOperator
from airflow.sensors.external_task_sensor import ExternalTaskSensor
import json
from airflow.operators.trigger_dagrun import TriggerDagRunOperator
from airflow.operators.email_operator import EmailOperator
from airflow.utils.trigger_rule import TriggerRule
sshHook = SSHHook(ssh_conn_id ='ssh_air')
default_args = {
'owner': 'tek_newsletter@163.com',
'email_on_failure': True,
'email_on_retry':True,
'start_date': datetime(2024, 1, 1),
'depends_on_past': False,
'retries': 6,
'retry_delay': timedelta(minutes=10),
}
dag = DAG('wf_dag_agents_market_newsletter', default_args=default_args,
schedule_interval="0 10 * * 5,1,3",
catchup=False,
dagrun_timeout=timedelta(minutes=600),
max_active_runs=3,
tags=['64cbd1bbced14209b5e3a879f89e8ab1','TK_Cust','市场News-Agents调度']
)
task_failed = EmailOperator (
dag=dag,
trigger_rule=TriggerRule.ONE_FAILED,
task_id="task_failed",
to=["tek_newsletter@163.com"],
cc=[""],
subject="agents_market_newsletter_failed",
html_content='<h3>您好agents_market_newsletter作业失败请及时处理" </h3>')
agents_recuriments = SSHOperator(
ssh_hook=sshHook,
task_id='agents_recuriments',
command='cd /data/airflow/etl/agents && python agents_recuriments.py',
params={'my_param':"agents_recuriments"},
depends_on_past=False,
retries=1,
dag=dag)
agents_market_summar = SSHOperator(
ssh_hook=sshHook,
task_id='agents_market_summar',
command='cd /data/airflow/etl/agents && python agents_market_summar.py',
params={'my_param':"agents_market_summar"},
depends_on_past=False,
retries=1,
dag=dag)
tk_product_prefer = SSHOperator(
ssh_hook=sshHook,
task_id='tk_product_prefer',
command='cd /data/airflow/etl/agents && python tk_product_prefer.py',
params={'my_param':"tk_product_prefer"},
depends_on_past=False,
retries=1,
dag=dag)
batch_gen_crawer = SSHOperator(
ssh_hook=sshHook,
task_id='batch_gen_crawer',
command='cd /data/airflow/etl/agents && python batch_gen_crawer.py',
params={'my_param':"batch_gen_crawer"},
depends_on_past=False,
retries=1,
dag=dag)
batch_gen_recruiment = SSHOperator(
ssh_hook=sshHook,
task_id='batch_gen_recruiment',
command='cd /data/airflow/etl/agents && python batch_gen_recruiment.py',
params={'my_param':"batch_gen_recruiment"},
depends_on_past=False,
retries=1,
dag=dag)
a_market_Intelligence_newsletter = SSHOperator(
ssh_hook=sshHook,
task_id='a_market_Intelligence_newsletter',
command='/data/airflow/etl/COM/run_psql.sh {{ ds_nodash }} {{params.my_param}} ',
params={'my_param':"a_market_Intelligence_newsletter_agi"},
depends_on_past=False,
retries=3,
dag=dag)
dysql_news_letter_info = SSHOperator(
ssh_hook=sshHook,
task_id='dysql_news_letter_info',
command='/data/airflow/etl/MART/run_psql.sh {{ ds_nodash }} {{params.my_param}} ',
params={'my_param':"dysql_news_letter_info"},
depends_on_past=False,
retries=3,
dag=dag)
agents_market_summar >> tk_product_prefer
agents_market_summar >> batch_gen_crawer
agents_recuriments >> batch_gen_recruiment
a_market_Intelligence_newsletter >> tk_product_prefer
tk_product_prefer >> dysql_news_letter_info
agents_market_summar >> a_market_Intelligence_newsletter
a_market_Intelligence_newsletter >> task_failed