model_script/monitor/supply_workflow_monitor.py

from datetime import datetime
from typing import Dict, Any, List, Optional

import pandas
import pandas as pd

from helper.MySQLHelper import MySQLHelper
from util import feishu_inform_util

fei_shu_webhook = "https://open.feishu.cn/open-apis/bot/v2/hook/c09712a8-22cd-4bfa-93a5-30ae7b1db11b"

mysql_helper = MySQLHelper(
    host="rm-t4na9qj85v7790tf84o.mysql.singapore.rds.aliyuncs.com",
    username="readonly",
    password="HdkZ4TDmeK6SQ3BRtJBk",
    database="aigc-admin-prod"
)

column_width_map = {
    "状态": "80px",
    "任务数": "80px",
    "记录数": "80px",

    "待处理": "80px",
    "运行中": "80px",
    "成功": "80px",
    "失败": "80px",
}

header_template = ["wathet", "turquoise", "yellow", "orange", "carmine", "violet", "purple", "indigo", "green"]
header_template_index = 0


def build_config_json() -> Dict[str, Any]:
    return {
        "width_mode": "default",
        "enable_forward": False,
    }


def build_markdown_element_json(content: str) -> Dict[str, Any]:
    return {
        "tag": "markdown",
        "element_id": "markdown",
        "content": content
    }


def build_table_element_json(df: pandas.DataFrame) -> Optional[Dict[str, Any]]:
    if df.empty:
        return None
    columns = []
    for column in df.columns:
        columns.append({
            "name": column,
            "display_name": column,
            "width": column_width_map.get(column, "auto"),
            "data_type": "text",
            "horizontal_align": "center",
            "vertical_align": "center",
        })

    rows = df.to_dict(orient="records")
    return {
        "tag": "table",
        "element_id": "table_detail",
        "margin": "2px 0px",
        "page_size": 10,
        "header_style": {
            "text_align": "center",
            "text_size": "normal",
            "background_style": "none",
            "text_color": "grey",
            "bold": True,
            "lines": 1
        },
        "columns": columns,
        "rows": rows
    }


def build_markdown_table_element_json(df: pandas.DataFrame) -> Dict[str, Any]:
    # 表头
    header = "| " + " | ".join(df.columns) + " |\n"
    # 分隔线
    separator = "| " + " | ".join(["---"] * len(df.columns)) + " |\n"
    # 数据行
    rows = "\n".join(
        "| " + " | ".join(str(val) for val in row) + " |"
        for row in df.to_numpy()  # 或 df.values
    )
    content = header + separator + rows
    return build_markdown_element_json(content)


def build_collapsible_panel_element_json(elements: List[Dict[str, Any]], title: str) -> Dict[str, Any]:
    return {
        "tag": "collapsible_panel",
        "expanded": False,
        "header": {
            "title": {
                "tag": "plain_text",
                "content": title
            },
            "vertical_align": "center",
        },
        "border": {
            "color": "grey",
            "corner_radius": "5px"
        },
        "elements": elements
    }


def build_hr_element_json() -> Dict[str, Any]:
    return {
        "tag": "hr",
        "element_id": "custom_id",
        "margin": "10px 0px"
    }


def build_text_tag_json(content: str, color: str = 'green') -> Dict[str, Any]:
    return {
        "tag": "text_tag",
        "element_id": "custom_id",
        "text": {
            "tag": "plain_text",
            "content": content
        },
        "color": color
    }


def build_sub_title_json(content: str) -> Dict[str, Any]:
    return {
        "tag": "plain_text",
        "content": content
    }


def build_header_json(content: str, template: str = 'green', sub_title: Dict[str, Any] = None, text_tag_list: List[Dict[str, Any]] = []) -> Dict[str, Any]:
    return {
        "title": {
            "content": content
        },
        "text_tag_list": text_tag_list,
        "template": template,
        "subtitle": sub_title
    }


def build_card_json(body_elements: List[Dict[str, Any]], header: Dict[str, Any], config: Dict[str, Any] = None) -> Dict[str, Any]:
    return {
        "schema": "2.0",
        "config": config,
        "header": header,
        "body": {
            "elements": body_elements
        }
    }


def supply_workflow_dashboard_stat(ts: int) -> List[Dict[str, Any]]:
    sql = f"""
            SELECT sw.name AS '策略名',
                   case
                       when sw.status = 1 then '开启中'
                       when sw.status = 0 then '已关闭'
                       else '未知' end AS 状态,
                   swt.init_cnt        AS '待处理',
                   swt.running_cnt     AS '运行中',
                   swt.success_cnt     AS '成功',
                   swt.fail_success    AS '失败'
            FROM supply_workflow sw
                     LEFT JOIN (
                                   SELECT workflow_id,
                                          COUNT(DISTINCT IF(task_status = 0, task_id, NULL)) AS init_cnt,
                                          COUNT(DISTINCT IF(task_status = 1, task_id, NULL)) AS running_cnt,
                                          COUNT(DISTINCT IF(task_status = 2, task_id, NULL)) AS success_cnt,
                                          COUNT(DISTINCT IF(task_status = 3, task_id, NULL)) AS fail_success
                                   FROM supply_workflow_task
                                   WHERE create_timestamp >= {ts}
                                   GROUP BY workflow_id
                               ) swt ON sw.id = swt.workflow_id
            ORDER BY sw.status DESC, sw.create_timestamp DESC;
            """
    return mysql_helper.execute_query(sql)


def workflow_task_status_stat(ts: int, workflow_id: str) -> List[Dict[str, Any]]:
    sql = f"""
            select workflow_id             as 'workflow_id',
                   task_status             as '任务状态',
                   count(distinct task_id) as '任务数',
                   count(task_input)       as '需求数'
            from (
                     select workflow_id         as workflow_id,
                            task_id             as task_id,
                            task_input          as task_input,
                            case
                                when task_status = 0 then '初始化'
                                when task_status = 1 then '运行中'
                                when task_status = 2 then '成功'
                                when task_status = 3 then '失败'
                                else '未知' end AS task_status
                    from supply_workflow sw
                        left join supply_workflow_task swt
                            on sw.id = swt.workflow_id
                     where sw.id = '{workflow_id}'
                       and swt.create_timestamp >= {ts}
                 ) as t
            group by t.task_status, t.workflow_id;
            """
    result = mysql_helper.execute_query(sql)
    return result


def task_exe_step_stat_query(ts: int, workflow_id: str) -> List[Dict[str, Any]]:
    sql = f'''
            select step_name AS '步骤名称',
                   status    as '执行状态',
                   error_msg as '错误原因',
                   count(1)  AS '任务数'
            from (
                     select swt.task_id,
                            step_name,
                            case
                                when status = 0 then '初始化'
                                when status = 1 then '运行中'
                                when status = 2 then '成功'
                                when status = 3 then '失败'
                                else '未知' end AS status,
                            case
                                when swtes.error_msg like '%Data too long%' then '数据超过字段长度限制'
                                when swtes.error_msg like '%Deadlock%' then '数据库死锁'
                                when (swtes.error_msg <> '' and swtes.error_msg is not null) then substring_index(swtes.error_msg, ',', 1)
                                else '' end  AS error_msg
                     from supply_workflow_task swt
                              left join supply_workflow_task_exe_step swtes on swt.task_id = swtes.task_id
                     where swt.create_timestamp >= {ts}
                       and swt.workflow_id = {workflow_id}
                 ) as t
            group by step_name, status, error_msg;
        '''
    return mysql_helper.execute_query(sql)


def crawler_and_produce_stat_query(ts: int, workflow_id: str) -> List[Dict[str, Any]]:
    sql = f'''
            select t.step      as '阶段',
                   t.status    as '状态',
                   t.error_msg as '失败原因',
                   t.task_cnt  as '任务数',
                   t.cnt       as '记录数'
            from (
                     select '创建爬取计划'                  as step,
                            workflow_id                     as workflow_id,
                            status                          as status,
                            error_msg                       as error_msg,
                            count(distinct task_id)         as task_cnt,
                            count(distinct crawler_plan_id) as cnt
                     from (
                              select swt.workflow_id       as workflow_id,
                                     swt.task_id           as task_id,
                                     swcpr.crawler_plan_id as crawler_plan_id,
                                     case
                                         when swcpr.status = 0 then '初始化'
                                         when swcpr.status = 1 then '运行中'
                                         when swcpr.status = 2 then '成功'
                                         when swcpr.status = 3 then '失败'
                                         when swt.task_status = 0 then '初始化'
                                         when swt.task_status = 1 then '运行中'
                                         when swt.task_status = 2 then '成功'
                                         when swt.task_status = 3 then '失败'
                                         else '未知' end   AS status,
                                     case
                                         when swcpr.error_msg like '%Data too long%' then '数据超过字段长度限制'
                                         when swcpr.error_msg like '%Deadlock%' then '数据库死锁'
                                         when (swcpr.error_msg <> '' AND swcpr.error_msg is not null) then substring_index(swcpr.error_msg, ',', 1)
                                         when swt.error_msg like '%Data too long%' then '数据超过字段长度限制'
                                         when swt.error_msg like '%Deadlock%' then '数据库死锁'
                                         when (swt.error_msg <> '' AND swt.error_msg is not null) then substring_index(swt.error_msg, ',', 1)
                                         else '' 
                                         end               AS error_msg
                              from supply_workflow_task swt
                                       left join supply_workflow_crawler_plan_record swcpr
                                                 on swt.task_id = swcpr.task_id
                              where swt.create_timestamp >= {ts}
                          ) as t
                     group by workflow_id, status, error_msg
                     union all
                     select '生成计划绑定'          as step,
                            workflow_id             as workflow_id,
                            status                  as status,
                            error_msg               as error_msg,
                            count(distinct task_id) as task_cnt,
                            count(1)                as cnt
                     from (
                              select swt.workflow_id     as workflow_id,
                                     swt.task_id         as task_id,
                                     case
                                         when swppr.status = 0 then '初始化'
                                         when swppr.status = 1 then '运行中'
                                         when swppr.status = 2 then '成功'
                                         when swppr.status = 3 then '失败'
                                         when swt.task_status = 0 then '初始化'
                                         when swt.task_status = 1 then '运行中'
                                         when swt.task_status = 2 then '成功'
                                         when swt.task_status = 3 then '失败'
                                         else '未知' end AS status,
                                     case
                                         when swppr.error_msg like '%Data too long%' then '数据超过字段长度限制'
                                         when swppr.error_msg like '%Deadlock%' then '数据库死锁'
                                         when (swppr.error_msg <> '' and swppr.error_msg is not null) then substring_index(swppr.error_msg, ',', 1)
                                         when swt.error_msg like '%Data too long%' then '数据超过字段长度限制'
                                         when swt.error_msg like '%Deadlock%' then '数据库死锁'
                                         when (swt.error_msg <> '' AND swt.error_msg is not null) then substring_index(swt.error_msg, ',', 1)
                                         else ''
                                         end             AS error_msg
                              from supply_workflow_task swt
                                       left join supply_workflow_produce_bind_record swppr
                                                 on swt.task_id = swppr.task_id
                              where swt.create_timestamp >= {ts}
                          ) as t
                     group by workflow_id, status, error_msg
                 ) as t
                     left join supply_workflow sw on t.workflow_id = sw.id
            where workflow_id = '{workflow_id}';
            '''
    return mysql_helper.execute_query(sql)


def workflow_monitor(workflow_id: str, workflow_name: str, status: int):
    today_midnight = datetime.now().replace(hour=0, minute=0, second=0, microsecond=0)

    start_dt_str = today_midnight.strftime('%Y-%m-%d %H:%M:%S')
    end_dt_str = datetime.now().strftime('%Y-%m-%d %H:%M:%S')
    timestamp_ms = int(today_midnight.timestamp() * 1000)

    task_status_stat = workflow_task_status_stat(timestamp_ms, workflow_id)

    exe_step_stat = task_exe_step_stat_query(timestamp_ms, workflow_id)
    status_stat = crawler_and_produce_stat_query(timestamp_ms, workflow_id)

    # 构建标题

    text_tag_list = []
    if status == 0:
        text_tag_list.append(build_text_tag_json("已关闭", "red"))
    elif status == 1:
        text_tag_list.append(build_text_tag_json("开启中", "green"))
    else:
        text_tag_list.append(build_text_tag_json("未知", "orange"))

    template = header_template[header_template_index % len(header_template)]
    sub_title = build_sub_title_json(f"{start_dt_str} - {end_dt_str}")
    header = build_header_json(f"【workflow策略】{workflow_name}", template, sub_title, text_tag_list)

    task_status_df = pd.DataFrame(task_status_stat)
    if "任务数" not in task_status_df.columns:
        task_info = f"**总任务数**: 0"
    else:
        # 先构建总任务数部分
        task_info = f"**总任务数**: {task_status_df['任务数'].sum()}"
        # 再逐行添加各状态的任务数
        for _, row in task_status_df.iterrows():
            task_info += f"\n**{row['任务状态']}任务数**: {row['任务数']}"

    elements = [build_markdown_element_json(task_info)]

    exe_step_table = build_table_element_json(pd.DataFrame(exe_step_stat))
    if exe_step_table is not None:
        elements.append(build_hr_element_json())
        elements.append(exe_step_table)

    status_stat_table = build_table_element_json(pd.DataFrame(status_stat))
    if status_stat_table is not None:
        elements.append(build_hr_element_json())
        elements.append(status_stat_table)

    config = build_config_json()

    card_json = build_card_json(elements, header, config)

    feishu_inform_util.send_card_msg_to_feishu(
        webhook=fei_shu_webhook,
        card_json=card_json
    )


def workflow_dashboard_monitor():
    today_midnight = datetime.now().replace(hour=0, minute=0, second=0, microsecond=0)
    result = supply_workflow_dashboard_stat(int(today_midnight.timestamp() * 1000))

    start_dt_str = today_midnight.strftime('%Y-%m-%d %H:%M')
    end_dt_str = datetime.now().strftime('%Y-%m-%d %H:%M')

    sub_title = build_sub_title_json(f"统计时间: {start_dt_str} - {end_dt_str}")
    header = build_header_json("【供给workflow】当日任务统计", "blue", sub_title, [])
    dashboard_stat = build_table_element_json(pd.DataFrame(result))
    elements = []
    if dashboard_stat is not None:
        elements.append(dashboard_stat)
    else:
        elements.append(build_markdown_element_json("**今日workflow还未创建新的任务，请关注**"))

    config = build_config_json()
    card_json = build_card_json(elements, header, config)

    feishu_inform_util.send_card_msg_to_feishu(
        webhook=fei_shu_webhook,
        card_json=card_json
    )

    # 总体情况发送到大群
    feishu_inform_util.send_card_msg_to_feishu(
        webhook="https://open.feishu.cn/open-apis/bot/v2/hook/9f5c5cce-5eb2-4731-b368-33926f5549f9",
        card_json=card_json
    )


def main():
    workflow_dashboard_monitor()

    global header_template_index
    workflows = mysql_helper.execute_query('select id, name, status from supply_workflow;')
    for workflow in workflows:
        try:
            workflow_monitor(workflow['id'], workflow['name'], workflow['status'])
        except Exception as e:
            print(f"【workflow策略】{workflow['name']} 监控异常: {e}")


if __name__ == '__main__':
    main()