LongArticleTaskServer/app/jobs/task_config.py

from dataclasses import dataclass


@dataclass
class TaskConfig:
    """任务配置"""

    timeout: int  # 超时时间（秒）
    max_concurrent: int = 5  # 最大并发数
    retry_times: int = 0  # 重试次数
    retryable: bool = True  # 是否可重试
    alert_on_failure: bool = True  # 失败时是否告警


class TaskStatus:
    """任务状态常量"""

    INIT = 0
    PROCESSING = 1
    SUCCESS = 2
    FAILED = 99


class TaskConstants:
    """任务系统常量"""

    # 默认配置
    DEFAULT_TIMEOUT = 1800
    DEFAULT_MAX_CONCURRENT = 5
    DEFAULT_RETRY_TIMES = 0

    # 数据库表名
    TASK_TABLE = "long_articles_task_manager"


# 所有任务的配置映射
TASK_CONFIGS = {
    # 监控类任务
    "check_kimi_balance": TaskConfig(
        timeout=20,
        max_concurrent=1,
        retry_times=3,
    ),
    "get_off_videos": TaskConfig(
        timeout=1800,
        max_concurrent=3,
    ),
    "check_publish_video_audit_status": TaskConfig(
        timeout=1800,
        max_concurrent=3,
    ),
    "outside_article_monitor": TaskConfig(
        timeout=3 * 3600,
        max_concurrent=2,
    ),
    "inner_article_monitor": TaskConfig(
        timeout=3600,
        max_concurrent=3,
    ),
    "task_processing_monitor": TaskConfig(
        timeout=300,
        max_concurrent=1,
    ),
    # 爬虫类任务
    "crawler_toutiao": TaskConfig(
        timeout=5 * 3600,
        max_concurrent=3,
        retry_times=2,
    ),
    "crawler_gzh_articles": TaskConfig(
        timeout=4 * 3600,
        max_concurrent=3,
        retry_times=2,
    ),
    "crawler_account_manager": TaskConfig(
        timeout=1800,
        max_concurrent=5,
    ),
    "crawler_detail_analysis": TaskConfig(
        timeout=3600,
        max_concurrent=3,
    ),
    # 数据处理类任务
    "daily_publish_articles_recycle": TaskConfig(
        timeout=3600,
        max_concurrent=2,
    ),
    "update_root_source_id": TaskConfig(
        timeout=3600,
        max_concurrent=2,
    ),
    "recycle_outside_account_articles": TaskConfig(
        timeout=3600,
        max_concurrent=2,
    ),
    "update_outside_account_article_root_source_id": TaskConfig(
        timeout=3600,
        max_concurrent=2,
    ),
    "fwh_daily_recycle": TaskConfig(
        timeout=3600,
        max_concurrent=2,
    ),
    # 算法类任务
    "article_pool_cold_start": TaskConfig(
        timeout=4 * 3600,
        max_concurrent=2,
        retry_times=1,
    ),
    "candidate_account_quality_analysis": TaskConfig(
        timeout=3600,
        max_concurrent=3,
    ),
    "article_pool_category_generation": TaskConfig(
        timeout=3600,
        max_concurrent=3,
    ),
    "account_category_analysis": TaskConfig(
        timeout=3600,
        max_concurrent=3,
    ),
    "update_limited_account_info": TaskConfig(
        timeout=3600,
        max_concurrent=2,
    ),
    # LLM 类任务
    "title_rewrite": TaskConfig(
        timeout=1800,
        max_concurrent=3,
        retry_times=2,
    ),
    "extract_title_features": TaskConfig(
        timeout=1800,
        max_concurrent=3,
        retry_times=2,
    ),
    # 统计分析类任务
    "update_account_read_rate_avg": TaskConfig(
        timeout=1800,
        max_concurrent=3,
    ),
    "update_account_read_avg": TaskConfig(
        timeout=1800,
        max_concurrent=3,
    ),
    "update_account_open_rate_avg": TaskConfig(
        timeout=1800,
        max_concurrent=3,
    ),
    # 自动化任务
    "auto_follow_account": TaskConfig(
        timeout=1800,
        max_concurrent=2,
    ),
    "get_follow_result": TaskConfig(
        timeout=1800,
        max_concurrent=2,
    ),
    "extract_reply_result": TaskConfig(
        timeout=1800,
        max_concurrent=2,
    ),
    # 合作账号任务
    "cooperate_accounts_monitor": TaskConfig(
        timeout=7200,
        max_concurrent=2,
    ),
    "cooperate_accounts_detail": TaskConfig(
        timeout=3600,
        max_concurrent=2,
    ),
    # 其他任务
    "mini_program_detail_process": TaskConfig(
        timeout=1800,
        max_concurrent=3,
    ),
}


def get_task_config(task_name: str) -> TaskConfig:
    """获取任务配置，如果不存在则返回默认配置"""
    return TASK_CONFIGS.get(
        task_name,
        TaskConfig(
            timeout=TaskConstants.DEFAULT_TIMEOUT,
            max_concurrent=TaskConstants.DEFAULT_MAX_CONCURRENT,
            retry_times=TaskConstants.DEFAULT_RETRY_TIMES,
        ),
    )


__all__ = [
    "TaskConfig",
    "TaskStatus",
    "TaskConstants",
    "TASK_CONFIGS",
    "get_task_config",
]