LongArticlesJob/applications/const/__init__.py

"""
@author: luojunhui
任务常量配置文件
"""


class ColdStartTaskConst:
    """
    冷启动任务常量配置
    """

    PUBLISHED_STATUS = 2  # 文章已发布状态
    INIT_STATUS = 1  # 文章初始状态
    BAD_STATUS = 0  # 低质量文章状态

    # 常量
    ACCOUNT_GOOD_STATUS = 1

    # 账号是否每日抓取
    ACCOUNT_DAILY_SCRAPE = 1
    ACCOUNT_NOT_DAILY_SCRAPE = 0

    # 默认值
    DEFAULT_VIEW_COUNT = 0
    DEFAULT_LIKE_COUNT = 0
    DEFAULT_ARTICLE_STATUS = 1
    DEFAULT_TIMESTAMP = 1717171200

    # 标题sensitivity
    TITLE_SENSITIVE = 1
    TITLE_NOT_SENSITIVE = 0

    # 文章联想深度
    ARTICLE_ASSOCIATION_MAX_DEPTH = 4

    # 相关分百分位阈值
    PERCENT_THRESHOLD = 95

    # 相关性分阈值
    CORRELATION_THRESHOLD = 0.5

    # 阅读量阈值
    READ_COUNT_THRESHOLD = 1000

    # 阅读均值倍数阈值
    READ_AVG_THRESHOLD = 1.3

    # 群发类型
    BULK_PUBLISH_TYPE = 9

    # 种子文章数量
    SEED_ARTICLE_LIMIT_NUM = 60


class updatePublishedMsgTaskConst:
    """
    更新已发布文章消息常量配置
    """

    # 爬虫详情接口返回code
    ARTICLE_ILLEGAL_CODE = 25012
    ARTICLE_DELETE_CODE = 25005
    ARTICLE_SUCCESS_CODE = 0
    ARTICLE_UNKNOWN_CODE = 10000

    # 请求爬虫详情接口状态码
    # 记录默认状态
    DEFAULT_STATUS = 0
    # 请求接口失败状态
    REQUEST_FAIL_STATUS = -1
    # 文章被删除状态
    DELETE_STATUS = -2
    # 未知原因无信息返回状态
    UNKNOWN_STATUS = -3
    # 文章违规状态
    ILLEGAL_STATUS = -4

    # 公众号类型（订阅号 or 服务号）
    # 订阅号
    SUBSCRIBE_TYPE_SET = {0, 1}
    # 服务号
    SERVICE_TYPE = 2
    # 监测周期（秒）
    MONITOR_PERIOD = 60 * 60 * 24 * 3

    # 新号抓文章周期
    NEW_ACCOUNT_CRAWL_PERIOD = 60 * 60 * 24 * 30

    # 订阅号，抓取失败失败率报警阈值
    SUBSCRIBE_FAIL_RATE_THRESHOLD = 0.3


class UpdateAccountReadRateTaskConst:
    """
    更新账号阅读率常量配置
    """

    # 阅读率统计周期（秒）
    STATISTICS_PERIOD = 31 * 24 * 60 * 60
    # 一天的秒数
    ONE_DAY_IN_SECONDS = 60 * 60 * 24
    # 相对变化率阈值
    RELATIVE_VALUE_THRESHOLD = 0.1
    # 发文类型
    UNLIMITED_PUBLISH_TYPE = 10002
    BULK_PUBLISH_TYPE = 9
    # 文章位置
    ARTICLE_INDEX_LIST = [1, 2, 3, 4, 5, 6, 7, 8]

    # 默认粉丝
    DEFAULT_FANS = 0

    # 最低粉丝量
    MIN_FANS = 1000


class UpdateAccountReadAvgTaskConst:
    """
    更新账号阅读均值常量配置
    """

    # 投流账号
    TOULIU_ACCOUNTS = {
        "gh_93e00e187787",
        "gh_ac43e43b253b",
        "gh_68e7fdc09fe4",
        "gh_77f36c109fb1",
        "gh_b181786a6c8c",
        "gh_1ee2e1b39ccf",
        "gh_d3f039c9db2b",
    }

    # 发文模式
    ARTICLES_DAILY = 1
    TOULIU = 2

    # 默认粉丝
    DEFAULT_FANS = 0

    # index list
    ARTICLE_INDEX_LIST = [1, 2, 3, 4, 5, 6, 7, 8]

    # 默认点赞
    DEFAULT_LIKE = 0

    # 状态
    USING_STATUS = 1
    NOT_USING_STATUS = 0

    # 统计周期
    STAT_PERIOD = 30

    # default upper_quantile, confidence = 0.95
    DEFAULT_UPPER_QUANTILE = 0.975

    ACCOUNT_READ_RATE_TABLE = "long_articles_read_rate"
    ACCOUNT_READ_AVG_TABLE = "account_avg_info_v3"


class WeixinVideoCrawlerConst:
    """
    微信视频抓取常量配置
    """

    # 账号抓取状态
    ACCOUNT_CRAWL_STATUS = 1
    ACCOUNT_DO_NOT_CRAWL_STATUS = 0

    # 默认最早抓取时间戳(2024-01-01)
    DEFAULT_TIMESTAMP = 1704038400

    # 搜索爬虫最大页数
    MAX_SEARCH_PAGE_NUM = 10

    # 抓取每一页的等待时间
    SLEEP_SECONDS = 5

    # 种子标题最低阅读均值倍数
    READ_AVG_MULTIPLE = 1.3

    # 种子标题最低阅读量
    MIN_READ_COUNT = 2000

    # 获取种子标题的统计周期
    STAT_PERIOD = 7 * 24 * 60 * 60

    # 接口请求成功code
    REQUEST_SUCCESS = 0
    PUBLISHED_ILLEGAL_TITLE_CODE = 1015

    # 是否需要扫描查询源账号
    NEED_SCAN_SOURCE_ACCOUNT = 1
    DO_NOT_NEED_SOURCE_ACCOUNT = 0

    # 视频审核状态长文库
    VIDEO_AUDIT_INIT_STATUS = 0
    VIDEO_AUDIT_SUCCESS_STATUS = 1
    VIDEO_AUDIT_FAIL_STATUS = 2
    VIDEO_TITLE_GENERATE_FAIL_STATUS = 4
    VIDEO_AUDIT_PROCESSING_STATUS = -1

    # 票圈视频审核状态, 1 审核中，2 不通过 3 待修改，4 自己可见 5 通过
    PQ_AUDIT_PROCESSING_STATUS = 1
    PQ_AUDIT_FAIL_STATUS = 2
    PQ_AUDIT_WAIT_STATUS = 3
    PQ_AUDIT_SELF_VISIBLE_STATUS = 4
    PQ_AUDIT_SUCCESS_STATUS = 5

    # 默认账号
    DEFAULT_ACCOUNT_UID = 76862180

    # 每天发送的审核视频数量
    MAX_VIDEO_NUM = 1000

    # 单次发布视频审核量
    MAX_VIDEO_NUM_PER_PUBLISH = 350

    # 标题状态
    TITLE_DEFAULT_STATUS = 0
    TITLE_EXIT_STATUS = 1
    TITLE_FESTIVAL_STATUS = 2
    TITLE_SHORT_STATUS = 3

    # 标题最短长度
    TITLE_MIN_LENGTH = 15

    # safe score
    TITLE_SAFE_SCORE_THRESHOLD = 7

    # Task Status
    INIT_STATUS = 0
    PROCESSING_STATUS = 1
    SUCCESS_STATUS = 2
    FAIL_STATUS = 99


class UpdateMiniProgramDetailConst(updatePublishedMsgTaskConst):
    """
    更新小程序详情常量配置
    """


# 账号联想
class AccountAssociationTaskConst:
    """
    账号联想任务常量配置
    """

    # 获取种子标题的统计周期
    STAT_PERIOD = 7 * 24 * 60 * 60

    # 阅读均值阈值
    READ_AVG_MULTIPLE = 1.3

    # 最小阅读量
    MIN_READ_COUNT = 2000

    # 种子数量限制
    SEED_TITLE_LIMIT = 100


# 从aigc获取文章
class ArticleCollectorConst:
    """
    文章采集任务常量配置
    """

    # 发送方式

    # 手动推送
    MANUAL_PUSH = 1
    # 自动群发
    BULK_AUTO_PUSH = 2
    # 无限流推送
    UNLIMITED_PUSH = 3

    # 文章状态
    # 初始状态
    INIT_STATUS = 0
    # 成功状态
    SUCCESS_STATUS = 1
    # 失败状态
    FAIL_STATUS = -1
    # 发布状态
    PUBLISHED_STATUS = 2

    # 爬虫接口
    ARTICLE_ILLEGAL_CODE = 25012
    ARTICLE_DELETE_CODE = 25005
    ARTICLE_SUCCESS_CODE = 0
    ARTICLE_UNKNOWN_CODE = 10000


class BaiduVideoCrawlerConst:
    """
    const for baidu video crawler
    """

    # account status
    BAIDU_ACCOUNT_GOOD_STATUS = 1
    BAIDU_ACCOUNT_BAD_STATUS = 0

    # earliest cursor, 2024-01-01 00:00:00
    DEFAULT_CURSOR = 17040384000000

    # no source account
    NO_SOURCE_ACCOUNT_STATUS = 0

    # timestamp To Cursor
    TIMESTAMP_TO_CURSOR = 10000

    # local path dir
    LOCAL_PATH_DIR = "static"


class TitleRewriteTaskConst:
    """
    title rewrite task const
    """

    # title rewrite status
    TITLE_REWRITE_INIT_STATUS = 0
    TITLE_REWRITE_SUCCESS_STATUS = 1
    TITLE_REWRITE_FAIL_STATUS = 99
    TITLE_REWRITE_LOCK_STATUS = 101

    # article status
    ARTICLE_AUDIT_PASSED_STATUS = 1
    ARTICLE_POSITIVE_STATUS = 0

    # title useful status
    TITLE_USEFUL_STATUS = 1

    # prompt version
    PROMPT_VERSION = "xx_250228"  # 信欣2025-02-28提供

    # block expire time 1h
    TITLE_REWRITE_LOCK_TIME = 60 * 60


class ChannelVideoCrawlerConst:
    """
    const for baidu video crawler
    """

    # account status
    CHANNEL_ACCOUNT_GOOD_STATUS = 1
    CHANNEL_ACCOUNT_BAD_STATUS = 0

    # earliest cursor, 2024-01-01 00:00:00
    DEFAULT_CURSOR = 1704038400

    # no source account
    NO_SOURCE_ACCOUNT_STATUS = 0

    # local path dir
    LOCAL_PATH_DIR = "static"

    # title length min
    MIN_TITLE_LENGTH = 10

    # max video length(second)
    MAX_VIDEO_LENGTH = 600

    # sleep second
    SLEEP_SECOND = 2


class ToutiaoVideoCrawlerConst:
    """
    const for toutiao video crawler
    """

    # platform
    PLATFORM = "toutiao"

    # account status
    TOUTIAO_ACCOUNT_GOOD_STATUS = 1
    TOUTIAO_ACCOUNT_BAD_STATUS = 0

    # earliest cursor, 2021-01-01 00:00:00
    DEFAULT_CURSOR = 1609430400

    # no source account
    NO_SOURCE_ACCOUNT_STATUS = 0

    # title length min
    MIN_TITLE_LENGTH = 10

    # max video length(second)
    MAX_VIDEO_LENGTH = 600

    # sleep second
    SLEEP_SECOND = 3


class SohuVideoCrawlerConst:
    """
    const for sohu video crawler
    """
    # platform
    PLATFORM = "sohu"

    # account status
    GET_RECOMMEND_INIT_STATUS = 0
    GET_RECOMMEND_SUCCESS_STATUS = 1
    GET_RECOMMEND_FAIL_STATUS = 99

    # title length min
    MIN_TITLE_LENGTH = 10

    # max video length(second)
    MAX_VIDEO_LENGTH = 600

    # sleep second
    SLEEP_SECOND = 3

    # 获取推荐的最低相关性分
    GET_RECOMMEND_THRESHOLD_SCORE = 0.6

    # 审核状态
    AUDIT_SUCCESS_STATUS = 1

    # 视频状态
    VIDEO_NOT_BAD_STATUS = 0

    # PAGE_LIST
    PAGE_LIST = [i for i in range(1, 8)]


class SingleVideoPoolPublishTaskConst:
    """
    const for single video pool publish task
    """

    TRANSFORM_INIT_STATUS = 0
    TRANSFORM_SUCCESS_STATUS = 1
    TRANSFORM_FAIL_STATUS = 99

    SUCCESS_STATUS = 2



class GoogleVideoUnderstandTaskConst:
    # task batch size
    BATCH_SIZE = 100

    # task status
    INIT_STATUS = 0
    PROCESSING_STATUS = 1
    SUCCESS_STATUS = 2
    FAIL_STATUS = 99

    # sleep seconds
    SLEEP_SECONDS = 60

    # max processing time
    MAX_PROCESSING_TIME = 3600

    # task info
    TABLE_NAME = "long_articles_new_video_cover"
    TASK_NAME = "extract_video_best_frame_as_cover"
    DIR_NAME = "static"
    POOL_SIZE = 15


class CategoryGenerationTaskConst:
    """
    const for category generation task
    """
    # MAX THREAD
    MAX_WORKERS = 10

    # task batch size
    BATCH_SIZE = 20

    # min batch
    MIN_BATCH_SIZE = 1

    # article status
    ARTICLE_GOOD_STATUS = 0

    # task status
    INIT_STATUS = 0
    PROCESSING_STATUS = 1
    SUCCESS_STATUS = 2
    FAIL_STATUS = 99

    # max processing time
    MAX_PROCESSING_TIME = 3600

    # task info
    VIDEO_TABLE_NAME = "publish_single_video_source"
    ARTICLE_TABLE_NAME = "crawler_meta_article"
    TASK_NAME = "generate_category_with_title"

    # article_status
    ARTICLE_INIT_STATUS = 1
    ARTICLE_PUBLISHED_STATUS = 2
    ARTICLE_BAD_STATUS = 0

    # limit score
    LIMIT_SCORE = 0.4