123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150 |
- """
- @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
- 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 * 7
- 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]
- 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
- 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 = 500
- # 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
|