const.py 3.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144
  1. """
  2. @author: luojunhui
  3. 任务常量配置文件
  4. """
  5. class coldStartTaskConst:
  6. """
  7. 冷启动任务常量配置
  8. """
  9. PUBLISHED_STATUS = 2 # 文章已发布状态
  10. INIT_STATUS = 1 # 文章初始状态
  11. BAD_STATUS = 0 # 低质量文章状态
  12. class updatePublishedMsgTaskConst:
  13. """
  14. 更新已发布文章消息常量配置
  15. """
  16. # 爬虫详情接口返回code
  17. ARTICLE_DELETE_CODE = 25005
  18. ARTICLE_SUCCESS_CODE = 0
  19. # 请求爬虫详情接口状态码
  20. # 记录默认状态
  21. DEFAULT_STATUS = 0
  22. # 请求接口失败状态
  23. REQUEST_FAIL_STATUS = -1
  24. # 文章被删除状态
  25. DELETE_STATUS = -2
  26. # 未知原因无信息返回状态
  27. UNKNOWN_STATUS = -3
  28. # 公众号类型(订阅号 or 服务号)
  29. # 订阅号
  30. SUBSCRIBE_TYPE_SET = {0, 1}
  31. # 服务号
  32. SERVICE_TYPE = 2
  33. class updateAccountReadRateTaskConst:
  34. """
  35. 更新账号阅读率常量配置
  36. """
  37. # 阅读率统计周期(秒)
  38. STATISTICS_PERIOD = 31 * 24 * 60 * 60
  39. # 一天的秒数
  40. ONE_DAY_IN_SECONDS = 60 * 60 * 24
  41. # 相对变化率阈值
  42. RELATIVE_VALUE_THRESHOLD = 0.1
  43. # 发文类型
  44. UNLIMITED_PUBLISH_TYPE = 10002
  45. BULK_PUBLISH_TYPE = 9
  46. # 文章位置
  47. ARTICLE_INDEX_LIST = [1, 2, 3, 4, 5, 6, 7, 8]
  48. class updateAccountReadAvgTaskConst:
  49. """
  50. 更新账号阅读均值常量配置
  51. """
  52. # 投流账号
  53. TOULIU_ACCOUNTS = {
  54. 'gh_93e00e187787',
  55. 'gh_ac43e43b253b',
  56. 'gh_68e7fdc09fe4',
  57. 'gh_77f36c109fb1',
  58. 'gh_b181786a6c8c',
  59. 'gh_1ee2e1b39ccf',
  60. 'gh_d3f039c9db2b'
  61. }
  62. # 发文模式
  63. ARTICLES_DAILY = 1
  64. TOULIU = 2
  65. class WeixinVideoCrawlerConst:
  66. """
  67. 微信视频抓取常量配置
  68. """
  69. # 账号抓取状态
  70. ACCOUNT_CRAWL_STATUS = 1
  71. ACCOUNT_DO_NOT_CRAWL_STATUS = 0
  72. # 默认最早抓取时间戳(2024-01-01)
  73. DEFAULT_TIMESTAMP = 1704038400
  74. # 搜索爬虫最大页数
  75. MAX_SEARCH_PAGE_NUM = 10
  76. # 抓取每一页的等待时间
  77. SLEEP_SECONDS = 5
  78. # 种子标题最低阅读均值倍数
  79. READ_AVG_MULTIPLE = 1.3
  80. # 种子标题最低阅读量
  81. MIN_READ_COUNT = 2000
  82. # 获取种子标题的统计周期
  83. STAT_PERIOD = 7 * 24 * 60 * 60
  84. # 接口请求成功code
  85. REQUEST_SUCCESS = 0
  86. PUBLISHED_ILLEGAL_TITLE_CODE = 1015
  87. # 是否需要扫描查询源账号
  88. NEED_SCAN_SOURCE_ACCOUNT = 1
  89. DO_NOT_NEED_SOURCE_ACCOUNT = 0
  90. # 视频审核状态长文库
  91. VIDEO_AUDIT_INIT_STATUS = 0
  92. VIDEO_AUDIT_SUCCESS_STATUS = 1
  93. VIDEO_AUDIT_FAIL_STATUS = 2
  94. VIDEO_TITLE_GENERATE_FAIL_STATUS = 4
  95. VIDEO_AUDIT_PROCESSING_STATUS = -1
  96. # 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
  97. PQ_AUDIT_PROCESSING_STATUS = 1
  98. PQ_AUDIT_FAIL_STATUS = 2
  99. PQ_AUDIT_WAIT_STATUS = 3
  100. PQ_AUDIT_SELF_VISIBLE_STATUS = 4
  101. PQ_AUDIT_SUCCESS_STATUS = 5
  102. # 默认账号
  103. DEFAULT_ACCOUNT_UID = 76862180
  104. # 每天发送的审核视频数量
  105. MAX_VIDEO_NUM = 500
  106. # 标题状态
  107. TITLE_DEFAULT_STATUS = 0
  108. TITLE_EXIT_STATUS = 1
  109. TITLE_FESTIVAL_STATUS = 2
  110. TITLE_SHORT_STATUS = 3
  111. # 标题最短长度
  112. TITLE_MIN_LENGTH = 15