const.py 3.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150
  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_ILLEGAL_CODE = 25012
  18. ARTICLE_DELETE_CODE = 25005
  19. ARTICLE_SUCCESS_CODE = 0
  20. ARTICLE_UNKNOWN_CODE = 10000
  21. # 请求爬虫详情接口状态码
  22. # 记录默认状态
  23. DEFAULT_STATUS = 0
  24. # 请求接口失败状态
  25. REQUEST_FAIL_STATUS = -1
  26. # 文章被删除状态
  27. DELETE_STATUS = -2
  28. # 未知原因无信息返回状态
  29. UNKNOWN_STATUS = -3
  30. # 文章违规状态
  31. ILLEGAL_STATUS = -4
  32. # 公众号类型(订阅号 or 服务号)
  33. # 订阅号
  34. SUBSCRIBE_TYPE_SET = {0, 1}
  35. # 服务号
  36. SERVICE_TYPE = 2
  37. # 监测周期(秒)
  38. MONITOR_PERIOD = 60 * 60 * 24 * 7
  39. class updateAccountReadRateTaskConst:
  40. """
  41. 更新账号阅读率常量配置
  42. """
  43. # 阅读率统计周期(秒)
  44. STATISTICS_PERIOD = 31 * 24 * 60 * 60
  45. # 一天的秒数
  46. ONE_DAY_IN_SECONDS = 60 * 60 * 24
  47. # 相对变化率阈值
  48. RELATIVE_VALUE_THRESHOLD = 0.1
  49. # 发文类型
  50. UNLIMITED_PUBLISH_TYPE = 10002
  51. BULK_PUBLISH_TYPE = 9
  52. # 文章位置
  53. ARTICLE_INDEX_LIST = [1, 2, 3, 4, 5, 6, 7, 8]
  54. class updateAccountReadAvgTaskConst:
  55. """
  56. 更新账号阅读均值常量配置
  57. """
  58. # 投流账号
  59. TOULIU_ACCOUNTS = {
  60. 'gh_93e00e187787',
  61. 'gh_ac43e43b253b',
  62. 'gh_68e7fdc09fe4',
  63. 'gh_77f36c109fb1',
  64. 'gh_b181786a6c8c',
  65. 'gh_1ee2e1b39ccf',
  66. 'gh_d3f039c9db2b'
  67. }
  68. # 发文模式
  69. ARTICLES_DAILY = 1
  70. TOULIU = 2
  71. class WeixinVideoCrawlerConst:
  72. """
  73. 微信视频抓取常量配置
  74. """
  75. # 账号抓取状态
  76. ACCOUNT_CRAWL_STATUS = 1
  77. ACCOUNT_DO_NOT_CRAWL_STATUS = 0
  78. # 默认最早抓取时间戳(2024-01-01)
  79. DEFAULT_TIMESTAMP = 1704038400
  80. # 搜索爬虫最大页数
  81. MAX_SEARCH_PAGE_NUM = 10
  82. # 抓取每一页的等待时间
  83. SLEEP_SECONDS = 5
  84. # 种子标题最低阅读均值倍数
  85. READ_AVG_MULTIPLE = 1.3
  86. # 种子标题最低阅读量
  87. MIN_READ_COUNT = 2000
  88. # 获取种子标题的统计周期
  89. STAT_PERIOD = 7 * 24 * 60 * 60
  90. # 接口请求成功code
  91. REQUEST_SUCCESS = 0
  92. PUBLISHED_ILLEGAL_TITLE_CODE = 1015
  93. # 是否需要扫描查询源账号
  94. NEED_SCAN_SOURCE_ACCOUNT = 1
  95. DO_NOT_NEED_SOURCE_ACCOUNT = 0
  96. # 视频审核状态长文库
  97. VIDEO_AUDIT_INIT_STATUS = 0
  98. VIDEO_AUDIT_SUCCESS_STATUS = 1
  99. VIDEO_AUDIT_FAIL_STATUS = 2
  100. VIDEO_TITLE_GENERATE_FAIL_STATUS = 4
  101. VIDEO_AUDIT_PROCESSING_STATUS = -1
  102. # 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
  103. PQ_AUDIT_PROCESSING_STATUS = 1
  104. PQ_AUDIT_FAIL_STATUS = 2
  105. PQ_AUDIT_WAIT_STATUS = 3
  106. PQ_AUDIT_SELF_VISIBLE_STATUS = 4
  107. PQ_AUDIT_SUCCESS_STATUS = 5
  108. # 默认账号
  109. DEFAULT_ACCOUNT_UID = 76862180
  110. # 每天发送的审核视频数量
  111. MAX_VIDEO_NUM = 500
  112. # 标题状态
  113. TITLE_DEFAULT_STATUS = 0
  114. TITLE_EXIT_STATUS = 1
  115. TITLE_FESTIVAL_STATUS = 2
  116. TITLE_SHORT_STATUS = 3
  117. # 标题最短长度
  118. TITLE_MIN_LENGTH = 15