__init__.py 7.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364
  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. # 常量
  13. ACCOUNT_GOOD_STATUS = 1
  14. # 账号是否每日抓取
  15. ACCOUNT_DAILY_SCRAPE = 1
  16. ACCOUNT_NOT_DAILY_SCRAPE = 0
  17. # 默认值
  18. DEFAULT_VIEW_COUNT = 0
  19. DEFAULT_LIKE_COUNT = 0
  20. DEFAULT_ARTICLE_STATUS = 1
  21. DEFAULT_TIMESTAMP = 1717171200
  22. # 标题sensitivity
  23. TITLE_SENSITIVE = 1
  24. TITLE_NOT_SENSITIVE = 0
  25. # 文章联想深度
  26. ARTICLE_ASSOCIATION_MAX_DEPTH = 4
  27. # 相关分百分位阈值
  28. PERCENT_THRESHOLD = 95
  29. # 相关性分阈值
  30. CORRELATION_THRESHOLD = 0.5
  31. # 阅读量阈值
  32. READ_COUNT_THRESHOLD = 1000
  33. # 阅读均值倍数阈值
  34. READ_AVG_THRESHOLD = 1.3
  35. # 群发类型
  36. BULK_PUBLISH_TYPE = 9
  37. # 种子文章数量
  38. SEED_ARTICLE_LIMIT_NUM = 60
  39. class updatePublishedMsgTaskConst:
  40. """
  41. 更新已发布文章消息常量配置
  42. """
  43. # 爬虫详情接口返回code
  44. ARTICLE_ILLEGAL_CODE = 25012
  45. ARTICLE_DELETE_CODE = 25005
  46. ARTICLE_SUCCESS_CODE = 0
  47. ARTICLE_UNKNOWN_CODE = 10000
  48. # 请求爬虫详情接口状态码
  49. # 记录默认状态
  50. DEFAULT_STATUS = 0
  51. # 请求接口失败状态
  52. REQUEST_FAIL_STATUS = -1
  53. # 文章被删除状态
  54. DELETE_STATUS = -2
  55. # 未知原因无信息返回状态
  56. UNKNOWN_STATUS = -3
  57. # 文章违规状态
  58. ILLEGAL_STATUS = -4
  59. # 公众号类型(订阅号 or 服务号)
  60. # 订阅号
  61. SUBSCRIBE_TYPE_SET = {0, 1}
  62. # 服务号
  63. SERVICE_TYPE = 2
  64. # 监测周期(秒)
  65. MONITOR_PERIOD = 60 * 60 * 24 * 3
  66. # 新号抓文章周期
  67. NEW_ACCOUNT_CRAWL_PERIOD = 60 * 60 * 24 * 30
  68. # 订阅号,抓取失败失败率报警阈值
  69. SUBSCRIBE_FAIL_RATE_THRESHOLD = 0.3
  70. class updateAccountReadRateTaskConst:
  71. """
  72. 更新账号阅读率常量配置
  73. """
  74. # 阅读率统计周期(秒)
  75. STATISTICS_PERIOD = 31 * 24 * 60 * 60
  76. # 一天的秒数
  77. ONE_DAY_IN_SECONDS = 60 * 60 * 24
  78. # 相对变化率阈值
  79. RELATIVE_VALUE_THRESHOLD = 0.1
  80. # 发文类型
  81. UNLIMITED_PUBLISH_TYPE = 10002
  82. BULK_PUBLISH_TYPE = 9
  83. # 文章位置
  84. ARTICLE_INDEX_LIST = [1, 2, 3, 4, 5, 6, 7, 8]
  85. class UpdateAccountReadAvgTaskConst:
  86. """
  87. 更新账号阅读均值常量配置
  88. """
  89. # 投流账号
  90. TOULIU_ACCOUNTS = {
  91. 'gh_93e00e187787',
  92. 'gh_ac43e43b253b',
  93. 'gh_68e7fdc09fe4',
  94. 'gh_77f36c109fb1',
  95. 'gh_b181786a6c8c',
  96. 'gh_1ee2e1b39ccf',
  97. 'gh_d3f039c9db2b'
  98. }
  99. # 发文模式
  100. ARTICLES_DAILY = 1
  101. TOULIU = 2
  102. # 默认粉丝
  103. DEFAULT_FANS = 0
  104. # index list
  105. ARTICLE_INDEX_LIST = [1, 2, 3, 4, 5, 6, 7, 8]
  106. # 默认点赞
  107. DEFAULT_LIKE = 0
  108. # 状态
  109. USING_STATUS = 1
  110. NOT_USING_STATUS = 0
  111. class WeixinVideoCrawlerConst:
  112. """
  113. 微信视频抓取常量配置
  114. """
  115. # 账号抓取状态
  116. ACCOUNT_CRAWL_STATUS = 1
  117. ACCOUNT_DO_NOT_CRAWL_STATUS = 0
  118. # 默认最早抓取时间戳(2024-01-01)
  119. DEFAULT_TIMESTAMP = 1704038400
  120. # 搜索爬虫最大页数
  121. MAX_SEARCH_PAGE_NUM = 10
  122. # 抓取每一页的等待时间
  123. SLEEP_SECONDS = 5
  124. # 种子标题最低阅读均值倍数
  125. READ_AVG_MULTIPLE = 1.3
  126. # 种子标题最低阅读量
  127. MIN_READ_COUNT = 2000
  128. # 获取种子标题的统计周期
  129. STAT_PERIOD = 7 * 24 * 60 * 60
  130. # 接口请求成功code
  131. REQUEST_SUCCESS = 0
  132. PUBLISHED_ILLEGAL_TITLE_CODE = 1015
  133. # 是否需要扫描查询源账号
  134. NEED_SCAN_SOURCE_ACCOUNT = 1
  135. DO_NOT_NEED_SOURCE_ACCOUNT = 0
  136. # 视频审核状态长文库
  137. VIDEO_AUDIT_INIT_STATUS = 0
  138. VIDEO_AUDIT_SUCCESS_STATUS = 1
  139. VIDEO_AUDIT_FAIL_STATUS = 2
  140. VIDEO_TITLE_GENERATE_FAIL_STATUS = 4
  141. VIDEO_AUDIT_PROCESSING_STATUS = -1
  142. # 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
  143. PQ_AUDIT_PROCESSING_STATUS = 1
  144. PQ_AUDIT_FAIL_STATUS = 2
  145. PQ_AUDIT_WAIT_STATUS = 3
  146. PQ_AUDIT_SELF_VISIBLE_STATUS = 4
  147. PQ_AUDIT_SUCCESS_STATUS = 5
  148. # 默认账号
  149. DEFAULT_ACCOUNT_UID = 76862180
  150. # 每天发送的审核视频数量
  151. MAX_VIDEO_NUM = 1000
  152. # 标题状态
  153. TITLE_DEFAULT_STATUS = 0
  154. TITLE_EXIT_STATUS = 1
  155. TITLE_FESTIVAL_STATUS = 2
  156. TITLE_SHORT_STATUS = 3
  157. # 标题最短长度
  158. TITLE_MIN_LENGTH = 15
  159. # safe score
  160. TITLE_SAFE_SCORE_THRESHOLD = 7
  161. class UpdateMiniProgramDetailConst(updatePublishedMsgTaskConst):
  162. """
  163. 更新小程序详情常量配置
  164. """
  165. # 账号联想
  166. class AccountAssociationTaskConst:
  167. """
  168. 账号联想任务常量配置
  169. """
  170. # 获取种子标题的统计周期
  171. STAT_PERIOD = 7 * 24 * 60 * 60
  172. # 阅读均值阈值
  173. READ_AVG_MULTIPLE = 1.3
  174. # 最小阅读量
  175. MIN_READ_COUNT = 2000
  176. # 种子数量限制
  177. SEED_TITLE_LIMIT = 100
  178. # 从aigc获取文章
  179. class ArticleCollectorConst:
  180. """
  181. 文章采集任务常量配置
  182. """
  183. # 发送方式
  184. # 手动推送
  185. MANUAL_PUSH = 1
  186. # 自动群发
  187. BULK_AUTO_PUSH = 2
  188. # 无限流推送
  189. UNLIMITED_PUSH = 3
  190. # 文章状态
  191. # 初始状态
  192. INIT_STATUS = 0
  193. # 成功状态
  194. SUCCESS_STATUS = 1
  195. # 失败状态
  196. FAIL_STATUS = -1
  197. # 发布状态
  198. PUBLISHED_STATUS = 2
  199. # 爬虫接口
  200. ARTICLE_ILLEGAL_CODE = 25012
  201. ARTICLE_DELETE_CODE = 25005
  202. ARTICLE_SUCCESS_CODE = 0
  203. ARTICLE_UNKNOWN_CODE = 10000
  204. class BaiduVideoCrawlerConst:
  205. """
  206. const for baidu video crawler
  207. """
  208. # account status
  209. BAIDU_ACCOUNT_GOOD_STATUS = 1
  210. BAIDU_ACCOUNT_BAD_STATUS = 0
  211. # earliest cursor, 2024-01-01 00:00:00
  212. DEFAULT_CURSOR = 17040384000000
  213. # no source account
  214. NO_SOURCE_ACCOUNT_STATUS = 0
  215. # timestamp To Cursor
  216. TIMESTAMP_TO_CURSOR = 10000
  217. # local path dir
  218. LOCAL_PATH_DIR = "static"
  219. class TitleRewriteTaskConst:
  220. """
  221. title rewrite task const
  222. """
  223. # title rewrite status
  224. TITLE_REWRITE_INIT_STATUS = 0
  225. TITLE_REWRITE_SUCCESS_STATUS = 1
  226. TITLE_REWRITE_FAIL_STATUS = 99
  227. TITLE_REWRITE_LOCK_STATUS = 101
  228. # article status
  229. ARTICLE_AUDIT_PASSED_STATUS = 1
  230. ARTICLE_POSITIVE_STATUS = 0
  231. # title useful status
  232. TITLE_USEFUL_STATUS = 1
  233. # prompt version
  234. PROMPT_VERSION = "xx_250228" # 信欣2025-02-28提供
  235. # block expire time 1h
  236. TITLE_REWRITE_LOCK_TIME = 60 * 60
  237. class ChannelVideoCrawlerConst:
  238. """
  239. const for baidu video crawler
  240. """
  241. # account status
  242. CHANNEL_ACCOUNT_GOOD_STATUS = 1
  243. CHANNEL_ACCOUNT_BAD_STATUS = 0
  244. # earliest cursor, 2024-01-01 00:00:00
  245. DEFAULT_CURSOR = 1704038400
  246. # no source account
  247. NO_SOURCE_ACCOUNT_STATUS = 0
  248. # local path dir
  249. LOCAL_PATH_DIR = "static"
  250. # title length min
  251. MIN_TITLE_LENGTH = 10
  252. # max video length(second)
  253. MAX_VIDEO_LENGTH = 600
  254. # sleep second
  255. SLEEP_SECOND = 2
  256. class ToutiaoVideoCrawlerConst:
  257. """
  258. const for toutiao video crawler
  259. """
  260. # platform
  261. PLATFORM = "toutiao"
  262. # account status
  263. TOUTIAO_ACCOUNT_GOOD_STATUS = 1
  264. TOUTIAO_ACCOUNT_BAD_STATUS = 0
  265. # earliest cursor, 2021-01-01 00:00:00
  266. DEFAULT_CURSOR = 1609430400
  267. # no source account
  268. NO_SOURCE_ACCOUNT_STATUS = 0
  269. # title length min
  270. MIN_TITLE_LENGTH = 10
  271. # max video length(second)
  272. MAX_VIDEO_LENGTH = 600
  273. # sleep second
  274. SLEEP_SECOND = 3