__init__.py 7.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370
  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. # 默认粉丝
  86. DEFAULT_FANS = 0
  87. # 最低粉丝量
  88. MIN_FANS = 1000
  89. class UpdateAccountReadAvgTaskConst:
  90. """
  91. 更新账号阅读均值常量配置
  92. """
  93. # 投流账号
  94. TOULIU_ACCOUNTS = {
  95. 'gh_93e00e187787',
  96. 'gh_ac43e43b253b',
  97. 'gh_68e7fdc09fe4',
  98. 'gh_77f36c109fb1',
  99. 'gh_b181786a6c8c',
  100. 'gh_1ee2e1b39ccf',
  101. 'gh_d3f039c9db2b'
  102. }
  103. # 发文模式
  104. ARTICLES_DAILY = 1
  105. TOULIU = 2
  106. # 默认粉丝
  107. DEFAULT_FANS = 0
  108. # index list
  109. ARTICLE_INDEX_LIST = [1, 2, 3, 4, 5, 6, 7, 8]
  110. # 默认点赞
  111. DEFAULT_LIKE = 0
  112. # 状态
  113. USING_STATUS = 1
  114. NOT_USING_STATUS = 0
  115. class WeixinVideoCrawlerConst:
  116. """
  117. 微信视频抓取常量配置
  118. """
  119. # 账号抓取状态
  120. ACCOUNT_CRAWL_STATUS = 1
  121. ACCOUNT_DO_NOT_CRAWL_STATUS = 0
  122. # 默认最早抓取时间戳(2024-01-01)
  123. DEFAULT_TIMESTAMP = 1704038400
  124. # 搜索爬虫最大页数
  125. MAX_SEARCH_PAGE_NUM = 10
  126. # 抓取每一页的等待时间
  127. SLEEP_SECONDS = 5
  128. # 种子标题最低阅读均值倍数
  129. READ_AVG_MULTIPLE = 1.3
  130. # 种子标题最低阅读量
  131. MIN_READ_COUNT = 2000
  132. # 获取种子标题的统计周期
  133. STAT_PERIOD = 7 * 24 * 60 * 60
  134. # 接口请求成功code
  135. REQUEST_SUCCESS = 0
  136. PUBLISHED_ILLEGAL_TITLE_CODE = 1015
  137. # 是否需要扫描查询源账号
  138. NEED_SCAN_SOURCE_ACCOUNT = 1
  139. DO_NOT_NEED_SOURCE_ACCOUNT = 0
  140. # 视频审核状态长文库
  141. VIDEO_AUDIT_INIT_STATUS = 0
  142. VIDEO_AUDIT_SUCCESS_STATUS = 1
  143. VIDEO_AUDIT_FAIL_STATUS = 2
  144. VIDEO_TITLE_GENERATE_FAIL_STATUS = 4
  145. VIDEO_AUDIT_PROCESSING_STATUS = -1
  146. # 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
  147. PQ_AUDIT_PROCESSING_STATUS = 1
  148. PQ_AUDIT_FAIL_STATUS = 2
  149. PQ_AUDIT_WAIT_STATUS = 3
  150. PQ_AUDIT_SELF_VISIBLE_STATUS = 4
  151. PQ_AUDIT_SUCCESS_STATUS = 5
  152. # 默认账号
  153. DEFAULT_ACCOUNT_UID = 76862180
  154. # 每天发送的审核视频数量
  155. MAX_VIDEO_NUM = 1000
  156. # 标题状态
  157. TITLE_DEFAULT_STATUS = 0
  158. TITLE_EXIT_STATUS = 1
  159. TITLE_FESTIVAL_STATUS = 2
  160. TITLE_SHORT_STATUS = 3
  161. # 标题最短长度
  162. TITLE_MIN_LENGTH = 15
  163. # safe score
  164. TITLE_SAFE_SCORE_THRESHOLD = 7
  165. class UpdateMiniProgramDetailConst(updatePublishedMsgTaskConst):
  166. """
  167. 更新小程序详情常量配置
  168. """
  169. # 账号联想
  170. class AccountAssociationTaskConst:
  171. """
  172. 账号联想任务常量配置
  173. """
  174. # 获取种子标题的统计周期
  175. STAT_PERIOD = 7 * 24 * 60 * 60
  176. # 阅读均值阈值
  177. READ_AVG_MULTIPLE = 1.3
  178. # 最小阅读量
  179. MIN_READ_COUNT = 2000
  180. # 种子数量限制
  181. SEED_TITLE_LIMIT = 100
  182. # 从aigc获取文章
  183. class ArticleCollectorConst:
  184. """
  185. 文章采集任务常量配置
  186. """
  187. # 发送方式
  188. # 手动推送
  189. MANUAL_PUSH = 1
  190. # 自动群发
  191. BULK_AUTO_PUSH = 2
  192. # 无限流推送
  193. UNLIMITED_PUSH = 3
  194. # 文章状态
  195. # 初始状态
  196. INIT_STATUS = 0
  197. # 成功状态
  198. SUCCESS_STATUS = 1
  199. # 失败状态
  200. FAIL_STATUS = -1
  201. # 发布状态
  202. PUBLISHED_STATUS = 2
  203. # 爬虫接口
  204. ARTICLE_ILLEGAL_CODE = 25012
  205. ARTICLE_DELETE_CODE = 25005
  206. ARTICLE_SUCCESS_CODE = 0
  207. ARTICLE_UNKNOWN_CODE = 10000
  208. class BaiduVideoCrawlerConst:
  209. """
  210. const for baidu video crawler
  211. """
  212. # account status
  213. BAIDU_ACCOUNT_GOOD_STATUS = 1
  214. BAIDU_ACCOUNT_BAD_STATUS = 0
  215. # earliest cursor, 2024-01-01 00:00:00
  216. DEFAULT_CURSOR = 17040384000000
  217. # no source account
  218. NO_SOURCE_ACCOUNT_STATUS = 0
  219. # timestamp To Cursor
  220. TIMESTAMP_TO_CURSOR = 10000
  221. # local path dir
  222. LOCAL_PATH_DIR = "static"
  223. class TitleRewriteTaskConst:
  224. """
  225. title rewrite task const
  226. """
  227. # title rewrite status
  228. TITLE_REWRITE_INIT_STATUS = 0
  229. TITLE_REWRITE_SUCCESS_STATUS = 1
  230. TITLE_REWRITE_FAIL_STATUS = 99
  231. TITLE_REWRITE_LOCK_STATUS = 101
  232. # article status
  233. ARTICLE_AUDIT_PASSED_STATUS = 1
  234. ARTICLE_POSITIVE_STATUS = 0
  235. # title useful status
  236. TITLE_USEFUL_STATUS = 1
  237. # prompt version
  238. PROMPT_VERSION = "xx_250228" # 信欣2025-02-28提供
  239. # block expire time 1h
  240. TITLE_REWRITE_LOCK_TIME = 60 * 60
  241. class ChannelVideoCrawlerConst:
  242. """
  243. const for baidu video crawler
  244. """
  245. # account status
  246. CHANNEL_ACCOUNT_GOOD_STATUS = 1
  247. CHANNEL_ACCOUNT_BAD_STATUS = 0
  248. # earliest cursor, 2024-01-01 00:00:00
  249. DEFAULT_CURSOR = 1704038400
  250. # no source account
  251. NO_SOURCE_ACCOUNT_STATUS = 0
  252. # local path dir
  253. LOCAL_PATH_DIR = "static"
  254. # title length min
  255. MIN_TITLE_LENGTH = 10
  256. # max video length(second)
  257. MAX_VIDEO_LENGTH = 600
  258. # sleep second
  259. SLEEP_SECOND = 2
  260. class ToutiaoVideoCrawlerConst:
  261. """
  262. const for toutiao video crawler
  263. """
  264. # platform
  265. PLATFORM = "toutiao"
  266. # account status
  267. TOUTIAO_ACCOUNT_GOOD_STATUS = 1
  268. TOUTIAO_ACCOUNT_BAD_STATUS = 0
  269. # earliest cursor, 2021-01-01 00:00:00
  270. DEFAULT_CURSOR = 1609430400
  271. # no source account
  272. NO_SOURCE_ACCOUNT_STATUS = 0
  273. # title length min
  274. MIN_TITLE_LENGTH = 10
  275. # max video length(second)
  276. MAX_VIDEO_LENGTH = 600
  277. # sleep second
  278. SLEEP_SECOND = 3