__init__.py 8.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451
  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. # 统计周期
  116. STAT_PERIOD = 30
  117. class WeixinVideoCrawlerConst:
  118. """
  119. 微信视频抓取常量配置
  120. """
  121. # 账号抓取状态
  122. ACCOUNT_CRAWL_STATUS = 1
  123. ACCOUNT_DO_NOT_CRAWL_STATUS = 0
  124. # 默认最早抓取时间戳(2024-01-01)
  125. DEFAULT_TIMESTAMP = 1704038400
  126. # 搜索爬虫最大页数
  127. MAX_SEARCH_PAGE_NUM = 10
  128. # 抓取每一页的等待时间
  129. SLEEP_SECONDS = 5
  130. # 种子标题最低阅读均值倍数
  131. READ_AVG_MULTIPLE = 1.3
  132. # 种子标题最低阅读量
  133. MIN_READ_COUNT = 2000
  134. # 获取种子标题的统计周期
  135. STAT_PERIOD = 7 * 24 * 60 * 60
  136. # 接口请求成功code
  137. REQUEST_SUCCESS = 0
  138. PUBLISHED_ILLEGAL_TITLE_CODE = 1015
  139. # 是否需要扫描查询源账号
  140. NEED_SCAN_SOURCE_ACCOUNT = 1
  141. DO_NOT_NEED_SOURCE_ACCOUNT = 0
  142. # 视频审核状态长文库
  143. VIDEO_AUDIT_INIT_STATUS = 0
  144. VIDEO_AUDIT_SUCCESS_STATUS = 1
  145. VIDEO_AUDIT_FAIL_STATUS = 2
  146. VIDEO_TITLE_GENERATE_FAIL_STATUS = 4
  147. VIDEO_AUDIT_PROCESSING_STATUS = -1
  148. # 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
  149. PQ_AUDIT_PROCESSING_STATUS = 1
  150. PQ_AUDIT_FAIL_STATUS = 2
  151. PQ_AUDIT_WAIT_STATUS = 3
  152. PQ_AUDIT_SELF_VISIBLE_STATUS = 4
  153. PQ_AUDIT_SUCCESS_STATUS = 5
  154. # 默认账号
  155. DEFAULT_ACCOUNT_UID = 76862180
  156. # 每天发送的审核视频数量
  157. MAX_VIDEO_NUM = 1000
  158. # 单次发布视频审核量
  159. MAX_VIDEO_NUM_PER_PUBLISH = 350
  160. # 标题状态
  161. TITLE_DEFAULT_STATUS = 0
  162. TITLE_EXIT_STATUS = 1
  163. TITLE_FESTIVAL_STATUS = 2
  164. TITLE_SHORT_STATUS = 3
  165. # 标题最短长度
  166. TITLE_MIN_LENGTH = 15
  167. # safe score
  168. TITLE_SAFE_SCORE_THRESHOLD = 7
  169. class UpdateMiniProgramDetailConst(updatePublishedMsgTaskConst):
  170. """
  171. 更新小程序详情常量配置
  172. """
  173. # 账号联想
  174. class AccountAssociationTaskConst:
  175. """
  176. 账号联想任务常量配置
  177. """
  178. # 获取种子标题的统计周期
  179. STAT_PERIOD = 7 * 24 * 60 * 60
  180. # 阅读均值阈值
  181. READ_AVG_MULTIPLE = 1.3
  182. # 最小阅读量
  183. MIN_READ_COUNT = 2000
  184. # 种子数量限制
  185. SEED_TITLE_LIMIT = 100
  186. # 从aigc获取文章
  187. class ArticleCollectorConst:
  188. """
  189. 文章采集任务常量配置
  190. """
  191. # 发送方式
  192. # 手动推送
  193. MANUAL_PUSH = 1
  194. # 自动群发
  195. BULK_AUTO_PUSH = 2
  196. # 无限流推送
  197. UNLIMITED_PUSH = 3
  198. # 文章状态
  199. # 初始状态
  200. INIT_STATUS = 0
  201. # 成功状态
  202. SUCCESS_STATUS = 1
  203. # 失败状态
  204. FAIL_STATUS = -1
  205. # 发布状态
  206. PUBLISHED_STATUS = 2
  207. # 爬虫接口
  208. ARTICLE_ILLEGAL_CODE = 25012
  209. ARTICLE_DELETE_CODE = 25005
  210. ARTICLE_SUCCESS_CODE = 0
  211. ARTICLE_UNKNOWN_CODE = 10000
  212. class BaiduVideoCrawlerConst:
  213. """
  214. const for baidu video crawler
  215. """
  216. # account status
  217. BAIDU_ACCOUNT_GOOD_STATUS = 1
  218. BAIDU_ACCOUNT_BAD_STATUS = 0
  219. # earliest cursor, 2024-01-01 00:00:00
  220. DEFAULT_CURSOR = 17040384000000
  221. # no source account
  222. NO_SOURCE_ACCOUNT_STATUS = 0
  223. # timestamp To Cursor
  224. TIMESTAMP_TO_CURSOR = 10000
  225. # local path dir
  226. LOCAL_PATH_DIR = "static"
  227. class TitleRewriteTaskConst:
  228. """
  229. title rewrite task const
  230. """
  231. # title rewrite status
  232. TITLE_REWRITE_INIT_STATUS = 0
  233. TITLE_REWRITE_SUCCESS_STATUS = 1
  234. TITLE_REWRITE_FAIL_STATUS = 99
  235. TITLE_REWRITE_LOCK_STATUS = 101
  236. # article status
  237. ARTICLE_AUDIT_PASSED_STATUS = 1
  238. ARTICLE_POSITIVE_STATUS = 0
  239. # title useful status
  240. TITLE_USEFUL_STATUS = 1
  241. # prompt version
  242. PROMPT_VERSION = "xx_250228" # 信欣2025-02-28提供
  243. # block expire time 1h
  244. TITLE_REWRITE_LOCK_TIME = 60 * 60
  245. class ChannelVideoCrawlerConst:
  246. """
  247. const for baidu video crawler
  248. """
  249. # account status
  250. CHANNEL_ACCOUNT_GOOD_STATUS = 1
  251. CHANNEL_ACCOUNT_BAD_STATUS = 0
  252. # earliest cursor, 2024-01-01 00:00:00
  253. DEFAULT_CURSOR = 1704038400
  254. # no source account
  255. NO_SOURCE_ACCOUNT_STATUS = 0
  256. # local path dir
  257. LOCAL_PATH_DIR = "static"
  258. # title length min
  259. MIN_TITLE_LENGTH = 10
  260. # max video length(second)
  261. MAX_VIDEO_LENGTH = 600
  262. # sleep second
  263. SLEEP_SECOND = 2
  264. class ToutiaoVideoCrawlerConst:
  265. """
  266. const for toutiao video crawler
  267. """
  268. # platform
  269. PLATFORM = "toutiao"
  270. # account status
  271. TOUTIAO_ACCOUNT_GOOD_STATUS = 1
  272. TOUTIAO_ACCOUNT_BAD_STATUS = 0
  273. # earliest cursor, 2021-01-01 00:00:00
  274. DEFAULT_CURSOR = 1609430400
  275. # no source account
  276. NO_SOURCE_ACCOUNT_STATUS = 0
  277. # title length min
  278. MIN_TITLE_LENGTH = 10
  279. # max video length(second)
  280. MAX_VIDEO_LENGTH = 600
  281. # sleep second
  282. SLEEP_SECOND = 3
  283. class SohuVideoCrawlerConst:
  284. """
  285. const for sohu video crawler
  286. """
  287. # platform
  288. PLATFORM = "sohu"
  289. # account status
  290. GET_RECOMMEND_INIT_STATUS = 0
  291. GET_RECOMMEND_SUCCESS_STATUS = 1
  292. GET_RECOMMEND_FAIL_STATUS = 99
  293. # title length min
  294. MIN_TITLE_LENGTH = 10
  295. # max video length(second)
  296. MAX_VIDEO_LENGTH = 600
  297. # sleep second
  298. SLEEP_SECOND = 3
  299. # 获取推荐的最低相关性分
  300. GET_RECOMMEND_THRESHOLD_SCORE = 0.6
  301. # 审核状态
  302. AUDIT_SUCCESS_STATUS = 1
  303. # 视频状态
  304. VIDEO_NOT_BAD_STATUS = 0
  305. # PAGE_LIST
  306. PAGE_LIST = [i for i in range(1, 8)]
  307. class SingleVideoPoolPublishTaskConst:
  308. """
  309. const for single video pool publish task
  310. """
  311. TRANSFORM_INIT_STATUS = 0
  312. TRANSFORM_SUCCESS_STATUS = 1
  313. TRANSFORM_FAIL_STATUS = 99
  314. class GoogleVideoUnderstandTaskConst:
  315. # task batch size
  316. BATCH_SIZE = 100
  317. # task status
  318. INIT_STATUS = 0
  319. PROCESSING_STATUS = 1
  320. SUCCESS_STATUS = 2
  321. FAIL_STATUS = 99
  322. # sleep seconds
  323. SLEEP_SECONDS = 60
  324. # max processing time
  325. MAX_PROCESSING_TIME = 3600
  326. # task info
  327. TABLE_NAME = "long_articles_new_video_cover"
  328. TASK_NAME = "extract_video_best_frame_as_cover"
  329. DIR_NAME = "static"
  330. POOL_SIZE = 15