query.json 2.1 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455
  1. {
  2. "crawlWorkCompleted": {
  3. "enable": true,
  4. "cron": "50 50 22 * * ?",
  5. "count": 2000.0,
  6. "excludeAttachment": [],
  7. "excludeSite": [],
  8. "maxAttachmentSize": 5242880.0,
  9. "###enable": "是否启用###",
  10. "###cron": "定时cron表达式.###",
  11. "###count": "每次处理的数量,默认每小时处理所以默认为2000,同时每次将重爬最旧的25%,按时间轮询25%.###",
  12. "###excludeAttachment": "忽略附件名称.###",
  13. "###excludeSite": "忽略附件位置.###",
  14. "###maxAttachmentSize": "最大附件大小.###"
  15. },
  16. "crawlWork": {
  17. "enable": true,
  18. "cron": "40 40 7-21 * * ?",
  19. "count": 100.0,
  20. "excludeAttachment": [],
  21. "excludeSite": [],
  22. "maxAttachmentSize": 5242880.0,
  23. "###enable": "是否启用###",
  24. "###cron": "定时cron表达式.###",
  25. "###count": "每次处理的数量,默认每小时处理所以默认为100,同时每次将重爬最旧的50%,按时间轮询50%.###",
  26. "###excludeAttachment": "忽略附件名称.###",
  27. "###excludeSite": "忽略附件位置.###",
  28. "###maxAttachmentSize": "最大附件大小.###"
  29. },
  30. "crawlCms": {
  31. "enable": true,
  32. "cron": "30 30 7-21 * * ?",
  33. "count": 100.0,
  34. "excludeAttachment": [],
  35. "maxAttachmentSize": 5242880.0,
  36. "###enable": "是否启用###",
  37. "###cron": "定时cron表达式.###",
  38. "###count": "每次处理的数量,默认每小时处理所以默认为100,同时每次将重爬最旧的50%,按时间轮询50%.###",
  39. "###excludeAttachment": "忽略附件名称.###",
  40. "###maxAttachmentSize": "最大附件大小.###"
  41. },
  42. "extractOffice": true,
  43. "extractPdf": true,
  44. "extractText": true,
  45. "extractImage": false,
  46. "tessLanguage": "chi_sim",
  47. "###crawlWorkCompleted": "已完成工作收集器设置.###",
  48. "###crawlWork": "工作收集器设置.###",
  49. "###crawlCms": "内容管理收集器设置.###",
  50. "###extractOffice": "抽取office中的文本.###",
  51. "###extractPdf": "抽取pdf中的文本.###",
  52. "###extractText": "抽取文本中的文本.###",
  53. "###extractImage": "抽取图像中的文本.###",
  54. "###tessLanguage": "tess使用语言.###"
  55. }