{ "crawlWorkCompleted": { "enable": true, "cron": "50 50 21 * * ?", "count": 500.0, "excludeAttachment": [], "excludeSite": [], "maxAttachmentSize": 5242880.0, "###enable": "是否启用###", "###cron": "定时cron表达式.###", "###count": "每次处理的数量,默认每小时处理所以默认为500,同时每次将重爬最旧的25%,按时间轮询25%.###", "###excludeAttachment": "忽略附件名称.###", "###excludeSite": "忽略附件位置.###", "###maxAttachmentSize": "最大附件大小.###" }, "crawlWork": { "enable": true, "cron": "40 40 10,12,14,16 * * ?", "count": 50.0, "excludeAttachment": [], "excludeSite": [], "maxAttachmentSize": 5242880.0, "###enable": "是否启用###", "###cron": "定时cron表达式.###", "###count": "每次处理的数量,默认每小时处理所以默认为50,同时每次将重爬最旧的50%,按时间轮询50%.###", "###excludeAttachment": "忽略附件名称.###", "###excludeSite": "忽略附件位置.###", "###maxAttachmentSize": "最大附件大小.###" }, "crawlCms": { "enable": true, "cron": "30 30 9,12,15,18 * * ?", "count": 30.0, "excludeAttachment": [], "maxAttachmentSize": 5242880.0, "###enable": "是否启用###", "###cron": "定时cron表达式.###", "###count": "每次处理的数量,默认每小时处理所以默认为30,同时每次将重爬最旧的50%,按时间轮询50%.###", "###excludeAttachment": "忽略附件名称.###", "###maxAttachmentSize": "最大附件大小.###" }, "extractOffice": true, "extractPdf": true, "extractText": true, "extractImage": false, "tessLanguage": "chi_sim", "###crawlWorkCompleted": "已完成工作收集器设置.###", "###crawlWork": "工作收集器设置.###", "###crawlCms": "内容管理收集器设置.###", "###extractOffice": "抽取office中的文本.###", "###extractPdf": "抽取pdf中的文本.###", "###extractText": "抽取文本中的文本.###", "###extractImage": "抽取图像中的文本.###", "###tessLanguage": "tess使用语言.###" }