core/SMILA.application/configuration/org.eclipse.smila.jobmanager/jobs.json

{
  "jobs":[
    {
      "name":"indexUpdate",
      "workflow":"indexUpdate",
      "parameters":{
        "tempStore":"temp"
      }
    },
    {
      "name":"indexUpdateFeed",
      "workflow":"indexUpdateFeed",
      "parameters":{
        "tempStore":"temp"
      }
    },
    {
      "name":"indexUpdateXml",
      "workflow":"indexUpdateXml",
      "parameters":{
        "tempStore":"temp",
        "xmlSnippetJobName":"indexUpdateXml"
      }
    },
    {
      "name":"crawlFilesystem",
      "workflow":"fileCrawling",
      "parameters":{
        "tempStore":"temp",
        "dataSource":"file",
        "rootFolder":"< EDIT: directory to crawl >",
        "jobToPushTo":"indexUpdate",
        "mapping":{
          "fileContent":"Content",
          "filePath":"Path",
          "fileName":"Filename",
          "fileSize":"Size",
          "fileExtension":"Extension",
          "fileLastModified":"LastModifiedDate"
        }
      }
    },
    {
      "name":"crawlSmilaWiki",
      "workflow":"webCrawling",
      "parameters":{
        "tempStore":"temp",
        "dataSource":"web",
        "jobToPushTo":"indexUpdate",
        "startUrl":"http://wiki.eclipse.org/SMILA",
        "linksPerBulk": 100,
        "filters":{
          "urlPatterns": {
             "include": ["http://wiki\\.eclipse\\.org/SMILA.*",
                "http://wiki\\.eclipse\\.org/Image:.*",
                "http://wiki\\.eclipse\\.org/images/.*"],
             "exclude": [".*\\?.*" ]
          }
        },
        "mapping": {
          "httpCharset": "Charset",
          "httpContenttype": "ContentType",
          "httpLastModified": "LastModifiedDate",
          "httpMimetype": "MimeType",
          "httpSize": "Size",
          "httpUrl": "Url",
          "httpContent": "Content"
        }
      }
    }
  ]
}