Distributed File Input

TODO CODE

public class SourceFileConfigPojo {
	//Pipeline fields:
	private String url = null; // (HDFS path - note security, relative or absolute (relative to /user/tomcat), path must be /user/tomcat/input/NAME/<path> where NAME must contain the community ids of the shared communities)

	// For all files
	public String type = null; // null == decide based on file type, otherwise "json", "xml", "tika", "Xsv" (eg "tsv", "csv")
	public String pathInclude = null; // optional regex (files only) - must match entire path
	public String pathExclude = null; // optional regex (files and directories) - must match entire path, note directories end with "/"
	public String renameAfterParse = null; // If "" deletes the file after processing; otherwise if non-null renames to specified string with escapes $path and $name
	
	//for xml/json/csv files
	public List<String> XmlRootLevelValues = null;
	public List<String> XmlIgnoreValues = null;
	public String XmlSourceName = null;
	public String XmlPrimaryKey = null;
	public Boolean XmlPreserveCase = null; // (default: false)
	public String XmlAttributePrefix = null; // (default: null - if enabled, attributes are converted into tags with this prefix)

}