Creating services using WSProxy

• Allow the service to be accessed using standard Web Service calls
• Allow the service to be deployed in standard containers such as Apache Tomcat
• Provide means to transparently interface the NaradaBrokering system to allow reading and writing data as a stream of messages

WSProxy

Creating a filter requires extending either the cgl.hpsearch.wsproxy.RunnableProxyWebService or the cgl.hpsearch.wsproxy.WrapperProxyWebService class. The description of each of these classes and differences is given here .

Example

import java.io.BufferedReader;
import java.io.BufferedWriter;
import java.io.FileWriter;
import java.io.IOException;
import java.io.InputStream;
import java.io.InputStreamReader;
import java.io.OutputStream;

import org.apache.log4j.Logger;

import cgl.hpsearch.wsproxy.WrapperProxyWebService;

/**
 * @author Harshawardhan Gadgil (hgadgil@grids.ucs.indiana.edu)
 */

public class PIFilterRunner extends WrapperProxyWebService {

	static Logger log = Logger.getLogger("PIFilterRunner");

	private BufferedReader br;
	private BufferedWriter bw;
	private String MinimumMagnitude;

	public void process() throws Exception {

		double MAGNITUDE_THRESHOLD = Double.parseDouble(MinimumMagnitude);
		
		// Step 1: Accumulate data
		String line;

		while (true) {
			try {
				line = br.readLine();
			}
			catch (IOException e) {
				line = null;
			}

			if (line == null) {// EOF reached...
				break;
			}
			else {
				System.out.println("READ:] " + line);
				// We do not write this line to output
				if (line.startsWith("#")) continue;
				if (line.startsWith("SOF")) continue;
				if (line.startsWith("EOF")) {
					// EOF reached... bail out...
					break;
				}

				// A sample line would look like this
				// "1932" "12" "20" "-115.990" "34.545" "3.01"
				line = line.replaceAll("\"\t\"", " ");
				line = line.substring(1, line.indexOf('"', 2));

				// Find the last token in the string. This is the magnitude !!
				String magnitude = line.substring(line.lastIndexOf(" "));
				double dbl = Double.parseDouble(magnitude);

				if (dbl >= MAGNITUDE_THRESHOLD) {
					try {
						bw.write(line + "\n");
						bw.flush();
					}
					catch (IOException e1) {
						log.error("", e1);
						return;
					}
				}
				bw.write(line + "\n");
				bw.flush();
			}

		}

		// MUST close the catalog file writer since we opened the file
		bw.close();

		// My jobz over...
		serviceFinished();
	}

	public void initService(InputStream[] arg0, OutputStream[] arg1) {

		MinimumMagnitude = getProperty("MinimumMagnitude");
		
		// Wrap streams for easy access
		br = new BufferedReader(new InputStreamReader(arg0[0]));
		bw = new BufferedWriter(new OutputStreamWriter(arg1[0]);
	}

	public String[] serviceExceptions() {
		return new String[0];
	}
}
	

NOTES

• Here, the PIFilterRunner service filters incoming data ("1932" "12" "20" "-115.990" "34.545" "3.01") where the individual fields are (year, month, day, longitude, latitude, magnitude) of observations from a GPS database. The input line contains these fields spearated by tab (\t) characters. Depending on the value of the service parameter MinimumMagnitude, the service only outputs those lines which have the value of the last column (magnitude >= MinimumMagnitude).
• Further, the filter service also removes the quotations and the tab chanracters and replaces them with a single space.
• The input data is also marked with special start of file (SOF) and end of file (EOF) which are also removed before writing the final output.
• Further note the initService(InputStream[] arg0, OutputStream[] arg1) function that takes in an array of input and output stream objects. These are automatically mapped to various URIs (Refer here for a list of supported URIs).
  This function also initializes various parameters specific to the service and other things such as wrapping input streams as Buffered streams for easy access to data.
• Finally note the String[] serviceExceptions() function that returns an array of possible execeptions that might be thrown by the service and which should have actions defined in the user's script. If no actions are defined the service simply ignores the error and continues. For most services we simply return an empty array of strings. An example of usage of this feature may be found in the Network Sensor Demo.

(back to Notes)