Like this article? We recommend
Like this article? We recommend
Like this article? We recommend
A Second Ferret
Now that we have the framework, using another type of ferret is simply a matter of creating another class that implements the Ferret interface and instantiating it within the application. The GoogleSpecificFinder class is shown in Listing 4.
Listing 4 GoogleSpecificFinder Class
package org.chase.ferrets; import com.google.soap.search.GoogleSearch; import com.google.soap.
This class also uses the four Ferret methods, but it doesn't use the same "brute-force" approach that GoogleFinder does. Instead of simply converting the result to a string wholesale, it instead picks out the name and URL of each listing. Also, if the result count is an estimate (and it usually is), GoogleSpecificFinder returns -1 to signal the lack of an exact count.
So, how do we use this new class? It's simple: We just replace GoogleFinder with it in the application, as shown in Listing 5.
Listing 5 Using GoogleSpecificFinder
import org.chase.research.Ferret;
import org.chase.ferrets.GoogleSpecificFinder;
public class ResearchProject {
public static void main(String[] args) {
if (args.length > 0) {
String searchTerm = args[0];
Ferret theFerret = new GoogleSpecificFinder();
System.out.println("Getting Results ...");
theFerret.find(searchTerm);
int numResults = theFerret.getNumResults();
...
Everything about the ResearchProject application is exactly as it was before, except that we're using GoogleSpecificFinder to instantiate the Ferret instead of GoogleFinder. The results, however, are very different:
Getting Results ... Result count not available. ========================= First Result: ========================= No Title Available http://www.hyperdrive.com/ ========================= All available results: ========================= No Title Available http://www.hyperdrive.com/ <b>Hyperdrive</b> http://www.hyperdriveuk.com/hyperdrive_/hyperdrive.html <b>Hyperdrive</b> Solutions http://www.hyperdrive.co.uk/ ...
Because the result count was not exact, we get a message rather than the estimate; for each result, we get only the title and the URL, rather than all of the information.
And we barely had to touch the base application.