1 package net.javacoding.jspider.mod.rule; 2 3 import net.javacoding.jspider.api.model.*; 4 import net.javacoding.jspider.core.SpiderContext; 5 import net.javacoding.jspider.core.rule.impl.BaseRuleImpl; 6 import net.javacoding.jspider.core.model.DecisionInternal; 7 8 import java.net.URL ; 9 10 13 public class TextHtmlMimeTypeOnlyRule extends BaseRuleImpl { 14 15 public Decision apply(SpiderContext context, Site currentSite, URL url) { 16 FetchedResource resource = (FetchedResource)context.getStorage().getResourceDAO().getResource(url); 17 String mime = resource.getMime(); 18 Decision decision = new DecisionInternal(Decision.RULE_IGNORE, "mimetype is '" + mime + "' - resource ignored"); 19 20 if ( mime == null ) { 21 decision = new DecisionInternal(Decision.RULE_ACCEPT, "mimetype is null - defaulted to text/html - accepted" ); 22 } else if (mime.toLowerCase().indexOf("text/html") > -1) { decision = new DecisionInternal(Decision.RULE_ACCEPT, "mimetype is '" + mime + "' - resource accepted"); 24 } 25 26 return decision; 27 } 28 } 29 | Popular Tags |