View Javadoc
   package net.sf.flock.parser;
   
   import java.text.SimpleDateFormat;
   import java.util.Date;
   import java.util.Iterator;
   import java.util.List;
   
   import net.sf.flock.FeedFactoryI;
   import net.sf.flock.FeedI;
  import net.sf.flock.FlockResourceException;
  import net.sf.flock.SubscriptionInfoI;
  
  import org.apache.log4j.LogManager;
  import org.apache.log4j.Logger;
  import org.jdom.Attribute;
  import org.jdom.Document;
  import org.jdom.Element;
  import org.jdom.Namespace;
  
  public class Rss20Parser implements FeedParserI {
  
  	private final static Logger LOGGER = LogManager.getLogger(Rss20Parser.class);
  
  	private SimpleDateFormat timeFormat = new SimpleDateFormat("yyyy.MM.dd H:mm");
  
  	/***
  	 * @see net.sf.flock.parser.FeedParserI#isSuitable(Document)
  	 */
  	public boolean isSuitable(Document doc) {
  		Element root = doc.getRootElement();
  		if (!"rss".equals(root.getName())) {
  			return false;
  		}
  		Attribute rssVersion = root.getAttribute("version");
  		if (rssVersion==null || rssVersion.getValue()==null) {
  			return false;
  		}
  		LOGGER.debug("Found RSS version " + rssVersion.getValue());
  		return rssVersion.getValue().startsWith("2.0");
  	}
  
  	/***
  	 * @see net.sf.flock.parser.FeedParserI#parse(FeedFactoryI,Document)
  	 */
  	public FeedI parse(SubscriptionInfoI subscriptionInfoI,FeedFactoryI feedFactory, Document doc) throws FlockResourceException {
  
  		Date dateParsed = new Date();
  		LOGGER.debug("start parsing.");
  
  		Element root = doc.getRootElement();
  
  		// Get the channel element (only one occurs)
  		Element channel = root.getChild("channel");
  
  		// 1 title element
  		FeedI feed = feedFactory.createFeed(subscriptionInfoI);
  
  		feed.setTitle( HTMLUtil.unescape( channel.getChildTextTrim("title") ) );
  
  		// 1 description element
  		// !!! feed.setDescription(channel.getChildTextTrim("description"));
  
  		// 1 link element
  		feed.setSite(ParserUtil.getURL(channel.getChildTextTrim("link")));
  
  		// 1 language element
  		// !!! feed.setLanguage(channel.getChildTextTrim("language"));
  
  		// 1..n item elements
  		List items = channel.getChildren("item");
  		Iterator i = items.iterator();
  		while (i.hasNext()) {
  			Element item = (Element) i.next();
  
  			// get description element
  			Element elDesc = item.getChild("description");
  			String strDesc = "";
  			if (elDesc != null) {
  				strDesc = elDesc.getTextTrim();
  			}
  
  			// get link element
  			Element elLink = item.getChild("link");
  			String strLink = "";
  			if (elLink != null) {
  				strLink = elLink.getTextTrim();
  				LOGGER.debug("found link '"+strLink+"'");
  			} else {
  				// try 'guid' 
  				//  ... http://radio.weblogs.com/0112015/rss.xml
  				elLink = item.getChild("guid");
  				if (elLink != null) {
  					strLink = elLink.getTextTrim();
  					LOGGER.debug("found guid '"+strLink+"'");
  				} else {
  					LOGGER.debug("no link/guid");
  				}
  				
  			}
 			
 			
 
 			Date pubDate = ParserUtil.parseDate(item.getChild("pubDate"));
 
 			if (pubDate==null) {
 				// try dc namespace
 				Namespace dcNS = ParserUtil.getNamespace(root, "dc");
 				if (dcNS!=null) {
 					pubDate = ParserUtil.parseDate( item.getChild("date", dcNS) );
 				}
 			}
 
 			LOGGER.debug("date :"+pubDate);
 
 
 			Date creationTime = pubDate!=null ? pubDate : dateParsed;
 			// !!! item.setFound(dateParsed);
 
 			Element elTitle = item.getChild("title");
 			LOGGER.debug("title '"+(elTitle==null ? "null" : elTitle.getTextTrim()) +"'");
 			String title =
 				elTitle!=null ? HTMLUtil.unescape( elTitle.getTextTrim() ) :
 				feed.getTitle() + " " + this.timeFormat.format(creationTime);
 			
 			feed.newItem(creationTime, title, strDesc, ParserUtil.getURL(strLink));
 		}
 
 		// 0..1 rating element
 
 		// 0..1 image element
 		/*
 		Element image = channel.getChild("image");
 		if (image != null) {
 		  ImageIF rssImage =
 		    cBuilder.makeImage(image.getChildTextTrim("title"),
 		                       getURL(image.getChildTextTrim("url")));
 		  Element imgWidth = image.getChild("width");
 		  if (imgWidth != null) {
 		    try {
 		      rssImage.setWidth(Integer.parseInt(imgWidth.getTextTrim()));
 		    } catch (NumberFormatException e) {
 		      logger.warn(e);
 		    }
 		  }
 		  Element imgHeight = image.getChild("height");
 		  if (imgHeight != null) {
 		    try {
 		      rssImage.setHeight(Integer.parseInt(imgHeight.getTextTrim()));
 		    } catch (NumberFormatException e) {
 		      logger.warn(e);
 		    }          
 		  }
 		  Element imgDescr = image.getChild("description");
 		  if (imgDescr != null) {
 		    rssImage.setDescription(imgDescr.getTextTrim());
 		  }
 		  chnl.setImage(rssImage);
 		}
 		
 		// 0..1 textinput element
 		Element txtinp = channel.getChild("textinput");
 		if (txtinp != null) {
 		  TextInputIF rssTextInput =
 		    cBuilder.makeTextInput(txtinp.getChild("title").getTextTrim(),
 		                           txtinp.getChild("description").getTextTrim(),
 		                           getURL(txtinp.getChild("link").getTextTrim()));
 		  chnl.setTextInput(rssTextInput);
 		}
 		
 		// 0..1 copyright element
 		Element copyright = channel.getChild("copyright");
 		if (copyright != null) {
 		  chnl.setCopyright(copyright.getTextTrim());
 		}
 		
 		// 0..1 pubDate element
 		Element pubDate = channel.getChild("pubDate");
 		// if (pubDate != null) {
 		//   chnl.setPubDate(pubDate.getTextTrim());
 		// }
 		
 		// 0..1 lastBuildDate element
 		Element lastBuildDate = channel.getChild("lastBuildDate");
 		// if (lastBuildDate != null) {
 		//   chnl.setLastBuildDate(lastBuildDate.getTextTrim());
 		// }
 		
 		// 0..1 docs element
 		
 		// 0..1 managingEditor element
 		Element managingEditor = channel.getChild("managingEditor");
 		if (managingEditor != null) {
 		  chnl.setCreator(managingEditor.getTextTrim());
 		}
 		
 		// 0..1 webMaster element
 		Element webMaster = channel.getChild("webMaster");
 		if (webMaster != null) {
 		  chnl.setPublisher(webMaster.getTextTrim());
 		}
 		
 		// 0..1 skipHours element
 		// 0..1 skipDays element
 		*/
 		return feed;
 	}
 
 }