View Javadoc
   package net.sf.flock.parser;
   
   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;
  import org.jdom.output.XMLOutputter;
  
  public class Rss10Parser implements FeedParserI {
  
  	private final static Logger LOGGER = LogManager.getLogger(Rss10Parser.class);
  
  	/***
  	 * @see net.sf.flock.parser.FeedParserI#isSuitable(Document)
  	 */
  	public boolean isSuitable(Document doc) {
  		Element root = doc.getRootElement();
  		
  		LOGGER.debug("root element "+root.getName());
  		
  		Namespace defNS = ParserUtil.getDefaultNS(root);
  		if (defNS == null) {
              Namespace rss10 = ParserUtil.getNamespaceFromURI(root,"http://purl.org/rss/1.0/");
  			if (rss10==null) 
                  return false;
  		}
  		if ("RDF".equals(root.getName())) {
  			return true;	
  		}
  		return false;
  	}
  
  	/***
  	 * @see net.sf.flock.parser.FeedParserI#parse(FeedFactoryI, Document)
  	 */
  	public FeedI parse(SubscriptionInfoI subscriptionInfoI,FeedFactoryI feedFactory, Document doc) throws FlockResourceException {
  
  		LOGGER.debug("start parsing.");
  
  		// Get the root element (must be rss)
  		Element root = doc.getRootElement();
  		Namespace defNS = ParserUtil.getDefaultNS(root);
  		if (defNS == null) {
              defNS = ParserUtil.getNamespaceFromURI(root,"http://purl.org/rss/1.0/");
              if (defNS==null) 
      			throw new FlockResourceException("No default namespace found.");
  		}
  		Namespace dcNS = ParserUtil.getNamespace(root, "dc");
  		// fall back to default name space
  		if (dcNS == null) {
  			dcNS = defNS;
  		}
  
  		FeedI feed = feedFactory.createFeed(subscriptionInfoI);
  
  		// Get the channel element (only one occurs)
  		Element channel = root.getChild("channel", defNS);
  
  		// title element
  		feed.setTitle( HTMLUtil.unescape( channel.getChildTextTrim("title", defNS) ) );
  
  		// description element
  		// !!! feed.setDescription(channel.getChildTextTrim("description", defNS));
  		
  		// link element
  		String link = channel.getChildTextTrim("link", defNS);
  		if ((link==null) || (link.length()==0)) {
  
  			for (Iterator iterator = channel.getAttributes().iterator();iterator.hasNext();) {
  				Attribute attr = (Attribute) iterator.next();
  				if ("about".equals(attr.getName())) {
  					LOGGER.debug("channel.about attribute found:"+attr.getNamespace()+":"+attr.getName()+" = "+attr.getValue());
  					link = attr.getValue();
  					break;
  				}
  				LOGGER.debug("channel attribute :"+attr.getNamespace()+":"+attr.getName()+" = "+attr.getValue());
  			}
  		}
  		feed.setSite(ParserUtil.getURL(link));
  
  		/*
  		// !!!
  
  		// creator element
  		Element creator = channel.getChild("creator", dcNS);
  		if (creator != null) {
  			feed.setCreator(creator.getTextTrim());
  		}
 
 		// publisher element
 		Element publisher = channel.getChild("publisher", dcNS);
 		if (publisher != null) {
 			feed.setPublisher(publisher.getTextTrim());
 		}
 
 		// language element
 		Element language = channel.getChild("language", dcNS);
 		if (language != null) {
 			feed.setLanguage(language.getTextTrim());
 		}
 
 		// rights element
 		Element copyright = channel.getChild("copyright", defNS);
 		if (copyright != null) {
 			feed.setCopyright(copyright.getTextTrim());
 		}
 		*/
 
 		long parseDate = System.currentTimeMillis();		
 
 		// item elements
 		List items = root.getChildren("item", defNS);
 		Iterator i = items.iterator();
 		while (i.hasNext()) {
 			Element item = (Element) i.next();
 			// get title element
 			Element elTitle = item.getChild("title", defNS);
 			String strTitle = "<No Title>";
 			if (elTitle != null) {
 				strTitle = HTMLUtil.unescape( elTitle.getTextTrim() );
 			}
 
 			LOGGER.debug("Item element found (" + strTitle + ").");
 
 			// get link element
 			Element elLink = item.getChild("link", defNS);
 			String strLink = "";
 			if (elLink != null) {
 				strLink = elLink.getTextTrim();
 			}
 
 			// get description element
 			Element elDesc = item.getChild("description", dcNS);
 			if (elDesc==null) {
 				// try to get it w/o namespace
 				elDesc = item.getChild("description", defNS);	
 				LOGGER.debug("description without namespace is "+ParserUtil.elementValue(elDesc,20));
 			} else {
 				LOGGER.debug("description with namespace is not null");
 			}
 
 			if (elDesc==null) {
 				elDesc = item.getChild("subtitle", defNS);
 				LOGGER.debug("subtitle is "+ParserUtil.elementValue(elDesc,20));
 			}			
 			String strDesc = "";
 			if (elDesc != null) {
 				
 				strDesc = elDesc.getTextTrim();
 			
 				if (strDesc.length()==0) {
 					LOGGER.debug("text empty, use XMLOutputter");
 					XMLOutputter output = new XMLOutputter("",false);
 					strDesc = output.outputString(elDesc);
 				}
 			}
 
 			// generate new RSS item (link to article)
 
 			/*
 			article.setFound(dateParsed);
 
 			// get creator element
 			Element elCreator = item.getChild("creator", dcNS);
 			if (elCreator != null) {
 				article.setCreator(elCreator.getTextTrim());
 			}
 			// get subject element
 			Element elSubject = item.getChild("subject", dcNS);
 			if (elSubject != null) {
 				// !!! Mulitple subject elements not handled currently
 				article.setSubject(elSubject.getTextTrim());
 			}
 			*/
 
 			// get date element
 			Date creationTime = ParserUtil.parseDate( item.getChild("date", dcNS) );
 
 			// TODO: this is bullshit :)
 			if (creationTime==null) {
 				creationTime = new Date(parseDate--);
 			}
 
 			feed.newItem(creationTime, strTitle, strDesc, ParserUtil.getURL(strLink));
 		}
 
 		/*
 		// image element
 		Element image = root.getChild("image", defNS);
 		if (image != null) {
 		  ImageIF rssImage =
 		    cBuilder.makeImage(image.getChildTextTrim("title", defNS),
 		                       getURL(image.getChildTextTrim("url", defNS)));
 		  Element imgWidth = image.getChild("width", defNS);
 		  if (imgWidth != null) {
 		    try {
 		      rssImage.setWidth(Integer.parseInt(imgWidth.getTextTrim()));
 		    } catch (NumberFormatException e) {
 		      logger.warn(e);
 		    }
 		  }
 		  Element imgHeight = image.getChild("height", defNS);
 		  if (imgHeight != null) {
 		    try {
 		      rssImage.setHeight(Integer.parseInt(imgHeight.getTextTrim()));
 		    } catch (NumberFormatException e) {
 		      logger.warn(e);
 		    }          
 		  }
 		  Element imgDescr = image.getChild("description", dcNS);
 		  if (imgDescr != null) {
 		    rssImage.setDescription(imgDescr.getTextTrim());
 		  }
 		  feed.setImage(rssImage);
 		}
 
 		// textinput element
 		Element txtinp = root.getChild("textinput", defNS);
 		if (txtinp != null) {
 		  String tiTitle = null;
 		  if (txtinp.getChild("title", defNS) != null) {
 		    tiTitle = txtinp.getChild("title", defNS).getTextTrim();
 		  }
 		  String tiDescr = null;
 		  if (txtinp.getChild("description", dcNS) != null) {
 		    tiDescr = txtinp.getChild("description", dcNS).getTextTrim();
 		  }
 		  URL tiLink = null;
 		  if (txtinp.getChild("link", defNS) != null) {
 		    tiLink = getURL(txtinp.getChild("link", defNS).getTextTrim());
 		  }
 		  TextInputIF rssTextInput =
 		    cBuilder.makeTextInput(tiTitle, tiDescr, tiLink);
 		  feed.setTextInput(rssTextInput);
 		}
 		*/
 
 		return feed;
 	}
 
 }