RSS_1_0_Parser


1   //
2   // Informa -- RSS Library for Java
3   // Copyright (c) 2002 by Niko Schmuck
4   //
5   // Niko Schmuck
6   // http://sourceforge.net/projects/informa
7   // mailto:niko_schmuck@users.sourceforge.net
8   //
9   // This library is free software.
10  //
11  // You may redistribute it and/or modify it under the terms of the GNU
12  // Lesser General Public License as published by the Free Software Foundation.
13  //
14  // Version 2.1 of the license should be included with this distribution in
15  // the file LICENSE. If the license is not included with this distribution,
16  // you may find a copy at the FSF web site at 'www.gnu.org' or 'www.fsf.org',
17  // or you may write to the Free Software Foundation, 675 Mass Ave, Cambridge,
18  // MA 02139 USA.
19  //
20  // This library is distributed in the hope that it will be useful,
21  // but WITHOUT ANY WARRANTY; without even the implied waranty of
22  // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
23  // Lesser General Public License for more details.
24  //
25  
26  
27  // $Id: RSS_1_0_Parser.java,v 1.39 2003/11/04 19:09:43 niko_schmuck Exp $
28  
29  package de.nava.informa.parsers;
30  
31  import java.net.URL  ;
32  import java.util.Date  ;
33  import java.util.Iterator  ;
34  import java.util.List  ;
35  
36  import org.apache.commons.logging.Log;
37  import org.apache.commons.logging.LogFactory;
38  
39  import org.jdom.Attribute;
40  import org.jdom.Element;
41  import org.jdom.Namespace;
42  
43  import de.nava.informa.core.ChannelBuilderIF;
44  import de.nava.informa.core.ChannelFormat;
45  import de.nava.informa.core.ChannelIF;
46  import de.nava.informa.core.ImageIF;
47  import de.nava.informa.core.ItemIF;
48  import de.nava.informa.core.ItemSourceIF;
49  import de.nava.informa.core.ParseException;
50  import de.nava.informa.core.TextInputIF;
51  import de.nava.informa.utils.ParserUtils;
52  
53  /**
54   * Parser which reads in document instances according to the RSS 1.0
55   * (RDF) specification and generates a news channel object.
56   *
57   * @author Niko Schmuck
58   */
59  class RSS_1_0_Parser {
60  
61    private static Log logger = LogFactory.getLog(RSS_1_0_Parser.class);
62  
63    static ChannelIF parse(ChannelBuilderIF cBuilder,
64                             Element root) throws ParseException {
65      if (cBuilder == null) {
66        throw new RuntimeException  ("Without builder no channel can " +
67                                   "be created.");
68      }
69      Date   dateParsed = new Date  ();
70      Namespace defNS = ParserUtils.getDefaultNS(root);
71      if (defNS == null) {
72        defNS = Namespace.NO_NAMESPACE;
73        logger.info("No default namespace found.");
74      }
75  
76      // RSS 1.0 Dublin Core Module namespace
77      Namespace dcNS = ParserUtils.getNamespace(root, "dc");
78      // fall back to default name space (for retrieving descriptions)
79      if (dcNS == null) {
80        dcNS = defNS;
81      }
82  
83      // RSS 1.0 Syndication Module namespace
84      Namespace syNS = ParserUtils.getNamespace(root, "sy");
85  
86      // RSS 1.0 Aggregation Module namespace
87      Namespace agNS = ParserUtils.getNamespace(root, "ag");
88  
89      // RSS 1.0 Administration Module namespace
90      Namespace adminNS = ParserUtils.getNamespace(root, "admin");
91  
92      // RSS 1.0 DCTerms Module namespace
93      Namespace dctermsNS = ParserUtils.getNamespace(root, "dcterms");
94  
95      // RSS 1.0 Annotation Module namespace
96      Namespace annotateNS = ParserUtils.getNamespace(root, "annotate");
97  
98      // RSS091 Module namespace
99      Namespace rss091NS = ParserUtils.getNamespace(root, "rss091");
100 
101     // Get the channel element (only one occurs)
102     Element channel = root.getChild("channel", defNS);
103     if (channel == null) {
104       logger.warn("Channel element could not be retrieved from feed.");
105       throw new ParseException("No channel element found in feed.");
106     }
107 
108     // ----------------------- read in channel information
109     // title element
110     ChannelIF chnl =
111       cBuilder.createChannel(channel, channel.getChildTextTrim("title", defNS));
112 
113     // set channel format
114     chnl.setFormat(ChannelFormat.RSS_1_0);
115 
116     // description element
117     chnl.setDescription(channel.getChildTextTrim("description", defNS));
118 
119     // link element
120     chnl.setSite(ParserUtils.getURL(channel.getChildTextTrim("link", defNS)));
121 
122     // creator element
123     Element creator = channel.getChild("creator", dcNS);
124     if (creator == null) {
125       creator = channel.getChild("managingEditor", rss091NS);
126     }
127     if (creator != null) {
128       chnl.setCreator(creator.getTextTrim());
129     }
130 
131     // publisher element
132     String   publisher = channel.getChildTextTrim("publisher", dcNS);
133     if (publisher == null) {
134       Element elErrorReportsTo = channel.getChild("errorReportsTo", adminNS);
135       if (elErrorReportsTo != null) {
136         publisher = elErrorReportsTo.getAttributeValue("resource",
137             ParserUtils.getNamespace(elErrorReportsTo, "rdf"));
138       }
139     }
140     if (publisher == null) {
141       publisher = channel.getChildTextTrim("webMaster", rss091NS);
142     }
143     chnl.setPublisher(publisher);
144 
145     // language element
146     Element language = channel.getChild("language", dcNS);
147     if (language == null) {
148       language = channel.getChild("language", rss091NS);
149     }
150     if (language != null) {
151       chnl.setLanguage(language.getTextTrim());
152     }
153 
154     // rights element
155     Element copyright = channel.getChild("rights", dcNS);
156     if (copyright == null) {
157       copyright = channel.getChild("copyright", rss091NS);
158     }
159     if (copyright != null) {
160       chnl.setCopyright(copyright.getTextTrim());
161     }
162 
163     // 0..1 Rating element
164     Element rating = channel.getChild("rating", rss091NS);
165     if (rating != null) {
166       chnl.setRating(rating.getTextTrim());
167     }
168 
169     // 0..1 Docs element
170     // use namespace URI
171     chnl.setDocs(defNS.getURI());
172 
173     // 0..1 pubDate element
174     Element pubDate = channel.getChild("date", dcNS);
175     if (pubDate == null) {
176       pubDate = channel.getChild("issued", dctermsNS);
177     }
178     if (pubDate == null) {
179       pubDate = channel.getChild("pubdate", rss091NS);
180     }
181     if (pubDate != null) {
182       chnl.setPubDate(ParserUtils.getDate(pubDate.getTextTrim()));
183     }
184 
185     // 0..1 lastBuildDate element
186     Element lastBuildDate = channel.getChild("lastBuildDate");
187     if (lastBuildDate == null) {
188       lastBuildDate = channel.getChild("modified", dctermsNS);
189     }
190     if (lastBuildDate == null) {
191       lastBuildDate = channel.getChild("lastBuildDate", rss091NS);
192     }
193     if (lastBuildDate != null) {
194       chnl.setLastBuildDate(ParserUtils.getDate(lastBuildDate.getTextTrim()));
195     }
196 
197     // RSS 1.0 Administration Module support
198 
199     // 0..1 generator element
200     Element elGenerator = channel.getChild("generatorAgent", adminNS);
201     if (elGenerator != null) {
202       Attribute generator = elGenerator.getAttribute("resource", ParserUtils.getNamespace(elGenerator, "rdf"));
203       if (generator != null) {
204         chnl.setGenerator(generator.getValue());
205       }
206     }
207 
208     // RSS 1.0 Syndication Module support
209 
210     // 0..1 update period element
211     Element updatePeriod = channel.getChild("updatePeriod", syNS);
212     if (updatePeriod != null) {
213       chnl.setUpdatePeriod(updatePeriod.getTextTrim());
214     }
215 
216     // 0..1 update frequency element
217     Element updateFrequency = channel.getChild("updateFrequency", syNS);
218     if (updateFrequency != null) {
219       chnl.setUpdateFrequency((new Integer  (updateFrequency.getTextTrim())).intValue());
220     }
221 
222     // 0..1 update base element
223     Element updateBase = channel.getChild("updateBase", syNS);
224     if (updateBase != null) {
225       chnl.setUpdateBase(ParserUtils.getDate(updateBase.getTextTrim()));
226     }
227 
228     if ((updatePeriod != null) && updateFrequency != null) {
229       int ttl = getTtl(chnl.getUpdatePeriod(), chnl.getUpdateFrequency());
230       chnl.setTtl(ttl);
231     }
232 
233     // item elements
234     List   items = root.getChildren("item", defNS);
235     Iterator   i = items.iterator();
236     while (i.hasNext()) {
237       Element item = (Element) i.next();
238 
239       // get title element
240       Element elTitle = item.getChild("title", defNS);
241       String   strTitle = "<No Title>";
242       if (elTitle != null) {
243         strTitle = elTitle.getTextTrim();
244       }
245       if (logger.isDebugEnabled()) {
246         logger.debug("Item element found (" + strTitle + ").");
247       }
248 
249       // get link element
250       Element elLink = item.getChild("link", defNS);
251       String   strLink = "";
252       if (elLink != null) {
253         strLink = elLink.getTextTrim();
254       }
255 
256       // get description element
257       Element elDesc = item.getChild("description", defNS);
258       if (elDesc == null) {
259         elDesc = item.getChild("description", dcNS);
260       }
261       String   strDesc = "";
262       if (elDesc != null) {
263         strDesc = elDesc.getTextTrim();
264       }
265 
266       // generate new RSS item (link to article)
267       ItemIF rssItem = cBuilder.createItem(item, chnl, strTitle, strDesc,
268                                            ParserUtils.getURL(strLink));
269       rssItem.setFound(dateParsed);
270 
271       // get creator element
272       Element elCreator = item.getChild("creator", dcNS);
273       if (elCreator != null) {
274         rssItem.setCreator(elCreator.getTextTrim());
275       }
276 
277       // get subject element
278       Element elSubject = item.getChild("subject", dcNS);
279       if (elSubject != null) {
280         // TODO: Mulitple subject elements not handled currently
281         rssItem.setSubject(elSubject.getTextTrim());
282       }
283 
284       // get date element
285       Element elDate = item.getChild("date", dcNS);
286       if (elDate != null) {
287         rssItem.setDate(ParserUtils.getDate(elDate.getTextTrim()));
288       }
289 
290       // get source element - default to Aggregation module, then try Dublin Core
291       String   sourceName = null;
292       String   sourceLocation = null;
293       Date   sourceTimestamp = null;
294 
295       Element elSourceURL = item.getChild("sourceURL", agNS);
296       if (elSourceURL == null) { //  No Aggregation module - try Dublin Core
297         elSourceURL = item.getChild("source", dcNS);
298         if (elSourceURL != null) {
299           sourceLocation = elSourceURL.getTextTrim();
300           sourceName = "Source";
301         }
302      } else {  // Aggregation module
303         sourceLocation = elSourceURL.getTextTrim();
304         Element elSourceName = item.getChild("source", agNS);
305         if (elSourceName != null) {
306           sourceName = elSourceName.getTextTrim();
307         }
308         Element elSourceTimestamp = item.getChild("timestamp", agNS);
309         if (elSourceTimestamp != null) {
310           sourceTimestamp = ParserUtils.getDate(elSourceTimestamp.getTextTrim());
311         }
312       }
313 
314       if (sourceLocation != null) {
315         ItemSourceIF itemSource = cBuilder.createItemSource(rssItem, sourceName, sourceLocation, sourceTimestamp);
316         rssItem.setSource(itemSource);
317       }
318 
319       // comments element - use Annotation module
320       Element elReference = item.getChild("reference", annotateNS);
321       if (elReference != null) {
322         Attribute resource = elReference.getAttribute("resource", ParserUtils.getNamespace(elReference, "rdf"));
323         if (resource != null) {
324           URL   resourceURL = ParserUtils.getURL(resource.getValue());
325           if (resourceURL != null) {
326             rssItem.setComments(resourceURL);
327           }
328         }
329       }
330 
331     }
332 
333     // image element
334     Element image = root.getChild("image", defNS);
335     if (image != null) {
336       ImageIF rssImage =
337         cBuilder.createImage(image.getChildTextTrim("title", defNS),
338                            ParserUtils.getURL(image.getChildTextTrim("url", defNS)),
339                            ParserUtils.getURL(image.getChildTextTrim("link", defNS)));
340       Element imgWidth = image.getChild("width", defNS);
341       if (imgWidth != null) {
342         try {
343           rssImage.setWidth(Integer.parseInt(imgWidth.getTextTrim()));
344         } catch (NumberFormatException   e) {
345           logger.warn(e);
346         }
347       }
348       Element imgHeight = image.getChild("height", defNS);
349       if (imgHeight != null) {
350         try {
351           rssImage.setHeight(Integer.parseInt(imgHeight.getTextTrim()));
352         } catch (NumberFormatException   e) {
353           logger.warn(e);
354         }
355       }
356       Element imgDescr = image.getChild("description", defNS);
357       if (imgDescr != null) {
358         rssImage.setDescription(imgDescr.getTextTrim());
359       }
360       chnl.setImage(rssImage);
361     }
362 
363     // textinput element
364     Element txtinp = root.getChild("textinput", defNS);
365     if (txtinp != null) {
366       String   tiTitle = null;
367       if (txtinp.getChild("title", defNS) != null) {
368         tiTitle = txtinp.getChild("title", defNS).getTextTrim();
369       }
370       String   tiDescr = null;
371       if (txtinp.getChild("description", defNS) != null) {
372         tiDescr = txtinp.getChild("description", defNS).getTextTrim();
373       }
374       String   tiName = null;
375       if (txtinp.getChild("name", defNS) != null) {
376         tiName = txtinp.getChild("name", defNS).getTextTrim();
377       }
378       URL   tiLink = null;
379       if (txtinp.getChild("link", defNS) != null) {
380         tiLink = ParserUtils.getURL(txtinp.getChild("link", defNS).getTextTrim());
381       }
382       TextInputIF rssTextInput =
383         cBuilder.createTextInput(tiTitle, tiDescr, tiName, tiLink);
384       chnl.setTextInput(rssTextInput);
385     }
386 
387     chnl.setLastUpdated(dateParsed);
388 
389     return chnl;
390   }
391 
392   /**
393    * Returns the Ttl value corresponding to updatePeriod and updateFrequency
394    * @param updatePeriod the update period
395    * @param updateFrequency the update frequency
396    * @return the Ttl value
397    */
398   private static int getTtl(final String   updatePeriod, int updateFrequency) {
399     int minutes = -1;
400     if (updatePeriod.equals(ChannelIF.UPDATE_HOURLY)) {
401       minutes = 60;
402     } else if (updatePeriod.equals(ChannelIF.UPDATE_DAILY)) {
403       minutes = 24 * 60;
404     } else if (updatePeriod.equals(ChannelIF.UPDATE_WEEKLY)) {
405       minutes = 168 * 60;
406     } else if (updatePeriod.equals(ChannelIF.UPDATE_MONTHLY)) {
407       minutes = 30 * 24 * 60;
408     } else if (updatePeriod.equals(ChannelIF.UPDATE_YEARLY)) {
409       minutes = 365 * 24 * 60;
410     } else {
411       minutes = 24 * 60;
412     }
413     return (minutes / updateFrequency);
414  }
415 }
416
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags