|
1 package org.mortbay.xml;
16
17 import java.io.File  ;
18 import java.io.IOException ;
19 import java.io.InputStream ;
20 import java.net.URL ;
21 import java.util.AbstractList ;
22 import java.util.ArrayList ;
23 import java.util.HashMap ;
24 import java.util.Iterator ;
25 import java.util.Map ;
26 import java.util.NoSuchElementException ;
27 import java.util.StringTokenizer ;
28
29 import javax.xml.parsers.SAXParser ;
30 import javax.xml.parsers.SAXParserFactory ;
31
32 import org.apache.commons.logging.Log;
33 import org.mortbay.log.LogFactory;
34 import org.mortbay.util.LazyList;
35 import org.mortbay.util.LogSupport;
36 import org.xml.sax.Attributes ;
37 import org.xml.sax.InputSource ;
38 import org.xml.sax.SAXException ;
39 import org.xml.sax.SAXParseException ;
40 import org.xml.sax.XMLReader ;
41 import org.xml.sax.helpers.DefaultHandler ;
42
43
44
54 public class XmlParser
55 {
56 private static Log log=LogFactory.getLog(XmlParser.class);
57 private Map _redirectMap=new HashMap ();
58 private SAXParser _parser;
59 private String _xpath;
60 private Object _xpaths;
61
62
63
66 public XmlParser()
67 {
68 try
69 {
70 SAXParserFactory factory=SAXParserFactory.newInstance();
71 boolean notValidating=Boolean.getBoolean("org.mortbay.xml.XmlParser.NotValidating");
72 factory.setValidating(!notValidating);
73 _parser=factory.newSAXParser();
74 try
75 {
76 if(!notValidating)
77 _parser.getXMLReader().setFeature("http://apache.org/xml/features/validation/schema",true);
78 }
79 catch(Exception e)
80 {
81 log.warn("Schema validation may not be supported");
82 log.debug("",e);
83 notValidating=true;
84 }
85 _parser.getXMLReader().setFeature("http://xml.org/sax/features/validation",!notValidating);
86 _parser.getXMLReader().setFeature("http://xml.org/sax/features/namespaces",!notValidating);
87 _parser.getXMLReader().setFeature("http://xml.org/sax/features/namespace-prefixes",!notValidating);
88 }
89 catch(Exception e)
90 {
91 log.warn(LogSupport.EXCEPTION,e);
92 throw new Error (e.toString());
93 }
94 }
95
96
97
100 public XmlParser(boolean validating)
101 {
102 try
103 {
104 SAXParserFactory factory=SAXParserFactory.newInstance();
105 factory.setValidating(validating);
106 _parser=factory.newSAXParser();
107 try
108 {
109 if(validating)
110 _parser.getXMLReader().setFeature("http://apache.org/xml/features/validation/schema",validating);
111 }
112 catch(Exception e)
113 {
114 if(validating)
115 log.warn("Schema validation may not be supported: ",e);
116 else
117 LogSupport.ignore(log,e);
118 }
119 _parser.getXMLReader().setFeature("http://xml.org/sax/features/validation",validating);
120 _parser.getXMLReader().setFeature("http://xml.org/sax/features/namespaces",validating);
121 _parser.getXMLReader().setFeature("http://xml.org/sax/features/namespace-prefixes",validating);
122 }
123 catch(Exception e)
124 {
125 log.warn(LogSupport.EXCEPTION,e);
126 throw new Error (e.toString());
127 }
128 }
129
130
131
135 public synchronized void redirectEntity(String name,URL entity)
136 {
137 if(entity!=null)
138 _redirectMap.put(name,entity);
139 }
140
141
142
146 public String getXpath()
147 {
148 return _xpath;
149 }
150
151
152
158 public void setXpath(String xpath)
159 {
160 _xpath = xpath;
161 StringTokenizer tok = new StringTokenizer (xpath,"| ");
162 while(tok.hasMoreTokens())
163 _xpaths=LazyList.add(_xpaths, tok.nextToken());
164 }
165
166
167 public synchronized Node parse(InputSource source) throws IOException ,SAXException
168 {
169 Handler handler=new Handler ();
170 XMLReader reader=_parser.getXMLReader();
171 reader.setContentHandler(handler);
172 reader.setErrorHandler(handler);
173 reader.setEntityResolver(handler);
174 if(log.isDebugEnabled())
175 log.debug("parsing: sid="+source.getSystemId()+",pid="+source.getPublicId());
176 _parser.parse(source,handler);
177 if(handler._error!=null)
178 throw handler._error;
179 Node doc=(Node)handler._top.get(0);
180 handler.clear();
181 return doc;
182 }
183
184
185
188 public synchronized Node parse(String url) throws IOException ,SAXException
189 {
190 if(log.isDebugEnabled())
191 log.debug("parse: "+url);
192 return parse(new InputSource (url));
193 }
194
195
196
199 public synchronized Node parse(File file) throws IOException ,SAXException
200 {
201 if(log.isDebugEnabled())
202 log.debug("parse: "+file);
203 return parse(new InputSource (file.toURL().toString()));
204 }
205
206
207
210 public synchronized Node parse(InputStream in) throws IOException ,SAXException
211 {
212 Handler handler=new Handler ();
213 XMLReader reader=_parser.getXMLReader();
214 reader.setContentHandler(handler);
215 reader.setErrorHandler(handler);
216 reader.setEntityResolver(handler);
217 _parser.parse(new InputSource (in),handler);
218 if(handler._error!=null)
219 throw handler._error;
220 Node doc=(Node)handler._top.get(0);
221 handler.clear();
222 return doc;
223 }
224
225
226
229 public synchronized Node parse(URL url) throws IOException ,SAXException
230 {
231 Node n=null;
232 InputStream is=url.openStream();
233 try
234 {
235 n=parse(is);
236 }
237 finally
238 {
239 try
240 {
241 is.close();
242 }
243 catch(Exception e)
244 {
245 }
249 }
250 return n;
251 }
252
253
254
255
256
257 private class NoopHandler extends DefaultHandler
258 {
259 Handler _next;
260 int _depth;
261
262 NoopHandler(Handler next)
263 {
264 this._next=next;
265 }
266
267
268 public void startElement(String uri,String localName,String qName,Attributes attrs) throws SAXException
269 {
270 _depth++;
271 }
272
273
274 public void endElement(String uri,String localName,String qName) throws SAXException
275 {
276 if (_depth==0)
277 _parser.getXMLReader().setContentHandler(_next);
278 else
279 _depth--;
280 }
281 }
282
283
284
285 private class Handler extends DefaultHandler
286 {
287 Node _top=new Node(null,null,null);
288 SAXParseException _error;
289 private Node _context=_top;
290 private NoopHandler _noop;
291
292 Handler()
293 {
294 _noop = new NoopHandler(this);
295 }
296
297
298 void clear()
299 {
300 _top=null;
301 _error=null;
302 _context=null;
303 }
304
305
306 public void startElement(String uri,String localName,String qName,Attributes attrs) throws SAXException
307 {
308 String name=(uri==null||uri.equals(""))?qName:localName;
309 Node node=new Node(_context,name,attrs);
310
311 if (_xpaths!=null)
313 {
314 String path=node.getPath();
315 boolean match=false;
316 for (int i=LazyList.size(_xpaths);!match&&i-->0;)
317 {
318 String xpath=(String )LazyList.get(_xpaths,i);
319
320 match=path.equals(xpath) ||
321 xpath.startsWith(path) && xpath.length()>path.length() && xpath.charAt(path.length())=='/';
322 }
323
324 if (match)
325 {
326 _context.add(node);
327 _context=node;
328 }
329 else
330 {
331 _parser.getXMLReader().setContentHandler(_noop);
332 }
333 }
334 else
335 {
336 _context.add(node);
337 _context=node;
338 }
339 }
340
341
342 public void endElement(String uri,String localName,String qName) throws SAXException
343 {
344 _context=_context._parent;
345 }
346
347
348 public void ignorableWhitespace(char buf[],int offset,int len) throws SAXException
349 {
350 }
351
352
353 public void characters(char buf[],int offset,int len) throws SAXException
354 {
355 _context.add(new String (buf,offset,len));
356 }
357
358
359 public void warning(SAXParseException ex)
360 {
361 log.debug(LogSupport.EXCEPTION,ex);
362 log.warn("WARNING@"+getLocationString(ex)+" : "+ex.toString());
363 }
364
365
366 public void error(SAXParseException ex) throws SAXException
367 {
368 if(_error==null)
370 _error=ex;
371 log.debug(LogSupport.EXCEPTION,ex);
372 log.warn("ERROR@"+getLocationString(ex)+" : "+ex.toString());
373 }
374
375
376 public void fatalError(SAXParseException ex) throws SAXException
377 {
378 _error=ex;
379 log.debug(LogSupport.EXCEPTION,ex);
380 log.warn("FATAL@"+getLocationString(ex)+" : "+ex.toString());
381 throw ex;
382 }
383
384
385 private String getLocationString(SAXParseException ex)
386 {
387 return ex.getSystemId()+" line:"+ex.getLineNumber()+" col:"+ex.getColumnNumber();
388 }
389
390
391 public InputSource resolveEntity(String pid,String sid)
392 {
393 if(log.isDebugEnabled())
394 log.debug("resolveEntity("+pid+", "+sid+")");
395 URL entity=null;
396 if(pid!=null)
397 entity=(URL )_redirectMap.get(pid);
398 if(entity==null)
399 entity=(URL )_redirectMap.get(sid);
400 if(entity==null)
401 {
402 String dtd=sid;
403 if(dtd.lastIndexOf('/')>=0)
404 dtd=dtd.substring(dtd.lastIndexOf('/')+1);
405 if(log.isDebugEnabled())
406 log.debug("Can't exact match entity in redirect map, trying "+dtd);
407 entity=(URL )_redirectMap.get(dtd);
408 }
409 if(entity!=null)
410 {
411 try
412 {
413 InputStream in=entity.openStream();
414 if(log.isDebugEnabled())
415 log.debug("Redirected entity "+sid+" --> "+entity);
416 InputSource is=new InputSource (in);
417 is.setSystemId(sid);
418 return is;
419 }
420 catch(IOException e)
421 {
422 LogSupport.ignore(log,e);
423 }
424 }
425 return null;
426 }
427 }
428
429
430
431
432
433
436 public static class Attribute
437 {
438 private String _name;
439 private String _value;
440
441 Attribute(String n,String v)
442 {
443 _name=n;
444 _value=v;
445 }
446
447 public String getName()
448 {
449 return _name;
450 }
451
452 public String getValue()
453 {
454 return _value;
455 }
456 }
457
458
459
460
461
462
465 public static class Node extends AbstractList
466 {
467 Node _parent;
468 private ArrayList _list;
469 private String _tag;
470 private Attribute[] _attrs;
471 private boolean _lastString=false;
472 private String _path;
473
474
475 Node(Node parent,String tag,Attributes attrs)
476 {
477 _parent=parent;
478 _tag=tag;
479 if(attrs!=null)
480 {
481 _attrs=new Attribute[attrs.getLength()];
482 for(int i=0;i<attrs.getLength();i++)
483 {
484 String name=attrs.getLocalName(i);
485 if(name==null||name.equals(""))
486 name=attrs.getQName(i);
487 _attrs[i]=new Attribute(name,attrs.getValue(i));
488 }
489 }
490 }
491
492
493 public Node getParent()
494 {
495 return _parent;
496 }
497
498
499 public String getTag()
500 {
501 return _tag;
502 }
503
504
505 public String getPath()
506 {
507 if (_path==null)
508 {
509 if (getParent()!=null && getParent().getTag()!=null)
510 _path= getParent().getPath()+"/"+_tag;
511 else
512 _path ="/"+_tag;
513 }
514 return _path;
515 }
516
517
518
521 public Attribute[] getAttributes()
522 {
523 return _attrs;
524 }
525
526
527
532 public String getAttribute(String name)
533 {
534 return getAttribute(name,null);
535 }
536
537
538
543 public String getAttribute(String name,String dft)
544 {
545 if(_attrs==null||name==null)
546 return dft;
547 for(int i=0;i<_attrs.length;i++)
548 if(name.equals(_attrs[i].getName()))
549 return _attrs[i].getValue();
550 return dft;
551 }
552
553
554
557 public int size()
558 {
559 if(_list!=null)
560 return _list.size();
561 return 0;
562 }
563
564
565
570 public Object get(int i)
571 {
572 if(_list!=null)
573 return _list.get(i);
574 return null;
575 }
576
577
578
584 public Node get(String tag)
585 {
586 if(_list!=null)
587 {
588 for(int i=0;i<_list.size();i++)
589 {
590 Object o=_list.get(i);
591 if(o instanceof Node)
592 {
593 Node n=(Node)o;
594 if(tag.equals(n._tag))
595 return n;
596 }
597 }
598 }
599 return null;
600 }
601
602
603 public void add(int i,Object o)
604 {
605 if(_list==null)
606 _list=new ArrayList ();
607 if(o instanceof String )
608 {
609 if(_lastString)
610 {
611 int last=_list.size()-1;
612 _list.set(last,(String )_list.get(last)+o);
613 }
614 else
615 _list.add(i,o);
616 _lastString=true;
617 }
618 else
619 {
620 _lastString=false;
621 _list.add(i,o);
622 }
623 }
624
625
626 public void clear()
627 {
628 if(_list!=null)
629 _list.clear();
630 _list=null;
631 }
632
633
634
642 public String getString(String tag,boolean tags,boolean trim)
643 {
644 Node node=get(tag);
645 if(node==null)
646 return null;
647 String s=node.toString(tags);
648 if(s!=null&&trim)
649 s=s.trim();
650 return s;
651 }
652
653
654 public synchronized String toString()
655 {
656 return toString(true);
657 }
658
659
660
665 public synchronized String toString(boolean tag)
666 {
667 StringBuffer buf=new StringBuffer ();
668 synchronized(buf)
669 {
670 toString(buf,tag);
671 return buf.toString();
672 }
673 }
674
675
676
681 public synchronized String toString(boolean tag,boolean trim)
682 {
683 String s=toString(tag);
684 if(s!=null&&trim)
685 s=s.trim();
686 return s;
687 }
688
689
690 private synchronized void toString(StringBuffer buf,boolean tag)
691 {
692 if(tag)
693 {
694 buf.append("<");
695 buf.append(_tag);
696 if(_attrs!=null)
697 {
698 for(int i=0;i<_attrs.length;i++)
699 {
700 buf.append(' ');
701 buf.append(_attrs[i].getName());
702 buf.append("=\"");
703 buf.append(_attrs[i].getValue());
704 buf.append("\"");
705 }
706 }
707 }
708 if(_list!=null)
709 {
710 if(tag)
711 buf.append(">");
712 for(int i=0;i<_list.size();i++)
713 {
714 Object o=_list.get(i);
715 if(o==null)
716 continue;
717 if(o instanceof Node)
718 ((Node)o).toString(buf,tag);
719 else
720 buf.append(o.toString());
721 }
722 if(tag)
723 {
724 buf.append("</");
725 buf.append(_tag);
726 buf.append(">");
727 }
728 }
729 else if(tag)
730 buf.append("/>");
731 }
732
733
734
740 public Iterator iterator(final String tag)
741 {
742 return new Iterator ()
743 {
744 int c=0;
745 Node _node;
746
747
748 public boolean hasNext()
749 {
750 if(_node!=null)
751 return true;
752 while(_list!=null&&c<_list.size())
753 {
754 Object o=_list.get(c);
755 if(o instanceof Node)
756 {
757 Node n=(Node)o;
758 if(tag.equals(n._tag))
759 {
760 _node=n;
761 return true;
762 }
763 }
764 c++;
765 }
766 return false;
767 }
768
769
770 public Object next()
771 {
772 try
773 {
774 if(hasNext())
775 return _node;
776 throw new NoSuchElementException ();
777 }
778 finally
779 {
780 _node=null;
781 c++;
782 }
783 }
784
785
786 public void remove()
787 {
788 throw new UnsupportedOperationException ("Not supported");
789 }
790 };
791 }
792 }
793 }
794 |
Popular Tags |
Java API By Example, From Geeks To Geeks.