DataUnformatFilter


1   /*-- 
2   
3    Copyright (C) 2000 Brett McLaughlin & Jason Hunter.
4    All rights reserved.
5    
6    Redistribution and use in source and binary forms, with or without
7    modification, are permitted provided that the following conditions
8    are met:
9    
10   1. Redistributions of source code must retain the above copyright
11      notice, this list of conditions, and the following disclaimer.
12   
13   2. Redistributions in binary form must reproduce the above copyright
14      notice, this list of conditions, and the disclaimer that follows 
15      these conditions in the documentation and/or other materials 
16      provided with the distribution.
17  
18   3. The name "JDOM" must not be used to endorse or promote products
19      derived from this software without prior written permission.  For
20      written permission, please contact license@jdom.org.
21   
22   4. Products derived from this software may not be called "JDOM", nor
23      may "JDOM" appear in their name, without prior written permission
24      from the JDOM Project Management (pm@jdom.org).
25   
26   In addition, we request (but do not require) that you include in the 
27   end-user documentation provided with the redistribution and/or in the 
28   software itself an acknowledgement equivalent to the following:
29       "This product includes software developed by the
30        JDOM Project (http://www.jdom.org/)."
31   Alternatively, the acknowledgment may be graphical using the logos 
32   available at http://www.jdom.org/images/logos.
33  
34   THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESSED OR IMPLIED
35   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
36   OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
37   DISCLAIMED.  IN NO EVENT SHALL THE JDOM AUTHORS OR THE PROJECT
38   CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
39   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
40   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
41   USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
42   ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
43   OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
44   OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
45   SUCH DAMAGE.
46  
47   This software consists of voluntary contributions made by many 
48   individuals on behalf of the JDOM Project and was originally 
49   created by Brett McLaughlin <brett@jdom.org> and 
50   Jason Hunter <jhunter@jdom.org>.  For more information on the 
51   JDOM Project, please see <http://www.jdom.org/>.
52   
53   */
54  package sax;
55  
56  import java.util.Stack  ;
57  
58  import org.xml.sax.Attributes  ;
59  import org.xml.sax.SAXException  ;
60  import org.xml.sax.XMLReader  ;
61  
62  
63  /**
64   * Filter for removing formatting from data- or field-oriented XML. 
65   *
66   * <i>Code and comments adapted from DataWriter-0.2, written
67   * by David Megginson and released into the public domain,
68   * without warranty.</i>
69   *
70   * <p>This filter removes leading and trailing whitespace from
71   * field-oriented XML without mixed content. Note that this class will
72   * likely not yield appropriate results for document-oriented XML like
73   * XHTML pages, which mix character data and elements together.</p>
74   *
75   * @see DataFormatFilter
76   */
77  public class DataUnformatFilter extends XMLFilterBase
78  {
79      
80      
81      
82      ////////////////////////////////////////////////////////////////////
83      // Constructors.
84      ////////////////////////////////////////////////////////////////////
85  
86  
87      /**
88       * Create a new filter.
89       */
90      public DataUnformatFilter()
91      {
92      }
93  
94  
95      /**
96       * Create a new filter.
97       *
98       * <p>Use the XMLReader provided as the source of events.</p>
99       *
100      * @param xmlreader The parent in the filter chain.
101      */
102     public DataUnformatFilter(XMLReader   xmlreader)
103     {
104         super(xmlreader);
105     }
106 
107 
108 
109     ////////////////////////////////////////////////////////////////////
110     // Public methods.
111     ////////////////////////////////////////////////////////////////////
112 
113 
114     /**
115      * Reset the filter so that it can be reused.
116      *
117      * <p>This method is especially useful if the filter failed
118      * with an exception the last time through.</p>
119      */
120     public void reset ()
121     {
122         state = SEEN_NOTHING;
123         stateStack = new Stack  ();
124         whitespace = new StringBuffer  ();
125     }
126 
127 
128 
129     ////////////////////////////////////////////////////////////////////
130     // Methods from org.xml.sax.ContentHandler.
131     ////////////////////////////////////////////////////////////////////
132 
133     
134     /**
135      * Filter a start document event. 
136      *
137      * <p>Reset state and pass the event on for further processing.</p>
138      *
139      * @exception org.xml.sax.SAXException If a filter
140      *            further down the chain raises an exception.
141      * @see org.xml.sax.ContentHandler#startDocument
142      */
143     public void startDocument ()
144     throws SAXException  
145     {
146         reset();
147         super.startDocument();
148     }
149 
150 
151     /**
152      * Filter a start element event.
153      *
154      * @param uri The element's Namespace URI.
155      * @param localName The element's local name.
156      * @param qName The element's qualified (prefixed) name.
157      * @param atts The element's attribute list.
158      * @exception org.xml.sax.SAXException If a filter
159      *            further down the chain raises an exception.
160      * @see org.xml.sax.ContentHandler#startElement
161      */
162     public void startElement (String   uri, String   localName,
163                               String   qName, Attributes   atts)
164     throws SAXException  
165     {
166         clearWhitespace();
167         stateStack.push(SEEN_ELEMENT);
168         state = SEEN_NOTHING;
169         super.startElement(uri, localName, qName, atts);
170     }
171 
172 
173     /**
174      * Filter an end element event.
175      *
176      * @param uri The element's Namespace URI.
177      * @param localName The element's local name.
178      * @param qName The element's qualified (prefixed) name.
179      * @exception org.xml.sax.SAXException If a filter
180      *            further down the chain raises an exception.
181      * @see org.xml.sax.ContentHandler#endElement
182      */
183     public void endElement (String   uri, String   localName, String   qName)
184     throws SAXException  
185     {
186         if (state == SEEN_ELEMENT) {
187             clearWhitespace();
188         } else {
189             emitWhitespace();
190         }
191         state = stateStack.pop();
192         super.endElement(uri, localName, qName);
193     }
194 
195 
196     /**
197      * Filter a character data event.
198      *
199      * @param ch The characters to write.
200      * @param start The starting position in the array.
201      * @param length The number of characters to use.
202      * @exception org.xml.sax.SAXException If a filter
203      *            further down the chain raises an exception.
204      * @see org.xml.sax.ContentHandler#characters
205      */
206     public void characters (char ch[], int start, int length)
207     throws SAXException  
208     {
209         if (state != SEEN_DATA) {
210 
211             /* Look for non-whitespace. */
212 
213             int end = start + length;
214             while (end-- > start) {
215                 if (!isXMLWhitespace(ch[end]))
216                     break;
217             }
218 
219             /*
220              * If all the characters are whitespace, save them for later.
221              * If we've got some data, emit any saved whitespace and update
222              * our state to show we've seen data.
223              */
224 
225             if (end < start) {
226                 saveWhitespace(ch, start, length);
227             } else {
228                 state = SEEN_DATA;
229                 emitWhitespace();
230             }
231         }
232 
233         /* Pass on everything inside a data field. */
234         
235         if (state == SEEN_DATA) {
236             super.characters(ch, start, length);
237         }
238     }
239     
240     
241      /**
242       * Filter an ignorable whitespace event.
243       *
244       * @param ch The array of characters to write.
245       * @param start The starting position in the array.
246       * @param length The number of characters to write.
247       * @exception org.xml.sax.SAXException If a filter
248       *            further down the chain raises an exception.
249       * @see org.xml.sax.ContentHandler#ignorableWhitespace
250       */
251     public void ignorableWhitespace (char ch[], int start, int length)
252     throws SAXException  
253     {
254         emitWhitespace();
255         // ignore
256     }
257     
258 
259     /**
260      * Filter a processing instruction event.
261      *
262      * @param target The PI target.
263      * @param data The PI data.
264      * @exception org.xml.sax.SAXException If a filter
265      *            further down the chain raises an exception.
266      * @see org.xml.sax.ContentHandler#processingInstruction
267      */
268     public void processingInstruction (String   target, String   data)
269     throws SAXException  
270     {
271         emitWhitespace();
272         super.processingInstruction(target, data);
273     }
274 
275 
276 
277     ////////////////////////////////////////////////////////////////////
278     // Internal methods.
279     ////////////////////////////////////////////////////////////////////
280 
281 
282     /**
283      * Saves trailing whitespace.
284      */
285     protected void saveWhitespace (char[] ch, int start, int length) {
286         whitespace.append(ch, start, length);
287     }
288 
289 
290     /**
291      * Passes saved whitespace down the filter chain.
292      */
293     protected void emitWhitespace ()
294     throws SAXException  
295     {
296         char[] data = new char[whitespace.length()];
297         whitespace.getChars(0, data.length, data, 0);
298         whitespace.setLength(0);
299         super.characters(data, 0, data.length);
300     }
301 
302 
303     /**
304      * Discards saved whitespace. 
305      */
306     protected void clearWhitespace () {
307         whitespace.setLength(0);
308     }
309 
310 
311     /**
312      * Returns <var>true</var> if character is XML whitespace.
313      */
314     private boolean isXMLWhitespace (char c)
315     {
316         return c == ' ' || c == '\t' || c == '\r' || c == '\n';
317     }
318 
319 
320 
321 
322     ////////////////////////////////////////////////////////////////////
323     // Constants.
324     ////////////////////////////////////////////////////////////////////
325 
326     private static final Object   SEEN_NOTHING = new Object  ();
327     private static final Object   SEEN_ELEMENT = new Object  ();
328     private static final Object   SEEN_DATA = new Object  ();
329 
330 
331     ////////////////////////////////////////////////////////////////////
332     // Internal state.
333     ////////////////////////////////////////////////////////////////////
334 
335     private Object   state = SEEN_NOTHING;
336     private Stack   stateStack = new Stack  ();
337 
338     private StringBuffer   whitespace = new StringBuffer  ();
339 
340 }
341 
342 // end of DataUnformatFilter.java
343
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags