RewritableLinkTransformer


1   /*
2    * WebSphinx web-crawling toolkit
3    *
4    * Copyright (c) 1998-2002 Carnegie Mellon University.  All rights
5    * reserved.
6    *
7    * Redistribution and use in source and binary forms, with or without
8    * modification, are permitted provided that the following conditions
9    * are met:
10   *
11   * 1. Redistributions of source code must retain the above copyright
12   *    notice, this list of conditions and the following disclaimer.
13   *
14   * 2. Redistributions in binary form must reproduce the above copyright
15   *    notice, this list of conditions and the following disclaimer in
16   *    the documentation and/or other materials provided with the
17   *    distribution.
18   *
19   * THIS SOFTWARE IS PROVIDED BY CARNEGIE MELLON UNIVERSITY ``AS IS'' AND
20   * ANY EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
21   * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22   * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL CARNEGIE MELLON UNIVERSITY
23   * NOR ITS EMPLOYEES BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24   * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25   * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26   * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27   * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28   * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29   * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30   *
31   */
32  
33  package websphinx;
34  
35  import java.io.*;
36  import java.net.URL  ;
37  
38  /**
39   * Transformer that remaps URLs in links in such a way
40   * that if the URL mapping changes during (or after) some
41   * HTML has been transformed, the HTML can be fixed up after
42   * the fact.  This class is used by Concatenator and Mirror,
43   * since in those operations, the URL mapping function
44   * changes as each page is written to the concatenation or
45   * mirror.
46   */
47  public class RewritableLinkTransformer extends LinkTransformer {
48  
49      private RewriteRegion head, tail;
50      private File file;
51      private boolean closed = false;
52  
53      /**
54       * Make a RewritableLinkTransformer.
55       * @param filename Filename to write to
56       */
57      public RewritableLinkTransformer (String   filename) throws IOException {
58          super (filename, true);
59          file = new File (filename);
60      }
61  
62      public void close () throws IOException {
63          super.close ();
64          closed = true;
65      }
66  
67      static final String   PLACEHOLDER = "@WEBSPHINX@";
68  
69      protected void handleLink (Link link) throws IOException {
70          URL   url = link.getURL ();
71  
72          Tag t = link.replaceHref (PLACEHOLDER);
73          String   s = t.toString();
74          int prefix = s.indexOf (PLACEHOLDER);
75          if (prefix != -1) {
76              int postfix = prefix + PLACEHOLDER.length();
77              
78              emit (s.substring (0, prefix));
79              
80              String   href = lookup (base, url);
81              RewriteRegion node = addURL (url, getFilePointer(), href.length());
82              emit (href);
83              
84              emit (s.substring (postfix));
85          } else {
86              emit (s);
87          }
88  
89          transformContents (link);
90          if (link.getEndTag () != null)
91             emit (link.getEndTag ());
92      }
93  
94      private RewriteRegion addURL (URL   url, long offset, int len) {
95          RewriteRegion node = new RewriteRegion ();
96          node.url = url;
97          node.offset = offset;
98          node.len = len;
99  
100         if (tail == null) {
101             head = tail = node;
102         }
103         else {
104             node.next = tail.next;
105             tail.next = node;
106             node.prev = tail;
107             if (node.next != null)
108                 node.next.prev = node;
109             tail = node;
110         }
111 
112         return node;
113     }
114 
115     static final int BUFFER_SIZE = 8;
116 
117     /**
118      * Rewrite the file, remapping all the URLs according to their
119      * current values from lookup().
120      */
121     public void rewrite () throws IOException {
122         flush ();
123         
124         if (head == null)
125             // no links to rewrite
126             return;
127 
128         RandomAccessFile raf = closed
129             ? Access.getAccess ().readWriteFile (file)
130             : getRandomAccessFile ();
131 
132         byte buf[] = new byte[BUFFER_SIZE];
133         long end = raf.length ();
134         long src = 0;
135         long dest = 0;
136         long left;
137         int n;
138         int growth = 0;
139         int shrinkage = 0;
140 
141         // Forward pass
142         //    Rewrite only URLs which are becoming shorter
143         raf.seek (dest);
144         for (RewriteRegion loc = head; loc != null; loc = loc.next) {
145             // loop invariant: file[0..dest-1] is rewritten,
146             // and next byte to copy to file[dest] is from file[src]
147             // and raf.getFilePointer() == dest
148             long diff = dest - src;
149 
150             String   href = lookup (base, loc.url);
151             loc.newHref = href;
152             loc.newLen = href.length ();
153 
154             if (loc.newLen > loc.len) {
155                 // new URL is longer than old URL
156                 // must postpone rewriting this until the backward pass
157                 growth += loc.newLen - loc.len;
158                 loc.offset += diff;
159                 continue;
160             }
161             else
162                 shrinkage += loc.len - loc.newLen;
163 
164             // rewrite up to loc
165             left = loc.offset - src;
166             while (left > BUFFER_SIZE) {
167                 raf.seek (src);
168                 raf.read (buf);
169                 raf.seek (dest);
170                 raf.write (buf);
171                 src += BUFFER_SIZE;
172                 dest += BUFFER_SIZE;
173                 left -= BUFFER_SIZE;
174             }
175             if (left > 0) {
176                 n = (int)left;
177                 raf.seek (src);
178                 raf.read (buf, 0, n);
179                 raf.seek (dest);
180                 raf.write (buf, 0, n);
181                 src += n;
182                 dest += n;
183                 left -= n;
184             }
185 
186             // write loc
187             raf.writeBytes (href);
188 
189             dest += loc.newLen;
190             src += loc.len;
191 
192             loc.offset += diff;
193             loc.len = loc.newLen;
194         }
195 
196         if (src > dest) {
197             // rewrite rest of file
198             while (true) {
199                 raf.seek (src);
200                 if ((n = raf.read (buf)) == -1)
201                     break;
202                 raf.seek (dest);
203                 raf.write (buf, 0, n);
204                 src += n;
205                 dest += n;
206             }
207         }
208         else
209             src = dest = end;
210 
211         src = dest;
212         dest += growth;
213         for (RewriteRegion loc = tail; loc != null; loc = loc.prev) {
214             // loop invariant: file[dest...end-1] is rewritten,
215             // and next byte to copy to file[dest] is from file[src]
216             long diff = dest - src;
217 
218             if (loc.newLen <= loc.len) {
219                 loc.offset += diff;
220                 continue;
221             }
222 
223             // rewrite back to loc
224             left = src - (loc.offset + loc.len);
225             while (left > BUFFER_SIZE) {
226                 src -= BUFFER_SIZE;
227                 dest -= BUFFER_SIZE;
228                 left -= BUFFER_SIZE;
229                 raf.seek (src);
230                 raf.read (buf);
231                 raf.seek (dest);
232                 raf.write (buf);
233             }
234             if (left > 0) {
235                 n = (int)left;
236                 src -= n;
237                 dest -= n;
238                 raf.seek (src);
239                 raf.read (buf, 0, n);
240                 raf.seek (dest);
241                 raf.write (buf, 0, n);
242             }
243 
244             // write loc
245             dest -= loc.newLen;
246             src -= loc.len;
247             raf.seek (dest);
248             raf.writeBytes (loc.newHref);
249 
250             loc.offset = dest;
251             loc.len = loc.newLen;
252         }
253 
254         if (src != dest)
255             System.err.println ("ASSERTION FAILURE: SRC=" + src + "!=dest=" + dest);
256 
257         if (shrinkage > growth) {
258             // overwrite the rest of the file with spaces
259             for (int i=0; i<BUFFER_SIZE; ++i)
260                 buf[i] = (byte)' ';
261             left = shrinkage - growth;
262             raf.seek (end - left);
263             while (left > BUFFER_SIZE) {
264                 raf.write (buf);
265                 left -= BUFFER_SIZE;
266             }
267             if (left > 0)
268                 raf.write (buf, 0, (int)left);
269         }
270         else
271             raf.seek (end + (growth - shrinkage));
272 
273 
274         if (closed)
275             raf.close ();
276     }
277 
278     /*
279      * Testing
280      *
281     public static void main (String[] args) throws Exception {
282         RewritableLinkTransformer unparser = new TestILTransformer (args[1]);
283         Link link = new Link (args[0]);
284         Page page = new Page (link);
285         System.out.println ("Writing " + page.toDescription());
286         unparser.writePage (page);
287         System.out.println ("Rewriting while open");
288         unparser.rewrite ();
289         unparser.close ();
290         System.out.println ("Rewriting after close");
291         unparser.rewrite ();
292     }
293      */
294 }
295 
296 class RewriteRegion {
297     URL   url;
298     long offset;
299     int len;
300 
301     String   newHref;
302     int newLen;
303 
304     RewriteRegion next;
305     RewriteRegion prev;
306 }
307 
308 /*
309  * Testing
310  *
311 class TestILTransformer extends RewritableLinkTransformer {
312     public TestILTransformer (String filename) throws IOException {
313         super (filename);
314     }
315 
316     final static String BIG_STRING =
317         "@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"
318        +"@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"
319        +"@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"
320        +"@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@";
321 
322     public String lookup (URL base, URL url) {
323         if (closed)
324             return super.lookup (base, url);
325         else if (Math.random() > 0.5)
326             return BIG_STRING.substring (0, url.toString().length()*2);
327         else
328             return "";
329     }
330 }
331  */
332
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags