1 //
2 // ========================================================================
3 // Copyright (c) 1995-2016 Mort Bay Consulting Pty. Ltd.
4 // ------------------------------------------------------------------------
5 // All rights reserved. This program and the accompanying materials
6 // are made available under the terms of the Eclipse Public License v1.0
7 // and Apache License v2.0 which accompanies this distribution.
8 //
9 // The Eclipse Public License is available at
10 // http://www.eclipse.org/legal/epl-v10.html
11 //
12 // The Apache License v2.0 is available at
13 // http://www.opensource.org/licenses/apache2.0.php
14 //
15 // You may elect to redistribute this code under either of these licenses.
16 // ========================================================================
17 //
18
19 package org.eclipse.jetty.servlets.gzip;
20
21 import java.io.IOException;
22 import java.io.OutputStream;
23 import java.io.OutputStreamWriter;
24 import java.io.PrintWriter;
25 import java.io.UnsupportedEncodingException;
26 import java.util.Set;
27 import java.util.zip.DeflaterOutputStream;
28 import java.util.zip.GZIPOutputStream;
29
30 import javax.servlet.AsyncEvent;
31 import javax.servlet.AsyncListener;
32 import javax.servlet.ServletContext;
33 import javax.servlet.ServletException;
34 import javax.servlet.http.HttpServletRequest;
35 import javax.servlet.http.HttpServletResponse;
36
37 import org.eclipse.jetty.http.HttpMethod;
38 import org.eclipse.jetty.http.MimeTypes;
39 import org.eclipse.jetty.http.pathmap.PathSpecSet;
40 import org.eclipse.jetty.server.Request;
41 import org.eclipse.jetty.server.handler.HandlerWrapper;
42 import org.eclipse.jetty.util.IncludeExclude;
43 import org.eclipse.jetty.util.RegexSet;
44 import org.eclipse.jetty.util.StringUtil;
45 import org.eclipse.jetty.util.URIUtil;
46 import org.eclipse.jetty.util.log.Log;
47 import org.eclipse.jetty.util.log.Logger;
48
49 /* ------------------------------------------------------------ */
50 /**
51 * GZIP Handler This handler will gzip the content of a response if:
52 * <ul>
53 * <li>The handler is mapped to a matching path</li>
54 * <li>The response status code is >=200 and <300
55 * <li>The content length is unknown or more than the <code>minGzipSize</code> initParameter or the minGzipSize is 0(default)</li>
56 * <li>The content-type matches one of the set of mimetypes to be compressed</li>
57 * <li>The content-type does NOT match one of the set of mimetypes AND setExcludeMimeTypes is <code>true</code></li>
58 * <li>No content-encoding is specified by the resource</li>
59 * </ul>
60 *
61 * <p>
62 * Compressing the content can greatly improve the network bandwidth usage, but at a cost of memory and CPU cycles. If this handler is used for static content,
63 * then use of efficient direct NIO may be prevented, thus use of the gzip mechanism of the <code>org.eclipse.jetty.servlet.DefaultServlet</code> is advised instead.
64 * </p>
65 */
66 public class GzipHandler extends HandlerWrapper
67 {
68 private static final Logger LOG = Log.getLogger(GzipHandler.class);
69
70 protected int _bufferSize = 8192;
71 protected int _minGzipSize = 256;
72 protected String _vary = "Accept-Encoding, User-Agent";
73
74 private final IncludeExclude<String> _agentPatterns=new IncludeExclude<>(RegexSet.class);
75 private final IncludeExclude<String> _methods = new IncludeExclude<>();
76 private final IncludeExclude<String> _paths = new IncludeExclude<String>(PathSpecSet.class);
77 private final IncludeExclude<String> _mimeTypes = new IncludeExclude<>();
78
79 /* ------------------------------------------------------------ */
80 /**
81 * Instantiates a new gzip handler.
82 */
83 public GzipHandler()
84 {
85 _methods.include(HttpMethod.GET.asString());
86 for (String type:MimeTypes.getKnownMimeTypes())
87 {
88 if ("image/svg+xml".equals(type))
89 _paths.exclude("*.svgz");
90 else if (type.startsWith("image/")||
91 type.startsWith("audio/")||
92 type.startsWith("video/"))
93 _mimeTypes.exclude(type);
94 }
95 _mimeTypes.exclude("application/compress");
96 _mimeTypes.exclude("application/zip");
97 _mimeTypes.exclude("application/gzip");
98 _mimeTypes.exclude("application/bzip2");
99 _mimeTypes.exclude("application/x-rar-compressed");
100 LOG.debug("{} mime types {}",this,_mimeTypes);
101
102 _agentPatterns.exclude(".*MSIE 6.0.*");
103 }
104
105 /* ------------------------------------------------------------ */
106 /**
107 * @param patterns Regular expressions matching user agents to exclude
108 */
109 public void addExcludedAgentPatterns(String... patterns)
110 {
111 _agentPatterns.exclude(patterns);
112 }
113
114 /* ------------------------------------------------------------ */
115 /**
116 * @param methods The methods to exclude in compression
117 */
118 public void addExcludedMethods(String... methods)
119 {
120 for (String m : methods)
121 _methods.exclude(m);
122 }
123
124 /* ------------------------------------------------------------ */
125 /**
126 * Set the mime types.
127 * @param types The mime types to exclude (without charset or other parameters).
128 * For backward compatibility the mimetypes may be comma separated strings, but this
129 * will not be supported in future versions.
130 */
131 public void addExcludedMimeTypes(String... types)
132 {
133 for (String t : types)
134 _mimeTypes.exclude(StringUtil.csvSplit(t));
135 }
136
137 /* ------------------------------------------------------------ */
138 /**
139 * Add path to excluded paths list.
140 * <p>
141 * There are 2 syntaxes supported, Servlet <code>url-pattern</code> based, and
142 * Regex based. This means that the initial characters on the path spec
143 * line are very strict, and determine the behavior of the path matching.
144 * <ul>
145 * <li>If the spec starts with <code>'^'</code> the spec is assumed to be
146 * a regex based path spec and will match with normal Java regex rules.</li>
147 * <li>If the spec starts with <code>'/'</code> then spec is assumed to be
148 * a Servlet url-pattern rules path spec for either an exact match
149 * or prefix based match.</li>
150 * <li>If the spec starts with <code>'*.'</code> then spec is assumed to be
151 * a Servlet url-pattern rules path spec for a suffix based match.</li>
152 * <li>All other syntaxes are unsupported</li>
153 * </ul>
154 * <p>
155 * Note: inclusion takes precedence over exclude.
156 *
157 * @param pathspecs Path specs (as per servlet spec) to exclude. If a
158 * ServletContext is available, the paths are relative to the context path,
159 * otherwise they are absolute.<br>
160 * For backward compatibility the pathspecs may be comma separated strings, but this
161 * will not be supported in future versions.
162 */
163 public void addExcludedPaths(String... pathspecs)
164 {
165 for (String p : pathspecs)
166 _paths.exclude(StringUtil.csvSplit(p));
167 }
168
169 /* ------------------------------------------------------------ */
170 /**
171 * @param patterns Regular expressions matching user agents to exclude
172 */
173 public void addIncludedAgentPatterns(String... patterns)
174 {
175 _agentPatterns.include(patterns);
176 }
177
178 /* ------------------------------------------------------------ */
179 /**
180 * @param methods The methods to include in compression
181 */
182 public void addIncludedMethods(String... methods)
183 {
184 for (String m : methods)
185 _methods.include(m);
186 }
187
188 /* ------------------------------------------------------------ */
189 /**
190 * Add included mime types. Inclusion takes precedence over
191 * exclusion.
192 * @param types The mime types to include (without charset or other parameters)
193 * For backward compatibility the mimetypes may be comma separated strings, but this
194 * will not be supported in future versions.
195 */
196 public void addIncludedMimeTypes(String... types)
197 {
198 for (String t : types)
199 _mimeTypes.include(StringUtil.csvSplit(t));
200 }
201
202 /* ------------------------------------------------------------ */
203 /**
204 * Add path specs to include.
205 * <p>
206 * There are 2 syntaxes supported, Servlet <code>url-pattern</code> based, and
207 * Regex based. This means that the initial characters on the path spec
208 * line are very strict, and determine the behavior of the path matching.
209 * <ul>
210 * <li>If the spec starts with <code>'^'</code> the spec is assumed to be
211 * a regex based path spec and will match with normal Java regex rules.</li>
212 * <li>If the spec starts with <code>'/'</code> then spec is assumed to be
213 * a Servlet url-pattern rules path spec for either an exact match
214 * or prefix based match.</li>
215 * <li>If the spec starts with <code>'*.'</code> then spec is assumed to be
216 * a Servlet url-pattern rules path spec for a suffix based match.</li>
217 * <li>All other syntaxes are unsupported</li>
218 * </ul>
219 * <p>
220 * Note: inclusion takes precedence over exclude.
221 *
222 * @param pathspecs Path specs (as per servlet spec) to include. If a
223 * ServletContext is available, the paths are relative to the context path,
224 * otherwise they are absolute
225 */
226 public void addIncludedPaths(String... pathspecs)
227 {
228 for (String p : pathspecs)
229 _paths.include(StringUtil.csvSplit(p));
230 }
231
232 /* ------------------------------------------------------------ */
233 public String[] getExcludedAgentPatterns()
234 {
235 Set<String> excluded=_agentPatterns.getExcluded();
236 return excluded.toArray(new String[excluded.size()]);
237 }
238
239 /* ------------------------------------------------------------ */
240 public String[] getExcludedMethods()
241 {
242 Set<String> excluded=_methods.getExcluded();
243 return excluded.toArray(new String[excluded.size()]);
244 }
245
246 /* ------------------------------------------------------------ */
247 public String[] getExcludedMimeTypes()
248 {
249 Set<String> excluded=_mimeTypes.getExcluded();
250 return excluded.toArray(new String[excluded.size()]);
251 }
252
253 /* ------------------------------------------------------------ */
254 public String[] getExcludedPaths()
255 {
256 Set<String> excluded=_paths.getExcluded();
257 return excluded.toArray(new String[excluded.size()]);
258 }
259
260 /* ------------------------------------------------------------ */
261 public String[] getIncludedAgentPatterns()
262 {
263 Set<String> includes=_agentPatterns.getIncluded();
264 return includes.toArray(new String[includes.size()]);
265 }
266
267 /* ------------------------------------------------------------ */
268 public String[] getIncludedMethods()
269 {
270 Set<String> includes=_methods.getIncluded();
271 return includes.toArray(new String[includes.size()]);
272 }
273
274 /* ------------------------------------------------------------ */
275 public String[] getIncludedMimeTypes()
276 {
277 Set<String> includes=_mimeTypes.getIncluded();
278 return includes.toArray(new String[includes.size()]);
279 }
280
281 /* ------------------------------------------------------------ */
282 public String[] getIncludedPaths()
283 {
284 Set<String> includes=_paths.getIncluded();
285 return includes.toArray(new String[includes.size()]);
286 }
287
288 /* ------------------------------------------------------------ */
289 /**
290 * Get the mime types.
291 *
292 * @return mime types to set
293 * @deprecated use {@link #getExcludedMimeTypes()} or {@link #getIncludedMimeTypes()} instead
294 */
295 @Deprecated
296 public Set<String> getMimeTypes()
297 {
298 throw new UnsupportedOperationException("Use getIncludedMimeTypes or getExcludedMimeTypes instead");
299 }
300
301 /* ------------------------------------------------------------ */
302 /**
303 * Set the mime types.
304 *
305 * @param mimeTypes
306 * the mime types to set
307 * @deprecated use {@link #setExcludedMimeTypes()} or {@link #setIncludedMimeTypes()} instead
308 */
309 @Deprecated
310 public void setMimeTypes(Set<String> mimeTypes)
311 {
312 throw new UnsupportedOperationException("Use setIncludedMimeTypes or setExcludedMimeTypes instead");
313 }
314
315 /* ------------------------------------------------------------ */
316 /**
317 * Set the mime types.
318 *
319 * @param mimeTypes
320 * the mime types to set
321 * @deprecated use {@link #setExcludedMimeTypes()} or {@link #setIncludedMimeTypes()} instead
322 */
323 @Deprecated
324 public void setMimeTypes(String mimeTypes)
325 {
326 throw new UnsupportedOperationException("Use setIncludedMimeTypes or setExcludedMimeTypes instead");
327 }
328
329 /* ------------------------------------------------------------ */
330 /**
331 * Set the mime types.
332 * @deprecated use {@link #setExcludedMimeTypes()} instead
333 */
334 @Deprecated
335 public void setExcludeMimeTypes(boolean exclude)
336 {
337 throw new UnsupportedOperationException("Use setExcludedMimeTypes instead");
338 }
339
340 /* ------------------------------------------------------------ */
341 /**
342 * Get the excluded user agents.
343 *
344 * @return excluded user agents
345 */
346 public Set<String> getExcluded()
347 {
348 return _agentPatterns.getExcluded();
349 }
350
351 /* ------------------------------------------------------------ */
352 /**
353 * Set the excluded user agents.
354 *
355 * @param excluded
356 * excluded user agents to set
357 */
358 public void setExcluded(Set<String> excluded)
359 {
360 _agentPatterns.getExcluded().clear();
361 _agentPatterns.getExcluded().addAll(excluded);
362 }
363
364 /* ------------------------------------------------------------ */
365 /**
366 * Set the excluded user agents.
367 *
368 * @param excluded
369 * excluded user agents to set
370 */
371 public void setExcluded(String excluded)
372 {
373 _agentPatterns.getExcluded().clear();
374
375 if (excluded != null)
376 {
377 _agentPatterns.exclude(StringUtil.csvSplit(excluded));
378 }
379 }
380
381 /* ------------------------------------------------------------ */
382 /**
383 * @return The value of the Vary header set if a response can be compressed.
384 */
385 public String getVary()
386 {
387 return _vary;
388 }
389
390 /* ------------------------------------------------------------ */
391 /**
392 * Set the value of the Vary header sent with responses that could be compressed.
393 * <p>
394 * By default it is set to 'Accept-Encoding, User-Agent' since IE6 is excluded by
395 * default from the excludedAgents. If user-agents are not to be excluded, then
396 * this can be set to 'Accept-Encoding'. Note also that shared caches may cache
397 * many copies of a resource that is varied by User-Agent - one per variation of the
398 * User-Agent, unless the cache does some normalization of the UA string.
399 * @param vary The value of the Vary header set if a response can be compressed.
400 */
401 public void setVary(String vary)
402 {
403 _vary = vary;
404 }
405
406 /* ------------------------------------------------------------ */
407 /**
408 * Get the buffer size.
409 *
410 * @return the buffer size
411 */
412 public int getBufferSize()
413 {
414 return _bufferSize;
415 }
416
417 /* ------------------------------------------------------------ */
418 /**
419 * Set the buffer size.
420 *
421 * @param bufferSize
422 * buffer size to set
423 */
424 public void setBufferSize(int bufferSize)
425 {
426 _bufferSize = bufferSize;
427 }
428
429 /* ------------------------------------------------------------ */
430 /**
431 * Get the minimum reponse size.
432 *
433 * @return minimum reponse size
434 */
435 public int getMinGzipSize()
436 {
437 return _minGzipSize;
438 }
439
440 /* ------------------------------------------------------------ */
441 /**
442 * Set the minimum reponse size.
443 *
444 * @param minGzipSize
445 * minimum reponse size
446 */
447 public void setMinGzipSize(int minGzipSize)
448 {
449 _minGzipSize = minGzipSize;
450 }
451
452 /* ------------------------------------------------------------ */
453 @Override
454 protected void doStart() throws Exception
455 {
456 super.doStart();
457 }
458
459 /* ------------------------------------------------------------ */
460 /**
461 * @see org.eclipse.jetty.server.handler.HandlerWrapper#handle(java.lang.String, org.eclipse.jetty.server.Request, javax.servlet.http.HttpServletRequest, javax.servlet.http.HttpServletResponse)
462 */
463 @Override
464 public void handle(String target, Request baseRequest, HttpServletRequest request, HttpServletResponse response) throws IOException, ServletException
465 {
466 if(_handler == null || !isStarted())
467 {
468 // do nothing
469 return;
470 }
471
472 if(isGzippable(baseRequest, request, response))
473 {
474 final CompressedResponseWrapper wrappedResponse = newGzipResponseWrapper(request,response);
475
476 boolean exceptional=true;
477 try
478 {
479 _handler.handle(target, baseRequest, request, wrappedResponse);
480 exceptional=false;
481 }
482 finally
483 {
484 if (request.isAsyncStarted())
485 {
486 request.getAsyncContext().addListener(new AsyncListener()
487 {
488
489 @Override
490 public void onTimeout(AsyncEvent event) throws IOException
491 {
492 }
493
494 @Override
495 public void onStartAsync(AsyncEvent event) throws IOException
496 {
497 }
498
499 @Override
500 public void onError(AsyncEvent event) throws IOException
501 {
502 }
503
504 @Override
505 public void onComplete(AsyncEvent event) throws IOException
506 {
507 try
508 {
509 wrappedResponse.finish();
510 }
511 catch(IOException e)
512 {
513 LOG.warn(e);
514 }
515 }
516 });
517 }
518 else if (exceptional && !response.isCommitted())
519 {
520 wrappedResponse.resetBuffer();
521 wrappedResponse.noCompression();
522 }
523 else
524 wrappedResponse.finish();
525 }
526 }
527 else
528 {
529 _handler.handle(target,baseRequest, request, response);
530 }
531 }
532
533 private boolean isGzippable(Request baseRequest, HttpServletRequest request, HttpServletResponse response)
534 {
535 String ae = request.getHeader("accept-encoding");
536 if (ae == null || !ae.contains("gzip"))
537 {
538 // Request not indicated for Gzip
539 return false;
540 }
541
542 if(response.containsHeader("Content-Encoding"))
543 {
544 // Response is already declared, can't gzip
545 LOG.debug("{} excluded as Content-Encoding already declared {}",this,request);
546 return false;
547 }
548
549 if(HttpMethod.HEAD.is(request.getMethod()))
550 {
551 // HEAD is never Gzip'd
552 LOG.debug("{} excluded by method {}",this,request);
553 return false;
554 }
555
556 // Exclude based on Request Method
557 if (!_methods.matches(baseRequest.getMethod()))
558 {
559 LOG.debug("{} excluded by method {}",this,request);
560 return false;
561 }
562
563 // Exclude based on Request Path
564 ServletContext context = baseRequest.getServletContext();
565 String path = context==null?baseRequest.getRequestURI():URIUtil.addPaths(baseRequest.getServletPath(),baseRequest.getPathInfo());
566
567 if(path != null && !_paths.matches(path))
568 {
569 LOG.debug("{} excluded by path {}",this,request);
570 return false;
571 }
572
573
574 // Exclude non compressible mime-types known from URI extension. - no Vary because no matter what client, this URI is always excluded
575 String mimeType = context==null?null:context.getMimeType(path);
576 if (mimeType!=null)
577 {
578 mimeType = MimeTypes.getContentTypeWithoutCharset(mimeType);
579 if (!_mimeTypes.matches(mimeType))
580 {
581 LOG.debug("{} excluded by path suffix mime type {}",this,request);
582 return false;
583 }
584 }
585
586 // Exclude on User Agent
587 String ua = request.getHeader("User-Agent");
588 if(ua != null && !_agentPatterns.matches(ua))
589 {
590 LOG.debug("{} excluded by user-agent {}",this,request);
591 return false;
592 }
593
594 return true;
595 }
596
597 /**
598 * Allows derived implementations to replace ResponseWrapper implementation.
599 *
600 * @param request the request
601 * @param response the response
602 * @return the gzip response wrapper
603 */
604 protected CompressedResponseWrapper newGzipResponseWrapper(HttpServletRequest request, HttpServletResponse response)
605 {
606 return new CompressedResponseWrapper(request,response)
607 {
608 {
609 super.setMimeTypes(GzipHandler.this._mimeTypes);
610 super.setBufferSize(GzipHandler.this._bufferSize);
611 super.setMinCompressSize(GzipHandler.this._minGzipSize);
612 }
613
614 @Override
615 protected AbstractCompressedStream newCompressedStream(HttpServletRequest request,HttpServletResponse response) throws IOException
616 {
617 return new AbstractCompressedStream("gzip",request,this,_vary)
618 {
619 @Override
620 protected DeflaterOutputStream createStream() throws IOException
621 {
622 return new GZIPOutputStream(_response.getOutputStream(),_bufferSize);
623 }
624 };
625 }
626
627 @Override
628 protected PrintWriter newWriter(OutputStream out,String encoding) throws UnsupportedEncodingException
629 {
630 return GzipHandler.this.newWriter(out,encoding);
631 }
632 };
633 }
634
635 /**
636 * Allows derived implementations to replace PrintWriter implementation.
637 *
638 * @param out the out
639 * @param encoding the encoding
640 * @return the prints the writer
641 * @throws UnsupportedEncodingException
642 */
643 protected PrintWriter newWriter(OutputStream out,String encoding) throws UnsupportedEncodingException
644 {
645 return encoding==null?new PrintWriter(out):new PrintWriter(new OutputStreamWriter(out,encoding));
646 }
647 }