1 // ========================================================================
2 // Copyright (c) 2006-2009 Mort Bay Consulting Pty. Ltd.
3 // ------------------------------------------------------------------------
4 // All rights reserved. This program and the accompanying materials
5 // are made available under the terms of the Eclipse Public License v1.0
6 // and Apache License v2.0 which accompanies this distribution.
7 // The Eclipse Public License is available at
8 // http://www.eclipse.org/legal/epl-v10.html
9 // The Apache License v2.0 is available at
10 // http://www.opensource.org/licenses/apache2.0.php
11 // You may elect to redistribute this code under either of these licenses.
12 // ========================================================================
13
14 package org.eclipse.jetty.http;
15
16 import java.io.UnsupportedEncodingException;
17
18 import org.eclipse.jetty.util.MultiMap;
19 import org.eclipse.jetty.util.StringUtil;
20 import org.eclipse.jetty.util.TypeUtil;
21 import org.eclipse.jetty.util.URIUtil;
22 import org.eclipse.jetty.util.UrlEncoded;
23 import org.eclipse.jetty.util.Utf8StringBuilder;
24
25
26 /* ------------------------------------------------------------ */
27 /** Http URI.
28 * Parse a HTTP URI from a string or byte array. Given a URI
29 * <code>http://user@host:port/path/info;param?query#fragment</code>
30 * this class will split it into the following undecoded optional elements:<ul>
31 * <li>{@link #getScheme()} - http:</li>
32 * <li>{@link #getAuthority()} - //name@host:port</li>
33 * <li>{@link #getHost()} - host</li>
34 * <li>{@link #getPort()} - port</li>
35 * <li>{@link #getPath()} - /path/info</li>
36 * <li>{@link #getParam()} - param</li>
37 * <li>{@link #getQuery()} - query</li>
38 * <li>{@link #getFragment()} - fragment</li>
39 * </ul>
40 *
41 */
42 public class HttpURI
43 {
44 private static final byte[] __empty={};
45 private final static int
46 START=0,
47 AUTH_OR_PATH=1,
48 SCHEME_OR_PATH=2,
49 AUTH=4,
50 IPV6=5,
51 PORT=6,
52 PATH=7,
53 PARAM=8,
54 QUERY=9,
55 ASTERISK=10;
56
57 boolean _partial=false;
58 byte[] _raw=__empty;
59 String _rawString;
60 int _scheme;
61 int _authority;
62 int _host;
63 int _port;
64 int _path;
65 int _param;
66 int _query;
67 int _fragment;
68 int _end;
69 boolean _encoded=false;
70
71 final Utf8StringBuilder _utf8b = new Utf8StringBuilder(64);
72
73 public HttpURI()
74 {
75
76 }
77
78 /* ------------------------------------------------------------ */
79 /**
80 * @param parsePartialAuth If True, parse auth without prior scheme, else treat all URIs starting with / as paths
81 */
82 public HttpURI(boolean parsePartialAuth)
83 {
84 _partial=parsePartialAuth;
85 }
86
87 public HttpURI(String raw)
88 {
89 _rawString=raw;
90 byte[] b = raw.getBytes();
91 parse(b,0,b.length);
92 }
93
94 public HttpURI(byte[] raw,int offset, int length)
95 {
96 parse2(raw,offset,length);
97 }
98
99 public void parse(String raw)
100 {
101 byte[] b = raw.getBytes();
102 parse2(b,0,b.length);
103 _rawString=raw;
104 }
105
106 public void parse(byte[] raw,int offset, int length)
107 {
108 _rawString=null;
109 parse2(raw,offset,length);
110 }
111
112 private void parse2(byte[] raw,int offset, int length)
113 {
114 _encoded=false;
115 _raw=raw;
116 int i=offset;
117 int e=offset+length;
118 int state=START;
119 int m=offset;
120 _end=offset+length;
121 _scheme=offset;
122 _authority=offset;
123 _host=offset;
124 _port=offset;
125 _path=offset;
126 _param=_end;
127 _query=_end;
128 _fragment=_end;
129 while (i<e)
130 {
131 char c=(char)(0xff&_raw[i]);
132 int s=i++;
133
134 state: switch (state)
135 {
136 case START:
137 {
138 m=s;
139 switch(c)
140 {
141 case '/':
142 state=AUTH_OR_PATH;
143 break;
144 case ';':
145 _param=s;
146 state=PARAM;
147 break;
148 case '?':
149 _param=s;
150 _query=s;
151 state=QUERY;
152 break;
153 case '#':
154 _param=s;
155 _query=s;
156 _fragment=s;
157 break;
158 case '*':
159 _path=s;
160 state=ASTERISK;
161 break;
162
163 default:
164 if (Character.isLetterOrDigit(c))
165 state=SCHEME_OR_PATH;
166 else
167 throw new IllegalArgumentException(StringUtil.toString(_raw,offset,length,URIUtil.__CHARSET));
168 }
169
170 continue;
171 }
172
173 case AUTH_OR_PATH:
174 {
175 if ((_partial||_scheme!=_authority) && c=='/')
176 {
177 _host=i;
178 _port=_end;
179 _path=_end;
180 state=AUTH;
181 }
182 else if (c==';' || c=='?' || c=='#')
183 {
184 i--;
185 state=PATH;
186 }
187 else
188 {
189 _host=m;
190 _port=m;
191 state=PATH;
192 }
193 continue;
194 }
195
196 case SCHEME_OR_PATH:
197 {
198 // short cut for http and https
199 if (length>6 && c=='t')
200 {
201 if (_raw[offset+3]==':')
202 {
203 s=offset+3;
204 i=offset+4;
205 c=':';
206 }
207 else if (_raw[offset+4]==':')
208 {
209 s=offset+4;
210 i=offset+5;
211 c=':';
212 }
213 else if (_raw[offset+5]==':')
214 {
215 s=offset+5;
216 i=offset+6;
217 c=':';
218 }
219 }
220
221 switch (c)
222 {
223 case ':':
224 {
225 m = i++;
226 _authority = m;
227 _path = m;
228 c = (char)(0xff & _raw[i]);
229 if (c == '/')
230 state = AUTH_OR_PATH;
231 else
232 {
233 _host = m;
234 _port = m;
235 state = PATH;
236 }
237 break;
238 }
239
240 case '/':
241 {
242 state = PATH;
243 break;
244 }
245
246 case ';':
247 {
248 _param = s;
249 state = PARAM;
250 break;
251 }
252
253 case '?':
254 {
255 _param = s;
256 _query = s;
257 state = QUERY;
258 break;
259 }
260
261 case '#':
262 {
263 _param = s;
264 _query = s;
265 _fragment = s;
266 break;
267 }
268 }
269 continue;
270 }
271
272 case AUTH:
273 {
274 switch (c)
275 {
276
277 case '/':
278 {
279 m = s;
280 _path = m;
281 _port = _path;
282 state = PATH;
283 break;
284 }
285 case '@':
286 {
287 _host = i;
288 break;
289 }
290 case ':':
291 {
292 _port = s;
293 state = PORT;
294 break;
295 }
296 case '[':
297 {
298 state = IPV6;
299 break;
300 }
301 }
302 continue;
303 }
304
305 case IPV6:
306 {
307 switch (c)
308 {
309 case '/':
310 {
311 throw new IllegalArgumentException("No closing ']' for " + StringUtil.toString(_raw,offset,length,URIUtil.__CHARSET));
312 }
313 case ']':
314 {
315 state = AUTH;
316 break;
317 }
318 }
319
320 continue;
321 }
322
323 case PORT:
324 {
325 if (c=='/')
326 {
327 m=s;
328 _path=m;
329 if (_port<=_authority)
330 _port=_path;
331 state=PATH;
332 }
333 continue;
334 }
335
336 case PATH:
337 {
338 switch (c)
339 {
340 case ';':
341 {
342 _param = s;
343 state = PARAM;
344 break;
345 }
346 case '?':
347 {
348 _param = s;
349 _query = s;
350 state = QUERY;
351 break;
352 }
353 case '#':
354 {
355 _param = s;
356 _query = s;
357 _fragment = s;
358 break state;
359 }
360 case '%':
361 {
362 _encoded=true;
363 }
364 }
365 continue;
366 }
367
368 case PARAM:
369 {
370 switch (c)
371 {
372 case '?':
373 {
374 _query = s;
375 state = QUERY;
376 break;
377 }
378 case '#':
379 {
380 _query = s;
381 _fragment = s;
382 break state;
383 }
384 }
385 continue;
386 }
387
388 case QUERY:
389 {
390 if (c=='#')
391 {
392 _fragment=s;
393 break state;
394 }
395 continue;
396 }
397
398 case ASTERISK:
399 {
400 throw new IllegalArgumentException("only '*'");
401 }
402 }
403 }
404 }
405
406 private String toUtf8String(int offset,int length)
407 {
408 _utf8b.reset();
409 _utf8b.append(_raw,offset,length);
410 return _utf8b.toString();
411 }
412
413 public String getScheme()
414 {
415 if (_scheme==_authority)
416 return null;
417 int l=_authority-_scheme;
418 if (l==5 &&
419 _raw[_scheme]=='h' &&
420 _raw[_scheme+1]=='t' &&
421 _raw[_scheme+2]=='t' &&
422 _raw[_scheme+3]=='p' )
423 return HttpSchemes.HTTP;
424 if (l==6 &&
425 _raw[_scheme]=='h' &&
426 _raw[_scheme+1]=='t' &&
427 _raw[_scheme+2]=='t' &&
428 _raw[_scheme+3]=='p' &&
429 _raw[_scheme+4]=='s' )
430 return HttpSchemes.HTTPS;
431
432 return toUtf8String(_scheme,_authority-_scheme-1);
433 }
434
435 public String getAuthority()
436 {
437 if (_authority==_path)
438 return null;
439 return toUtf8String(_authority,_path-_authority);
440 }
441
442 public String getHost()
443 {
444 if (_host==_port)
445 return null;
446 return toUtf8String(_host,_port-_host);
447 }
448
449 public int getPort()
450 {
451 if (_port==_path)
452 return -1;
453 return TypeUtil.parseInt(_raw, _port+1, _path-_port-1,10);
454 }
455
456 public String getPath()
457 {
458 if (_path==_param)
459 return null;
460 return toUtf8String(_path,_param-_path);
461 }
462
463 public String getDecodedPath()
464 {
465 if (_path==_param)
466 return null;
467
468 int length = _param-_path;
469 byte[] bytes=null;
470 int n=0;
471
472 for (int i=_path;i<_param;i++)
473 {
474 byte b = _raw[i];
475
476 if (b=='%' && (i+2)<_param)
477 {
478 b=(byte)(0xff&TypeUtil.parseInt(_raw,i+1,2,16));
479 i+=2;
480 }
481 else if (bytes==null)
482 {
483 n++;
484 continue;
485 }
486
487 if (bytes==null)
488 {
489 bytes=new byte[length];
490 System.arraycopy(_raw,_path,bytes,0,n);
491 }
492
493 bytes[n++]=b;
494 }
495
496 if (bytes==null)
497 return toUtf8String(_path,length);
498
499 _utf8b.reset();
500 _utf8b.append(bytes,0,n);
501 return _utf8b.toString();
502 }
503
504 public String getPathAndParam()
505 {
506 if (_path==_query)
507 return null;
508 return toUtf8String(_path,_query-_path);
509 }
510
511 public String getCompletePath()
512 {
513 if (_path==_end)
514 return null;
515 return toUtf8String(_path,_end-_path);
516 }
517
518 public String getParam()
519 {
520 if (_param==_query)
521 return null;
522 return toUtf8String(_param+1,_query-_param-1);
523 }
524
525 public String getQuery()
526 {
527 if (_query==_fragment)
528 return null;
529 return toUtf8String(_query+1,_fragment-_query-1);
530 }
531
532 public String getQuery(String encoding)
533 {
534 if (_query==_fragment)
535 return null;
536 return StringUtil.toString(_raw,_query+1,_fragment-_query-1,encoding);
537 }
538
539 public boolean hasQuery()
540 {
541 return (_fragment>_query);
542 }
543
544 public String getFragment()
545 {
546 if (_fragment==_end)
547 return null;
548 return toUtf8String(_fragment+1,_end-_fragment-1);
549 }
550
551 public void decodeQueryTo(MultiMap parameters)
552 {
553 if (_query==_fragment)
554 return;
555 _utf8b.reset();
556 UrlEncoded.decodeUtf8To(_raw,_query+1,_fragment-_query-1,parameters,_utf8b);
557 }
558
559 public void decodeQueryTo(MultiMap parameters, String encoding)
560 throws UnsupportedEncodingException
561 {
562 if (_query==_fragment)
563 return;
564
565 if (encoding==null || StringUtil.isUTF8(encoding))
566 UrlEncoded.decodeUtf8To(_raw,_query+1,_fragment-_query-1,parameters);
567 else
568 UrlEncoded.decodeTo(toUtf8String(_query+1,_fragment-_query-1),parameters,encoding);
569 }
570
571 public void clear()
572 {
573 _scheme=_authority=_host=_port=_path=_param=_query=_fragment=_end=0;
574 _raw=__empty;
575 _rawString="";
576 _encoded=false;
577 }
578
579 public String toString()
580 {
581 if (_rawString==null)
582 _rawString=toUtf8String(_scheme,_end-_scheme);
583 return _rawString;
584 }
585
586 public void writeTo(Utf8StringBuilder buf)
587 {
588 buf.append(_raw,_scheme,_end-_scheme);
589 }
590
591 }