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 _portValue;
65 int _path;
66 int _param;
67 int _query;
68 int _fragment;
69 int _end;
70 boolean _encoded=false;
71
72 final Utf8StringBuilder _utf8b = new Utf8StringBuilder(64);
73
74 public HttpURI()
75 {
76
77 }
78
79 /* ------------------------------------------------------------ */
80 /**
81 * @param parsePartialAuth If True, parse auth without prior scheme, else treat all URIs starting with / as paths
82 */
83 public HttpURI(boolean parsePartialAuth)
84 {
85 _partial=parsePartialAuth;
86 }
87
88 public HttpURI(String raw)
89 {
90 _rawString=raw;
91 byte[] b = raw.getBytes();
92 parse(b,0,b.length);
93 }
94
95 public HttpURI(byte[] raw,int offset, int length)
96 {
97 parse2(raw,offset,length);
98 }
99
100 public void parse(String raw)
101 {
102 byte[] b = raw.getBytes();
103 parse2(b,0,b.length);
104 _rawString=raw;
105 }
106
107 public void parse(byte[] raw,int offset, int length)
108 {
109 _rawString=null;
110 parse2(raw,offset,length);
111 }
112
113 private void parse2(byte[] raw,int offset, int length)
114 {
115 _encoded=false;
116 _raw=raw;
117 int i=offset;
118 int e=offset+length;
119 int state=START;
120 int m=offset;
121 _end=offset+length;
122 _scheme=offset;
123 _authority=offset;
124 _host=offset;
125 _port=offset;
126 _portValue=-1;
127 _path=offset;
128 _param=_end;
129 _query=_end;
130 _fragment=_end;
131 while (i<e)
132 {
133 char c=(char)(0xff&_raw[i]);
134 int s=i++;
135
136 state: switch (state)
137 {
138 case START:
139 {
140 m=s;
141 switch(c)
142 {
143 case '/':
144 state=AUTH_OR_PATH;
145 break;
146 case ';':
147 _param=s;
148 state=PARAM;
149 break;
150 case '?':
151 _param=s;
152 _query=s;
153 state=QUERY;
154 break;
155 case '#':
156 _param=s;
157 _query=s;
158 _fragment=s;
159 break;
160 case '*':
161 _path=s;
162 state=ASTERISK;
163 break;
164
165 default:
166 if (Character.isLetterOrDigit(c))
167 state=SCHEME_OR_PATH;
168 else
169 throw new IllegalArgumentException("!(SCHEME|PATH|AUTH):"+StringUtil.toString(_raw,offset,length,URIUtil.__CHARSET));
170 }
171
172 continue;
173 }
174
175 case AUTH_OR_PATH:
176 {
177 if ((_partial||_scheme!=_authority) && c=='/')
178 {
179 _host=i;
180 _port=_end;
181 _path=_end;
182 state=AUTH;
183 }
184 else if (c==';' || c=='?' || c=='#')
185 {
186 i--;
187 state=PATH;
188 }
189 else
190 {
191 _host=m;
192 _port=m;
193 state=PATH;
194 }
195 continue;
196 }
197
198 case SCHEME_OR_PATH:
199 {
200 // short cut for http and https
201 if (length>6 && c=='t')
202 {
203 if (_raw[offset+3]==':')
204 {
205 s=offset+3;
206 i=offset+4;
207 c=':';
208 }
209 else if (_raw[offset+4]==':')
210 {
211 s=offset+4;
212 i=offset+5;
213 c=':';
214 }
215 else if (_raw[offset+5]==':')
216 {
217 s=offset+5;
218 i=offset+6;
219 c=':';
220 }
221 }
222
223 switch (c)
224 {
225 case ':':
226 {
227 m = i++;
228 _authority = m;
229 _path = m;
230 c = (char)(0xff & _raw[i]);
231 if (c == '/')
232 state = AUTH_OR_PATH;
233 else
234 {
235 _host = m;
236 _port = m;
237 state = PATH;
238 }
239 break;
240 }
241
242 case '/':
243 {
244 state = PATH;
245 break;
246 }
247
248 case ';':
249 {
250 _param = s;
251 state = PARAM;
252 break;
253 }
254
255 case '?':
256 {
257 _param = s;
258 _query = s;
259 state = QUERY;
260 break;
261 }
262
263 case '#':
264 {
265 _param = s;
266 _query = s;
267 _fragment = s;
268 break;
269 }
270 }
271 continue;
272 }
273
274 case AUTH:
275 {
276 switch (c)
277 {
278
279 case '/':
280 {
281 m = s;
282 _path = m;
283 _port = _path;
284 state = PATH;
285 break;
286 }
287 case '@':
288 {
289 _host = i;
290 break;
291 }
292 case ':':
293 {
294 _port = s;
295 state = PORT;
296 break;
297 }
298 case '[':
299 {
300 state = IPV6;
301 break;
302 }
303 }
304 continue;
305 }
306
307 case IPV6:
308 {
309 switch (c)
310 {
311 case '/':
312 {
313 throw new IllegalArgumentException("No closing ']' for " + StringUtil.toString(_raw,offset,length,URIUtil.__CHARSET));
314 }
315 case ']':
316 {
317 state = AUTH;
318 break;
319 }
320 }
321
322 continue;
323 }
324
325 case PORT:
326 {
327 if (c=='/')
328 {
329 m=s;
330 _path=m;
331 if (_port<=_authority)
332 _port=_path;
333 state=PATH;
334 }
335 continue;
336 }
337
338 case PATH:
339 {
340 switch (c)
341 {
342 case ';':
343 {
344 _param = s;
345 state = PARAM;
346 break;
347 }
348 case '?':
349 {
350 _param = s;
351 _query = s;
352 state = QUERY;
353 break;
354 }
355 case '#':
356 {
357 _param = s;
358 _query = s;
359 _fragment = s;
360 break state;
361 }
362 case '%':
363 {
364 _encoded=true;
365 }
366 }
367 continue;
368 }
369
370 case PARAM:
371 {
372 switch (c)
373 {
374 case '?':
375 {
376 _query = s;
377 state = QUERY;
378 break;
379 }
380 case '#':
381 {
382 _query = s;
383 _fragment = s;
384 break state;
385 }
386 }
387 continue;
388 }
389
390 case QUERY:
391 {
392 if (c=='#')
393 {
394 _fragment=s;
395 break state;
396 }
397 continue;
398 }
399
400 case ASTERISK:
401 {
402 throw new IllegalArgumentException("only '*'");
403 }
404 }
405 }
406
407 if (_port<_path)
408 _portValue=TypeUtil.parseInt(_raw, _port+1, _path-_port-1,10);
409 }
410
411 private String toUtf8String(int offset,int length)
412 {
413 _utf8b.reset();
414 _utf8b.append(_raw,offset,length);
415 return _utf8b.toString();
416 }
417
418 public String getScheme()
419 {
420 if (_scheme==_authority)
421 return null;
422 int l=_authority-_scheme;
423 if (l==5 &&
424 _raw[_scheme]=='h' &&
425 _raw[_scheme+1]=='t' &&
426 _raw[_scheme+2]=='t' &&
427 _raw[_scheme+3]=='p' )
428 return HttpSchemes.HTTP;
429 if (l==6 &&
430 _raw[_scheme]=='h' &&
431 _raw[_scheme+1]=='t' &&
432 _raw[_scheme+2]=='t' &&
433 _raw[_scheme+3]=='p' &&
434 _raw[_scheme+4]=='s' )
435 return HttpSchemes.HTTPS;
436
437 return toUtf8String(_scheme,_authority-_scheme-1);
438 }
439
440 public String getAuthority()
441 {
442 if (_authority==_path)
443 return null;
444 return toUtf8String(_authority,_path-_authority);
445 }
446
447 public String getHost()
448 {
449 if (_host==_port)
450 return null;
451 return toUtf8String(_host,_port-_host);
452 }
453
454 public int getPort()
455 {
456 return _portValue;
457 }
458
459 public String getPath()
460 {
461 if (_path==_param)
462 return null;
463 return toUtf8String(_path,_param-_path);
464 }
465
466 public String getDecodedPath()
467 {
468 if (_path==_param)
469 return null;
470
471 int length = _param-_path;
472 byte[] bytes=null;
473 int n=0;
474
475 for (int i=_path;i<_param;i++)
476 {
477 byte b = _raw[i];
478
479 if (b=='%')
480 {
481 if ((i+2)>=_param)
482 throw new IllegalArgumentException("Bad % encoding: "+this);
483 b=(byte)(0xff&TypeUtil.parseInt(_raw,i+1,2,16));
484 i+=2;
485 }
486 else if (bytes==null)
487 {
488 n++;
489 continue;
490 }
491
492 if (bytes==null)
493 {
494 bytes=new byte[length];
495 System.arraycopy(_raw,_path,bytes,0,n);
496 }
497
498 bytes[n++]=b;
499 }
500
501 if (bytes==null)
502 return toUtf8String(_path,length);
503
504 _utf8b.reset();
505 _utf8b.append(bytes,0,n);
506 return _utf8b.toString();
507 }
508
509 public String getPathAndParam()
510 {
511 if (_path==_query)
512 return null;
513 return toUtf8String(_path,_query-_path);
514 }
515
516 public String getCompletePath()
517 {
518 if (_path==_end)
519 return null;
520 return toUtf8String(_path,_end-_path);
521 }
522
523 public String getParam()
524 {
525 if (_param==_query)
526 return null;
527 return toUtf8String(_param+1,_query-_param-1);
528 }
529
530 public String getQuery()
531 {
532 if (_query==_fragment)
533 return null;
534 return toUtf8String(_query+1,_fragment-_query-1);
535 }
536
537 public String getQuery(String encoding)
538 {
539 if (_query==_fragment)
540 return null;
541 return StringUtil.toString(_raw,_query+1,_fragment-_query-1,encoding);
542 }
543
544 public boolean hasQuery()
545 {
546 return (_fragment>_query);
547 }
548
549 public String getFragment()
550 {
551 if (_fragment==_end)
552 return null;
553 return toUtf8String(_fragment+1,_end-_fragment-1);
554 }
555
556 public void decodeQueryTo(MultiMap parameters)
557 {
558 if (_query==_fragment)
559 return;
560 _utf8b.reset();
561 UrlEncoded.decodeUtf8To(_raw,_query+1,_fragment-_query-1,parameters,_utf8b);
562 }
563
564 public void decodeQueryTo(MultiMap parameters, String encoding)
565 throws UnsupportedEncodingException
566 {
567 if (_query==_fragment)
568 return;
569
570 if (encoding==null || StringUtil.isUTF8(encoding))
571 UrlEncoded.decodeUtf8To(_raw,_query+1,_fragment-_query-1,parameters);
572 else
573 UrlEncoded.decodeTo(toUtf8String(_query+1,_fragment-_query-1),parameters,encoding);
574 }
575
576 public void clear()
577 {
578 _scheme=_authority=_host=_port=_path=_param=_query=_fragment=_end=0;
579 _raw=__empty;
580 _rawString="";
581 _encoded=false;
582 }
583
584 @Override
585 public String toString()
586 {
587 if (_rawString==null)
588 _rawString=toUtf8String(_scheme,_end-_scheme);
589 return _rawString;
590 }
591
592 public void writeTo(Utf8StringBuilder buf)
593 {
594 buf.append(_raw,_scheme,_end-_scheme);
595 }
596
597 }