1 /*
2 * Copyright 2000-2006 Sun Microsystems, Inc. All Rights Reserved.
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 *
5 * This code is free software; you can redistribute it and/or modify it
6 * under the terms of the GNU General Public License version 2 only, as
7 * published by the Free Software Foundation. Sun designates this
8 * particular file as subject to the "Classpath" exception as provided
9 * by Sun in the LICENSE file that accompanied this code.
10 *
11 * This code is distributed in the hope that it will be useful, but WITHOUT
12 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
13 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
14 * version 2 for more details (a copy is included in the LICENSE file that
15 * accompanied this code).
16 *
17 * You should have received a copy of the GNU General Public License version
18 * 2 along with this work; if not, write to the Free Software Foundation,
19 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
20 *
21 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
22 * CA 95054 USA or visit www.sun.com if you need additional information or
23 * have any questions.
24 */
25
26 package java.net;
27
28 import java.io.IOException;
29 import java.io.InvalidObjectException;
30 import java.io.ObjectInputStream;
31 import java.io.ObjectOutputStream;
32 import java.io.Serializable;
33 import java.nio.ByteBuffer;
34 import java.nio.CharBuffer;
35 import java.nio.charset.CharsetDecoder;
36 import java.nio.charset.CharsetEncoder;
37 import java.nio.charset.CoderResult;
38 import java.nio.charset.CodingErrorAction;
39 import java.nio.charset.CharacterCodingException;
40 import java.text.Normalizer;
41 import sun.nio.cs.ThreadLocalCoders;
42
43 import java.lang.Character; // for javadoc
44 import java.lang.NullPointerException; // for javadoc
45
46
47 /**
48 * Represents a Uniform Resource Identifier (URI) reference.
49 *
50 * <p> Aside from some minor deviations noted below, an instance of this
51 * class represents a URI reference as defined by
52 * <a href="http://www.ietf.org/rfc/rfc2396.txt""><i>RFC 2396: Uniform
53 * Resource Identifiers (URI): Generic Syntax</i></a>, amended by <a
54 * href="http://www.ietf.org/rfc/rfc2732.txt"><i>RFC 2732: Format for
55 * Literal IPv6 Addresses in URLs</i></a>. The Literal IPv6 address format
56 * also supports scope_ids. The syntax and usage of scope_ids is described
57 * <a href="Inet6Address.html#scoped">here</a>.
58 * This class provides constructors for creating URI instances from
59 * their components or by parsing their string forms, methods for accessing the
60 * various components of an instance, and methods for normalizing, resolving,
61 * and relativizing URI instances. Instances of this class are immutable.
62 *
63 *
64 * <h4> URI syntax and components </h4>
65 *
66 * At the highest level a URI reference (hereinafter simply "URI") in string
67 * form has the syntax
68 *
69 * <blockquote>
70 * [<i>scheme</i><tt><b>:</b></tt><i></i>]<i>scheme-specific-part</i>[<tt><b>#</b></tt><i>fragment</i>]
71 * </blockquote>
72 *
73 * where square brackets [...] delineate optional components and the characters
74 * <tt><b>:</b></tt> and <tt><b>#</b></tt> stand for themselves.
75 *
76 * <p> An <i>absolute</i> URI specifies a scheme; a URI that is not absolute is
77 * said to be <i>relative</i>. URIs are also classified according to whether
78 * they are <i>opaque</i> or <i>hierarchical</i>.
79 *
80 * <p> An <i>opaque</i> URI is an absolute URI whose scheme-specific part does
81 * not begin with a slash character (<tt>'/'</tt>). Opaque URIs are not
82 * subject to further parsing. Some examples of opaque URIs are:
83 *
84 * <blockquote><table cellpadding=0 cellspacing=0 summary="layout">
85 * <tr><td><tt>mailto:java-net@java.sun.com</tt><td></tr>
86 * <tr><td><tt>news:comp.lang.java</tt><td></tr>
87 * <tr><td><tt>urn:isbn:096139210x</tt></td></tr>
88 * </table></blockquote>
89 *
90 * <p> A <i>hierarchical</i> URI is either an absolute URI whose
91 * scheme-specific part begins with a slash character, or a relative URI, that
92 * is, a URI that does not specify a scheme. Some examples of hierarchical
93 * URIs are:
94 *
95 * <blockquote>
96 * <tt>http://java.sun.com/j2se/1.3/</tt><br>
97 * <tt>docs/guide/collections/designfaq.html#28</tt><br>
98 * <tt>../../../demo/jfc/SwingSet2/src/SwingSet2.java</tt><br>
99 * <tt>file:///~/calendar</tt>
100 * </blockquote>
101 *
102 * <p> A hierarchical URI is subject to further parsing according to the syntax
103 *
104 * <blockquote>
105 * [<i>scheme</i><tt><b>:</b></tt>][<tt><b>//</b></tt><i>authority</i>][<i>path</i>][<tt><b>?</b></tt><i>query</i>][<tt><b>#</b></tt><i>fragment</i>]
106 * </blockquote>
107 *
108 * where the characters <tt><b>:</b></tt>, <tt><b>/</b></tt>,
109 * <tt><b>?</b></tt>, and <tt><b>#</b></tt> stand for themselves. The
110 * scheme-specific part of a hierarchical URI consists of the characters
111 * between the scheme and fragment components.
112 *
113 * <p> The authority component of a hierarchical URI is, if specified, either
114 * <i>server-based</i> or <i>registry-based</i>. A server-based authority
115 * parses according to the familiar syntax
116 *
117 * <blockquote>
118 * [<i>user-info</i><tt><b>@</b></tt>]<i>host</i>[<tt><b>:</b></tt><i>port</i>]
119 * </blockquote>
120 *
121 * where the characters <tt><b>@</b></tt> and <tt><b>:</b></tt> stand for
122 * themselves. Nearly all URI schemes currently in use are server-based. An
123 * authority component that does not parse in this way is considered to be
124 * registry-based.
125 *
126 * <p> The path component of a hierarchical URI is itself said to be absolute
127 * if it begins with a slash character (<tt>'/'</tt>); otherwise it is
128 * relative. The path of a hierarchical URI that is either absolute or
129 * specifies an authority is always absolute.
130 *
131 * <p> All told, then, a URI instance has the following nine components:
132 *
133 * <blockquote><table summary="Describes the components of a URI:scheme,scheme-specific-part,authority,user-info,host,port,path,query,fragment">
134 * <tr><th><i>Component</i></th><th><i>Type</i></th></tr>
135 * <tr><td>scheme</td><td><tt>String</tt></td></tr>
136 * <tr><td>scheme-specific-part </td><td><tt>String</tt></td></tr>
137 * <tr><td>authority</td><td><tt>String</tt></td></tr>
138 * <tr><td>user-info</td><td><tt>String</tt></td></tr>
139 * <tr><td>host</td><td><tt>String</tt></td></tr>
140 * <tr><td>port</td><td><tt>int</tt></td></tr>
141 * <tr><td>path</td><td><tt>String</tt></td></tr>
142 * <tr><td>query</td><td><tt>String</tt></td></tr>
143 * <tr><td>fragment</td><td><tt>String</tt></td></tr>
144 * </table></blockquote>
145 *
146 * In a given instance any particular component is either <i>undefined</i> or
147 * <i>defined</i> with a distinct value. Undefined string components are
148 * represented by <tt>null</tt>, while undefined integer components are
149 * represented by <tt>-1</tt>. A string component may be defined to have the
150 * empty string as its value; this is not equivalent to that component being
151 * undefined.
152 *
153 * <p> Whether a particular component is or is not defined in an instance
154 * depends upon the type of the URI being represented. An absolute URI has a
155 * scheme component. An opaque URI has a scheme, a scheme-specific part, and
156 * possibly a fragment, but has no other components. A hierarchical URI always
157 * has a path (though it may be empty) and a scheme-specific-part (which at
158 * least contains the path), and may have any of the other components. If the
159 * authority component is present and is server-based then the host component
160 * will be defined and the user-information and port components may be defined.
161 *
162 *
163 * <h4> Operations on URI instances </h4>
164 *
165 * The key operations supported by this class are those of
166 * <i>normalization</i>, <i>resolution</i>, and <i>relativization</i>.
167 *
168 * <p> <i>Normalization</i> is the process of removing unnecessary <tt>"."</tt>
169 * and <tt>".."</tt> segments from the path component of a hierarchical URI.
170 * Each <tt>"."</tt> segment is simply removed. A <tt>".."</tt> segment is
171 * removed only if it is preceded by a non-<tt>".."</tt> segment.
172 * Normalization has no effect upon opaque URIs.
173 *
174 * <p> <i>Resolution</i> is the process of resolving one URI against another,
175 * <i>base</i> URI. The resulting URI is constructed from components of both
176 * URIs in the manner specified by RFC 2396, taking components from the
177 * base URI for those not specified in the original. For hierarchical URIs,
178 * the path of the original is resolved against the path of the base and then
179 * normalized. The result, for example, of resolving
180 *
181 * <blockquote>
182 * <tt>docs/guide/collections/designfaq.html#28 </tt>(1)
183 * </blockquote>
184 *
185 * against the base URI <tt>http://java.sun.com/j2se/1.3/</tt> is the result
186 * URI
187 *
188 * <blockquote>
189 * <tt>http://java.sun.com/j2se/1.3/docs/guide/collections/designfaq.html#28</tt>
190 * </blockquote>
191 *
192 * Resolving the relative URI
193 *
194 * <blockquote>
195 * <tt>../../../demo/jfc/SwingSet2/src/SwingSet2.java </tt>(2)
196 * </blockquote>
197 *
198 * against this result yields, in turn,
199 *
200 * <blockquote>
201 * <tt>http://java.sun.com/j2se/1.3/demo/jfc/SwingSet2/src/SwingSet2.java</tt>
202 * </blockquote>
203 *
204 * Resolution of both absolute and relative URIs, and of both absolute and
205 * relative paths in the case of hierarchical URIs, is supported. Resolving
206 * the URI <tt>file:///~calendar</tt> against any other URI simply yields the
207 * original URI, since it is absolute. Resolving the relative URI (2) above
208 * against the relative base URI (1) yields the normalized, but still relative,
209 * URI
210 *
211 * <blockquote>
212 * <tt>demo/jfc/SwingSet2/src/SwingSet2.java</tt>
213 * </blockquote>
214 *
215 * <p> <i>Relativization</i>, finally, is the inverse of resolution: For any
216 * two normalized URIs <i>u</i> and <i>v</i>,
217 *
218 * <blockquote>
219 * <i>u</i><tt>.relativize(</tt><i>u</i><tt>.resolve(</tt><i>v</i><tt>)).equals(</tt><i>v</i><tt>)</tt> and<br>
220 * <i>u</i><tt>.resolve(</tt><i>u</i><tt>.relativize(</tt><i>v</i><tt>)).equals(</tt><i>v</i><tt>)</tt> .<br>
221 * </blockquote>
222 *
223 * This operation is often useful when constructing a document containing URIs
224 * that must be made relative to the base URI of the document wherever
225 * possible. For example, relativizing the URI
226 *
227 * <blockquote>
228 * <tt>http://java.sun.com/j2se/1.3/docs/guide/index.html</tt>
229 * </blockquote>
230 *
231 * against the base URI
232 *
233 * <blockquote>
234 * <tt>http://java.sun.com/j2se/1.3</tt>
235 * </blockquote>
236 *
237 * yields the relative URI <tt>docs/guide/index.html</tt>.
238 *
239 *
240 * <h4> Character categories </h4>
241 *
242 * RFC 2396 specifies precisely which characters are permitted in the
243 * various components of a URI reference. The following categories, most of
244 * which are taken from that specification, are used below to describe these
245 * constraints:
246 *
247 * <blockquote><table cellspacing=2 summary="Describes categories alpha,digit,alphanum,unreserved,punct,reserved,escaped,and other">
248 * <tr><th valign=top><i>alpha</i></th>
249 * <td>The US-ASCII alphabetic characters,
250 * <tt>'A'</tt> through <tt>'Z'</tt>
251 * and <tt>'a'</tt> through <tt>'z'</tt></td></tr>
252 * <tr><th valign=top><i>digit</i></th>
253 * <td>The US-ASCII decimal digit characters,
254 * <tt>'0'</tt> through <tt>'9'</tt></td></tr>
255 * <tr><th valign=top><i>alphanum</i></th>
256 * <td>All <i>alpha</i> and <i>digit</i> characters</td></tr>
257 * <tr><th valign=top><i>unreserved</i> </th>
258 * <td>All <i>alphanum</i> characters together with those in the string
259 * <tt>"_-!.~'()*"</tt></td></tr>
260 * <tr><th valign=top><i>punct</i></th>
261 * <td>The characters in the string <tt>",;:$&+="</tt></td></tr>
262 * <tr><th valign=top><i>reserved</i></th>
263 * <td>All <i>punct</i> characters together with those in the string
264 * <tt>"?/[]@"</tt></td></tr>
265 * <tr><th valign=top><i>escaped</i></th>
266 * <td>Escaped octets, that is, triplets consisting of the percent
267 * character (<tt>'%'</tt>) followed by two hexadecimal digits
268 * (<tt>'0'</tt>-<tt>'9'</tt>, <tt>'A'</tt>-<tt>'F'</tt>, and
269 * <tt>'a'</tt>-<tt>'f'</tt>)</td></tr>
270 * <tr><th valign=top><i>other</i></th>
271 * <td>The Unicode characters that are not in the US-ASCII character set,
272 * are not control characters (according to the {@link
273 * java.lang.Character#isISOControl(char) Character.isISOControl}
274 * method), and are not space characters (according to the {@link
275 * java.lang.Character#isSpaceChar(char) Character.isSpaceChar}
276 * method) <i>(<b>Deviation from RFC 2396</b>, which is
277 * limited to US-ASCII)</i></td></tr>
278 * </table></blockquote>
279 *
280 * <p><a name="legal-chars"></a> The set of all legal URI characters consists of
281 * the <i>unreserved</i>, <i>reserved</i>, <i>escaped</i>, and <i>other</i>
282 * characters.
283 *
284 *
285 * <h4> Escaped octets, quotation, encoding, and decoding </h4>
286 *
287 * RFC 2396 allows escaped octets to appear in the user-info, path, query, and
288 * fragment components. Escaping serves two purposes in URIs:
289 *
290 * <ul>
291 *
292 * <li><p> To <i>encode</i> non-US-ASCII characters when a URI is required to
293 * conform strictly to RFC 2396 by not containing any <i>other</i>
294 * characters. </p></li>
295 *
296 * <li><p> To <i>quote</i> characters that are otherwise illegal in a
297 * component. The user-info, path, query, and fragment components differ
298 * slightly in terms of which characters are considered legal and illegal.
299 * </p></li>
300 *
301 * </ul>
302 *
303 * These purposes are served in this class by three related operations:
304 *
305 * <ul>
306 *
307 * <li><p><a name="encode"></a> A character is <i>encoded</i> by replacing it
308 * with the sequence of escaped octets that represent that character in the
309 * UTF-8 character set. The Euro currency symbol (<tt>'\u20AC'</tt>),
310 * for example, is encoded as <tt>"%E2%82%AC"</tt>. <i>(<b>Deviation from
311 * RFC 2396</b>, which does not specify any particular character
312 * set.)</i> </p></li>
313 *
314 * <li><p><a name="quote"></a> An illegal character is <i>quoted</i> simply by
315 * encoding it. The space character, for example, is quoted by replacing it
316 * with <tt>"%20"</tt>. UTF-8 contains US-ASCII, hence for US-ASCII
317 * characters this transformation has exactly the effect required by
318 * RFC 2396. </p></li>
319 *
320 * <li><p><a name="decode"></a>
321 * A sequence of escaped octets is <i>decoded</i> by
322 * replacing it with the sequence of characters that it represents in the
323 * UTF-8 character set. UTF-8 contains US-ASCII, hence decoding has the
324 * effect of de-quoting any quoted US-ASCII characters as well as that of
325 * decoding any encoded non-US-ASCII characters. If a <a
326 * href="../nio/charset/CharsetDecoder.html#ce">decoding error</a> occurs
327 * when decoding the escaped octets then the erroneous octets are replaced by
328 * <tt>'\uFFFD'</tt>, the Unicode replacement character. </p></li>
329 *
330 * </ul>
331 *
332 * These operations are exposed in the constructors and methods of this class
333 * as follows:
334 *
335 * <ul>
336 *
337 * <li><p> The {@link #URI(java.lang.String) <code>single-argument
338 * constructor</code>} requires any illegal characters in its argument to be
339 * quoted and preserves any escaped octets and <i>other</i> characters that
340 * are present. </p></li>
341 *
342 * <li><p> The {@link
343 * #URI(java.lang.String,java.lang.String,java.lang.String,int,java.lang.String,java.lang.String,java.lang.String)
344 * <code>multi-argument constructors</code>} quote illegal characters as
345 * required by the components in which they appear. The percent character
346 * (<tt>'%'</tt>) is always quoted by these constructors. Any <i>other</i>
347 * characters are preserved. </p></li>
348 *
349 * <li><p> The {@link #getRawUserInfo() getRawUserInfo}, {@link #getRawPath()
350 * getRawPath}, {@link #getRawQuery() getRawQuery}, {@link #getRawFragment()
351 * getRawFragment}, {@link #getRawAuthority() getRawAuthority}, and {@link
352 * #getRawSchemeSpecificPart() getRawSchemeSpecificPart} methods return the
353 * values of their corresponding components in raw form, without interpreting
354 * any escaped octets. The strings returned by these methods may contain
355 * both escaped octets and <i>other</i> characters, and will not contain any
356 * illegal characters. </p></li>
357 *
358 * <li><p> The {@link #getUserInfo() getUserInfo}, {@link #getPath()
359 * getPath}, {@link #getQuery() getQuery}, {@link #getFragment()
360 * getFragment}, {@link #getAuthority() getAuthority}, and {@link
361 * #getSchemeSpecificPart() getSchemeSpecificPart} methods decode any escaped
362 * octets in their corresponding components. The strings returned by these
363 * methods may contain both <i>other</i> characters and illegal characters,
364 * and will not contain any escaped octets. </p></li>
365 *
366 * <li><p> The {@link #toString() toString} method returns a URI string with
367 * all necessary quotation but which may contain <i>other</i> characters.
368 * </p></li>
369 *
370 * <li><p> The {@link #toASCIIString() toASCIIString} method returns a fully
371 * quoted and encoded URI string that does not contain any <i>other</i>
372 * characters. </p></li>
373 *
374 * </ul>
375 *
376 *
377 * <h4> Identities </h4>
378 *
379 * For any URI <i>u</i>, it is always the case that
380 *
381 * <blockquote>
382 * <tt>new URI(</tt><i>u</i><tt>.toString()).equals(</tt><i>u</i><tt>)</tt> .
383 * </blockquote>
384 *
385 * For any URI <i>u</i> that does not contain redundant syntax such as two
386 * slashes before an empty authority (as in <tt>file:///tmp/</tt> ) or a
387 * colon following a host name but no port (as in
388 * <tt>http://java.sun.com:</tt> ), and that does not encode characters
389 * except those that must be quoted, the following identities also hold:
390 *
391 * <blockquote>
392 * <tt>new URI(</tt><i>u</i><tt>.getScheme(),<br>
393 * </tt><i>u</i><tt>.getSchemeSpecificPart(),<br>
394 * </tt><i>u</i><tt>.getFragment())<br>
395 * .equals(</tt><i>u</i><tt>)</tt>
396 * </blockquote>
397 *
398 * in all cases,
399 *
400 * <blockquote>
401 * <tt>new URI(</tt><i>u</i><tt>.getScheme(),<br>
402 * </tt><i>u</i><tt>.getUserInfo(), </tt><i>u</i><tt>.getAuthority(),<br>
403 * </tt><i>u</i><tt>.getPath(), </tt><i>u</i><tt>.getQuery(),<br>
404 * </tt><i>u</i><tt>.getFragment())<br>
405 * .equals(</tt><i>u</i><tt>)</tt>
406 * </blockquote>
407 *
408 * if <i>u</i> is hierarchical, and
409 *
410 * <blockquote>
411 * <tt>new URI(</tt><i>u</i><tt>.getScheme(),<br>
412 * </tt><i>u</i><tt>.getUserInfo(), </tt><i>u</i><tt>.getHost(), </tt><i>u</i><tt>.getPort(),<br>
413 * </tt><i>u</i><tt>.getPath(), </tt><i>u</i><tt>.getQuery(),<br>
414 * </tt><i>u</i><tt>.getFragment())<br>
415 * .equals(</tt><i>u</i><tt>)</tt>
416 * </blockquote>
417 *
418 * if <i>u</i> is hierarchical and has either no authority or a server-based
419 * authority.
420 *
421 *
422 * <h4> URIs, URLs, and URNs </h4>
423 *
424 * A URI is a uniform resource <i>identifier</i> while a URL is a uniform
425 * resource <i>locator</i>. Hence every URL is a URI, abstractly speaking, but
426 * not every URI is a URL. This is because there is another subcategory of
427 * URIs, uniform resource <i>names</i> (URNs), which name resources but do not
428 * specify how to locate them. The <tt>mailto</tt>, <tt>news</tt>, and
429 * <tt>isbn</tt> URIs shown above are examples of URNs.
430 *
431 * <p> The conceptual distinction between URIs and URLs is reflected in the
432 * differences between this class and the {@link URL} class.
433 *
434 * <p> An instance of this class represents a URI reference in the syntactic
435 * sense defined by RFC 2396. A URI may be either absolute or relative.
436 * A URI string is parsed according to the generic syntax without regard to the
437 * scheme, if any, that it specifies. No lookup of the host, if any, is
438 * performed, and no scheme-dependent stream handler is constructed. Equality,
439 * hashing, and comparison are defined strictly in terms of the character
440 * content of the instance. In other words, a URI instance is little more than
441 * a structured string that supports the syntactic, scheme-independent
442 * operations of comparison, normalization, resolution, and relativization.
443 *
444 * <p> An instance of the {@link URL} class, by contrast, represents the
445 * syntactic components of a URL together with some of the information required
446 * to access the resource that it describes. A URL must be absolute, that is,
447 * it must always specify a scheme. A URL string is parsed according to its
448 * scheme. A stream handler is always established for a URL, and in fact it is
449 * impossible to create a URL instance for a scheme for which no handler is
450 * available. Equality and hashing depend upon both the scheme and the
451 * Internet address of the host, if any; comparison is not defined. In other
452 * words, a URL is a structured string that supports the syntactic operation of
453 * resolution as well as the network I/O operations of looking up the host and
454 * opening a connection to the specified resource.
455 *
456 *
457 * @author Mark Reinhold
458 * @since 1.4
459 *
460 * @see <a href="http://ietf.org/rfc/rfc2279.txt"><i>RFC 2279: UTF-8, a
461 * transformation format of ISO 10646</i></a>, <br><a
462 * href="http://www.ietf.org/rfc/rfc2373.txt"><i>RFC 2373: IPv6 Addressing
463 * Architecture</i></a>, <br><a
464 * href="http://www.ietf.org/rfc/rfc2396.txt""><i>RFC 2396: Uniform
465 * Resource Identifiers (URI): Generic Syntax</i></a>, <br><a
466 * href="http://www.ietf.org/rfc/rfc2732.txt"><i>RFC 2732: Format for
467 * Literal IPv6 Addresses in URLs</i></a>, <br><a
468 * href="URISyntaxException.html">URISyntaxException</a>
469 */
470
471 public final class URI
472 implements Comparable<URI>, Serializable
473 {
474
475 // Note: Comments containing the word "ASSERT" indicate places where a
476 // throw of an InternalError should be replaced by an appropriate assertion
477 // statement once asserts are enabled in the build.
478
479 static final long serialVersionUID = -6052424284110960213L;
480
481
482 // -- Properties and components of this instance --
483
484 // Components of all URIs: [<scheme>:]<scheme-specific-part>[#<fragment>]
485 private transient String scheme; // null ==> relative URI
486 private transient String fragment;
487
488 // Hierarchical URI components: [//<authority>]<path>[?<query>]
489 private transient String authority; // Registry or server
490
491 // Server-based authority: [<userInfo>@]<host>[:<port>]
492 private transient String userInfo;
493 private transient String host; // null ==> registry-based
494 private transient int port = -1; // -1 ==> undefined
495
496 // Remaining components of hierarchical URIs
497 private transient String path; // null ==> opaque
498 private transient String query;
499
500 // The remaining fields may be computed on demand
501
502 private volatile transient String schemeSpecificPart;
503 private volatile transient int hash; // Zero ==> undefined
504
505 private volatile transient String decodedUserInfo = null;
506 private volatile transient String decodedAuthority = null;
507 private volatile transient String decodedPath = null;
508 private volatile transient String decodedQuery = null;
509 private volatile transient String decodedFragment = null;
510 private volatile transient String decodedSchemeSpecificPart = null;
511
512 /**
513 * The string form of this URI.
514 *
515 * @serial
516 */
517 private volatile String string; // The only serializable field
518
519
520
521 // -- Constructors and factories --
522
523 private URI() { } // Used internally
524
525 /**
526 * Constructs a URI by parsing the given string.
527 *
528 * <p> This constructor parses the given string exactly as specified by the
529 * grammar in <a
530 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
531 * Appendix A, <b><i>except for the following deviations:</i></b> </p>
532 *
533 * <ul type=disc>
534 *
535 * <li><p> An empty authority component is permitted as long as it is
536 * followed by a non-empty path, a query component, or a fragment
537 * component. This allows the parsing of URIs such as
538 * <tt>"file:///foo/bar"</tt>, which seems to be the intent of
539 * RFC 2396 although the grammar does not permit it. If the
540 * authority component is empty then the user-information, host, and port
541 * components are undefined. </p></li>
542 *
543 * <li><p> Empty relative paths are permitted; this seems to be the
544 * intent of RFC 2396 although the grammar does not permit it. The
545 * primary consequence of this deviation is that a standalone fragment
546 * such as <tt>"#foo"</tt> parses as a relative URI with an empty path
547 * and the given fragment, and can be usefully <a
548 * href="#resolve-frag">resolved</a> against a base URI.
549 *
550 * <li><p> IPv4 addresses in host components are parsed rigorously, as
551 * specified by <a
552 * href="http://www.ietf.org/rfc/rfc2732.txt">RFC 2732</a>: Each
553 * element of a dotted-quad address must contain no more than three
554 * decimal digits. Each element is further constrained to have a value
555 * no greater than 255. </p></li>
556 *
557 * <li> <p> Hostnames in host components that comprise only a single
558 * domain label are permitted to start with an <i>alphanum</i>
559 * character. This seems to be the intent of <a
560 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>
561 * section 3.2.2 although the grammar does not permit it. The
562 * consequence of this deviation is that the authority component of a
563 * hierarchical URI such as <tt>s://123</tt>, will parse as a server-based
564 * authority. </p></li>
565 *
566 * <li><p> IPv6 addresses are permitted for the host component. An IPv6
567 * address must be enclosed in square brackets (<tt>'['</tt> and
568 * <tt>']'</tt>) as specified by <a
569 * href="http://www.ietf.org/rfc/rfc2732.txt">RFC 2732</a>. The
570 * IPv6 address itself must parse according to <a
571 * href="http://www.ietf.org/rfc/rfc2373.txt">RFC 2373</a>. IPv6
572 * addresses are further constrained to describe no more than sixteen
573 * bytes of address information, a constraint implicit in RFC 2373
574 * but not expressible in the grammar. </p></li>
575 *
576 * <li><p> Characters in the <i>other</i> category are permitted wherever
577 * RFC 2396 permits <i>escaped</i> octets, that is, in the
578 * user-information, path, query, and fragment components, as well as in
579 * the authority component if the authority is registry-based. This
580 * allows URIs to contain Unicode characters beyond those in the US-ASCII
581 * character set. </p></li>
582 *
583 * </ul>
584 *
585 * @param str The string to be parsed into a URI
586 *
587 * @throws NullPointerException
588 * If <tt>str</tt> is <tt>null</tt>
589 *
590 * @throws URISyntaxException
591 * If the given string violates RFC 2396, as augmented
592 * by the above deviations
593 */
594 public URI(String str) throws URISyntaxException {
595 new Parser(str).parse(false);
596 }
597
598 /**
599 * Constructs a hierarchical URI from the given components.
600 *
601 * <p> If a scheme is given then the path, if also given, must either be
602 * empty or begin with a slash character (<tt>'/'</tt>). Otherwise a
603 * component of the new URI may be left undefined by passing <tt>null</tt>
604 * for the corresponding parameter or, in the case of the <tt>port</tt>
605 * parameter, by passing <tt>-1</tt>.
606 *
607 * <p> This constructor first builds a URI string from the given components
608 * according to the rules specified in <a
609 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
610 * section 5.2, step 7: </p>
611 *
612 * <ol>
613 *
614 * <li><p> Initially, the result string is empty. </p></li>
615 *
616 * <li><p> If a scheme is given then it is appended to the result,
617 * followed by a colon character (<tt>':'</tt>). </p></li>
618 *
619 * <li><p> If user information, a host, or a port are given then the
620 * string <tt>"//"</tt> is appended. </p></li>
621 *
622 * <li><p> If user information is given then it is appended, followed by
623 * a commercial-at character (<tt>'@'</tt>). Any character not in the
624 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
625 * categories is <a href="#quote">quoted</a>. </p></li>
626 *
627 * <li><p> If a host is given then it is appended. If the host is a
628 * literal IPv6 address but is not enclosed in square brackets
629 * (<tt>'['</tt> and <tt>']'</tt>) then the square brackets are added.
630 * </p></li>
631 *
632 * <li><p> If a port number is given then a colon character
633 * (<tt>':'</tt>) is appended, followed by the port number in decimal.
634 * </p></li>
635 *
636 * <li><p> If a path is given then it is appended. Any character not in
637 * the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
638 * categories, and not equal to the slash character (<tt>'/'</tt>) or the
639 * commercial-at character (<tt>'@'</tt>), is quoted. </p></li>
640 *
641 * <li><p> If a query is given then a question-mark character
642 * (<tt>'?'</tt>) is appended, followed by the query. Any character that
643 * is not a <a href="#legal-chars">legal URI character</a> is quoted.
644 * </p></li>
645 *
646 * <li><p> Finally, if a fragment is given then a hash character
647 * (<tt>'#'</tt>) is appended, followed by the fragment. Any character
648 * that is not a legal URI character is quoted. </p></li>
649 *
650 * </ol>
651 *
652 * <p> The resulting URI string is then parsed as if by invoking the {@link
653 * #URI(String)} constructor and then invoking the {@link
654 * #parseServerAuthority()} method upon the result; this may cause a {@link
655 * URISyntaxException} to be thrown. </p>
656 *
657 * @param scheme Scheme name
658 * @param userInfo User name and authorization information
659 * @param host Host name
660 * @param port Port number
661 * @param path Path
662 * @param query Query
663 * @param fragment Fragment
664 *
665 * @throws URISyntaxException
666 * If both a scheme and a path are given but the path is relative,
667 * if the URI string constructed from the given components violates
668 * RFC 2396, or if the authority component of the string is
669 * present but cannot be parsed as a server-based authority
670 */
671 public URI(String scheme,
672 String userInfo, String host, int port,
673 String path, String query, String fragment)
674 throws URISyntaxException
675 {
676 String s = toString(scheme, null,
677 null, userInfo, host, port,
678 path, query, fragment);
679 checkPath(s, scheme, path);
680 new Parser(s).parse(true);
681 }
682
683 /**
684 * Constructs a hierarchical URI from the given components.
685 *
686 * <p> If a scheme is given then the path, if also given, must either be
687 * empty or begin with a slash character (<tt>'/'</tt>). Otherwise a
688 * component of the new URI may be left undefined by passing <tt>null</tt>
689 * for the corresponding parameter.
690 *
691 * <p> This constructor first builds a URI string from the given components
692 * according to the rules specified in <a
693 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
694 * section 5.2, step 7: </p>
695 *
696 * <ol>
697 *
698 * <li><p> Initially, the result string is empty. </p></li>
699 *
700 * <li><p> If a scheme is given then it is appended to the result,
701 * followed by a colon character (<tt>':'</tt>). </p></li>
702 *
703 * <li><p> If an authority is given then the string <tt>"//"</tt> is
704 * appended, followed by the authority. If the authority contains a
705 * literal IPv6 address then the address must be enclosed in square
706 * brackets (<tt>'['</tt> and <tt>']'</tt>). Any character not in the
707 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
708 * categories, and not equal to the commercial-at character
709 * (<tt>'@'</tt>), is <a href="#quote">quoted</a>. </p></li>
710 *
711 * <li><p> If a path is given then it is appended. Any character not in
712 * the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
713 * categories, and not equal to the slash character (<tt>'/'</tt>) or the
714 * commercial-at character (<tt>'@'</tt>), is quoted. </p></li>
715 *
716 * <li><p> If a query is given then a question-mark character
717 * (<tt>'?'</tt>) is appended, followed by the query. Any character that
718 * is not a <a href="#legal-chars">legal URI character</a> is quoted.
719 * </p></li>
720 *
721 * <li><p> Finally, if a fragment is given then a hash character
722 * (<tt>'#'</tt>) is appended, followed by the fragment. Any character
723 * that is not a legal URI character is quoted. </p></li>
724 *
725 * </ol>
726 *
727 * <p> The resulting URI string is then parsed as if by invoking the {@link
728 * #URI(String)} constructor and then invoking the {@link
729 * #parseServerAuthority()} method upon the result; this may cause a {@link
730 * URISyntaxException} to be thrown. </p>
731 *
732 * @param scheme Scheme name
733 * @param authority Authority
734 * @param path Path
735 * @param query Query
736 * @param fragment Fragment
737 *
738 * @throws URISyntaxException
739 * If both a scheme and a path are given but the path is relative,
740 * if the URI string constructed from the given components violates
741 * RFC 2396, or if the authority component of the string is
742 * present but cannot be parsed as a server-based authority
743 */
744 public URI(String scheme,
745 String authority,
746 String path, String query, String fragment)
747 throws URISyntaxException
748 {
749 String s = toString(scheme, null,
750 authority, null, null, -1,
751 path, query, fragment);
752 checkPath(s, scheme, path);
753 new Parser(s).parse(false);
754 }
755
756 /**
757 * Constructs a hierarchical URI from the given components.
758 *
759 * <p> A component may be left undefined by passing <tt>null</tt>.
760 *
761 * <p> This convenience constructor works as if by invoking the
762 * seven-argument constructor as follows:
763 *
764 * <blockquote><tt>
765 * new {@link #URI(String, String, String, int, String, String, String)
766 * URI}(scheme, null, host, -1, path, null, fragment);
767 * </tt></blockquote>
768 *
769 * @param scheme Scheme name
770 * @param host Host name
771 * @param path Path
772 * @param fragment Fragment
773 *
774 * @throws URISyntaxException
775 * If the URI string constructed from the given components
776 * violates RFC 2396
777 */
778 public URI(String scheme, String host, String path, String fragment)
779 throws URISyntaxException
780 {
781 this(scheme, null, host, -1, path, null, fragment);
782 }
783
784 /**
785 * Constructs a URI from the given components.
786 *
787 * <p> A component may be left undefined by passing <tt>null</tt>.
788 *
789 * <p> This constructor first builds a URI in string form using the given
790 * components as follows: </p>
791 *
792 * <ol>
793 *
794 * <li><p> Initially, the result string is empty. </p></li>
795 *
796 * <li><p> If a scheme is given then it is appended to the result,
797 * followed by a colon character (<tt>':'</tt>). </p></li>
798 *
799 * <li><p> If a scheme-specific part is given then it is appended. Any
800 * character that is not a <a href="#legal-chars">legal URI character</a>
801 * is <a href="#quote">quoted</a>. </p></li>
802 *
803 * <li><p> Finally, if a fragment is given then a hash character
804 * (<tt>'#'</tt>) is appended to the string, followed by the fragment.
805 * Any character that is not a legal URI character is quoted. </p></li>
806 *
807 * </ol>
808 *
809 * <p> The resulting URI string is then parsed in order to create the new
810 * URI instance as if by invoking the {@link #URI(String)} constructor;
811 * this may cause a {@link URISyntaxException} to be thrown. </p>
812 *
813 * @param scheme Scheme name
814 * @param ssp Scheme-specific part
815 * @param fragment Fragment
816 *
817 * @throws URISyntaxException
818 * If the URI string constructed from the given components
819 * violates RFC 2396
820 */
821 public URI(String scheme, String ssp, String fragment)
822 throws URISyntaxException
823 {
824 new Parser(toString(scheme, ssp,
825 null, null, null, -1,
826 null, null, fragment))
827 .parse(false);
828 }
829
830 /**
831 * Creates a URI by parsing the given string.
832 *
833 * <p> This convenience factory method works as if by invoking the {@link
834 * #URI(String)} constructor; any {@link URISyntaxException} thrown by the
835 * constructor is caught and wrapped in a new {@link
836 * IllegalArgumentException} object, which is then thrown.
837 *
838 * <p> This method is provided for use in situations where it is known that
839 * the given string is a legal URI, for example for URI constants declared
840 * within in a program, and so it would be considered a programming error
841 * for the string not to parse as such. The constructors, which throw
842 * {@link URISyntaxException} directly, should be used situations where a
843 * URI is being constructed from user input or from some other source that
844 * may be prone to errors. </p>
845 *
846 * @param str The string to be parsed into a URI
847 * @return The new URI
848 *
849 * @throws NullPointerException
850 * If <tt>str</tt> is <tt>null</tt>
851 *
852 * @throws IllegalArgumentException
853 * If the given string violates RFC 2396
854 */
855 public static URI create(String str) {
856 try {
857 return new URI(str);
858 } catch (URISyntaxException x) {
859 IllegalArgumentException y = new IllegalArgumentException();
860 y.initCause(x);
861 throw y;
862 }
863 }
864
865
866 // -- Operations --
867
868 /**
869 * Attempts to parse this URI's authority component, if defined, into
870 * user-information, host, and port components.
871 *
872 * <p> If this URI's authority component has already been recognized as
873 * being server-based then it will already have been parsed into
874 * user-information, host, and port components. In this case, or if this
875 * URI has no authority component, this method simply returns this URI.
876 *
877 * <p> Otherwise this method attempts once more to parse the authority
878 * component into user-information, host, and port components, and throws
879 * an exception describing why the authority component could not be parsed
880 * in that way.
881 *
882 * <p> This method is provided because the generic URI syntax specified in
883 * <a href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>
884 * cannot always distinguish a malformed server-based authority from a
885 * legitimate registry-based authority. It must therefore treat some
886 * instances of the former as instances of the latter. The authority
887 * component in the URI string <tt>"//foo:bar"</tt>, for example, is not a
888 * legal server-based authority but it is legal as a registry-based
889 * authority.
890 *
891 * <p> In many common situations, for example when working URIs that are
892 * known to be either URNs or URLs, the hierarchical URIs being used will
893 * always be server-based. They therefore must either be parsed as such or
894 * treated as an error. In these cases a statement such as
895 *
896 * <blockquote>
897 * <tt>URI </tt><i>u</i><tt> = new URI(str).parseServerAuthority();</tt>
898 * </blockquote>
899 *
900 * <p> can be used to ensure that <i>u</i> always refers to a URI that, if
901 * it has an authority component, has a server-based authority with proper
902 * user-information, host, and port components. Invoking this method also
903 * ensures that if the authority could not be parsed in that way then an
904 * appropriate diagnostic message can be issued based upon the exception
905 * that is thrown. </p>
906 *
907 * @return A URI whose authority field has been parsed
908 * as a server-based authority
909 *
910 * @throws URISyntaxException
911 * If the authority component of this URI is defined
912 * but cannot be parsed as a server-based authority
913 * according to RFC 2396
914 */
915 public URI parseServerAuthority()
916 throws URISyntaxException
917 {
918 // We could be clever and cache the error message and index from the
919 // exception thrown during the original parse, but that would require
920 // either more fields or a more-obscure representation.
921 if ((host != null) || (authority == null))
922 return this;
923 defineString();
924 new Parser(string).parse(true);
925 return this;
926 }
927
928 /**
929 * Normalizes this URI's path.
930 *
931 * <p> If this URI is opaque, or if its path is already in normal form,
932 * then this URI is returned. Otherwise a new URI is constructed that is
933 * identical to this URI except that its path is computed by normalizing
934 * this URI's path in a manner consistent with <a
935 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
936 * section 5.2, step 6, sub-steps c through f; that is:
937 * </p>
938 *
939 * <ol>
940 *
941 * <li><p> All <tt>"."</tt> segments are removed. </p></li>
942 *
943 * <li><p> If a <tt>".."</tt> segment is preceded by a non-<tt>".."</tt>
944 * segment then both of these segments are removed. This step is
945 * repeated until it is no longer applicable. </p></li>
946 *
947 * <li><p> If the path is relative, and if its first segment contains a
948 * colon character (<tt>':'</tt>), then a <tt>"."</tt> segment is
949 * prepended. This prevents a relative URI with a path such as
950 * <tt>"a:b/c/d"</tt> from later being re-parsed as an opaque URI with a
951 * scheme of <tt>"a"</tt> and a scheme-specific part of <tt>"b/c/d"</tt>.
952 * <b><i>(Deviation from RFC 2396)</i></b> </p></li>
953 *
954 * </ol>
955 *
956 * <p> A normalized path will begin with one or more <tt>".."</tt> segments
957 * if there were insufficient non-<tt>".."</tt> segments preceding them to
958 * allow their removal. A normalized path will begin with a <tt>"."</tt>
959 * segment if one was inserted by step 3 above. Otherwise, a normalized
960 * path will not contain any <tt>"."</tt> or <tt>".."</tt> segments. </p>
961 *
962 * @return A URI equivalent to this URI,
963 * but whose path is in normal form
964 */
965 public URI normalize() {
966 return normalize(this);
967 }
968
969 /**
970 * Resolves the given URI against this URI.
971 *
972 * <p> If the given URI is already absolute, or if this URI is opaque, then
973 * the given URI is returned.
974 *
975 * <p><a name="resolve-frag"></a> If the given URI's fragment component is
976 * defined, its path component is empty, and its scheme, authority, and
977 * query components are undefined, then a URI with the given fragment but
978 * with all other components equal to those of this URI is returned. This
979 * allows a URI representing a standalone fragment reference, such as
980 * <tt>"#foo"</tt>, to be usefully resolved against a base URI.
981 *
982 * <p> Otherwise this method constructs a new hierarchical URI in a manner
983 * consistent with <a
984 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
985 * section 5.2; that is: </p>
986 *
987 * <ol>
988 *
989 * <li><p> A new URI is constructed with this URI's scheme and the given
990 * URI's query and fragment components. </p></li>
991 *
992 * <li><p> If the given URI has an authority component then the new URI's
993 * authority and path are taken from the given URI. </p></li>
994 *
995 * <li><p> Otherwise the new URI's authority component is copied from
996 * this URI, and its path is computed as follows: </p></li>
997 *
998 * <ol type=a>
999 *
1000 * <li><p> If the given URI's path is absolute then the new URI's path
1001 * is taken from the given URI. </p></li>
1002 *
1003 * <li><p> Otherwise the given URI's path is relative, and so the new
1004 * URI's path is computed by resolving the path of the given URI
1005 * against the path of this URI. This is done by concatenating all but
1006 * the last segment of this URI's path, if any, with the given URI's
1007 * path and then normalizing the result as if by invoking the {@link
1008 * #normalize() normalize} method. </p></li>
1009 *
1010 * </ol>
1011 *
1012 * </ol>
1013 *
1014 * <p> The result of this method is absolute if, and only if, either this
1015 * URI is absolute or the given URI is absolute. </p>
1016 *
1017 * @param uri The URI to be resolved against this URI
1018 * @return The resulting URI
1019 *
1020 * @throws NullPointerException
1021 * If <tt>uri</tt> is <tt>null</tt>
1022 */
1023 public URI resolve(URI uri) {
1024 return resolve(this, uri);
1025 }
1026
1027 /**
1028 * Constructs a new URI by parsing the given string and then resolving it
1029 * against this URI.
1030 *
1031 * <p> This convenience method works as if invoking it were equivalent to
1032 * evaluating the expression <tt>{@link #resolve(java.net.URI)
1033 * resolve}(URI.{@link #create(String) create}(str))</tt>. </p>
1034 *
1035 * @param str The string to be parsed into a URI
1036 * @return The resulting URI
1037 *
1038 * @throws NullPointerException
1039 * If <tt>str</tt> is <tt>null</tt>
1040 *
1041 * @throws IllegalArgumentException
1042 * If the given string violates RFC 2396
1043 */
1044 public URI resolve(String str) {
1045 return resolve(URI.create(str));
1046 }
1047
1048 /**
1049 * Relativizes the given URI against this URI.
1050 *
1051 * <p> The relativization of the given URI against this URI is computed as
1052 * follows: </p>
1053 *
1054 * <ol>
1055 *
1056 * <li><p> If either this URI or the given URI are opaque, or if the
1057 * scheme and authority components of the two URIs are not identical, or
1058 * if the path of this URI is not a prefix of the path of the given URI,
1059 * then the given URI is returned. </p></li>
1060 *
1061 * <li><p> Otherwise a new relative hierarchical URI is constructed with
1062 * query and fragment components taken from the given URI and with a path
1063 * component computed by removing this URI's path from the beginning of
1064 * the given URI's path. </p></li>
1065 *
1066 * </ol>
1067 *
1068 * @param uri The URI to be relativized against this URI
1069 * @return The resulting URI
1070 *
1071 * @throws NullPointerException
1072 * If <tt>uri</tt> is <tt>null</tt>
1073 */
1074 public URI relativize(URI uri) {
1075 return relativize(this, uri);
1076 }
1077
1078 /**
1079 * Constructs a URL from this URI.
1080 *
1081 * <p> This convenience method works as if invoking it were equivalent to
1082 * evaluating the expression <tt>new URL(this.toString())</tt> after
1083 * first checking that this URI is absolute. </p>
1084 *
1085 * @return A URL constructed from this URI
1086 *
1087 * @throws IllegalArgumentException
1088 * If this URL is not absolute
1089 *
1090 * @throws MalformedURLException
1091 * If a protocol handler for the URL could not be found,
1092 * or if some other error occurred while constructing the URL
1093 */
1094 public URL toURL()
1095 throws MalformedURLException {
1096 if (!isAbsolute())
1097 throw new IllegalArgumentException("URI is not absolute");
1098 return new URL(toString());
1099 }
1100
1101 // -- Component access methods --
1102
1103 /**
1104 * Returns the scheme component of this URI.
1105 *
1106 * <p> The scheme component of a URI, if defined, only contains characters
1107 * in the <i>alphanum</i> category and in the string <tt>"-.+"</tt>. A
1108 * scheme always starts with an <i>alpha</i> character. <p>
1109 *
1110 * The scheme component of a URI cannot contain escaped octets, hence this
1111 * method does not perform any decoding.
1112 *
1113 * @return The scheme component of this URI,
1114 * or <tt>null</tt> if the scheme is undefined
1115 */
1116 public String getScheme() {
1117 return scheme;
1118 }
1119
1120 /**
1121 * Tells whether or not this URI is absolute.
1122 *
1123 * <p> A URI is absolute if, and only if, it has a scheme component. </p>
1124 *
1125 * @return <tt>true</tt> if, and only if, this URI is absolute
1126 */
1127 public boolean isAbsolute() {
1128 return scheme != null;
1129 }
1130
1131 /**
1132 * Tells whether or not this URI is opaque.
1133 *
1134 * <p> A URI is opaque if, and only if, it is absolute and its
1135 * scheme-specific part does not begin with a slash character ('/').
1136 * An opaque URI has a scheme, a scheme-specific part, and possibly
1137 * a fragment; all other components are undefined. </p>
1138 *
1139 * @return <tt>true</tt> if, and only if, this URI is opaque
1140 */
1141 public boolean isOpaque() {
1142 return path == null;
1143 }
1144
1145 /**
1146 * Returns the raw scheme-specific part of this URI. The scheme-specific
1147 * part is never undefined, though it may be empty.
1148 *
1149 * <p> The scheme-specific part of a URI only contains legal URI
1150 * characters. </p>
1151 *
1152 * @return The raw scheme-specific part of this URI
1153 * (never <tt>null</tt>)
1154 */
1155 public String getRawSchemeSpecificPart() {
1156 defineSchemeSpecificPart();
1157 return schemeSpecificPart;
1158 }
1159
1160 /**
1161 * Returns the decoded scheme-specific part of this URI.
1162 *
1163 * <p> The string returned by this method is equal to that returned by the
1164 * {@link #getRawSchemeSpecificPart() getRawSchemeSpecificPart} method
1165 * except that all sequences of escaped octets are <a
1166 * href="#decode">decoded</a>. </p>
1167 *
1168 * @return The decoded scheme-specific part of this URI
1169 * (never <tt>null</tt>)
1170 */
1171 public String getSchemeSpecificPart() {
1172 if (decodedSchemeSpecificPart == null)
1173 decodedSchemeSpecificPart = decode(getRawSchemeSpecificPart());
1174 return decodedSchemeSpecificPart;
1175 }
1176
1177 /**
1178 * Returns the raw authority component of this URI.
1179 *
1180 * <p> The authority component of a URI, if defined, only contains the
1181 * commercial-at character (<tt>'@'</tt>) and characters in the
1182 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, and <i>other</i>
1183 * categories. If the authority is server-based then it is further
1184 * constrained to have valid user-information, host, and port
1185 * components. </p>
1186 *
1187 * @return The raw authority component of this URI,
1188 * or <tt>null</tt> if the authority is undefined
1189 */
1190 public String getRawAuthority() {
1191 return authority;
1192 }
1193
1194 /**
1195 * Returns the decoded authority component of this URI.
1196 *
1197 * <p> The string returned by this method is equal to that returned by the
1198 * {@link #getRawAuthority() getRawAuthority} method except that all
1199 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1200 *
1201 * @return The decoded authority component of this URI,
1202 * or <tt>null</tt> if the authority is undefined
1203 */
1204 public String getAuthority() {
1205 if (decodedAuthority == null)
1206 decodedAuthority = decode(authority);
1207 return decodedAuthority;
1208 }
1209
1210 /**
1211 * Returns the raw user-information component of this URI.
1212 *
1213 * <p> The user-information component of a URI, if defined, only contains
1214 * characters in the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, and
1215 * <i>other</i> categories. </p>
1216 *
1217 * @return The raw user-information component of this URI,
1218 * or <tt>null</tt> if the user information is undefined
1219 */
1220 public String getRawUserInfo() {
1221 return userInfo;
1222 }
1223
1224 /**
1225 * Returns the decoded user-information component of this URI.
1226 *
1227 * <p> The string returned by this method is equal to that returned by the
1228 * {@link #getRawUserInfo() getRawUserInfo} method except that all
1229 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1230 *
1231 * @return The decoded user-information component of this URI,
1232 * or <tt>null</tt> if the user information is undefined
1233 */
1234 public String getUserInfo() {
1235 if ((decodedUserInfo == null) && (userInfo != null))
1236 decodedUserInfo = decode(userInfo);
1237 return decodedUserInfo;
1238 }
1239
1240 /**
1241 * Returns the host component of this URI.
1242 *
1243 * <p> The host component of a URI, if defined, will have one of the
1244 * following forms: </p>
1245 *
1246 * <ul type=disc>
1247 *
1248 * <li><p> A domain name consisting of one or more <i>labels</i>
1249 * separated by period characters (<tt>'.'</tt>), optionally followed by
1250 * a period character. Each label consists of <i>alphanum</i> characters
1251 * as well as hyphen characters (<tt>'-'</tt>), though hyphens never
1252 * occur as the first or last characters in a label. The rightmost
1253 * label of a domain name consisting of two or more labels, begins
1254 * with an <i>alpha</i> character. </li>
1255 *
1256 * <li><p> A dotted-quad IPv4 address of the form
1257 * <i>digit</i><tt>+.</tt><i>digit</i><tt>+.</tt><i>digit</i><tt>+.</tt><i>digit</i><tt>+</tt>,
1258 * where no <i>digit</i> sequence is longer than three characters and no
1259 * sequence has a value larger than 255. </p></li>
1260 *
1261 * <li><p> An IPv6 address enclosed in square brackets (<tt>'['</tt> and
1262 * <tt>']'</tt>) and consisting of hexadecimal digits, colon characters
1263 * (<tt>':'</tt>), and possibly an embedded IPv4 address. The full
1264 * syntax of IPv6 addresses is specified in <a
1265 * href="http://www.ietf.org/rfc/rfc2373.txt"><i>RFC 2373: IPv6
1266 * Addressing Architecture</i></a>. </p></li>
1267 *
1268 * </ul>
1269 *
1270 * The host component of a URI cannot contain escaped octets, hence this
1271 * method does not perform any decoding.
1272 *
1273 * @return The host component of this URI,
1274 * or <tt>null</tt> if the host is undefined
1275 */
1276 public String getHost() {
1277 return host;
1278 }
1279
1280 /**
1281 * Returns the port number of this URI.
1282 *
1283 * <p> The port component of a URI, if defined, is a non-negative
1284 * integer. </p>
1285 *
1286 * @return The port component of this URI,
1287 * or <tt>-1</tt> if the port is undefined
1288 */
1289 public int getPort() {
1290 return port;
1291 }
1292
1293 /**
1294 * Returns the raw path component of this URI.
1295 *
1296 * <p> The path component of a URI, if defined, only contains the slash
1297 * character (<tt>'/'</tt>), the commercial-at character (<tt>'@'</tt>),
1298 * and characters in the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>,
1299 * and <i>other</i> categories. </p>
1300 *
1301 * @return The path component of this URI,
1302 * or <tt>null</tt> if the path is undefined
1303 */
1304 public String getRawPath() {
1305 return path;
1306 }
1307
1308 /**
1309 * Returns the decoded path component of this URI.
1310 *
1311 * <p> The string returned by this method is equal to that returned by the
1312 * {@link #getRawPath() getRawPath} method except that all sequences of
1313 * escaped octets are <a href="#decode">decoded</a>. </p>
1314 *
1315 * @return The decoded path component of this URI,
1316 * or <tt>null</tt> if the path is undefined
1317 */
1318 public String getPath() {
1319 if ((decodedPath == null) && (path != null))
1320 decodedPath = decode(path);
1321 return decodedPath;
1322 }
1323
1324 /**
1325 * Returns the raw query component of this URI.
1326 *
1327 * <p> The query component of a URI, if defined, only contains legal URI
1328 * characters. </p>
1329 *
1330 * @return The raw query component of this URI,
1331 * or <tt>null</tt> if the query is undefined
1332 */
1333 public String getRawQuery() {
1334 return query;
1335 }
1336
1337 /**
1338 * Returns the decoded query component of this URI.
1339 *
1340 * <p> The string returned by this method is equal to that returned by the
1341 * {@link #getRawQuery() getRawQuery} method except that all sequences of
1342 * escaped octets are <a href="#decode">decoded</a>. </p>
1343 *
1344 * @return The decoded query component of this URI,
1345 * or <tt>null</tt> if the query is undefined
1346 */
1347 public String getQuery() {
1348 if ((decodedQuery == null) && (query != null))
1349 decodedQuery = decode(query);
1350 return decodedQuery;
1351 }
1352
1353 /**
1354 * Returns the raw fragment component of this URI.
1355 *
1356 * <p> The fragment component of a URI, if defined, only contains legal URI
1357 * characters. </p>
1358 *
1359 * @return The raw fragment component of this URI,
1360 * or <tt>null</tt> if the fragment is undefined
1361 */
1362 public String getRawFragment() {
1363 return fragment;
1364 }
1365
1366 /**
1367 * Returns the decoded fragment component of this URI.
1368 *
1369 * <p> The string returned by this method is equal to that returned by the
1370 * {@link #getRawFragment() getRawFragment} method except that all
1371 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1372 *
1373 * @return The decoded fragment component of this URI,
1374 * or <tt>null</tt> if the fragment is undefined
1375 */
1376 public String getFragment() {
1377 if ((decodedFragment == null) && (fragment != null))
1378 decodedFragment = decode(fragment);
1379 return decodedFragment;
1380 }
1381
1382
1383 // -- Equality, comparison, hash code, toString, and serialization --
1384
1385 /**
1386 * Tests this URI for equality with another object.
1387 *
1388 * <p> If the given object is not a URI then this method immediately
1389 * returns <tt>false</tt>.
1390 *
1391 * <p> For two URIs to be considered equal requires that either both are
1392 * opaque or both are hierarchical. Their schemes must either both be
1393 * undefined or else be equal without regard to case. Their fragments
1394 * must either both be undefined or else be equal.
1395 *
1396 * <p> For two opaque URIs to be considered equal, their scheme-specific
1397 * parts must be equal.
1398 *
1399 * <p> For two hierarchical URIs to be considered equal, their paths must
1400 * be equal and their queries must either both be undefined or else be
1401 * equal. Their authorities must either both be undefined, or both be
1402 * registry-based, or both be server-based. If their authorities are
1403 * defined and are registry-based, then they must be equal. If their
1404 * authorities are defined and are server-based, then their hosts must be
1405 * equal without regard to case, their port numbers must be equal, and
1406 * their user-information components must be equal.
1407 *
1408 * <p> When testing the user-information, path, query, fragment, authority,
1409 * or scheme-specific parts of two URIs for equality, the raw forms rather
1410 * than the encoded forms of these components are compared and the
1411 * hexadecimal digits of escaped octets are compared without regard to
1412 * case.
1413 *
1414 * <p> This method satisfies the general contract of the {@link
1415 * java.lang.Object#equals(Object) Object.equals} method. </p>
1416 *
1417 * @param ob The object to which this object is to be compared
1418 *
1419 * @return <tt>true</tt> if, and only if, the given object is a URI that
1420 * is identical to this URI
1421 */
1422 public boolean equals(Object ob) {
1423 if (ob == this)
1424 return true;
1425 if (!(ob instanceof URI))
1426 return false;
1427 URI that = (URI)ob;
1428 if (this.isOpaque() != that.isOpaque()) return false;
1429 if (!equalIgnoringCase(this.scheme, that.scheme)) return false;
1430 if (!equal(this.fragment, that.fragment)) return false;
1431
1432 // Opaque
1433 if (this.isOpaque())
1434 return equal(this.schemeSpecificPart, that.schemeSpecificPart);
1435
1436 // Hierarchical
1437 if (!equal(this.path, that.path)) return false;
1438 if (!equal(this.query, that.query)) return false;
1439
1440 // Authorities
1441 if (this.authority == that.authority) return true;
1442 if (this.host != null) {
1443 // Server-based
1444 if (!equal(this.userInfo, that.userInfo)) return false;
1445 if (!equalIgnoringCase(this.host, that.host)) return false;
1446 if (this.port != that.port) return false;
1447 } else if (this.authority != null) {
1448 // Registry-based
1449 if (!equal(this.authority, that.authority)) return false;
1450 } else if (this.authority != that.authority) {
1451 return false;
1452 }
1453
1454 return true;
1455 }
1456
1457 /**
1458 * Returns a hash-code value for this URI. The hash code is based upon all
1459 * of the URI's components, and satisfies the general contract of the
1460 * {@link java.lang.Object#hashCode() Object.hashCode} method.
1461 *
1462 * @return A hash-code value for this URI
1463 */
1464 public int hashCode() {
1465 if (hash != 0)
1466 return hash;
1467 int h = hashIgnoringCase(0, scheme);
1468 h = hash(h, fragment);
1469 if (isOpaque()) {
1470 h = hash(h, schemeSpecificPart);
1471 } else {
1472 h = hash(h, path);
1473 h = hash(h, query);
1474 if (host != null) {
1475 h = hash(h, userInfo);
1476 h = hashIgnoringCase(h, host);
1477 h += 1949 * port;
1478 } else {
1479 h = hash(h, authority);
1480 }
1481 }
1482 hash = h;
1483 return h;
1484 }
1485
1486 /**
1487 * Compares this URI to another object, which must be a URI.
1488 *
1489 * <p> When comparing corresponding components of two URIs, if one
1490 * component is undefined but the other is defined then the first is
1491 * considered to be less than the second. Unless otherwise noted, string
1492 * components are ordered according to their natural, case-sensitive
1493 * ordering as defined by the {@link java.lang.String#compareTo(Object)
1494 * String.compareTo} method. String components that are subject to
1495 * encoding are compared by comparing their raw forms rather than their
1496 * encoded forms.
1497 *
1498 * <p> The ordering of URIs is defined as follows: </p>
1499 *
1500 * <ul type=disc>
1501 *
1502 * <li><p> Two URIs with different schemes are ordered according the
1503 * ordering of their schemes, without regard to case. </p></li>
1504 *
1505 * <li><p> A hierarchical URI is considered to be less than an opaque URI
1506 * with an identical scheme. </p></li>
1507 *
1508 * <li><p> Two opaque URIs with identical schemes are ordered according
1509 * to the ordering of their scheme-specific parts. </p></li>
1510 *
1511 * <li><p> Two opaque URIs with identical schemes and scheme-specific
1512 * parts are ordered according to the ordering of their
1513 * fragments. </p></li>
1514 *
1515 * <li><p> Two hierarchical URIs with identical schemes are ordered
1516 * according to the ordering of their authority components: </p></li>
1517 *
1518 * <ul type=disc>
1519 *
1520 * <li><p> If both authority components are server-based then the URIs
1521 * are ordered according to their user-information components; if these
1522 * components are identical then the URIs are ordered according to the
1523 * ordering of their hosts, without regard to case; if the hosts are
1524 * identical then the URIs are ordered according to the ordering of
1525 * their ports. </p></li>
1526 *
1527 * <li><p> If one or both authority components are registry-based then
1528 * the URIs are ordered according to the ordering of their authority
1529 * components. </p></li>
1530 *
1531 * </ul>
1532 *
1533 * <li><p> Finally, two hierarchical URIs with identical schemes and
1534 * authority components are ordered according to the ordering of their
1535 * paths; if their paths are identical then they are ordered according to
1536 * the ordering of their queries; if the queries are identical then they
1537 * are ordered according to the order of their fragments. </p></li>
1538 *
1539 * </ul>
1540 *
1541 * <p> This method satisfies the general contract of the {@link
1542 * java.lang.Comparable#compareTo(Object) Comparable.compareTo}
1543 * method. </p>
1544 *
1545 * @param that
1546 * The object to which this URI is to be compared
1547 *
1548 * @return A negative integer, zero, or a positive integer as this URI is
1549 * less than, equal to, or greater than the given URI
1550 *
1551 * @throws ClassCastException
1552 * If the given object is not a URI
1553 */
1554 public int compareTo(URI that) {
1555 int c;
1556
1557 if ((c = compareIgnoringCase(this.scheme, that.scheme)) != 0)
1558 return c;
1559
1560 if (this.isOpaque()) {
1561 if (that.isOpaque()) {
1562 // Both opaque
1563 if ((c = compare(this.schemeSpecificPart,
1564 that.schemeSpecificPart)) != 0)
1565 return c;
1566 return compare(this.fragment, that.fragment);
1567 }
1568 return +1; // Opaque > hierarchical
1569 } else if (that.isOpaque()) {
1570 return -1; // Hierarchical < opaque
1571 }
1572
1573 // Hierarchical
1574 if ((this.host != null) && (that.host != null)) {
1575 // Both server-based
1576 if ((c = compare(this.userInfo, that.userInfo)) != 0)
1577 return c;
1578 if ((c = compareIgnoringCase(this.host, that.host)) != 0)
1579 return c;
1580 if ((c = this.port - that.port) != 0)
1581 return c;
1582 } else {
1583 // If one or both authorities are registry-based then we simply
1584 // compare them in the usual, case-sensitive way. If one is
1585 // registry-based and one is server-based then the strings are
1586 // guaranteed to be unequal, hence the comparison will never return
1587 // zero and the compareTo and equals methods will remain
1588 // consistent.
1589 if ((c = compare(this.authority, that.authority)) != 0) return c;
1590 }
1591
1592 if ((c = compare(this.path, that.path)) != 0) return c;
1593 if ((c = compare(this.query, that.query)) != 0) return c;
1594 return compare(this.fragment, that.fragment);
1595 }
1596
1597 /**
1598 * Returns the content of this URI as a string.
1599 *
1600 * <p> If this URI was created by invoking one of the constructors in this
1601 * class then a string equivalent to the original input string, or to the
1602 * string computed from the originally-given components, as appropriate, is
1603 * returned. Otherwise this URI was created by normalization, resolution,
1604 * or relativization, and so a string is constructed from this URI's
1605 * components according to the rules specified in <a
1606 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
1607 * section 5.2, step 7. </p>
1608 *
1609 * @return The string form of this URI
1610 */
1611 public String toString() {
1612 defineString();
1613 return string;
1614 }
1615
1616 /**
1617 * Returns the content of this URI as a US-ASCII string.
1618 *
1619 * <p> If this URI does not contain any characters in the <i>other</i>
1620 * category then an invocation of this method will return the same value as
1621 * an invocation of the {@link #toString() toString} method. Otherwise
1622 * this method works as if by invoking that method and then <a
1623 * href="#encode">encoding</a> the result. </p>
1624 *
1625 * @return The string form of this URI, encoded as needed
1626 * so that it only contains characters in the US-ASCII
1627 * charset
1628 */
1629 public String toASCIIString() {
1630 defineString();
1631 return encode(string);
1632 }
1633
1634
1635 // -- Serialization support --
1636
1637 /**
1638 * Saves the content of this URI to the given serial stream.
1639 *
1640 * <p> The only serializable field of a URI instance is its <tt>string</tt>
1641 * field. That field is given a value, if it does not have one already,
1642 * and then the {@link java.io.ObjectOutputStream#defaultWriteObject()}
1643 * method of the given object-output stream is invoked. </p>
1644 *
1645 * @param os The object-output stream to which this object
1646 * is to be written
1647 */
1648 private void writeObject(ObjectOutputStream os)
1649 throws IOException
1650 {
1651 defineString();
1652 os.defaultWriteObject(); // Writes the string field only
1653 }
1654
1655 /**
1656 * Reconstitutes a URI from the given serial stream.
1657 *
1658 * <p> The {@link java.io.ObjectInputStream#defaultReadObject()} method is
1659 * invoked to read the value of the <tt>string</tt> field. The result is
1660 * then parsed in the usual way.
1661 *
1662 * @param is The object-input stream from which this object
1663 * is being read
1664 */
1665 private void readObject(ObjectInputStream is)
1666 throws ClassNotFoundException, IOException
1667 {
1668 port = -1; // Argh
1669 is.defaultReadObject();
1670 try {
1671 new Parser(string).parse(false);
1672 } catch (URISyntaxException x) {
1673 IOException y = new InvalidObjectException("Invalid URI");
1674 y.initCause(x);
1675 throw y;
1676 }
1677 }
1678
1679
1680 // -- End of public methods --
1681
1682
1683 // -- Utility methods for string-field comparison and hashing --
1684
1685 // These methods return appropriate values for null string arguments,
1686 // thereby simplifying the equals, hashCode, and compareTo methods.
1687 //
1688 // The case-ignoring methods should only be applied to strings whose
1689 // characters are all known to be US-ASCII. Because of this restriction,
1690 // these methods are faster than the similar methods in the String class.
1691
1692 // US-ASCII only
1693 private static int toLower(char c) {
1694 if ((c >= 'A') && (c <= 'Z'))
1695 return c + ('a' - 'A');
1696 return c;
1697 }
1698
1699 private static boolean equal(String s, String t) {
1700 if (s == t) return true;
1701 if ((s != null) && (t != null)) {
1702 if (s.length() != t.length())
1703 return false;
1704 if (s.indexOf('%') < 0)
1705 return s.equals(t);
1706 int n = s.length();
1707 for (int i = 0; i < n;) {
1708 char c = s.charAt(i);
1709 char d = t.charAt(i);
1710 if (c != '%') {
1711 if (c != d)
1712 return false;
1713 i++;
1714 continue;
1715 }
1716 i++;
1717 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1718 return false;
1719 i++;
1720 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1721 return false;
1722 i++;
1723 }
1724 return true;
1725 }
1726 return false;
1727 }
1728
1729 // US-ASCII only
1730 private static boolean equalIgnoringCase(String s, String t) {
1731 if (s == t) return true;
1732 if ((s != null) && (t != null)) {
1733 int n = s.length();
1734 if (t.length() != n)
1735 return false;
1736 for (int i = 0; i < n; i++) {
1737 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1738 return false;
1739 }
1740 return true;
1741 }
1742 return false;
1743 }
1744
1745 private static int hash(int hash, String s) {
1746 if (s == null) return hash;
1747 return hash * 127 + s.hashCode();
1748 }
1749
1750 // US-ASCII only
1751 private static int hashIgnoringCase(int hash, String s) {
1752 if (s == null) return hash;
1753 int h = hash;
1754 int n = s.length();
1755 for (int i = 0; i < n; i++)
1756 h = 31 * h + toLower(s.charAt(i));
1757 return h;
1758 }
1759
1760 private static int compare(String s, String t) {
1761 if (s == t) return 0;
1762 if (s != null) {
1763 if (t != null)
1764 return s.compareTo(t);
1765 else
1766 return +1;
1767 } else {
1768 return -1;
1769 }
1770 }
1771
1772 // US-ASCII only
1773 private static int compareIgnoringCase(String s, String t) {
1774 if (s == t) return 0;
1775 if (s != null) {
1776 if (t != null) {
1777 int sn = s.length();
1778 int tn = t.length();
1779 int n = sn < tn ? sn : tn;
1780 for (int i = 0; i < n; i++) {
1781 int c = toLower(s.charAt(i)) - toLower(t.charAt(i));
1782 if (c != 0)
1783 return c;
1784 }
1785 return sn - tn;
1786 }
1787 return +1;
1788 } else {
1789 return -1;
1790 }
1791 }
1792
1793
1794 // -- String construction --
1795
1796 // If a scheme is given then the path, if given, must be absolute
1797 //
1798 private static void checkPath(String s, String scheme, String path)
1799 throws URISyntaxException
1800 {
1801 if (scheme != null) {
1802 if ((path != null)
1803 && ((path.length() > 0) && (path.charAt(0) != '/')))
1804 throw new URISyntaxException(s,
1805 "Relative path in absolute URI");
1806 }
1807 }
1808
1809 private void appendAuthority(StringBuffer sb,
1810 String authority,
1811 String userInfo,
1812 String host,
1813 int port)
1814 {
1815 if (host != null) {
1816 sb.append("//");
1817 if (userInfo != null) {
1818 sb.append(quote(userInfo, L_USERINFO, H_USERINFO));
1819 sb.append('@');
1820 }
1821 boolean needBrackets = ((host.indexOf(':') >= 0)
1822 && !host.startsWith("[")
1823 && !host.endsWith("]"));
1824 if (needBrackets) sb.append('[');
1825 sb.append(host);
1826 if (needBrackets) sb.append(']');
1827 if (port != -1) {
1828 sb.append(':');
1829 sb.append(port);
1830 }
1831 } else if (authority != null) {
1832 sb.append("//");
1833 if (authority.startsWith("[")) {
1834 int end = authority.indexOf("]");
1835 if (end != -1 && authority.indexOf(":")!=-1) {
1836 String doquote, dontquote;
1837 if (end == authority.length()) {
1838 dontquote = authority;
1839 doquote = "";
1840 } else {
1841 dontquote = authority.substring(0,end+1);
1842 doquote = authority.substring(end+1);
1843 }
1844 sb.append (dontquote);
1845 sb.append(quote(doquote,
1846 L_REG_NAME | L_SERVER,
1847 H_REG_NAME | H_SERVER));
1848 }
1849 } else {
1850 sb.append(quote(authority,
1851 L_REG_NAME | L_SERVER,
1852 H_REG_NAME | H_SERVER));
1853 }
1854 }
1855 }
1856
1857 private void appendSchemeSpecificPart(StringBuffer sb,
1858 String opaquePart,
1859 String authority,
1860 String userInfo,
1861 String host,
1862 int port,
1863 String path,
1864 String query)
1865 {
1866 if (opaquePart != null) {
1867 /* check if SSP begins with an IPv6 address
1868 * because we must not quote a literal IPv6 address
1869 */
1870 if (opaquePart.startsWith("//[")) {
1871 int end = opaquePart.indexOf("]");
1872 if (end != -1 && opaquePart.indexOf(":")!=-1) {
1873 String doquote, dontquote;
1874 if (end == opaquePart.length()) {
1875 dontquote = opaquePart;
1876 doquote = "";
1877 } else {
1878 dontquote = opaquePart.substring(0,end+1);
1879 doquote = opaquePart.substring(end+1);
1880 }
1881 sb.append (dontquote);
1882 sb.append(quote(doquote, L_URIC, H_URIC));
1883 }
1884 } else {
1885 sb.append(quote(opaquePart, L_URIC, H_URIC));
1886 }
1887 } else {
1888 appendAuthority(sb, authority, userInfo, host, port);
1889 if (path != null)
1890 sb.append(quote(path, L_PATH, H_PATH));
1891 if (query != null) {
1892 sb.append('?');
1893 sb.append(quote(query, L_URIC, H_URIC));
1894 }
1895 }
1896 }
1897
1898 private void appendFragment(StringBuffer sb, String fragment) {
1899 if (fragment != null) {
1900 sb.append('#');
1901 sb.append(quote(fragment, L_URIC, H_URIC));
1902 }
1903 }
1904
1905 private String toString(String scheme,
1906 String opaquePart,
1907 String authority,
1908 String userInfo,
1909 String host,
1910 int port,
1911 String path,
1912 String query,
1913 String fragment)
1914 {
1915 StringBuffer sb = new StringBuffer();
1916 if (scheme != null) {
1917 sb.append(scheme);
1918 sb.append(':');
1919 }
1920 appendSchemeSpecificPart(sb, opaquePart,
1921 authority, userInfo, host, port,
1922 path, query);
1923 appendFragment(sb, fragment);
1924 return sb.toString();
1925 }
1926
1927 private void defineSchemeSpecificPart() {
1928 if (schemeSpecificPart != null) return;
1929 StringBuffer sb = new StringBuffer();
1930 appendSchemeSpecificPart(sb, null, getAuthority(), getUserInfo(),
1931 host, port, getPath(), getQuery());
1932 if (sb.length() == 0) return;
1933 schemeSpecificPart = sb.toString();
1934 }
1935
1936 private void defineString() {
1937 if (string != null) return;
1938
1939 StringBuffer sb = new StringBuffer();
1940 if (scheme != null) {
1941 sb.append(scheme);
1942 sb.append(':');
1943 }
1944 if (isOpaque()) {
1945 sb.append(schemeSpecificPart);
1946 } else {
1947 if (host != null) {
1948 sb.append("//");
1949 if (userInfo != null) {
1950 sb.append(userInfo);
1951 sb.append('@');
1952 }
1953 boolean needBrackets = ((host.indexOf(':') >= 0)
1954 && !host.startsWith("[")
1955 && !host.endsWith("]"));
1956 if (needBrackets) sb.append('[');
1957 sb.append(host);
1958 if (needBrackets) sb.append(']');
1959 if (port != -1) {
1960 sb.append(':');
1961 sb.append(port);
1962 }
1963 } else if (authority != null) {
1964 sb.append("//");
1965 sb.append(authority);
1966 }
1967 if (path != null)
1968 sb.append(path);
1969 if (query != null) {
1970 sb.append('?');
1971 sb.append(query);
1972 }
1973 }
1974 if (fragment != null) {
1975 sb.append('#');
1976 sb.append(fragment);
1977 }
1978 string = sb.toString();
1979 }
1980
1981
1982 // -- Normalization, resolution, and relativization --
1983
1984 // RFC2396 5.2 (6)
1985 private static String resolvePath(String base, String child,
1986 boolean absolute)
1987 {
1988 int i = base.lastIndexOf('/');
1989 int cn = child.length();
1990 String path = "";
1991
1992 if (cn == 0) {
1993 // 5.2 (6a)
1994 if (i >= 0)
1995 path = base.substring(0, i + 1);
1996 } else {
1997 StringBuffer sb = new StringBuffer(base.length() + cn);
1998 // 5.2 (6a)
1999 if (i >= 0)
2000 sb.append(base.substring(0, i + 1));
2001 // 5.2 (6b)
2002 sb.append(child);
2003 path = sb.toString();
2004 }
2005
2006 // 5.2 (6c-f)
2007 String np = normalize(path);
2008
2009 // 5.2 (6g): If the result is absolute but the path begins with "../",
2010 // then we simply leave the path as-is
2011
2012 return np;
2013 }
2014
2015 // RFC2396 5.2
2016 private static URI resolve(URI base, URI child) {
2017 // check if child if opaque first so that NPE is thrown
2018 // if child is null.
2019 if (child.isOpaque() || base.isOpaque())
2020 return child;
2021
2022 // 5.2 (2): Reference to current document (lone fragment)
2023 if ((child.scheme == null) && (child.authority == null)
2024 && child.path.equals("") && (child.fragment != null)
2025 && (child.query == null)) {
2026 if ((base.fragment != null)
2027 && child.fragment.equals(base.fragment)) {
2028 return base;
2029 }
2030 URI ru = new URI();
2031 ru.scheme = base.scheme;
2032 ru.authority = base.authority;
2033 ru.userInfo = base.userInfo;
2034 ru.host = base.host;
2035 ru.port = base.port;
2036 ru.path = base.path;
2037 ru.fragment = child.fragment;
2038 ru.query = base.query;
2039 return ru;
2040 }
2041
2042 // 5.2 (3): Child is absolute
2043 if (child.scheme != null)
2044 return child;
2045
2046 URI ru = new URI(); // Resolved URI
2047 ru.scheme = base.scheme;
2048 ru.query = child.query;
2049 ru.fragment = child.fragment;
2050
2051 // 5.2 (4): Authority
2052 if (child.authority == null) {
2053 ru.authority = base.authority;
2054 ru.host = base.host;
2055 ru.userInfo = base.userInfo;
2056 ru.port = base.port;
2057
2058 String cp = (child.path == null) ? "" : child.path;
2059 if ((cp.length() > 0) && (cp.charAt(0) == '/')) {
2060 // 5.2 (5): Child path is absolute
2061 ru.path = child.path;
2062 } else {
2063 // 5.2 (6): Resolve relative path
2064 ru.path = resolvePath(base.path, cp, base.isAbsolute());
2065 }
2066 } else {
2067 ru.authority = child.authority;
2068 ru.host = child.host;
2069 ru.userInfo = child.userInfo;
2070 ru.host = child.host;
2071 ru.port = child.port;
2072 ru.path = child.path;
2073 }
2074
2075 // 5.2 (7): Recombine (nothing to do here)
2076 return ru;
2077 }
2078
2079 // If the given URI's path is normal then return the URI;
2080 // o.w., return a new URI containing the normalized path.
2081 //
2082 private static URI normalize(URI u) {
2083 if (u.isOpaque() || (u.path == null) || (u.path.length() == 0))
2084 return u;
2085
2086 String np = normalize(u.path);
2087 if (np == u.path)
2088 return u;
2089
2090 URI v = new URI();
2091 v.scheme = u.scheme;
2092 v.fragment = u.fragment;
2093 v.authority = u.authority;
2094 v.userInfo = u.userInfo;
2095 v.host = u.host;
2096 v.port = u.port;
2097 v.path = np;
2098 v.query = u.query;
2099 return v;
2100 }
2101
2102 // If both URIs are hierarchical, their scheme and authority components are
2103 // identical, and the base path is a prefix of the child's path, then
2104 // return a relative URI that, when resolved against the base, yields the
2105 // child; otherwise, return the child.
2106 //
2107 private static URI relativize(URI base, URI child) {
2108 // check if child if opaque first so that NPE is thrown
2109 // if child is null.
2110 if (child.isOpaque() || base.isOpaque())
2111 return child;
2112 if (!equalIgnoringCase(base.scheme, child.scheme)
2113 || !equal(base.authority, child.authority))
2114 return child;
2115
2116 String bp = normalize(base.path);
2117 String cp = normalize(child.path);
2118 if (!bp.equals(cp)) {
2119 if (!bp.endsWith("/"))
2120 bp = bp + "/";
2121 if (!cp.startsWith(bp))
2122 return child;
2123 }
2124
2125 URI v = new URI();
2126 v.path = cp.substring(bp.length());
2127 v.query = child.query;
2128 v.fragment = child.fragment;
2129 return v;
2130 }
2131
2132
2133
2134 // -- Path normalization --
2135
2136 // The following algorithm for path normalization avoids the creation of a
2137 // string object for each segment, as well as the use of a string buffer to
2138 // compute the final result, by using a single char array and editing it in
2139 // place. The array is first split into segments, replacing each slash
2140 // with '\0' and creating a segment-index array, each element of which is
2141 // the index of the first char in the corresponding segment. We then walk
2142 // through both arrays, removing ".", "..", and other segments as necessary
2143 // by setting their entries in the index array to -1. Finally, the two
2144 // arrays are used to rejoin the segments and compute the final result.
2145 //
2146 // This code is based upon src/solaris/native/java/io/canonicalize_md.c
2147
2148
2149 // Check the given path to see if it might need normalization. A path
2150 // might need normalization if it contains duplicate slashes, a "."
2151 // segment, or a ".." segment. Return -1 if no further normalization is
2152 // possible, otherwise return the number of segments found.
2153 //
2154 // This method takes a string argument rather than a char array so that
2155 // this test can be performed without invoking path.toCharArray().
2156 //
2157 static private int needsNormalization(String path) {
2158 boolean normal = true;
2159 int ns = 0; // Number of segments
2160 int end = path.length() - 1; // Index of last char in path
2161 int p = 0; // Index of next char in path
2162
2163 // Skip initial slashes
2164 while (p <= end) {
2165 if (path.charAt(p) != '/') break;
2166 p++;
2167 }
2168 if (p > 1) normal = false;
2169
2170 // Scan segments
2171 while (p <= end) {
2172
2173 // Looking at "." or ".." ?
2174 if ((path.charAt(p) == '.')
2175 && ((p == end)
2176 || ((path.charAt(p + 1) == '/')
2177 || ((path.charAt(p + 1) == '.')
2178 && ((p + 1 == end)
2179 || (path.charAt(p + 2) == '/')))))) {
2180 normal = false;
2181 }
2182 ns++;
2183
2184 // Find beginning of next segment
2185 while (p <= end) {
2186 if (path.charAt(p++) != '/')
2187 continue;
2188
2189 // Skip redundant slashes
2190 while (p <= end) {
2191 if (path.charAt(p) != '/') break;
2192 normal = false;
2193 p++;
2194 }
2195
2196 break;
2197 }
2198 }
2199
2200 return normal ? -1 : ns;
2201 }
2202
2203
2204 // Split the given path into segments, replacing slashes with nulls and
2205 // filling in the given segment-index array.
2206 //
2207 // Preconditions:
2208 // segs.length == Number of segments in path
2209 //
2210 // Postconditions:
2211 // All slashes in path replaced by '\0'
2212 // segs[i] == Index of first char in segment i (0 <= i < segs.length)
2213 //
2214 static private void split(char[] path, int[] segs) {
2215 int end = path.length - 1; // Index of last char in path
2216 int p = 0; // Index of next char in path
2217 int i = 0; // Index of current segment
2218
2219 // Skip initial slashes
2220 while (p <= end) {
2221 if (path[p] != '/') break;
2222 path[p] = '\0';
2223 p++;
2224 }
2225
2226 while (p <= end) {
2227
2228 // Note start of segment
2229 segs[i++] = p++;
2230
2231 // Find beginning of next segment
2232 while (p <= end) {
2233 if (path[p++] != '/')
2234 continue;
2235 path[p - 1] = '\0';
2236
2237 // Skip redundant slashes
2238 while (p <= end) {
2239 if (path[p] != '/') break;
2240 path[p++] = '\0';
2241 }
2242 break;
2243 }
2244 }
2245
2246 if (i != segs.length)
2247 throw new InternalError(); // ASSERT
2248 }
2249
2250
2251 // Join the segments in the given path according to the given segment-index
2252 // array, ignoring those segments whose index entries have been set to -1,
2253 // and inserting slashes as needed. Return the length of the resulting
2254 // path.
2255 //
2256 // Preconditions:
2257 // segs[i] == -1 implies segment i is to be ignored
2258 // path computed by split, as above, with '\0' having replaced '/'
2259 //
2260 // Postconditions:
2261 // path[0] .. path[return value] == Resulting path
2262 //
2263 static private int join(char[] path, int[] segs) {
2264 int ns = segs.length; // Number of segments
2265 int end = path.length - 1; // Index of last char in path
2266 int p = 0; // Index of next path char to write
2267
2268 if (path[p] == '\0') {
2269 // Restore initial slash for absolute paths
2270 path[p++] = '/';
2271 }
2272
2273 for (int i = 0; i < ns; i++) {
2274 int q = segs[i]; // Current segment
2275 if (q == -1)
2276 // Ignore this segment
2277 continue;
2278
2279 if (p == q) {
2280 // We're already at this segment, so just skip to its end
2281 while ((p <= end) && (path[p] != '\0'))
2282 p++;
2283 if (p <= end) {
2284 // Preserve trailing slash
2285 path[p++] = '/';
2286 }
2287 } else if (p < q) {
2288 // Copy q down to p
2289 while ((q <= end) && (path[q] != '\0'))
2290 path[p++] = path[q++];
2291 if (q <= end) {
2292 // Preserve trailing slash
2293 path[p++] = '/';
2294 }
2295 } else
2296 throw new InternalError(); // ASSERT false
2297 }
2298
2299 return p;
2300 }
2301
2302
2303 // Remove "." segments from the given path, and remove segment pairs
2304 // consisting of a non-".." segment followed by a ".." segment.
2305 //
2306 private static void removeDots(char[] path, int[] segs) {
2307 int ns = segs.length;
2308 int end = path.length - 1;
2309
2310 for (int i = 0; i < ns; i++) {
2311 int dots = 0; // Number of dots found (0, 1, or 2)
2312
2313 // Find next occurrence of "." or ".."
2314 do {
2315 int p = segs[i];
2316 if (path[p] == '.') {
2317 if (p == end) {
2318 dots = 1;
2319 break;
2320 } else if (path[p + 1] == '\0') {
2321 dots = 1;
2322 break;
2323 } else if ((path[p + 1] == '.')
2324 && ((p + 1 == end)
2325 || (path[p + 2] == '\0'))) {
2326 dots = 2;
2327 break;
2328 }
2329 }
2330 i++;
2331 } while (i < ns);
2332 if ((i > ns) || (dots == 0))
2333 break;
2334
2335 if (dots == 1) {
2336 // Remove this occurrence of "."
2337 segs[i] = -1;
2338 } else {
2339 // If there is a preceding non-".." segment, remove both that
2340 // segment and this occurrence of ".."; otherwise, leave this
2341 // ".." segment as-is.
2342 int j;
2343 for (j = i - 1; j >= 0; j--) {
2344 if (segs[j] != -1) break;
2345 }
2346 if (j >= 0) {
2347 int q = segs[j];
2348 if (!((path[q] == '.')
2349 && (path[q + 1] == '.')
2350 && (path[q + 2] == '\0'))) {
2351 segs[i] = -1;
2352 segs[j] = -1;
2353 }
2354 }
2355 }
2356 }
2357 }
2358
2359
2360 // DEVIATION: If the normalized path is relative, and if the first
2361 // segment could be parsed as a scheme name, then prepend a "." segment
2362 //
2363 private static void maybeAddLeadingDot(char[] path, int[] segs) {
2364
2365 if (path[0] == '\0')
2366 // The path is absolute
2367 return;
2368
2369 int ns = segs.length;
2370 int f = 0; // Index of first segment
2371 while (f < ns) {
2372 if (segs[f] >= 0)
2373 break;
2374 f++;
2375 }
2376 if ((f >= ns) || (f == 0))
2377 // The path is empty, or else the original first segment survived,
2378 // in which case we already know that no leading "." is needed
2379 return;
2380
2381 int p = segs[f];
2382 while ((p < path.length) && (path[p] != ':') && (path[p] != '\0')) p++;
2383 if (p >= path.length || path[p] == '\0')
2384 // No colon in first segment, so no "." needed
2385 return;
2386
2387 // At this point we know that the first segment is unused,
2388 // hence we can insert a "." segment at that position
2389 path[0] = '.';
2390 path[1] = '\0';
2391 segs[0] = 0;
2392 }
2393
2394
2395 // Normalize the given path string. A normal path string has no empty
2396 // segments (i.e., occurrences of "//"), no segments equal to ".", and no
2397 // segments equal to ".." that are preceded by a segment not equal to "..".
2398 // In contrast to Unix-style pathname normalization, for URI paths we
2399 // always retain trailing slashes.
2400 //
2401 private static String normalize(String ps) {
2402
2403 // Does this path need normalization?
2404 int ns = needsNormalization(ps); // Number of segments
2405 if (ns < 0)
2406 // Nope -- just return it
2407 return ps;
2408
2409 char[] path = ps.toCharArray(); // Path in char-array form
2410
2411 // Split path into segments
2412 int[] segs = new int[ns]; // Segment-index array
2413 split(path, segs);
2414
2415 // Remove dots
2416 removeDots(path, segs);
2417
2418 // Prevent scheme-name confusion
2419 maybeAddLeadingDot(path, segs);
2420
2421 // Join the remaining segments and return the result
2422 String s = new String(path, 0, join(path, segs));
2423 if (s.equals(ps)) {
2424 // string was already normalized
2425 return ps;
2426 }
2427 return s;
2428 }
2429
2430
2431
2432 // -- Character classes for parsing --
2433
2434 // RFC2396 precisely specifies which characters in the US-ASCII charset are
2435 // permissible in the various components of a URI reference. We here
2436 // define a set of mask pairs to aid in enforcing these restrictions. Each
2437 // mask pair consists of two longs, a low mask and a high mask. Taken
2438 // together they represent a 128-bit mask, where bit i is set iff the
2439 // character with value i is permitted.
2440 //
2441 // This approach is more efficient than sequentially searching arrays of
2442 // permitted characters. It could be made still more efficient by
2443 // precompiling the mask information so that a character's presence in a
2444 // given mask could be determined by a single table lookup.
2445
2446 // Compute the low-order mask for the characters in the given string
2447 private static long lowMask(String chars) {
2448 int n = chars.length();
2449 long m = 0;
2450 for (int i = 0; i < n; i++) {
2451 char c = chars.charAt(i);
2452 if (c < 64)
2453 m |= (1L << c);
2454 }
2455 return m;
2456 }
2457
2458 // Compute the high-order mask for the characters in the given string
2459 private static long highMask(String chars) {
2460 int n = chars.length();
2461 long m = 0;
2462 for (int i = 0; i < n; i++) {
2463 char c = chars.charAt(i);
2464 if ((c >= 64) && (c < 128))
2465 m |= (1L << (c - 64));
2466 }
2467 return m;
2468 }
2469
2470 // Compute a low-order mask for the characters
2471 // between first and last, inclusive
2472 private static long lowMask(char first, char last) {
2473 long m = 0;
2474 int f = Math.max(Math.min(first, 63), 0);
2475 int l = Math.max(Math.min(last, 63), 0);
2476 for (int i = f; i <= l; i++)
2477 m |= 1L << i;
2478 return m;
2479 }
2480
2481 // Compute a high-order mask for the characters
2482 // between first and last, inclusive
2483 private static long highMask(char first, char last) {
2484 long m = 0;
2485 int f = Math.max(Math.min(first, 127), 64) - 64;
2486 int l = Math.max(Math.min(last, 127), 64) - 64;
2487 for (int i = f; i <= l; i++)
2488 m |= 1L << i;
2489 return m;
2490 }
2491
2492 // Tell whether the given character is permitted by the given mask pair
2493 private static boolean match(char c, long lowMask, long highMask) {
2494 if (c < 64)
2495 return ((1L << c) & lowMask) != 0;
2496 if (c < 128)
2497 return ((1L << (c - 64)) & highMask) != 0;
2498 return false;
2499 }
2500
2501 // Character-class masks, in reverse order from RFC2396 because
2502 // initializers for static fields cannot make forward references.
2503
2504 // digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" |
2505 // "8" | "9"
2506 private static final long L_DIGIT = lowMask('0', '9');
2507 private static final long H_DIGIT = 0L;
2508
2509 // upalpha = "A" | "B" | "C" | "D" | "E" | "F" | "G" | "H" | "I" |
2510 // "J" | "K" | "L" | "M" | "N" | "O" | "P" | "Q" | "R" |
2511 // "S" | "T" | "U" | "V" | "W" | "X" | "Y" | "Z"
2512 private static final long L_UPALPHA = 0L;
2513 private static final long H_UPALPHA = highMask('A', 'Z');
2514
2515 // lowalpha = "a" | "b" | "c" | "d" | "e" | "f" | "g" | "h" | "i" |
2516 // "j" | "k" | "l" | "m" | "n" | "o" | "p" | "q" | "r" |
2517 // "s" | "t" | "u" | "v" | "w" | "x" | "y" | "z"
2518 private static final long L_LOWALPHA = 0L;
2519 private static final long H_LOWALPHA = highMask('a', 'z');
2520
2521 // alpha = lowalpha | upalpha
2522 private static final long L_ALPHA = L_LOWALPHA | L_UPALPHA;
2523 private static final long H_ALPHA = H_LOWALPHA | H_UPALPHA;
2524
2525 // alphanum = alpha | digit
2526 private static final long L_ALPHANUM = L_DIGIT | L_ALPHA;
2527 private static final long H_ALPHANUM = H_DIGIT | H_ALPHA;
2528
2529 // hex = digit | "A" | "B" | "C" | "D" | "E" | "F" |
2530 // "a" | "b" | "c" | "d" | "e" | "f"
2531 private static final long L_HEX = L_DIGIT;
2532 private static final long H_HEX = highMask('A', 'F') | highMask('a', 'f');
2533
2534 // mark = "-" | "_" | "." | "!" | "~" | "*" | "'" |
2535 // "(" | ")"
2536 private static final long L_MARK = lowMask("-_.!~*'()");
2537 private static final long H_MARK = highMask("-_.!~*'()");
2538
2539 // unreserved = alphanum | mark
2540 private static final long L_UNRESERVED = L_ALPHANUM | L_MARK;
2541 private static final long H_UNRESERVED = H_ALPHANUM | H_MARK;
2542
2543 // reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" |
2544 // "$" | "," | "[" | "]"
2545 // Added per RFC2732: "[", "]"
2546 private static final long L_RESERVED = lowMask(";/?:@&=+$,[]");
2547 private static final long H_RESERVED = highMask(";/?:@&=+$,[]");
2548
2549 // The zero'th bit is used to indicate that escape pairs and non-US-ASCII
2550 // characters are allowed; this is handled by the scanEscape method below.
2551 private static final long L_ESCAPED = 1L;
2552 private static final long H_ESCAPED = 0L;
2553
2554 // uric = reserved | unreserved | escaped
2555 private static final long L_URIC = L_RESERVED | L_UNRESERVED | L_ESCAPED;
2556 private static final long H_URIC = H_RESERVED | H_UNRESERVED | H_ESCAPED;
2557
2558 // pchar = unreserved | escaped |
2559 // ":" | "@" | "&" | "=" | "+" | "$" | ","
2560 private static final long L_PCHAR
2561 = L_UNRESERVED | L_ESCAPED | lowMask(":@&=+$,");
2562 private static final long H_PCHAR
2563 = H_UNRESERVED | H_ESCAPED | highMask(":@&=+$,");
2564
2565 // All valid path characters
2566 private static final long L_PATH = L_PCHAR | lowMask(";/");
2567 private static final long H_PATH = H_PCHAR | highMask(";/");
2568
2569 // Dash, for use in domainlabel and toplabel
2570 private static final long L_DASH = lowMask("-");
2571 private static final long H_DASH = highMask("-");
2572
2573 // Dot, for use in hostnames
2574 private static final long L_DOT = lowMask(".");
2575 private static final long H_DOT = highMask(".");
2576
2577 // userinfo = *( unreserved | escaped |
2578 // ";" | ":" | "&" | "=" | "+" | "$" | "," )
2579 private static final long L_USERINFO
2580 = L_UNRESERVED | L_ESCAPED | lowMask(";:&=+$,");
2581 private static final long H_USERINFO
2582 = H_UNRESERVED | H_ESCAPED | highMask(";:&=+$,");
2583
2584 // reg_name = 1*( unreserved | escaped | "$" | "," |
2585 // ";" | ":" | "@" | "&" | "=" | "+" )
2586 private static final long L_REG_NAME
2587 = L_UNRESERVED | L_ESCAPED | lowMask("$,;:@&=+");
2588 private static final long H_REG_NAME
2589 = H_UNRESERVED | H_ESCAPED | highMask("$,;:@&=+");
2590
2591 // All valid characters for server-based authorities
2592 private static final long L_SERVER
2593 = L_USERINFO | L_ALPHANUM | L_DASH | lowMask(".:@[]");
2594 private static final long H_SERVER
2595 = H_USERINFO | H_ALPHANUM | H_DASH | highMask(".:@[]");
2596
2597 // Special case of server authority that represents an IPv6 address
2598 // In this case, a % does not signify an escape sequence
2599 private static final long L_SERVER_PERCENT
2600 = L_SERVER | lowMask("%");
2601 private static final long H_SERVER_PERCENT
2602 = H_SERVER | highMask("%");
2603 private static final long L_LEFT_BRACKET = lowMask("[");
2604 private static final long H_LEFT_BRACKET = highMask("[");
2605
2606 // scheme = alpha *( alpha | digit | "+" | "-" | "." )
2607 private static final long L_SCHEME = L_ALPHA | L_DIGIT | lowMask("+-.");
2608 private static final long H_SCHEME = H_ALPHA | H_DIGIT | highMask("+-.");
2609
2610 // uric_no_slash = unreserved | escaped | ";" | "?" | ":" | "@" |
2611 // "&" | "=" | "+" | "$" | ","
2612 private static final long L_URIC_NO_SLASH
2613 = L_UNRESERVED | L_ESCAPED | lowMask(";?:@&=+$,");
2614 private static final long H_URIC_NO_SLASH
2615 = H_UNRESERVED | H_ESCAPED | highMask(";?:@&=+$,");
2616
2617
2618 // -- Escaping and encoding --
2619
2620 private final static char[] hexDigits = {
2621 '0', '1', '2', '3', '4', '5', '6', '7',
2622 '8', '9', 'A', 'B', 'C', 'D', 'E', 'F'
2623 };
2624
2625 private static void appendEscape(StringBuffer sb, byte b) {
2626 sb.append('%');
2627 sb.append(hexDigits[(b >> 4) & 0x0f]);
2628 sb.append(hexDigits[(b >> 0) & 0x0f]);
2629 }
2630
2631 private static void appendEncoded(StringBuffer sb, char c) {
2632 ByteBuffer bb = null;
2633 try {
2634 bb = ThreadLocalCoders.encoderFor("UTF-8")
2635 .encode(CharBuffer.wrap("" + c));
2636 } catch (CharacterCodingException x) {
2637 assert false;
2638 }
2639 while (bb.hasRemaining()) {
2640 int b = bb.get() & 0xff;
2641 if (b >= 0x80)
2642 appendEscape(sb, (byte)b);
2643 else
2644 sb.append((char)b);
2645 }
2646 }
2647
2648 // Quote any characters in s that are not permitted
2649 // by the given mask pair
2650 //
2651 private static String quote(String s, long lowMask, long highMask) {
2652 int n = s.length();
2653 StringBuffer sb = null;
2654 boolean allowNonASCII = ((lowMask & L_ESCAPED) != 0);
2655 for (int i = 0; i < s.length(); i++) {
2656 char c = s.charAt(i);
2657 if (c < '\u0080') {
2658 if (!match(c, lowMask, highMask)) {
2659 if (sb == null) {
2660 sb = new StringBuffer();
2661 sb.append(s.substring(0, i));
2662 }
2663 appendEscape(sb, (byte)c);
2664 } else {
2665 if (sb != null)
2666 sb.append(c);
2667 }
2668 } else if (allowNonASCII
2669 && (Character.isSpaceChar(c)
2670 || Character.isISOControl(c))) {
2671 if (sb == null) {
2672 sb = new StringBuffer();
2673 sb.append(s.substring(0, i));
2674 }
2675 appendEncoded(sb, c);
2676 } else {
2677 if (sb != null)
2678 sb.append(c);
2679 }
2680 }
2681 return (sb == null) ? s : sb.toString();
2682 }
2683
2684 // Encodes all characters >= \u0080 into escaped, normalized UTF-8 octets,
2685 // assuming that s is otherwise legal
2686 //
2687 private static String encode(String s) {
2688 int n = s.length();
2689 if (n == 0)
2690 return s;
2691
2692 // First check whether we actually need to encode
2693 for (int i = 0;;) {
2694 if (s.charAt(i) >= '\u0080')
2695 break;
2696 if (++i >= n)
2697 return s;
2698 }
2699
2700 String ns = Normalizer.normalize(s, Normalizer.Form.NFC);
2701 ByteBuffer bb = null;
2702 try {
2703 bb = ThreadLocalCoders.encoderFor("UTF-8")
2704 .encode(CharBuffer.wrap(ns));
2705 } catch (CharacterCodingException x) {
2706 assert false;
2707 }
2708
2709 StringBuffer sb = new StringBuffer();
2710 while (bb.hasRemaining()) {
2711 int b = bb.get() & 0xff;
2712 if (b >= 0x80)
2713 appendEscape(sb, (byte)b);
2714 else
2715 sb.append((char)b);
2716 }
2717 return sb.toString();
2718 }
2719
2720 private static int decode(char c) {
2721 if ((c >= '0') && (c <= '9'))
2722 return c - '0';
2723 if ((c >= 'a') && (c <= 'f'))
2724 return c - 'a' + 10;
2725 if ((c >= 'A') && (c <= 'F'))
2726 return c - 'A' + 10;
2727 assert false;
2728 return -1;
2729 }
2730
2731 private static byte decode(char c1, char c2) {
2732 return (byte)( ((decode(c1) & 0xf) << 4)
2733 | ((decode(c2) & 0xf) << 0));
2734 }
2735
2736 // Evaluates all escapes in s, applying UTF-8 decoding if needed. Assumes
2737 // that escapes are well-formed syntactically, i.e., of the form %XX. If a
2738 // sequence of escaped octets is not valid UTF-8 then the erroneous octets
2739 // are replaced with '\uFFFD'.
2740 // Exception: any "%" found between "[]" is left alone. It is an IPv6 literal
2741 // with a scope_id
2742 //
2743 private static String decode(String s) {
2744 if (s == null)
2745 return s;
2746 int n = s.length();
2747 if (n == 0)
2748 return s;
2749 if (s.indexOf('%') < 0)
2750 return s;
2751
2752 StringBuffer sb = new StringBuffer(n);
2753 ByteBuffer bb = ByteBuffer.allocate(n);
2754 CharBuffer cb = CharBuffer.allocate(n);
2755 CharsetDecoder dec = ThreadLocalCoders.decoderFor("UTF-8")
2756 .onMalformedInput(CodingErrorAction.REPLACE)
2757 .onUnmappableCharacter(CodingErrorAction.REPLACE);
2758
2759 // This is not horribly efficient, but it will do for now
2760 char c = s.charAt(0);
2761 boolean betweenBrackets = false;
2762
2763 for (int i = 0; i < n;) {
2764 assert c == s.charAt(i); // Loop invariant
2765 if (c == '[') {
2766 betweenBrackets = true;
2767 } else if (betweenBrackets && c == ']') {
2768 betweenBrackets = false;
2769 }
2770 if (c != '%' || betweenBrackets) {
2771 sb.append(c);
2772 if (++i >= n)
2773 break;
2774 c = s.charAt(i);
2775 continue;
2776 }
2777 bb.clear();
2778 int ui = i;
2779 for (;;) {
2780 assert (n - i >= 2);
2781 bb.put(decode(s.charAt(++i), s.charAt(++i)));
2782 if (++i >= n)
2783 break;
2784 c = s.charAt(i);
2785 if (c != '%')
2786 break;
2787 }
2788 bb.flip();
2789 cb.clear();
2790 dec.reset();
2791 CoderResult cr = dec.decode(bb, cb, true);
2792 assert cr.isUnderflow();
2793 cr = dec.flush(cb);
2794 assert cr.isUnderflow();
2795 sb.append(cb.flip().toString());
2796 }
2797
2798 return sb.toString();
2799 }
2800
2801
2802 // -- Parsing --
2803
2804 // For convenience we wrap the input URI string in a new instance of the
2805 // following internal class. This saves always having to pass the input
2806 // string as an argument to each internal scan/parse method.
2807
2808 private class Parser {
2809
2810 private String input; // URI input string
2811 private boolean requireServerAuthority = false;
2812
2813 Parser(String s) {
2814 input = s;
2815 string = s;
2816 }
2817
2818 // -- Methods for throwing URISyntaxException in various ways --
2819
2820 private void fail(String reason) throws URISyntaxException {
2821 throw new URISyntaxException(input, reason);
2822 }
2823
2824 private void fail(String reason, int p) throws URISyntaxException {
2825 throw new URISyntaxException(input, reason, p);
2826 }
2827
2828 private void failExpecting(String expected, int p)
2829 throws URISyntaxException
2830 {
2831 fail("Expected " + expected, p);
2832 }
2833
2834 private void failExpecting(String expected, String prior, int p)
2835 throws URISyntaxException
2836 {
2837 fail("Expected " + expected + " following " + prior, p);
2838 }
2839
2840
2841 // -- Simple access to the input string --
2842
2843 // Return a substring of the input string
2844 //
2845 private String substring(int start, int end) {
2846 return input.substring(start, end);
2847 }
2848
2849 // Return the char at position p,
2850 // assuming that p < input.length()
2851 //
2852 private char charAt(int p) {
2853 return input.charAt(p);
2854 }
2855
2856 // Tells whether start < end and, if so, whether charAt(start) == c
2857 //
2858 private boolean at(int start, int end, char c) {
2859 return (start < end) && (charAt(start) == c);
2860 }
2861
2862 // Tells whether start + s.length() < end and, if so,
2863 // whether the chars at the start position match s exactly
2864 //
2865 private boolean at(int start, int end, String s) {
2866 int p = start;
2867 int sn = s.length();
2868 if (sn > end - p)
2869 return false;
2870 int i = 0;
2871 while (i < sn) {
2872 if (charAt(p++) != s.charAt(i)) {
2873 break;
2874 }
2875 i++;
2876 }
2877 return (i == sn);
2878 }
2879
2880
2881 // -- Scanning --
2882
2883 // The various scan and parse methods that follow use a uniform
2884 // convention of taking the current start position and end index as
2885 // their first two arguments. The start is inclusive while the end is
2886 // exclusive, just as in the String class, i.e., a start/end pair
2887 // denotes the left-open interval [start, end) of the input string.
2888 //
2889 // These methods never proceed past the end position. They may return
2890 // -1 to indicate outright failure, but more often they simply return
2891 // the position of the first char after the last char scanned. Thus
2892 // a typical idiom is
2893 //
2894 // int p = start;
2895 // int q = scan(p, end, ...);
2896 // if (q > p)
2897 // // We scanned something
2898 // ...;
2899 // else if (q == p)
2900 // // We scanned nothing
2901 // ...;
2902 // else if (q == -1)
2903 // // Something went wrong
2904 // ...;
2905
2906
2907 // Scan a specific char: If the char at the given start position is
2908 // equal to c, return the index of the next char; otherwise, return the
2909 // start position.
2910 //
2911 private int scan(int start, int end, char c) {
2912 if ((start < end) && (charAt(start) == c))
2913 return start + 1;
2914 return start;
2915 }
2916
2917 // Scan forward from the given start position. Stop at the first char
2918 // in the err string (in which case -1 is returned), or the first char
2919 // in the stop string (in which case the index of the preceding char is
2920 // returned), or the end of the input string (in which case the length
2921 // of the input string is returned). May return the start position if
2922 // nothing matches.
2923 //
2924 private int scan(int start, int end, String err, String stop) {
2925 int p = start;
2926 while (p < end) {
2927 char c = charAt(p);
2928 if (err.indexOf(c) >= 0)
2929 return -1;
2930 if (stop.indexOf(c) >= 0)
2931 break;
2932 p++;
2933 }
2934 return p;
2935 }
2936
2937 // Scan a potential escape sequence, starting at the given position,
2938 // with the given first char (i.e., charAt(start) == c).
2939 //
2940 // This method assumes that if escapes are allowed then visible
2941 // non-US-ASCII chars are also allowed.
2942 //
2943 private int scanEscape(int start, int n, char first)
2944 throws URISyntaxException
2945 {
2946 int p = start;
2947 char c = first;
2948 if (c == '%') {
2949 // Process escape pair
2950 if ((p + 3 <= n)
2951 && match(charAt(p + 1), L_HEX, H_HEX)
2952 && match(charAt(p + 2), L_HEX, H_HEX)) {
2953 return p + 3;
2954 }
2955 fail("Malformed escape pair", p);
2956 } else if ((c > 128)
2957 && !Character.isSpaceChar(c)
2958 && !Character.isISOControl(c)) {
2959 // Allow unescaped but visible non-US-ASCII chars
2960 return p + 1;
2961 }
2962 return p;
2963 }
2964
2965 // Scan chars that match the given mask pair
2966 //
2967 private int scan(int start, int n, long lowMask, long highMask)
2968 throws URISyntaxException
2969 {
2970 int p = start;
2971 while (p < n) {
2972 char c = charAt(p);
2973 if (match(c, lowMask, highMask)) {
2974 p++;
2975 continue;
2976 }
2977 if ((lowMask & L_ESCAPED) != 0) {
2978 int q = scanEscape(p, n, c);
2979 if (q > p) {
2980 p = q;
2981 continue;
2982 }
2983 }
2984 break;
2985 }
2986 return p;
2987 }
2988
2989 // Check that each of the chars in [start, end) matches the given mask
2990 //
2991 private void checkChars(int start, int end,
2992 long lowMask, long highMask,
2993 String what)
2994 throws URISyntaxException
2995 {
2996 int p = scan(start, end, lowMask, highMask);
2997 if (p < end)
2998 fail("Illegal character in " + what, p);
2999 }
3000
3001 // Check that the char at position p matches the given mask
3002 //
3003 private void checkChar(int p,
3004 long lowMask, long highMask,
3005 String what)
3006 throws URISyntaxException
3007 {
3008 checkChars(p, p + 1, lowMask, highMask, what);
3009 }
3010
3011
3012 // -- Parsing --
3013
3014 // [<scheme>:]<scheme-specific-part>[#<fragment>]
3015 //
3016 void parse(boolean rsa) throws URISyntaxException {
3017 requireServerAuthority = rsa;
3018 int ssp; // Start of scheme-specific part
3019 int n = input.length();
3020 int p = scan(0, n, "/?#", ":");
3021 if ((p >= 0) && at(p, n, ':')) {
3022 if (p == 0)
3023 failExpecting("scheme name", 0);
3024 checkChar(0, L_ALPHA, H_ALPHA, "scheme name");
3025 checkChars(1, p, L_SCHEME, H_SCHEME, "scheme name");
3026 scheme = substring(0, p);
3027 p++; // Skip ':'
3028 ssp = p;
3029 if (at(p, n, '/')) {
3030 p = parseHierarchical(p, n);
3031 } else {
3032 int q = scan(p, n, "", "#");
3033 if (q <= p)
3034 failExpecting("scheme-specific part", p);
3035 checkChars(p, q, L_URIC, H_URIC, "opaque part");
3036 p = q;
3037 }
3038 } else {
3039 ssp = 0;
3040 p = parseHierarchical(0, n);
3041 }
3042 schemeSpecificPart = substring(ssp, p);
3043 if (at(p, n, '#')) {
3044 checkChars(p + 1, n, L_URIC, H_URIC, "fragment");
3045 fragment = substring(p + 1, n);
3046 p = n;
3047 }
3048 if (p < n)
3049 fail("end of URI", p);
3050 }
3051
3052 // [//authority]<path>[?<query>]
3053 //
3054 // DEVIATION from RFC2396: We allow an empty authority component as
3055 // long as it's followed by a non-empty path, query component, or
3056 // fragment component. This is so that URIs such as "file:///foo/bar"
3057 // will parse. This seems to be the intent of RFC2396, though the
3058 // grammar does not permit it. If the authority is empty then the
3059 // userInfo, host, and port components are undefined.
3060 //
3061 // DEVIATION from RFC2396: We allow empty relative paths. This seems
3062 // to be the intent of RFC2396, but the grammar does not permit it.
3063 // The primary consequence of this deviation is that "#f" parses as a
3064 // relative URI with an empty path.
3065 //
3066 private int parseHierarchical(int start, int n)
3067 throws URISyntaxException
3068 {
3069 int p = start;
3070 if (at(p, n, '/') && at(p + 1, n, '/')) {
3071 p += 2;
3072 int q = scan(p, n, "", "/?#");
3073 if (q > p) {
3074 p = parseAuthority(p, q);
3075 } else if (q < n) {
3076 // DEVIATION: Allow empty authority prior to non-empty
3077 // path, query component or fragment identifier
3078 } else
3079 failExpecting("authority", p);
3080 }
3081 int q = scan(p, n, "", "?#"); // DEVIATION: May be empty
3082 checkChars(p, q, L_PATH, H_PATH, "path");
3083 path = substring(p, q);
3084 p = q;
3085 if (at(p, n, '?')) {
3086 p++;
3087 q = scan(p, n, "", "#");
3088 checkChars(p, q, L_URIC, H_URIC, "query");
3089 query = substring(p, q);
3090 p = q;
3091 }
3092 return p;
3093 }
3094
3095 // authority = server | reg_name
3096 //
3097 // Ambiguity: An authority that is a registry name rather than a server
3098 // might have a prefix that parses as a server. We use the fact that
3099 // the authority component is always followed by '/' or the end of the
3100 // input string to resolve this: If the complete authority did not
3101 // parse as a server then we try to parse it as a registry name.
3102 //
3103 private int parseAuthority(int start, int n)
3104 throws URISyntaxException
3105 {
3106 int p = start;
3107 int q = p;
3108 URISyntaxException ex = null;
3109
3110 boolean serverChars;
3111 boolean regChars;
3112
3113 if (scan(p, n, "", "]") > p) {
3114 // contains a literal IPv6 address, therefore % is allowed
3115 serverChars = (scan(p, n, L_SERVER_PERCENT, H_SERVER_PERCENT) == n);
3116 } else {
3117 serverChars = (scan(p, n, L_SERVER, H_SERVER) == n);
3118 }
3119 regChars = (scan(p, n, L_REG_NAME, H_REG_NAME) == n);
3120
3121 if (regChars && !serverChars) {
3122 // Must be a registry-based authority
3123 authority = substring(p, n);
3124 return n;
3125 }
3126
3127 if (serverChars) {
3128 // Might be (probably is) a server-based authority, so attempt
3129 // to parse it as such. If the attempt fails, try to treat it
3130 // as a registry-based authority.
3131 try {
3132 q = parseServer(p, n);
3133 if (q < n)
3134 failExpecting("end of authority", q);
3135 authority = substring(p, n);
3136 } catch (URISyntaxException x) {
3137 // Undo results of failed parse
3138 userInfo = null;
3139 host = null;
3140 port = -1;
3141 if (requireServerAuthority) {
3142 // If we're insisting upon a server-based authority,
3143 // then just re-throw the exception
3144 throw x;
3145 } else {
3146 // Save the exception in case it doesn't parse as a
3147 // registry either
3148 ex = x;
3149 q = p;
3150 }
3151 }
3152 }
3153
3154 if (q < n) {
3155 if (regChars) {
3156 // Registry-based authority
3157 authority = substring(p, n);
3158 } else if (ex != null) {
3159 // Re-throw exception; it was probably due to
3160 // a malformed IPv6 address
3161 throw ex;
3162 } else {
3163 fail("Illegal character in authority", q);
3164 }
3165 }
3166
3167 return n;
3168 }
3169
3170
3171 // [<userinfo>@]<host>[:<port>]
3172 //
3173 private int parseServer(int start, int n)
3174 throws URISyntaxException
3175 {
3176 int p = start;
3177 int q;
3178
3179 // userinfo
3180 q = scan(p, n, "/?#", "@");
3181 if ((q >= p) && at(q, n, '@')) {
3182 checkChars(p, q, L_USERINFO, H_USERINFO, "user info");
3183 userInfo = substring(p, q);
3184 p = q + 1; // Skip '@'
3185 }
3186
3187 // hostname, IPv4 address, or IPv6 address
3188 if (at(p, n, '[')) {
3189 // DEVIATION from RFC2396: Support IPv6 addresses, per RFC2732
3190 p++;
3191 q = scan(p, n, "/?#", "]");
3192 if ((q > p) && at(q, n, ']')) {
3193 // look for a "%" scope id
3194 int r = scan (p, q, "", "%");
3195 if (r > p) {
3196 parseIPv6Reference(p, r);
3197 if (r+1 == q) {
3198 fail ("scope id expected");
3199 }
3200 checkChars (r+1, q, L_ALPHANUM, H_ALPHANUM,
3201 "scope id");
3202 } else {
3203 parseIPv6Reference(p, q);
3204 }
3205 host = substring(p-1, q+1);
3206 p = q + 1;
3207 } else {
3208 failExpecting("closing bracket for IPv6 address", q);
3209 }
3210 } else {
3211 q = parseIPv4Address(p, n);
3212 if (q <= p)
3213 q = parseHostname(p, n);
3214 p = q;
3215 }
3216
3217 // port
3218 if (at(p, n, ':')) {
3219 p++;
3220 q = scan(p, n, "", "/");
3221 if (q > p) {
3222 checkChars(p, q, L_DIGIT, H_DIGIT, "port number");
3223 try {
3224 port = Integer.parseInt(substring(p, q));
3225 } catch (NumberFormatException x) {
3226 fail("Malformed port number", p);
3227 }
3228 p = q;
3229 }
3230 }
3231 if (p < n)
3232 failExpecting("port number", p);
3233
3234 return p;
3235 }
3236
3237 // Scan a string of decimal digits whose value fits in a byte
3238 //
3239 private int scanByte(int start, int n)
3240 throws URISyntaxException
3241 {
3242 int p = start;
3243 int q = scan(p, n, L_DIGIT, H_DIGIT);
3244 if (q <= p) return q;
3245 if (Integer.parseInt(substring(p, q)) > 255) return p;
3246 return q;
3247 }
3248
3249 // Scan an IPv4 address.
3250 //
3251 // If the strict argument is true then we require that the given
3252 // interval contain nothing besides an IPv4 address; if it is false
3253 // then we only require that it start with an IPv4 address.
3254 //
3255 // If the interval does not contain or start with (depending upon the
3256 // strict argument) a legal IPv4 address characters then we return -1
3257 // immediately; otherwise we insist that these characters parse as a
3258 // legal IPv4 address and throw an exception on failure.
3259 //
3260 // We assume that any string of decimal digits and dots must be an IPv4
3261 // address. It won't parse as a hostname anyway, so making that
3262 // assumption here allows more meaningful exceptions to be thrown.
3263 //
3264 private int scanIPv4Address(int start, int n, boolean strict)
3265 throws URISyntaxException
3266 {
3267 int p = start;
3268 int q;
3269 int m = scan(p, n, L_DIGIT | L_DOT, H_DIGIT | H_DOT);
3270 if ((m <= p) || (strict && (m != n)))
3271 return -1;
3272 for (;;) {
3273 // Per RFC2732: At most three digits per byte
3274 // Further constraint: Each element fits in a byte
3275 if ((q = scanByte(p, m)) <= p) break; p = q;
3276 if ((q = scan(p, m, '.')) <= p) break; p = q;
3277 if ((q = scanByte(p, m)) <= p) break; p = q;
3278 if ((q = scan(p, m, '.')) <= p) break; p = q;
3279 if ((q = scanByte(p, m)) <= p) break; p = q;
3280 if ((q = scan(p, m, '.')) <= p) break; p = q;
3281 if ((q = scanByte(p, m)) <= p) break; p = q;
3282 if (q < m) break;
3283 return q;
3284 }
3285 fail("Malformed IPv4 address", q);
3286 return -1;
3287 }
3288
3289 // Take an IPv4 address: Throw an exception if the given interval
3290 // contains anything except an IPv4 address
3291 //
3292 private int takeIPv4Address(int start, int n, String expected)
3293 throws URISyntaxException
3294 {
3295 int p = scanIPv4Address(start, n, true);
3296 if (p <= start)
3297 failExpecting(expected, start);
3298 return p;
3299 }
3300
3301 // Attempt to parse an IPv4 address, returning -1 on failure but
3302 // allowing the given interval to contain [:<characters>] after
3303 // the IPv4 address.
3304 //
3305 private int parseIPv4Address(int start, int n) {
3306 int p;
3307
3308 try {
3309 p = scanIPv4Address(start, n, false);
3310 } catch (URISyntaxException x) {
3311 return -1;
3312 } catch (NumberFormatException nfe) {
3313 return -1;
3314 }
3315
3316 if (p > start && p < n) {
3317 // IPv4 address is followed by something - check that
3318 // it's a ":" as this is the only valid character to
3319 // follow an address.
3320 if (charAt(p) != ':') {
3321 p = -1;
3322 }
3323 }
3324
3325 if (p > start)
3326 host = substring(start, p);
3327
3328 return p;
3329 }
3330
3331 // hostname = domainlabel [ "." ] | 1*( domainlabel "." ) toplabel [ "." ]
3332 // domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum
3333 // toplabel = alpha | alpha *( alphanum | "-" ) alphanum
3334 //
3335 private int parseHostname(int start, int n)
3336 throws URISyntaxException
3337 {
3338 int p = start;
3339 int q;
3340 int l = -1; // Start of last parsed label
3341
3342 do {
3343 // domainlabel = alphanum [ *( alphanum | "-" ) alphanum ]
3344 q = scan(p, n, L_ALPHANUM, H_ALPHANUM);
3345 if (q <= p)
3346 break;
3347 l = p;
3348 if (q > p) {
3349 p = q;
3350 q = scan(p, n, L_ALPHANUM | L_DASH, H_ALPHANUM | H_DASH);
3351 if (q > p) {
3352 if (charAt(q - 1) == '-')
3353 fail("Illegal character in hostname", q - 1);
3354 p = q;
3355 }
3356 }
3357 q = scan(p, n, '.');
3358 if (q <= p)
3359 break;
3360 p = q;
3361 } while (p < n);
3362
3363 if ((p < n) && !at(p, n, ':'))
3364 fail("Illegal character in hostname", p);
3365
3366 if (l < 0)
3367 failExpecting("hostname", start);
3368
3369 // for a fully qualified hostname check that the rightmost
3370 // label starts with an alpha character.
3371 if (l > start && !match(charAt(l), L_ALPHA, H_ALPHA)) {
3372 fail("Illegal character in hostname", l);
3373 }
3374
3375 host = substring(start, p);
3376 return p;
3377 }
3378
3379
3380 // IPv6 address parsing, from RFC2373: IPv6 Addressing Architecture
3381 //
3382 // Bug: The grammar in RFC2373 Appendix B does not allow addresses of
3383 // the form ::12.34.56.78, which are clearly shown in the examples
3384 // earlier in the document. Here is the original grammar:
3385 //
3386 // IPv6address = hexpart [ ":" IPv4address ]
3387 // hexpart = hexseq | hexseq "::" [ hexseq ] | "::" [ hexseq ]
3388 // hexseq = hex4 *( ":" hex4)
3389 // hex4 = 1*4HEXDIG
3390 //
3391 // We therefore use the following revised grammar:
3392 //
3393 // IPv6address = hexseq [ ":" IPv4address ]
3394 // | hexseq [ "::" [ hexpost ] ]
3395 // | "::" [ hexpost ]
3396 // hexpost = hexseq | hexseq ":" IPv4address | IPv4address
3397 // hexseq = hex4 *( ":" hex4)
3398 // hex4 = 1*4HEXDIG
3399 //
3400 // This covers all and only the following cases:
3401 //
3402 // hexseq
3403 // hexseq : IPv4address
3404 // hexseq ::
3405 // hexseq :: hexseq
3406 // hexseq :: hexseq : IPv4address
3407 // hexseq :: IPv4address
3408 // :: hexseq
3409 // :: hexseq : IPv4address
3410 // :: IPv4address
3411 // ::
3412 //
3413 // Additionally we constrain the IPv6 address as follows :-
3414 //
3415 // i. IPv6 addresses without compressed zeros should contain
3416 // exactly 16 bytes.
3417 //
3418 // ii. IPv6 addresses with compressed zeros should contain
3419 // less than 16 bytes.
3420
3421 private int ipv6byteCount = 0;
3422
3423 private int parseIPv6Reference(int start, int n)
3424 throws URISyntaxException
3425 {
3426 int p = start;
3427 int q;
3428 boolean compressedZeros = false;
3429
3430 q = scanHexSeq(p, n);
3431
3432 if (q > p) {
3433 p = q;
3434 if (at(p, n, "::")) {
3435 compressedZeros = true;
3436 p = scanHexPost(p + 2, n);
3437 } else if (at(p, n, ':')) {
3438 p = takeIPv4Address(p + 1, n, "IPv4 address");
3439 ipv6byteCount += 4;
3440 }
3441 } else if (at(p, n, "::")) {
3442 compressedZeros = true;
3443 p = scanHexPost(p + 2, n);
3444 }
3445 if (p < n)
3446 fail("Malformed IPv6 address", start);
3447 if (ipv6byteCount > 16)
3448 fail("IPv6 address too long", start);
3449 if (!compressedZeros && ipv6byteCount < 16)
3450 fail("IPv6 address too short", start);
3451 if (compressedZeros && ipv6byteCount == 16)
3452 fail("Malformed IPv6 address", start);
3453
3454 return p;
3455 }
3456
3457 private int scanHexPost(int start, int n)
3458 throws URISyntaxException
3459 {
3460 int p = start;
3461 int q;
3462
3463 if (p == n)
3464 return p;
3465
3466 q = scanHexSeq(p, n);
3467 if (q > p) {
3468 p = q;
3469 if (at(p, n, ':')) {
3470 p++;
3471 p = takeIPv4Address(p, n, "hex digits or IPv4 address");
3472 ipv6byteCount += 4;
3473 }
3474 } else {
3475 p = takeIPv4Address(p, n, "hex digits or IPv4 address");
3476 ipv6byteCount += 4;
3477 }
3478 return p;
3479 }
3480
3481 // Scan a hex sequence; return -1 if one could not be scanned
3482 //
3483 private int scanHexSeq(int start, int n)
3484 throws URISyntaxException
3485 {
3486 int p = start;
3487 int q;
3488
3489 q = scan(p, n, L_HEX, H_HEX);
3490 if (q <= p)
3491 return -1;
3492 if (at(q, n, '.')) // Beginning of IPv4 address
3493 return -1;
3494 if (q > p + 4)
3495 fail("IPv6 hexadecimal digit sequence too long", p);
3496 ipv6byteCount += 2;
3497 p = q;
3498 while (p < n) {
3499 if (!at(p, n, ':'))
3500 break;
3501 if (at(p + 1, n, ':'))
3502 break; // "::"
3503 p++;
3504 q = scan(p, n, L_HEX, H_HEX);
3505 if (q <= p)
3506 failExpecting("digits for an IPv6 address", p);
3507 if (at(q, n, '.')) { // Beginning of IPv4 address
3508 p--;
3509 break;
3510 }
3511 if (q > p + 4)
3512 fail("IPv6 hexadecimal digit sequence too long", p);
3513 ipv6byteCount += 2;
3514 p = q;
3515 }
3516
3517 return p;
3518 }
3519
3520 }
3521
3522 }
Save This Page