1 /*
2 * Copyright (c) 2000, 2017, Oracle and/or its affiliates. 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. Oracle designates this
8 * particular file as subject to the "Classpath" exception as provided
9 * by Oracle 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 Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
22 * or visit www.oracle.com if you need additional information or have any
23 * 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.CoderResult;
37 import java.nio.charset.CodingErrorAction;
38 import java.nio.charset.CharacterCodingException;
39 import java.text.Normalizer;
40 import jdk.internal.loader.URLClassPath;
41 import jdk.internal.misc.JavaNetUriAccess;
42 import jdk.internal.misc.SharedSecrets;
43 import sun.nio.cs.ThreadLocalCoders;
44
45 import java.lang.Character; // for javadoc
46 import java.lang.NullPointerException; // for javadoc
47
48
49 /**
50 * Represents a Uniform Resource Identifier (URI) reference.
51 *
52 * <p> Aside from some minor deviations noted below, an instance of this
53 * class represents a URI reference as defined by
54 * <a href="http://www.ietf.org/rfc/rfc2396.txt"><i>RFC 2396: Uniform
55 * Resource Identifiers (URI): Generic Syntax</i></a>, amended by <a
56 * href="http://www.ietf.org/rfc/rfc2732.txt"><i>RFC 2732: Format for
57 * Literal IPv6 Addresses in URLs</i></a>. The Literal IPv6 address format
58 * also supports scope_ids. The syntax and usage of scope_ids is described
59 * <a href="Inet6Address.html#scoped">here</a>.
60 * This class provides constructors for creating URI instances from
61 * their components or by parsing their string forms, methods for accessing the
62 * various components of an instance, and methods for normalizing, resolving,
63 * and relativizing URI instances. Instances of this class are immutable.
64 *
65 *
66 * <h3> URI syntax and components </h3>
67 *
68 * At the highest level a URI reference (hereinafter simply "URI") in string
69 * form has the syntax
70 *
71 * <blockquote>
72 * [<i>scheme</i><b>{@code :}</b>]<i>scheme-specific-part</i>[<b>{@code #}</b><i>fragment</i>]
73 * </blockquote>
74 *
75 * where square brackets [...] delineate optional components and the characters
76 * <b>{@code :}</b> and <b>{@code #}</b> stand for themselves.
77 *
78 * <p> An <i>absolute</i> URI specifies a scheme; a URI that is not absolute is
79 * said to be <i>relative</i>. URIs are also classified according to whether
80 * they are <i>opaque</i> or <i>hierarchical</i>.
81 *
82 * <p> An <i>opaque</i> URI is an absolute URI whose scheme-specific part does
83 * not begin with a slash character ({@code '/'}). Opaque URIs are not
84 * subject to further parsing. Some examples of opaque URIs are:
85 *
86 * <blockquote><ul style="list-style-type:none">
87 * <li>{@code mailto:java-net@java.sun.com}</li>
88 * <li>{@code news:comp.lang.java}</li>
89 * <li>{@code urn:isbn:096139210x}</li>
90 * </ul></blockquote>
91 *
92 * <p> A <i>hierarchical</i> URI is either an absolute URI whose
93 * scheme-specific part begins with a slash character, or a relative URI, that
94 * is, a URI that does not specify a scheme. Some examples of hierarchical
95 * URIs are:
96 *
97 * <blockquote>
98 * {@code http://example.com/languages/java/}<br>
99 * {@code sample/a/index.html#28}<br>
100 * {@code ../../demo/b/index.html}<br>
101 * {@code file:///~/calendar}
102 * </blockquote>
103 *
104 * <p> A hierarchical URI is subject to further parsing according to the syntax
105 *
106 * <blockquote>
107 * [<i>scheme</i><b>{@code :}</b>][<b>{@code //}</b><i>authority</i>][<i>path</i>][<b>{@code ?}</b><i>query</i>][<b>{@code #}</b><i>fragment</i>]
108 * </blockquote>
109 *
110 * where the characters <b>{@code :}</b>, <b>{@code /}</b>,
111 * <b>{@code ?}</b>, and <b>{@code #}</b> stand for themselves. The
112 * scheme-specific part of a hierarchical URI consists of the characters
113 * between the scheme and fragment components.
114 *
115 * <p> The authority component of a hierarchical URI is, if specified, either
116 * <i>server-based</i> or <i>registry-based</i>. A server-based authority
117 * parses according to the familiar syntax
118 *
119 * <blockquote>
120 * [<i>user-info</i><b>{@code @}</b>]<i>host</i>[<b>{@code :}</b><i>port</i>]
121 * </blockquote>
122 *
123 * where the characters <b>{@code @}</b> and <b>{@code :}</b> stand for
124 * themselves. Nearly all URI schemes currently in use are server-based. An
125 * authority component that does not parse in this way is considered to be
126 * registry-based.
127 *
128 * <p> The path component of a hierarchical URI is itself said to be absolute
129 * if it begins with a slash character ({@code '/'}); otherwise it is
130 * relative. The path of a hierarchical URI that is either absolute or
131 * specifies an authority is always absolute.
132 *
133 * <p> All told, then, a URI instance has the following nine components:
134 *
135 * <table class="striped" style="margin-left:2em">
136 * <caption style="display:none">Describes the components of a URI:scheme,scheme-specific-part,authority,user-info,host,port,path,query,fragment</caption>
137 * <thead>
138 * <tr><th scope="col">Component</th><th scope="col">Type</th></tr>
139 * </thead>
140 * <tbody style="text-align:left">
141 * <tr><th scope="row">scheme</th><td>{@code String}</td></tr>
142 * <tr><th scope="row">scheme-specific-part</th><td>{@code String}</td></tr>
143 * <tr><th scope="row">authority</th><td>{@code String}</td></tr>
144 * <tr><th scope="row">user-info</th><td>{@code String}</td></tr>
145 * <tr><th scope="row">host</th><td>{@code String}</td></tr>
146 * <tr><th scope="row">port</th><td>{@code int}</td></tr>
147 * <tr><th scope="row">path</th><td>{@code String}</td></tr>
148 * <tr><th scope="row">query</th><td>{@code String}</td></tr>
149 * <tr><th scope="row">fragment</th><td>{@code String}</td></tr>
150 * </tbody>
151 * </table>
152 *
153 * In a given instance any particular component is either <i>undefined</i> or
154 * <i>defined</i> with a distinct value. Undefined string components are
155 * represented by {@code null}, while undefined integer components are
156 * represented by {@code -1}. A string component may be defined to have the
157 * empty string as its value; this is not equivalent to that component being
158 * undefined.
159 *
160 * <p> Whether a particular component is or is not defined in an instance
161 * depends upon the type of the URI being represented. An absolute URI has a
162 * scheme component. An opaque URI has a scheme, a scheme-specific part, and
163 * possibly a fragment, but has no other components. A hierarchical URI always
164 * has a path (though it may be empty) and a scheme-specific-part (which at
165 * least contains the path), and may have any of the other components. If the
166 * authority component is present and is server-based then the host component
167 * will be defined and the user-information and port components may be defined.
168 *
169 *
170 * <h4> Operations on URI instances </h4>
171 *
172 * The key operations supported by this class are those of
173 * <i>normalization</i>, <i>resolution</i>, and <i>relativization</i>.
174 *
175 * <p> <i>Normalization</i> is the process of removing unnecessary {@code "."}
176 * and {@code ".."} segments from the path component of a hierarchical URI.
177 * Each {@code "."} segment is simply removed. A {@code ".."} segment is
178 * removed only if it is preceded by a non-{@code ".."} segment.
179 * Normalization has no effect upon opaque URIs.
180 *
181 * <p> <i>Resolution</i> is the process of resolving one URI against another,
182 * <i>base</i> URI. The resulting URI is constructed from components of both
183 * URIs in the manner specified by RFC 2396, taking components from the
184 * base URI for those not specified in the original. For hierarchical URIs,
185 * the path of the original is resolved against the path of the base and then
186 * normalized. The result, for example, of resolving
187 *
188 * <blockquote>
189 * {@code sample/a/index.html#28}
190 *
191 * (1)
192 * </blockquote>
193 *
194 * against the base URI {@code http://example.com/languages/java/} is the result
195 * URI
196 *
197 * <blockquote>
198 * {@code http://example.com/languages/java/sample/a/index.html#28}
199 * </blockquote>
200 *
201 * Resolving the relative URI
202 *
203 * <blockquote>
204 * {@code ../../demo/b/index.html} (2)
205 * </blockquote>
206 *
207 * against this result yields, in turn,
208 *
209 * <blockquote>
210 * {@code http://example.com/languages/java/demo/b/index.html}
211 * </blockquote>
212 *
213 * Resolution of both absolute and relative URIs, and of both absolute and
214 * relative paths in the case of hierarchical URIs, is supported. Resolving
215 * the URI {@code file:///~calendar} against any other URI simply yields the
216 * original URI, since it is absolute. Resolving the relative URI (2) above
217 * against the relative base URI (1) yields the normalized, but still relative,
218 * URI
219 *
220 * <blockquote>
221 * {@code demo/b/index.html}
222 * </blockquote>
223 *
224 * <p> <i>Relativization</i>, finally, is the inverse of resolution: For any
225 * two normalized URIs <i>u</i> and <i>v</i>,
226 *
227 * <blockquote>
228 * <i>u</i>{@code .relativize(}<i>u</i>{@code .resolve(}<i>v</i>{@code )).equals(}<i>v</i>{@code )} and<br>
229 * <i>u</i>{@code .resolve(}<i>u</i>{@code .relativize(}<i>v</i>{@code )).equals(}<i>v</i>{@code )} .<br>
230 * </blockquote>
231 *
232 * This operation is often useful when constructing a document containing URIs
233 * that must be made relative to the base URI of the document wherever
234 * possible. For example, relativizing the URI
235 *
236 * <blockquote>
237 * {@code http://example.com/languages/java/sample/a/index.html#28}
238 * </blockquote>
239 *
240 * against the base URI
241 *
242 * <blockquote>
243 * {@code http://example.com/languages/java/}
244 * </blockquote>
245 *
246 * yields the relative URI {@code sample/a/index.html#28}.
247 *
248 *
249 * <h4> Character categories </h4>
250 *
251 * RFC 2396 specifies precisely which characters are permitted in the
252 * various components of a URI reference. The following categories, most of
253 * which are taken from that specification, are used below to describe these
254 * constraints:
255 *
256 * <table class="striped" style="margin-left:2em">
257 * <caption style="display:none">Describes categories alpha,digit,alphanum,unreserved,punct,reserved,escaped,and other</caption>
258 * <thead>
259 * <tr><th scope="col">Category</th><th scope="col">Description</th></tr>
260 * </thead>
261 * <tbody style="text-align:left">
262 * <tr><th scope="row" style="vertical-align:top">alpha</th>
263 * <td>The US-ASCII alphabetic characters,
264 * {@code 'A'} through {@code 'Z'}
265 * and {@code 'a'} through {@code 'z'}</td></tr>
266 * <tr><th scope="row" style="vertical-align:top">digit</th>
267 * <td>The US-ASCII decimal digit characters,
268 * {@code '0'} through {@code '9'}</td></tr>
269 * <tr><th scope="row" style="vertical-align:top">alphanum</th>
270 * <td>All <i>alpha</i> and <i>digit</i> characters</td></tr>
271 * <tr><th scope="row" style="vertical-align:top">unreserved</th>
272 * <td>All <i>alphanum</i> characters together with those in the string
273 * {@code "_-!.~'()*"}</td></tr>
274 * <tr><th scope="row" style="vertical-align:top">punct</th>
275 * <td>The characters in the string {@code ",;:$&+="}</td></tr>
276 * <tr><th scope="row" style="vertical-align:top">reserved</th>
277 * <td>All <i>punct</i> characters together with those in the string
278 * {@code "?/[]@"}</td></tr>
279 * <tr><th scope="row" style="vertical-align:top">escaped</th>
280 * <td>Escaped octets, that is, triplets consisting of the percent
281 * character ({@code '%'}) followed by two hexadecimal digits
282 * ({@code '0'}-{@code '9'}, {@code 'A'}-{@code 'F'}, and
283 * {@code 'a'}-{@code 'f'})</td></tr>
284 * <tr><th scope="row" style="vertical-align:top">other</th>
285 * <td>The Unicode characters that are not in the US-ASCII character set,
286 * are not control characters (according to the {@link
287 * java.lang.Character#isISOControl(char) Character.isISOControl}
288 * method), and are not space characters (according to the {@link
289 * java.lang.Character#isSpaceChar(char) Character.isSpaceChar}
290 * method) <i>(<b>Deviation from RFC 2396</b>, which is
291 * limited to US-ASCII)</i></td></tr>
292 * </tbody>
293 * </table>
294 *
295 * <p><a id="legal-chars"></a> The set of all legal URI characters consists of
296 * the <i>unreserved</i>, <i>reserved</i>, <i>escaped</i>, and <i>other</i>
297 * characters.
298 *
299 *
300 * <h4> Escaped octets, quotation, encoding, and decoding </h4>
301 *
302 * RFC 2396 allows escaped octets to appear in the user-info, path, query, and
303 * fragment components. Escaping serves two purposes in URIs:
304 *
305 * <ul>
306 *
307 * <li><p> To <i>encode</i> non-US-ASCII characters when a URI is required to
308 * conform strictly to RFC 2396 by not containing any <i>other</i>
309 * characters. </p></li>
310 *
311 * <li><p> To <i>quote</i> characters that are otherwise illegal in a
312 * component. The user-info, path, query, and fragment components differ
313 * slightly in terms of which characters are considered legal and illegal.
314 * </p></li>
315 *
316 * </ul>
317 *
318 * These purposes are served in this class by three related operations:
319 *
320 * <ul>
321 *
322 * <li><p><a id="encode"></a> A character is <i>encoded</i> by replacing it
323 * with the sequence of escaped octets that represent that character in the
324 * UTF-8 character set. The Euro currency symbol ({@code '\u005Cu20AC'}),
325 * for example, is encoded as {@code "%E2%82%AC"}. <i>(<b>Deviation from
326 * RFC 2396</b>, which does not specify any particular character
327 * set.)</i> </p></li>
328 *
329 * <li><p><a id="quote"></a> An illegal character is <i>quoted</i> simply by
330 * encoding it. The space character, for example, is quoted by replacing it
331 * with {@code "%20"}. UTF-8 contains US-ASCII, hence for US-ASCII
332 * characters this transformation has exactly the effect required by
333 * RFC 2396. </p></li>
334 *
335 * <li><p><a id="decode"></a>
336 * A sequence of escaped octets is <i>decoded</i> by
337 * replacing it with the sequence of characters that it represents in the
338 * UTF-8 character set. UTF-8 contains US-ASCII, hence decoding has the
339 * effect of de-quoting any quoted US-ASCII characters as well as that of
340 * decoding any encoded non-US-ASCII characters. If a <a
341 * href="../nio/charset/CharsetDecoder.html#ce">decoding error</a> occurs
342 * when decoding the escaped octets then the erroneous octets are replaced by
343 * {@code '\u005CuFFFD'}, the Unicode replacement character. </p></li>
344 *
345 * </ul>
346 *
347 * These operations are exposed in the constructors and methods of this class
348 * as follows:
349 *
350 * <ul>
351 *
352 * <li><p> The {@linkplain #URI(java.lang.String) single-argument
353 * constructor} requires any illegal characters in its argument to be
354 * quoted and preserves any escaped octets and <i>other</i> characters that
355 * are present. </p></li>
356 *
357 * <li><p> The {@linkplain
358 * #URI(java.lang.String,java.lang.String,java.lang.String,int,java.lang.String,java.lang.String,java.lang.String)
359 * multi-argument constructors} quote illegal characters as
360 * required by the components in which they appear. The percent character
361 * ({@code '%'}) is always quoted by these constructors. Any <i>other</i>
362 * characters are preserved. </p></li>
363 *
364 * <li><p> The {@link #getRawUserInfo() getRawUserInfo}, {@link #getRawPath()
365 * getRawPath}, {@link #getRawQuery() getRawQuery}, {@link #getRawFragment()
366 * getRawFragment}, {@link #getRawAuthority() getRawAuthority}, and {@link
367 * #getRawSchemeSpecificPart() getRawSchemeSpecificPart} methods return the
368 * values of their corresponding components in raw form, without interpreting
369 * any escaped octets. The strings returned by these methods may contain
370 * both escaped octets and <i>other</i> characters, and will not contain any
371 * illegal characters. </p></li>
372 *
373 * <li><p> The {@link #getUserInfo() getUserInfo}, {@link #getPath()
374 * getPath}, {@link #getQuery() getQuery}, {@link #getFragment()
375 * getFragment}, {@link #getAuthority() getAuthority}, and {@link
376 * #getSchemeSpecificPart() getSchemeSpecificPart} methods decode any escaped
377 * octets in their corresponding components. The strings returned by these
378 * methods may contain both <i>other</i> characters and illegal characters,
379 * and will not contain any escaped octets. </p></li>
380 *
381 * <li><p> The {@link #toString() toString} method returns a URI string with
382 * all necessary quotation but which may contain <i>other</i> characters.
383 * </p></li>
384 *
385 * <li><p> The {@link #toASCIIString() toASCIIString} method returns a fully
386 * quoted and encoded URI string that does not contain any <i>other</i>
387 * characters. </p></li>
388 *
389 * </ul>
390 *
391 *
392 * <h4> Identities </h4>
393 *
394 * For any URI <i>u</i>, it is always the case that
395 *
396 * <blockquote>
397 * {@code new URI(}<i>u</i>{@code .toString()).equals(}<i>u</i>{@code )} .
398 * </blockquote>
399 *
400 * For any URI <i>u</i> that does not contain redundant syntax such as two
401 * slashes before an empty authority (as in {@code file:///tmp/} ) or a
402 * colon following a host name but no port (as in
403 * {@code http://java.sun.com:} ), and that does not encode characters
404 * except those that must be quoted, the following identities also hold:
405 * <pre>
406 * new URI(<i>u</i>.getScheme(),
407 * <i>u</i>.getSchemeSpecificPart(),
408 * <i>u</i>.getFragment())
409 * .equals(<i>u</i>)</pre>
410 * in all cases,
411 * <pre>
412 * new URI(<i>u</i>.getScheme(),
413 * <i>u</i>.getAuthority(),
414 * <i>u</i>.getPath(), <i>u</i>.getQuery(),
415 * <i>u</i>.getFragment())
416 * .equals(<i>u</i>)</pre>
417 * if <i>u</i> is hierarchical, and
418 * <pre>
419 * new URI(<i>u</i>.getScheme(),
420 * <i>u</i>.getUserInfo(), <i>u</i>.getHost(), <i>u</i>.getPort(),
421 * <i>u</i>.getPath(), <i>u</i>.getQuery(),
422 * <i>u</i>.getFragment())
423 * .equals(<i>u</i>)</pre>
424 * if <i>u</i> is hierarchical and has either no authority or a server-based
425 * authority.
426 *
427 *
428 * <h4> URIs, URLs, and URNs </h4>
429 *
430 * A URI is a uniform resource <i>identifier</i> while a URL is a uniform
431 * resource <i>locator</i>. Hence every URL is a URI, abstractly speaking, but
432 * not every URI is a URL. This is because there is another subcategory of
433 * URIs, uniform resource <i>names</i> (URNs), which name resources but do not
434 * specify how to locate them. The {@code mailto}, {@code news}, and
435 * {@code isbn} URIs shown above are examples of URNs.
436 *
437 * <p> The conceptual distinction between URIs and URLs is reflected in the
438 * differences between this class and the {@link URL} class.
439 *
440 * <p> An instance of this class represents a URI reference in the syntactic
441 * sense defined by RFC 2396. A URI may be either absolute or relative.
442 * A URI string is parsed according to the generic syntax without regard to the
443 * scheme, if any, that it specifies. No lookup of the host, if any, is
444 * performed, and no scheme-dependent stream handler is constructed. Equality,
445 * hashing, and comparison are defined strictly in terms of the character
446 * content of the instance. In other words, a URI instance is little more than
447 * a structured string that supports the syntactic, scheme-independent
448 * operations of comparison, normalization, resolution, and relativization.
449 *
450 * <p> An instance of the {@link URL} class, by contrast, represents the
451 * syntactic components of a URL together with some of the information required
452 * to access the resource that it describes. A URL must be absolute, that is,
453 * it must always specify a scheme. A URL string is parsed according to its
454 * scheme. A stream handler is always established for a URL, and in fact it is
455 * impossible to create a URL instance for a scheme for which no handler is
456 * available. Equality and hashing depend upon both the scheme and the
457 * Internet address of the host, if any; comparison is not defined. In other
458 * words, a URL is a structured string that supports the syntactic operation of
459 * resolution as well as the network I/O operations of looking up the host and
460 * opening a connection to the specified resource.
461 *
462 *
463 * @author Mark Reinhold
464 * @since 1.4
465 *
466 * @see <a href="http://www.ietf.org/rfc/rfc2279.txt"><i>RFC 2279: UTF-8, a
467 * transformation format of ISO 10646</i></a>, <br><a
468 * href="http://www.ietf.org/rfc/rfc2373.txt"><i>RFC 2373: IPv6 Addressing
469 * Architecture</i></a>, <br><a
470 * href="http://www.ietf.org/rfc/rfc2396.txt"><i>RFC 2396: Uniform
471 * Resource Identifiers (URI): Generic Syntax</i></a>, <br><a
472 * href="http://www.ietf.org/rfc/rfc2732.txt"><i>RFC 2732: Format for
473 * Literal IPv6 Addresses in URLs</i></a>, <br><a
474 * href="URISyntaxException.html">URISyntaxException</a>
475 */
476
477 public final class URI
478 implements Comparable<URI>, Serializable
479 {
480
481 // Note: Comments containing the word "ASSERT" indicate places where a
482 // throw of an InternalError should be replaced by an appropriate assertion
483 // statement once asserts are enabled in the build.
484
485 static final long serialVersionUID = -6052424284110960213L;
486
487
488 // -- Properties and components of this instance --
489
490 // Components of all URIs: [<scheme>:]<scheme-specific-part>[#<fragment>]
491 private transient String scheme; // null ==> relative URI
492 private transient String fragment;
493
494 // Hierarchical URI components: [//<authority>]<path>[?<query>]
495 private transient String authority; // Registry or server
496
497 // Server-based authority: [<userInfo>@]<host>[:<port>]
498 private transient String userInfo;
499 private transient String host; // null ==> registry-based
500 private transient int port = -1; // -1 ==> undefined
501
502 // Remaining components of hierarchical URIs
503 private transient String path; // null ==> opaque
504 private transient String query;
505
506 // The remaining fields may be computed on demand, which is safe even in
507 // the face of multiple threads racing to initialize them
508 private transient String schemeSpecificPart;
509 private transient int hash; // Zero ==> undefined
510
511 private transient String decodedUserInfo;
512 private transient String decodedAuthority;
513 private transient String decodedPath;
514 private transient String decodedQuery;
515 private transient String decodedFragment;
516 private transient String decodedSchemeSpecificPart;
517
518 /**
519 * The string form of this URI.
520 *
521 * @serial
522 */
523 private volatile String string; // The only serializable field
524
525
526
527 // -- Constructors and factories --
528
529 private URI() { } // Used internally
530
531 /**
532 * Constructs a URI by parsing the given string.
533 *
534 * <p> This constructor parses the given string exactly as specified by the
535 * grammar in <a
536 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
537 * Appendix A, <b><i>except for the following deviations:</i></b> </p>
538 *
539 * <ul>
540 *
541 * <li><p> An empty authority component is permitted as long as it is
542 * followed by a non-empty path, a query component, or a fragment
543 * component. This allows the parsing of URIs such as
544 * {@code "file:///foo/bar"}, which seems to be the intent of
545 * RFC 2396 although the grammar does not permit it. If the
546 * authority component is empty then the user-information, host, and port
547 * components are undefined. </p></li>
548 *
549 * <li><p> Empty relative paths are permitted; this seems to be the
550 * intent of RFC 2396 although the grammar does not permit it. The
551 * primary consequence of this deviation is that a standalone fragment
552 * such as {@code "#foo"} parses as a relative URI with an empty path
553 * and the given fragment, and can be usefully <a
554 * href="#resolve-frag">resolved</a> against a base URI.
555 *
556 * <li><p> IPv4 addresses in host components are parsed rigorously, as
557 * specified by <a
558 * href="http://www.ietf.org/rfc/rfc2732.txt">RFC 2732</a>: Each
559 * element of a dotted-quad address must contain no more than three
560 * decimal digits. Each element is further constrained to have a value
561 * no greater than 255. </p></li>
562 *
563 * <li> <p> Hostnames in host components that comprise only a single
564 * domain label are permitted to start with an <i>alphanum</i>
565 * character. This seems to be the intent of <a
566 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>
567 * section 3.2.2 although the grammar does not permit it. The
568 * consequence of this deviation is that the authority component of a
569 * hierarchical URI such as {@code s://123}, will parse as a server-based
570 * authority. </p></li>
571 *
572 * <li><p> IPv6 addresses are permitted for the host component. An IPv6
573 * address must be enclosed in square brackets ({@code '['} and
574 * {@code ']'}) as specified by <a
575 * href="http://www.ietf.org/rfc/rfc2732.txt">RFC 2732</a>. The
576 * IPv6 address itself must parse according to <a
577 * href="http://www.ietf.org/rfc/rfc2373.txt">RFC 2373</a>. IPv6
578 * addresses are further constrained to describe no more than sixteen
579 * bytes of address information, a constraint implicit in RFC 2373
580 * but not expressible in the grammar. </p></li>
581 *
582 * <li><p> Characters in the <i>other</i> category are permitted wherever
583 * RFC 2396 permits <i>escaped</i> octets, that is, in the
584 * user-information, path, query, and fragment components, as well as in
585 * the authority component if the authority is registry-based. This
586 * allows URIs to contain Unicode characters beyond those in the US-ASCII
587 * character set. </p></li>
588 *
589 * </ul>
590 *
591 * @param str The string to be parsed into a URI
592 *
593 * @throws NullPointerException
594 * If {@code str} is {@code null}
595 *
596 * @throws URISyntaxException
597 * If the given string violates RFC 2396, as augmented
598 * by the above deviations
599 */
600 public URI(String str) throws URISyntaxException {
601 new Parser(str).parse(false);
602 }
603
604 /**
605 * Constructs a hierarchical URI from the given components.
606 *
607 * <p> If a scheme is given then the path, if also given, must either be
608 * empty or begin with a slash character ({@code '/'}). Otherwise a
609 * component of the new URI may be left undefined by passing {@code null}
610 * for the corresponding parameter or, in the case of the {@code port}
611 * parameter, by passing {@code -1}.
612 *
613 * <p> This constructor first builds a URI string from the given components
614 * according to the rules specified in <a
615 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
616 * section 5.2, step 7: </p>
617 *
618 * <ol>
619 *
620 * <li><p> Initially, the result string is empty. </p></li>
621 *
622 * <li><p> If a scheme is given then it is appended to the result,
623 * followed by a colon character ({@code ':'}). </p></li>
624 *
625 * <li><p> If user information, a host, or a port are given then the
626 * string {@code "//"} is appended. </p></li>
627 *
628 * <li><p> If user information is given then it is appended, followed by
629 * a commercial-at character ({@code '@'}). Any character not in the
630 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
631 * categories is <a href="#quote">quoted</a>. </p></li>
632 *
633 * <li><p> If a host is given then it is appended. If the host is a
634 * literal IPv6 address but is not enclosed in square brackets
635 * ({@code '['} and {@code ']'}) then the square brackets are added.
636 * </p></li>
637 *
638 * <li><p> If a port number is given then a colon character
639 * ({@code ':'}) is appended, followed by the port number in decimal.
640 * </p></li>
641 *
642 * <li><p> If a path is given then it is appended. Any character not in
643 * the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
644 * categories, and not equal to the slash character ({@code '/'}) or the
645 * commercial-at character ({@code '@'}), is quoted. </p></li>
646 *
647 * <li><p> If a query is given then a question-mark character
648 * ({@code '?'}) is appended, followed by the query. Any character that
649 * is not a <a href="#legal-chars">legal URI character</a> is quoted.
650 * </p></li>
651 *
652 * <li><p> Finally, if a fragment is given then a hash character
653 * ({@code '#'}) is appended, followed by the fragment. Any character
654 * that is not a legal URI character is quoted. </p></li>
655 *
656 * </ol>
657 *
658 * <p> The resulting URI string is then parsed as if by invoking the {@link
659 * #URI(String)} constructor and then invoking the {@link
660 * #parseServerAuthority()} method upon the result; this may cause a {@link
661 * URISyntaxException} to be thrown. </p>
662 *
663 * @param scheme Scheme name
664 * @param userInfo User name and authorization information
665 * @param host Host name
666 * @param port Port number
667 * @param path Path
668 * @param query Query
669 * @param fragment Fragment
670 *
671 * @throws URISyntaxException
672 * If both a scheme and a path are given but the path is relative,
673 * if the URI string constructed from the given components violates
674 * RFC 2396, or if the authority component of the string is
675 * present but cannot be parsed as a server-based authority
676 */
677 public URI(String scheme,
678 String userInfo, String host, int port,
679 String path, String query, String fragment)
680 throws URISyntaxException
681 {
682 String s = toString(scheme, null,
683 null, userInfo, host, port,
684 path, query, fragment);
685 checkPath(s, scheme, path);
686 new Parser(s).parse(true);
687 }
688
689 /**
690 * Constructs a hierarchical URI from the given components.
691 *
692 * <p> If a scheme is given then the path, if also given, must either be
693 * empty or begin with a slash character ({@code '/'}). Otherwise a
694 * component of the new URI may be left undefined by passing {@code null}
695 * for the corresponding parameter.
696 *
697 * <p> This constructor first builds a URI string from the given components
698 * according to the rules specified in <a
699 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
700 * section 5.2, step 7: </p>
701 *
702 * <ol>
703 *
704 * <li><p> Initially, the result string is empty. </p></li>
705 *
706 * <li><p> If a scheme is given then it is appended to the result,
707 * followed by a colon character ({@code ':'}). </p></li>
708 *
709 * <li><p> If an authority is given then the string {@code "//"} is
710 * appended, followed by the authority. If the authority contains a
711 * literal IPv6 address then the address must be enclosed in square
712 * brackets ({@code '['} and {@code ']'}). Any character not in the
713 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
714 * categories, and not equal to the commercial-at character
715 * ({@code '@'}), is <a href="#quote">quoted</a>. </p></li>
716 *
717 * <li><p> If a path is given then it is appended. Any character not in
718 * the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
719 * categories, and not equal to the slash character ({@code '/'}) or the
720 * commercial-at character ({@code '@'}), is quoted. </p></li>
721 *
722 * <li><p> If a query is given then a question-mark character
723 * ({@code '?'}) is appended, followed by the query. Any character that
724 * is not a <a href="#legal-chars">legal URI character</a> is quoted.
725 * </p></li>
726 *
727 * <li><p> Finally, if a fragment is given then a hash character
728 * ({@code '#'}) is appended, followed by the fragment. Any character
729 * that is not a legal URI character is quoted. </p></li>
730 *
731 * </ol>
732 *
733 * <p> The resulting URI string is then parsed as if by invoking the {@link
734 * #URI(String)} constructor and then invoking the {@link
735 * #parseServerAuthority()} method upon the result; this may cause a {@link
736 * URISyntaxException} to be thrown. </p>
737 *
738 * @param scheme Scheme name
739 * @param authority Authority
740 * @param path Path
741 * @param query Query
742 * @param fragment Fragment
743 *
744 * @throws URISyntaxException
745 * If both a scheme and a path are given but the path is relative,
746 * if the URI string constructed from the given components violates
747 * RFC 2396, or if the authority component of the string is
748 * present but cannot be parsed as a server-based authority
749 */
750 public URI(String scheme,
751 String authority,
752 String path, String query, String fragment)
753 throws URISyntaxException
754 {
755 String s = toString(scheme, null,
756 authority, null, null, -1,
757 path, query, fragment);
758 checkPath(s, scheme, path);
759 new Parser(s).parse(false);
760 }
761
762 /**
763 * Constructs a hierarchical URI from the given components.
764 *
765 * <p> A component may be left undefined by passing {@code null}.
766 *
767 * <p> This convenience constructor works as if by invoking the
768 * seven-argument constructor as follows:
769 *
770 * <blockquote>
771 * {@code new} {@link #URI(String, String, String, int, String, String, String)
772 * URI}{@code (scheme, null, host, -1, path, null, fragment);}
773 * </blockquote>
774 *
775 * @param scheme Scheme name
776 * @param host Host name
777 * @param path Path
778 * @param fragment Fragment
779 *
780 * @throws URISyntaxException
781 * If the URI string constructed from the given components
782 * violates RFC 2396
783 */
784 public URI(String scheme, String host, String path, String fragment)
785 throws URISyntaxException
786 {
787 this(scheme, null, host, -1, path, null, fragment);
788 }
789
790 /**
791 * Constructs a URI from the given components.
792 *
793 * <p> A component may be left undefined by passing {@code null}.
794 *
795 * <p> This constructor first builds a URI in string form using the given
796 * components as follows: </p>
797 *
798 * <ol>
799 *
800 * <li><p> Initially, the result string is empty. </p></li>
801 *
802 * <li><p> If a scheme is given then it is appended to the result,
803 * followed by a colon character ({@code ':'}). </p></li>
804 *
805 * <li><p> If a scheme-specific part is given then it is appended. Any
806 * character that is not a <a href="#legal-chars">legal URI character</a>
807 * is <a href="#quote">quoted</a>. </p></li>
808 *
809 * <li><p> Finally, if a fragment is given then a hash character
810 * ({@code '#'}) is appended to the string, followed by the fragment.
811 * Any character that is not a legal URI character is quoted. </p></li>
812 *
813 * </ol>
814 *
815 * <p> The resulting URI string is then parsed in order to create the new
816 * URI instance as if by invoking the {@link #URI(String)} constructor;
817 * this may cause a {@link URISyntaxException} to be thrown. </p>
818 *
819 * @param scheme Scheme name
820 * @param ssp Scheme-specific part
821 * @param fragment Fragment
822 *
823 * @throws URISyntaxException
824 * If the URI string constructed from the given components
825 * violates RFC 2396
826 */
827 public URI(String scheme, String ssp, String fragment)
828 throws URISyntaxException
829 {
830 new Parser(toString(scheme, ssp,
831 null, null, null, -1,
832 null, null, fragment))
833 .parse(false);
834 }
835
836 /**
837 * Constructs a simple URI consisting of only a scheme and a pre-validated
838 * path. Provides a fast-path for some internal cases.
839 */
840 URI(String scheme, String path) {
841 assert validSchemeAndPath(scheme, path);
842 this.scheme = scheme;
843 this.path = path;
844 }
845
846 private static boolean validSchemeAndPath(String scheme, String path) {
847 try {
848 URI u = new URI(scheme + ":" + path);
849 return scheme.equals(u.scheme) && path.equals(u.path);
850 } catch (URISyntaxException e) {
851 return false;
852 }
853 }
854
855 /**
856 * Creates a URI by parsing the given string.
857 *
858 * <p> This convenience factory method works as if by invoking the {@link
859 * #URI(String)} constructor; any {@link URISyntaxException} thrown by the
860 * constructor is caught and wrapped in a new {@link
861 * IllegalArgumentException} object, which is then thrown.
862 *
863 * <p> This method is provided for use in situations where it is known that
864 * the given string is a legal URI, for example for URI constants declared
865 * within in a program, and so it would be considered a programming error
866 * for the string not to parse as such. The constructors, which throw
867 * {@link URISyntaxException} directly, should be used situations where a
868 * URI is being constructed from user input or from some other source that
869 * may be prone to errors. </p>
870 *
871 * @param str The string to be parsed into a URI
872 * @return The new URI
873 *
874 * @throws NullPointerException
875 * If {@code str} is {@code null}
876 *
877 * @throws IllegalArgumentException
878 * If the given string violates RFC 2396
879 */
880 public static URI create(String str) {
881 try {
882 return new URI(str);
883 } catch (URISyntaxException x) {
884 throw new IllegalArgumentException(x.getMessage(), x);
885 }
886 }
887
888
889 // -- Operations --
890
891 /**
892 * Attempts to parse this URI's authority component, if defined, into
893 * user-information, host, and port components.
894 *
895 * <p> If this URI's authority component has already been recognized as
896 * being server-based then it will already have been parsed into
897 * user-information, host, and port components. In this case, or if this
898 * URI has no authority component, this method simply returns this URI.
899 *
900 * <p> Otherwise this method attempts once more to parse the authority
901 * component into user-information, host, and port components, and throws
902 * an exception describing why the authority component could not be parsed
903 * in that way.
904 *
905 * <p> This method is provided because the generic URI syntax specified in
906 * <a href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>
907 * cannot always distinguish a malformed server-based authority from a
908 * legitimate registry-based authority. It must therefore treat some
909 * instances of the former as instances of the latter. The authority
910 * component in the URI string {@code "//foo:bar"}, for example, is not a
911 * legal server-based authority but it is legal as a registry-based
912 * authority.
913 *
914 * <p> In many common situations, for example when working URIs that are
915 * known to be either URNs or URLs, the hierarchical URIs being used will
916 * always be server-based. They therefore must either be parsed as such or
917 * treated as an error. In these cases a statement such as
918 *
919 * <blockquote>
920 * {@code URI }<i>u</i>{@code = new URI(str).parseServerAuthority();}
921 * </blockquote>
922 *
923 * <p> can be used to ensure that <i>u</i> always refers to a URI that, if
924 * it has an authority component, has a server-based authority with proper
925 * user-information, host, and port components. Invoking this method also
926 * ensures that if the authority could not be parsed in that way then an
927 * appropriate diagnostic message can be issued based upon the exception
928 * that is thrown. </p>
929 *
930 * @return A URI whose authority field has been parsed
931 * as a server-based authority
932 *
933 * @throws URISyntaxException
934 * If the authority component of this URI is defined
935 * but cannot be parsed as a server-based authority
936 * according to RFC 2396
937 */
938 public URI parseServerAuthority()
939 throws URISyntaxException
940 {
941 // We could be clever and cache the error message and index from the
942 // exception thrown during the original parse, but that would require
943 // either more fields or a more-obscure representation.
944 if ((host != null) || (authority == null))
945 return this;
946 new Parser(toString()).parse(true);
947 return this;
948 }
949
950 /**
951 * Normalizes this URI's path.
952 *
953 * <p> If this URI is opaque, or if its path is already in normal form,
954 * then this URI is returned. Otherwise a new URI is constructed that is
955 * identical to this URI except that its path is computed by normalizing
956 * this URI's path in a manner consistent with <a
957 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
958 * section 5.2, step 6, sub-steps c through f; that is:
959 * </p>
960 *
961 * <ol>
962 *
963 * <li><p> All {@code "."} segments are removed. </p></li>
964 *
965 * <li><p> If a {@code ".."} segment is preceded by a non-{@code ".."}
966 * segment then both of these segments are removed. This step is
967 * repeated until it is no longer applicable. </p></li>
968 *
969 * <li><p> If the path is relative, and if its first segment contains a
970 * colon character ({@code ':'}), then a {@code "."} segment is
971 * prepended. This prevents a relative URI with a path such as
972 * {@code "a:b/c/d"} from later being re-parsed as an opaque URI with a
973 * scheme of {@code "a"} and a scheme-specific part of {@code "b/c/d"}.
974 * <b><i>(Deviation from RFC 2396)</i></b> </p></li>
975 *
976 * </ol>
977 *
978 * <p> A normalized path will begin with one or more {@code ".."} segments
979 * if there were insufficient non-{@code ".."} segments preceding them to
980 * allow their removal. A normalized path will begin with a {@code "."}
981 * segment if one was inserted by step 3 above. Otherwise, a normalized
982 * path will not contain any {@code "."} or {@code ".."} segments. </p>
983 *
984 * @return A URI equivalent to this URI,
985 * but whose path is in normal form
986 */
987 public URI normalize() {
988 return normalize(this);
989 }
990
991 /**
992 * Resolves the given URI against this URI.
993 *
994 * <p> If the given URI is already absolute, or if this URI is opaque, then
995 * the given URI is returned.
996 *
997 * <p><a id="resolve-frag"></a> If the given URI's fragment component is
998 * defined, its path component is empty, and its scheme, authority, and
999 * query components are undefined, then a URI with the given fragment but
1000 * with all other components equal to those of this URI is returned. This
1001 * allows a URI representing a standalone fragment reference, such as
1002 * {@code "#foo"}, to be usefully resolved against a base URI.
1003 *
1004 * <p> Otherwise this method constructs a new hierarchical URI in a manner
1005 * consistent with <a
1006 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
1007 * section 5.2; that is: </p>
1008 *
1009 * <ol>
1010 *
1011 * <li><p> A new URI is constructed with this URI's scheme and the given
1012 * URI's query and fragment components. </p></li>
1013 *
1014 * <li><p> If the given URI has an authority component then the new URI's
1015 * authority and path are taken from the given URI. </p></li>
1016 *
1017 * <li><p> Otherwise the new URI's authority component is copied from
1018 * this URI, and its path is computed as follows: </p>
1019 *
1020 * <ol>
1021 *
1022 * <li><p> If the given URI's path is absolute then the new URI's path
1023 * is taken from the given URI. </p></li>
1024 *
1025 * <li><p> Otherwise the given URI's path is relative, and so the new
1026 * URI's path is computed by resolving the path of the given URI
1027 * against the path of this URI. This is done by concatenating all but
1028 * the last segment of this URI's path, if any, with the given URI's
1029 * path and then normalizing the result as if by invoking the {@link
1030 * #normalize() normalize} method. </p></li>
1031 *
1032 * </ol></li>
1033 *
1034 * </ol>
1035 *
1036 * <p> The result of this method is absolute if, and only if, either this
1037 * URI is absolute or the given URI is absolute. </p>
1038 *
1039 * @param uri The URI to be resolved against this URI
1040 * @return The resulting URI
1041 *
1042 * @throws NullPointerException
1043 * If {@code uri} is {@code null}
1044 */
1045 public URI resolve(URI uri) {
1046 return resolve(this, uri);
1047 }
1048
1049 /**
1050 * Constructs a new URI by parsing the given string and then resolving it
1051 * against this URI.
1052 *
1053 * <p> This convenience method works as if invoking it were equivalent to
1054 * evaluating the expression {@link #resolve(java.net.URI)
1055 * resolve}{@code (URI.}{@link #create(String) create}{@code (str))}. </p>
1056 *
1057 * @param str The string to be parsed into a URI
1058 * @return The resulting URI
1059 *
1060 * @throws NullPointerException
1061 * If {@code str} is {@code null}
1062 *
1063 * @throws IllegalArgumentException
1064 * If the given string violates RFC 2396
1065 */
1066 public URI resolve(String str) {
1067 return resolve(URI.create(str));
1068 }
1069
1070 /**
1071 * Relativizes the given URI against this URI.
1072 *
1073 * <p> The relativization of the given URI against this URI is computed as
1074 * follows: </p>
1075 *
1076 * <ol>
1077 *
1078 * <li><p> If either this URI or the given URI are opaque, or if the
1079 * scheme and authority components of the two URIs are not identical, or
1080 * if the path of this URI is not a prefix of the path of the given URI,
1081 * then the given URI is returned. </p></li>
1082 *
1083 * <li><p> Otherwise a new relative hierarchical URI is constructed with
1084 * query and fragment components taken from the given URI and with a path
1085 * component computed by removing this URI's path from the beginning of
1086 * the given URI's path. </p></li>
1087 *
1088 * </ol>
1089 *
1090 * @param uri The URI to be relativized against this URI
1091 * @return The resulting URI
1092 *
1093 * @throws NullPointerException
1094 * If {@code uri} is {@code null}
1095 */
1096 public URI relativize(URI uri) {
1097 return relativize(this, uri);
1098 }
1099
1100 /**
1101 * Constructs a URL from this URI.
1102 *
1103 * <p> This convenience method works as if invoking it were equivalent to
1104 * evaluating the expression {@code new URL(this.toString())} after
1105 * first checking that this URI is absolute. </p>
1106 *
1107 * @return A URL constructed from this URI
1108 *
1109 * @throws IllegalArgumentException
1110 * If this URL is not absolute
1111 *
1112 * @throws MalformedURLException
1113 * If a protocol handler for the URL could not be found,
1114 * or if some other error occurred while constructing the URL
1115 */
1116 public URL toURL() throws MalformedURLException {
1117 return URL.fromURI(this);
1118 }
1119
1120 // -- Component access methods --
1121
1122 /**
1123 * Returns the scheme component of this URI.
1124 *
1125 * <p> The scheme component of a URI, if defined, only contains characters
1126 * in the <i>alphanum</i> category and in the string {@code "-.+"}. A
1127 * scheme always starts with an <i>alpha</i> character. <p>
1128 *
1129 * The scheme component of a URI cannot contain escaped octets, hence this
1130 * method does not perform any decoding.
1131 *
1132 * @return The scheme component of this URI,
1133 * or {@code null} if the scheme is undefined
1134 */
1135 public String getScheme() {
1136 return scheme;
1137 }
1138
1139 /**
1140 * Tells whether or not this URI is absolute.
1141 *
1142 * <p> A URI is absolute if, and only if, it has a scheme component. </p>
1143 *
1144 * @return {@code true} if, and only if, this URI is absolute
1145 */
1146 public boolean isAbsolute() {
1147 return scheme != null;
1148 }
1149
1150 /**
1151 * Tells whether or not this URI is opaque.
1152 *
1153 * <p> A URI is opaque if, and only if, it is absolute and its
1154 * scheme-specific part does not begin with a slash character ('/').
1155 * An opaque URI has a scheme, a scheme-specific part, and possibly
1156 * a fragment; all other components are undefined. </p>
1157 *
1158 * @return {@code true} if, and only if, this URI is opaque
1159 */
1160 public boolean isOpaque() {
1161 return path == null;
1162 }
1163
1164 /**
1165 * Returns the raw scheme-specific part of this URI. The scheme-specific
1166 * part is never undefined, though it may be empty.
1167 *
1168 * <p> The scheme-specific part of a URI only contains legal URI
1169 * characters. </p>
1170 *
1171 * @return The raw scheme-specific part of this URI
1172 * (never {@code null})
1173 */
1174 public String getRawSchemeSpecificPart() {
1175 String part = schemeSpecificPart;
1176 if (part != null) {
1177 return part;
1178 }
1179
1180 String s = string;
1181 if (s != null) {
1182 // if string is defined, components will have been parsed
1183 int start = 0;
1184 int end = s.length();
1185 if (scheme != null) {
1186 start = scheme.length() + 1;
1187 }
1188 if (fragment != null) {
1189 end -= fragment.length() + 1;
1190 }
1191 if (path != null && path.length() == end - start) {
1192 part = path;
1193 } else {
1194 part = s.substring(start, end);
1195 }
1196 } else {
1197 StringBuilder sb = new StringBuilder();
1198 appendSchemeSpecificPart(sb, null, getAuthority(), getUserInfo(),
1199 host, port, getPath(), getQuery());
1200 part = sb.toString();
1201 }
1202 return schemeSpecificPart = part;
1203 }
1204
1205 /**
1206 * Returns the decoded scheme-specific part of this URI.
1207 *
1208 * <p> The string returned by this method is equal to that returned by the
1209 * {@link #getRawSchemeSpecificPart() getRawSchemeSpecificPart} method
1210 * except that all sequences of escaped octets are <a
1211 * href="#decode">decoded</a>. </p>
1212 *
1213 * @return The decoded scheme-specific part of this URI
1214 * (never {@code null})
1215 */
1216 public String getSchemeSpecificPart() {
1217 String part = decodedSchemeSpecificPart;
1218 if (part == null) {
1219 decodedSchemeSpecificPart = part = decode(getRawSchemeSpecificPart());
1220 }
1221 return part;
1222 }
1223
1224 /**
1225 * Returns the raw authority component of this URI.
1226 *
1227 * <p> The authority component of a URI, if defined, only contains the
1228 * commercial-at character ({@code '@'}) and characters in the
1229 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, and <i>other</i>
1230 * categories. If the authority is server-based then it is further
1231 * constrained to have valid user-information, host, and port
1232 * components. </p>
1233 *
1234 * @return The raw authority component of this URI,
1235 * or {@code null} if the authority is undefined
1236 */
1237 public String getRawAuthority() {
1238 return authority;
1239 }
1240
1241 /**
1242 * Returns the decoded authority component of this URI.
1243 *
1244 * <p> The string returned by this method is equal to that returned by the
1245 * {@link #getRawAuthority() getRawAuthority} method except that all
1246 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1247 *
1248 * @return The decoded authority component of this URI,
1249 * or {@code null} if the authority is undefined
1250 */
1251 public String getAuthority() {
1252 String auth = decodedAuthority;
1253 if ((auth == null) && (authority != null)) {
1254 decodedAuthority = auth = decode(authority);
1255 }
1256 return auth;
1257 }
1258
1259 /**
1260 * Returns the raw user-information component of this URI.
1261 *
1262 * <p> The user-information component of a URI, if defined, only contains
1263 * characters in the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, and
1264 * <i>other</i> categories. </p>
1265 *
1266 * @return The raw user-information component of this URI,
1267 * or {@code null} if the user information is undefined
1268 */
1269 public String getRawUserInfo() {
1270 return userInfo;
1271 }
1272
1273 /**
1274 * Returns the decoded user-information component of this URI.
1275 *
1276 * <p> The string returned by this method is equal to that returned by the
1277 * {@link #getRawUserInfo() getRawUserInfo} method except that all
1278 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1279 *
1280 * @return The decoded user-information component of this URI,
1281 * or {@code null} if the user information is undefined
1282 */
1283 public String getUserInfo() {
1284 String user = decodedUserInfo;
1285 if ((user == null) && (userInfo != null)) {
1286 decodedUserInfo = user = decode(userInfo);
1287 }
1288 return user;
1289 }
1290
1291 /**
1292 * Returns the host component of this URI.
1293 *
1294 * <p> The host component of a URI, if defined, will have one of the
1295 * following forms: </p>
1296 *
1297 * <ul>
1298 *
1299 * <li><p> A domain name consisting of one or more <i>labels</i>
1300 * separated by period characters ({@code '.'}), optionally followed by
1301 * a period character. Each label consists of <i>alphanum</i> characters
1302 * as well as hyphen characters ({@code '-'}), though hyphens never
1303 * occur as the first or last characters in a label. The rightmost
1304 * label of a domain name consisting of two or more labels, begins
1305 * with an <i>alpha</i> character. </li>
1306 *
1307 * <li><p> A dotted-quad IPv4 address of the form
1308 * <i>digit</i>{@code +.}<i>digit</i>{@code +.}<i>digit</i>{@code +.}<i>digit</i>{@code +},
1309 * where no <i>digit</i> sequence is longer than three characters and no
1310 * sequence has a value larger than 255. </p></li>
1311 *
1312 * <li><p> An IPv6 address enclosed in square brackets ({@code '['} and
1313 * {@code ']'}) and consisting of hexadecimal digits, colon characters
1314 * ({@code ':'}), and possibly an embedded IPv4 address. The full
1315 * syntax of IPv6 addresses is specified in <a
1316 * href="http://www.ietf.org/rfc/rfc2373.txt"><i>RFC 2373: IPv6
1317 * Addressing Architecture</i></a>. </p></li>
1318 *
1319 * </ul>
1320 *
1321 * The host component of a URI cannot contain escaped octets, hence this
1322 * method does not perform any decoding.
1323 *
1324 * @return The host component of this URI,
1325 * or {@code null} if the host is undefined
1326 */
1327 public String getHost() {
1328 return host;
1329 }
1330
1331 /**
1332 * Returns the port number of this URI.
1333 *
1334 * <p> The port component of a URI, if defined, is a non-negative
1335 * integer. </p>
1336 *
1337 * @return The port component of this URI,
1338 * or {@code -1} if the port is undefined
1339 */
1340 public int getPort() {
1341 return port;
1342 }
1343
1344 /**
1345 * Returns the raw path component of this URI.
1346 *
1347 * <p> The path component of a URI, if defined, only contains the slash
1348 * character ({@code '/'}), the commercial-at character ({@code '@'}),
1349 * and characters in the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>,
1350 * and <i>other</i> categories. </p>
1351 *
1352 * @return The path component of this URI,
1353 * or {@code null} if the path is undefined
1354 */
1355 public String getRawPath() {
1356 return path;
1357 }
1358
1359 /**
1360 * Returns the decoded path component of this URI.
1361 *
1362 * <p> The string returned by this method is equal to that returned by the
1363 * {@link #getRawPath() getRawPath} method except that all sequences of
1364 * escaped octets are <a href="#decode">decoded</a>. </p>
1365 *
1366 * @return The decoded path component of this URI,
1367 * or {@code null} if the path is undefined
1368 */
1369 public String getPath() {
1370 String decoded = decodedPath;
1371 if ((decoded == null) && (path != null)) {
1372 decodedPath = decoded = decode(path);
1373 }
1374 return decoded;
1375 }
1376
1377 /**
1378 * Returns the raw query component of this URI.
1379 *
1380 * <p> The query component of a URI, if defined, only contains legal URI
1381 * characters. </p>
1382 *
1383 * @return The raw query component of this URI,
1384 * or {@code null} if the query is undefined
1385 */
1386 public String getRawQuery() {
1387 return query;
1388 }
1389
1390 /**
1391 * Returns the decoded query component of this URI.
1392 *
1393 * <p> The string returned by this method is equal to that returned by the
1394 * {@link #getRawQuery() getRawQuery} method except that all sequences of
1395 * escaped octets are <a href="#decode">decoded</a>. </p>
1396 *
1397 * @return The decoded query component of this URI,
1398 * or {@code null} if the query is undefined
1399 */
1400 public String getQuery() {
1401 String decoded = decodedQuery;
1402 if ((decoded == null) && (query != null)) {
1403 decodedQuery = decoded = decode(query, false);
1404 }
1405 return decoded;
1406 }
1407
1408 /**
1409 * Returns the raw fragment component of this URI.
1410 *
1411 * <p> The fragment component of a URI, if defined, only contains legal URI
1412 * characters. </p>
1413 *
1414 * @return The raw fragment component of this URI,
1415 * or {@code null} if the fragment is undefined
1416 */
1417 public String getRawFragment() {
1418 return fragment;
1419 }
1420
1421 /**
1422 * Returns the decoded fragment component of this URI.
1423 *
1424 * <p> The string returned by this method is equal to that returned by the
1425 * {@link #getRawFragment() getRawFragment} method except that all
1426 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1427 *
1428 * @return The decoded fragment component of this URI,
1429 * or {@code null} if the fragment is undefined
1430 */
1431 public String getFragment() {
1432 String decoded = decodedFragment;
1433 if ((decoded == null) && (fragment != null)) {
1434 decodedFragment = decoded = decode(fragment, false);
1435 }
1436 return decoded;
1437 }
1438
1439
1440 // -- Equality, comparison, hash code, toString, and serialization --
1441
1442 /**
1443 * Tests this URI for equality with another object.
1444 *
1445 * <p> If the given object is not a URI then this method immediately
1446 * returns {@code false}.
1447 *
1448 * <p> For two URIs to be considered equal requires that either both are
1449 * opaque or both are hierarchical. Their schemes must either both be
1450 * undefined or else be equal without regard to case. Their fragments
1451 * must either both be undefined or else be equal.
1452 *
1453 * <p> For two opaque URIs to be considered equal, their scheme-specific
1454 * parts must be equal.
1455 *
1456 * <p> For two hierarchical URIs to be considered equal, their paths must
1457 * be equal and their queries must either both be undefined or else be
1458 * equal. Their authorities must either both be undefined, or both be
1459 * registry-based, or both be server-based. If their authorities are
1460 * defined and are registry-based, then they must be equal. If their
1461 * authorities are defined and are server-based, then their hosts must be
1462 * equal without regard to case, their port numbers must be equal, and
1463 * their user-information components must be equal.
1464 *
1465 * <p> When testing the user-information, path, query, fragment, authority,
1466 * or scheme-specific parts of two URIs for equality, the raw forms rather
1467 * than the encoded forms of these components are compared and the
1468 * hexadecimal digits of escaped octets are compared without regard to
1469 * case.
1470 *
1471 * <p> This method satisfies the general contract of the {@link
1472 * java.lang.Object#equals(Object) Object.equals} method. </p>
1473 *
1474 * @param ob The object to which this object is to be compared
1475 *
1476 * @return {@code true} if, and only if, the given object is a URI that
1477 * is identical to this URI
1478 */
1479 public boolean equals(Object ob) {
1480 if (ob == this)
1481 return true;
1482 if (!(ob instanceof URI))
1483 return false;
1484 URI that = (URI)ob;
1485 if (this.isOpaque() != that.isOpaque()) return false;
1486 if (!equalIgnoringCase(this.scheme, that.scheme)) return false;
1487 if (!equal(this.fragment, that.fragment)) return false;
1488
1489 // Opaque
1490 if (this.isOpaque())
1491 return equal(this.schemeSpecificPart, that.schemeSpecificPart);
1492
1493 // Hierarchical
1494 if (!equal(this.path, that.path)) return false;
1495 if (!equal(this.query, that.query)) return false;
1496
1497 // Authorities
1498 if (this.authority == that.authority) return true;
1499 if (this.host != null) {
1500 // Server-based
1501 if (!equal(this.userInfo, that.userInfo)) return false;
1502 if (!equalIgnoringCase(this.host, that.host)) return false;
1503 if (this.port != that.port) return false;
1504 } else if (this.authority != null) {
1505 // Registry-based
1506 if (!equal(this.authority, that.authority)) return false;
1507 } else if (this.authority != that.authority) {
1508 return false;
1509 }
1510
1511 return true;
1512 }
1513
1514 /**
1515 * Returns a hash-code value for this URI. The hash code is based upon all
1516 * of the URI's components, and satisfies the general contract of the
1517 * {@link java.lang.Object#hashCode() Object.hashCode} method.
1518 *
1519 * @return A hash-code value for this URI
1520 */
1521 public int hashCode() {
1522 int h = hash;
1523 if (h == 0) {
1524 h = hashIgnoringCase(0, scheme);
1525 h = hash(h, fragment);
1526 if (isOpaque()) {
1527 h = hash(h, schemeSpecificPart);
1528 } else {
1529 h = hash(h, path);
1530 h = hash(h, query);
1531 if (host != null) {
1532 h = hash(h, userInfo);
1533 h = hashIgnoringCase(h, host);
1534 h += 1949 * port;
1535 } else {
1536 h = hash(h, authority);
1537 }
1538 }
1539 if (h != 0) {
1540 hash = h;
1541 }
1542 }
1543 return h;
1544 }
1545
1546 /**
1547 * Compares this URI to another object, which must be a URI.
1548 *
1549 * <p> When comparing corresponding components of two URIs, if one
1550 * component is undefined but the other is defined then the first is
1551 * considered to be less than the second. Unless otherwise noted, string
1552 * components are ordered according to their natural, case-sensitive
1553 * ordering as defined by the {@link java.lang.String#compareTo(Object)
1554 * String.compareTo} method. String components that are subject to
1555 * encoding are compared by comparing their raw forms rather than their
1556 * encoded forms.
1557 *
1558 * <p> The ordering of URIs is defined as follows: </p>
1559 *
1560 * <ul>
1561 *
1562 * <li><p> Two URIs with different schemes are ordered according the
1563 * ordering of their schemes, without regard to case. </p></li>
1564 *
1565 * <li><p> A hierarchical URI is considered to be less than an opaque URI
1566 * with an identical scheme. </p></li>
1567 *
1568 * <li><p> Two opaque URIs with identical schemes are ordered according
1569 * to the ordering of their scheme-specific parts. </p></li>
1570 *
1571 * <li><p> Two opaque URIs with identical schemes and scheme-specific
1572 * parts are ordered according to the ordering of their
1573 * fragments. </p></li>
1574 *
1575 * <li><p> Two hierarchical URIs with identical schemes are ordered
1576 * according to the ordering of their authority components: </p>
1577 *
1578 * <ul>
1579 *
1580 * <li><p> If both authority components are server-based then the URIs
1581 * are ordered according to their user-information components; if these
1582 * components are identical then the URIs are ordered according to the
1583 * ordering of their hosts, without regard to case; if the hosts are
1584 * identical then the URIs are ordered according to the ordering of
1585 * their ports. </p></li>
1586 *
1587 * <li><p> If one or both authority components are registry-based then
1588 * the URIs are ordered according to the ordering of their authority
1589 * components. </p></li>
1590 *
1591 * </ul></li>
1592 *
1593 * <li><p> Finally, two hierarchical URIs with identical schemes and
1594 * authority components are ordered according to the ordering of their
1595 * paths; if their paths are identical then they are ordered according to
1596 * the ordering of their queries; if the queries are identical then they
1597 * are ordered according to the order of their fragments. </p></li>
1598 *
1599 * </ul>
1600 *
1601 * <p> This method satisfies the general contract of the {@link
1602 * java.lang.Comparable#compareTo(Object) Comparable.compareTo}
1603 * method. </p>
1604 *
1605 * @param that
1606 * The object to which this URI is to be compared
1607 *
1608 * @return A negative integer, zero, or a positive integer as this URI is
1609 * less than, equal to, or greater than the given URI
1610 *
1611 * @throws ClassCastException
1612 * If the given object is not a URI
1613 */
1614 public int compareTo(URI that) {
1615 int c;
1616
1617 if ((c = compareIgnoringCase(this.scheme, that.scheme)) != 0)
1618 return c;
1619
1620 if (this.isOpaque()) {
1621 if (that.isOpaque()) {
1622 // Both opaque
1623 if ((c = compare(this.schemeSpecificPart,
1624 that.schemeSpecificPart)) != 0)
1625 return c;
1626 return compare(this.fragment, that.fragment);
1627 }
1628 return +1; // Opaque > hierarchical
1629 } else if (that.isOpaque()) {
1630 return -1; // Hierarchical < opaque
1631 }
1632
1633 // Hierarchical
1634 if ((this.host != null) && (that.host != null)) {
1635 // Both server-based
1636 if ((c = compare(this.userInfo, that.userInfo)) != 0)
1637 return c;
1638 if ((c = compareIgnoringCase(this.host, that.host)) != 0)
1639 return c;
1640 if ((c = this.port - that.port) != 0)
1641 return c;
1642 } else {
1643 // If one or both authorities are registry-based then we simply
1644 // compare them in the usual, case-sensitive way. If one is
1645 // registry-based and one is server-based then the strings are
1646 // guaranteed to be unequal, hence the comparison will never return
1647 // zero and the compareTo and equals methods will remain
1648 // consistent.
1649 if ((c = compare(this.authority, that.authority)) != 0) return c;
1650 }
1651
1652 if ((c = compare(this.path, that.path)) != 0) return c;
1653 if ((c = compare(this.query, that.query)) != 0) return c;
1654 return compare(this.fragment, that.fragment);
1655 }
1656
1657 /**
1658 * Returns the content of this URI as a string.
1659 *
1660 * <p> If this URI was created by invoking one of the constructors in this
1661 * class then a string equivalent to the original input string, or to the
1662 * string computed from the originally-given components, as appropriate, is
1663 * returned. Otherwise this URI was created by normalization, resolution,
1664 * or relativization, and so a string is constructed from this URI's
1665 * components according to the rules specified in <a
1666 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
1667 * section 5.2, step 7. </p>
1668 *
1669 * @return The string form of this URI
1670 */
1671 public String toString() {
1672 String s = string;
1673 if (s == null) {
1674 s = defineString();
1675 }
1676 return s;
1677 }
1678
1679 private String defineString() {
1680 String s = string;
1681 if (s != null) {
1682 return s;
1683 }
1684
1685 StringBuilder sb = new StringBuilder();
1686 if (scheme != null) {
1687 sb.append(scheme);
1688 sb.append(':');
1689 }
1690 if (isOpaque()) {
1691 sb.append(schemeSpecificPart);
1692 } else {
1693 if (host != null) {
1694 sb.append("//");
1695 if (userInfo != null) {
1696 sb.append(userInfo);
1697 sb.append('@');
1698 }
1699 boolean needBrackets = ((host.indexOf(':') >= 0)
1700 && !host.startsWith("[")
1701 && !host.endsWith("]"));
1702 if (needBrackets) sb.append('[');
1703 sb.append(host);
1704 if (needBrackets) sb.append(']');
1705 if (port != -1) {
1706 sb.append(':');
1707 sb.append(port);
1708 }
1709 } else if (authority != null) {
1710 sb.append("//");
1711 sb.append(authority);
1712 }
1713 if (path != null)
1714 sb.append(path);
1715 if (query != null) {
1716 sb.append('?');
1717 sb.append(query);
1718 }
1719 }
1720 if (fragment != null) {
1721 sb.append('#');
1722 sb.append(fragment);
1723 }
1724 return string = sb.toString();
1725 }
1726
1727 /**
1728 * Returns the content of this URI as a US-ASCII string.
1729 *
1730 * <p> If this URI does not contain any characters in the <i>other</i>
1731 * category then an invocation of this method will return the same value as
1732 * an invocation of the {@link #toString() toString} method. Otherwise
1733 * this method works as if by invoking that method and then <a
1734 * href="#encode">encoding</a> the result. </p>
1735 *
1736 * @return The string form of this URI, encoded as needed
1737 * so that it only contains characters in the US-ASCII
1738 * charset
1739 */
1740 public String toASCIIString() {
1741 return encode(toString());
1742 }
1743
1744
1745 // -- Serialization support --
1746
1747 /**
1748 * Saves the content of this URI to the given serial stream.
1749 *
1750 * <p> The only serializable field of a URI instance is its {@code string}
1751 * field. That field is given a value, if it does not have one already,
1752 * and then the {@link java.io.ObjectOutputStream#defaultWriteObject()}
1753 * method of the given object-output stream is invoked. </p>
1754 *
1755 * @param os The object-output stream to which this object
1756 * is to be written
1757 */
1758 private void writeObject(ObjectOutputStream os)
1759 throws IOException
1760 {
1761 defineString();
1762 os.defaultWriteObject(); // Writes the string field only
1763 }
1764
1765 /**
1766 * Reconstitutes a URI from the given serial stream.
1767 *
1768 * <p> The {@link java.io.ObjectInputStream#defaultReadObject()} method is
1769 * invoked to read the value of the {@code string} field. The result is
1770 * then parsed in the usual way.
1771 *
1772 * @param is The object-input stream from which this object
1773 * is being read
1774 */
1775 private void readObject(ObjectInputStream is)
1776 throws ClassNotFoundException, IOException
1777 {
1778 port = -1; // Argh
1779 is.defaultReadObject();
1780 try {
1781 new Parser(string).parse(false);
1782 } catch (URISyntaxException x) {
1783 IOException y = new InvalidObjectException("Invalid URI");
1784 y.initCause(x);
1785 throw y;
1786 }
1787 }
1788
1789
1790 // -- End of public methods --
1791
1792
1793 // -- Utility methods for string-field comparison and hashing --
1794
1795 // These methods return appropriate values for null string arguments,
1796 // thereby simplifying the equals, hashCode, and compareTo methods.
1797 //
1798 // The case-ignoring methods should only be applied to strings whose
1799 // characters are all known to be US-ASCII. Because of this restriction,
1800 // these methods are faster than the similar methods in the String class.
1801
1802 // US-ASCII only
1803 private static int toLower(char c) {
1804 if ((c >= 'A') && (c <= 'Z'))
1805 return c + ('a' - 'A');
1806 return c;
1807 }
1808
1809 // US-ASCII only
1810 private static int toUpper(char c) {
1811 if ((c >= 'a') && (c <= 'z'))
1812 return c - ('a' - 'A');
1813 return c;
1814 }
1815
1816 private static boolean equal(String s, String t) {
1817 if (s == t) return true;
1818 if ((s != null) && (t != null)) {
1819 if (s.length() != t.length())
1820 return false;
1821 if (s.indexOf('%') < 0)
1822 return s.equals(t);
1823 int n = s.length();
1824 for (int i = 0; i < n;) {
1825 char c = s.charAt(i);
1826 char d = t.charAt(i);
1827 if (c != '%') {
1828 if (c != d)
1829 return false;
1830 i++;
1831 continue;
1832 }
1833 if (d != '%')
1834 return false;
1835 i++;
1836 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1837 return false;
1838 i++;
1839 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1840 return false;
1841 i++;
1842 }
1843 return true;
1844 }
1845 return false;
1846 }
1847
1848 // US-ASCII only
1849 private static boolean equalIgnoringCase(String s, String t) {
1850 if (s == t) return true;
1851 if ((s != null) && (t != null)) {
1852 int n = s.length();
1853 if (t.length() != n)
1854 return false;
1855 for (int i = 0; i < n; i++) {
1856 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1857 return false;
1858 }
1859 return true;
1860 }
1861 return false;
1862 }
1863
1864 private static int hash(int hash, String s) {
1865 if (s == null) return hash;
1866 return s.indexOf('%') < 0 ? hash * 127 + s.hashCode()
1867 : normalizedHash(hash, s);
1868 }
1869
1870
1871 private static int normalizedHash(int hash, String s) {
1872 int h = 0;
1873 for (int index = 0; index < s.length(); index++) {
1874 char ch = s.charAt(index);
1875 h = 31 * h + ch;
1876 if (ch == '%') {
1877 /*
1878 * Process the next two encoded characters
1879 */
1880 for (int i = index + 1; i < index + 3; i++)
1881 h = 31 * h + toUpper(s.charAt(i));
1882 index += 2;
1883 }
1884 }
1885 return hash * 127 + h;
1886 }
1887
1888 // US-ASCII only
1889 private static int hashIgnoringCase(int hash, String s) {
1890 if (s == null) return hash;
1891 int h = hash;
1892 int n = s.length();
1893 for (int i = 0; i < n; i++)
1894 h = 31 * h + toLower(s.charAt(i));
1895 return h;
1896 }
1897
1898 private static int compare(String s, String t) {
1899 if (s == t) return 0;
1900 if (s != null) {
1901 if (t != null)
1902 return s.compareTo(t);
1903 else
1904 return +1;
1905 } else {
1906 return -1;
1907 }
1908 }
1909
1910 // US-ASCII only
1911 private static int compareIgnoringCase(String s, String t) {
1912 if (s == t) return 0;
1913 if (s != null) {
1914 if (t != null) {
1915 int sn = s.length();
1916 int tn = t.length();
1917 int n = sn < tn ? sn : tn;
1918 for (int i = 0; i < n; i++) {
1919 int c = toLower(s.charAt(i)) - toLower(t.charAt(i));
1920 if (c != 0)
1921 return c;
1922 }
1923 return sn - tn;
1924 }
1925 return +1;
1926 } else {
1927 return -1;
1928 }
1929 }
1930
1931
1932 // -- String construction --
1933
1934 // If a scheme is given then the path, if given, must be absolute
1935 //
1936 private static void checkPath(String s, String scheme, String path)
1937 throws URISyntaxException
1938 {
1939 if (scheme != null) {
1940 if ((path != null)
1941 && ((path.length() > 0) && (path.charAt(0) != '/')))
1942 throw new URISyntaxException(s,
1943 "Relative path in absolute URI");
1944 }
1945 }
1946
1947 private void appendAuthority(StringBuilder sb,
1948 String authority,
1949 String userInfo,
1950 String host,
1951 int port)
1952 {
1953 if (host != null) {
1954 sb.append("//");
1955 if (userInfo != null) {
1956 sb.append(quote(userInfo, L_USERINFO, H_USERINFO));
1957 sb.append('@');
1958 }
1959 boolean needBrackets = ((host.indexOf(':') >= 0)
1960 && !host.startsWith("[")
1961 && !host.endsWith("]"));
1962 if (needBrackets) sb.append('[');
1963 sb.append(host);
1964 if (needBrackets) sb.append(']');
1965 if (port != -1) {
1966 sb.append(':');
1967 sb.append(port);
1968 }
1969 } else if (authority != null) {
1970 sb.append("//");
1971 if (authority.startsWith("[")) {
1972 // authority should (but may not) contain an embedded IPv6 address
1973 int end = authority.indexOf(']');
1974 String doquote = authority, dontquote = "";
1975 if (end != -1 && authority.indexOf(':') != -1) {
1976 // the authority contains an IPv6 address
1977 if (end == authority.length()) {
1978 dontquote = authority;
1979 doquote = "";
1980 } else {
1981 dontquote = authority.substring(0 , end + 1);
1982 doquote = authority.substring(end + 1);
1983 }
1984 }
1985 sb.append(dontquote);
1986 sb.append(quote(doquote,
1987 L_REG_NAME | L_SERVER,
1988 H_REG_NAME | H_SERVER));
1989 } else {
1990 sb.append(quote(authority,
1991 L_REG_NAME | L_SERVER,
1992 H_REG_NAME | H_SERVER));
1993 }
1994 }
1995 }
1996
1997 private void appendSchemeSpecificPart(StringBuilder sb,
1998 String opaquePart,
1999 String authority,
2000 String userInfo,
2001 String host,
2002 int port,
2003 String path,
2004 String query)
2005 {
2006 if (opaquePart != null) {
2007 /* check if SSP begins with an IPv6 address
2008 * because we must not quote a literal IPv6 address
2009 */
2010 if (opaquePart.startsWith("//[")) {
2011 int end = opaquePart.indexOf(']');
2012 if (end != -1 && opaquePart.indexOf(':')!=-1) {
2013 String doquote, dontquote;
2014 if (end == opaquePart.length()) {
2015 dontquote = opaquePart;
2016 doquote = "";
2017 } else {
2018 dontquote = opaquePart.substring(0,end+1);
2019 doquote = opaquePart.substring(end+1);
2020 }
2021 sb.append (dontquote);
2022 sb.append(quote(doquote, L_URIC, H_URIC));
2023 }
2024 } else {
2025 sb.append(quote(opaquePart, L_URIC, H_URIC));
2026 }
2027 } else {
2028 appendAuthority(sb, authority, userInfo, host, port);
2029 if (path != null)
2030 sb.append(quote(path, L_PATH, H_PATH));
2031 if (query != null) {
2032 sb.append('?');
2033 sb.append(quote(query, L_URIC, H_URIC));
2034 }
2035 }
2036 }
2037
2038 private void appendFragment(StringBuilder sb, String fragment) {
2039 if (fragment != null) {
2040 sb.append('#');
2041 sb.append(quote(fragment, L_URIC, H_URIC));
2042 }
2043 }
2044
2045 private String toString(String scheme,
2046 String opaquePart,
2047 String authority,
2048 String userInfo,
2049 String host,
2050 int port,
2051 String path,
2052 String query,
2053 String fragment)
2054 {
2055 StringBuilder sb = new StringBuilder();
2056 if (scheme != null) {
2057 sb.append(scheme);
2058 sb.append(':');
2059 }
2060 appendSchemeSpecificPart(sb, opaquePart,
2061 authority, userInfo, host, port,
2062 path, query);
2063 appendFragment(sb, fragment);
2064 return sb.toString();
2065 }
2066
2067 // -- Normalization, resolution, and relativization --
2068
2069 // RFC2396 5.2 (6)
2070 private static String resolvePath(String base, String child,
2071 boolean absolute)
2072 {
2073 int i = base.lastIndexOf('/');
2074 int cn = child.length();
2075 String path = "";
2076
2077 if (cn == 0) {
2078 // 5.2 (6a)
2079 if (i >= 0)
2080 path = base.substring(0, i + 1);
2081 } else {
2082 StringBuilder sb = new StringBuilder(base.length() + cn);
2083 // 5.2 (6a)
2084 if (i >= 0)
2085 sb.append(base, 0, i + 1);
2086 // 5.2 (6b)
2087 sb.append(child);
2088 path = sb.toString();
2089 }
2090
2091 // 5.2 (6c-f)
2092 String np = normalize(path);
2093
2094 // 5.2 (6g): If the result is absolute but the path begins with "../",
2095 // then we simply leave the path as-is
2096
2097 return np;
2098 }
2099
2100 // RFC2396 5.2
2101 private static URI resolve(URI base, URI child) {
2102 // check if child if opaque first so that NPE is thrown
2103 // if child is null.
2104 if (child.isOpaque() || base.isOpaque())
2105 return child;
2106
2107 // 5.2 (2): Reference to current document (lone fragment)
2108 if ((child.scheme == null) && (child.authority == null)
2109 && child.path.isEmpty() && (child.fragment != null)
2110 && (child.query == null)) {
2111 if ((base.fragment != null)
2112 && child.fragment.equals(base.fragment)) {
2113 return base;
2114 }
2115 URI ru = new URI();
2116 ru.scheme = base.scheme;
2117 ru.authority = base.authority;
2118 ru.userInfo = base.userInfo;
2119 ru.host = base.host;
2120 ru.port = base.port;
2121 ru.path = base.path;
2122 ru.fragment = child.fragment;
2123 ru.query = base.query;
2124 return ru;
2125 }
2126
2127 // 5.2 (3): Child is absolute
2128 if (child.scheme != null)
2129 return child;
2130
2131 URI ru = new URI(); // Resolved URI
2132 ru.scheme = base.scheme;
2133 ru.query = child.query;
2134 ru.fragment = child.fragment;
2135
2136 // 5.2 (4): Authority
2137 if (child.authority == null) {
2138 ru.authority = base.authority;
2139 ru.host = base.host;
2140 ru.userInfo = base.userInfo;
2141 ru.port = base.port;
2142
2143 String cp = (child.path == null) ? "" : child.path;
2144 if ((cp.length() > 0) && (cp.charAt(0) == '/')) {
2145 // 5.2 (5): Child path is absolute
2146 ru.path = child.path;
2147 } else {
2148 // 5.2 (6): Resolve relative path
2149 ru.path = resolvePath(base.path, cp, base.isAbsolute());
2150 }
2151 } else {
2152 ru.authority = child.authority;
2153 ru.host = child.host;
2154 ru.userInfo = child.userInfo;
2155 ru.host = child.host;
2156 ru.port = child.port;
2157 ru.path = child.path;
2158 }
2159
2160 // 5.2 (7): Recombine (nothing to do here)
2161 return ru;
2162 }
2163
2164 // If the given URI's path is normal then return the URI;
2165 // o.w., return a new URI containing the normalized path.
2166 //
2167 private static URI normalize(URI u) {
2168 if (u.isOpaque() || (u.path == null) || (u.path.length() == 0))
2169 return u;
2170
2171 String np = normalize(u.path);
2172 if (np == u.path)
2173 return u;
2174
2175 URI v = new URI();
2176 v.scheme = u.scheme;
2177 v.fragment = u.fragment;
2178 v.authority = u.authority;
2179 v.userInfo = u.userInfo;
2180 v.host = u.host;
2181 v.port = u.port;
2182 v.path = np;
2183 v.query = u.query;
2184 return v;
2185 }
2186
2187 // If both URIs are hierarchical, their scheme and authority components are
2188 // identical, and the base path is a prefix of the child's path, then
2189 // return a relative URI that, when resolved against the base, yields the
2190 // child; otherwise, return the child.
2191 //
2192 private static URI relativize(URI base, URI child) {
2193 // check if child if opaque first so that NPE is thrown
2194 // if child is null.
2195 if (child.isOpaque() || base.isOpaque())
2196 return child;
2197 if (!equalIgnoringCase(base.scheme, child.scheme)
2198 || !equal(base.authority, child.authority))
2199 return child;
2200
2201 String bp = normalize(base.path);
2202 String cp = normalize(child.path);
2203 if (!bp.equals(cp)) {
2204 if (!bp.endsWith("/"))
2205 bp = bp + "/";
2206 if (!cp.startsWith(bp))
2207 return child;
2208 }
2209
2210 URI v = new URI();
2211 v.path = cp.substring(bp.length());
2212 v.query = child.query;
2213 v.fragment = child.fragment;
2214 return v;
2215 }
2216
2217
2218
2219 // -- Path normalization --
2220
2221 // The following algorithm for path normalization avoids the creation of a
2222 // string object for each segment, as well as the use of a string buffer to
2223 // compute the final result, by using a single char array and editing it in
2224 // place. The array is first split into segments, replacing each slash
2225 // with '\0' and creating a segment-index array, each element of which is
2226 // the index of the first char in the corresponding segment. We then walk
2227 // through both arrays, removing ".", "..", and other segments as necessary
2228 // by setting their entries in the index array to -1. Finally, the two
2229 // arrays are used to rejoin the segments and compute the final result.
2230 //
2231 // This code is based upon src/solaris/native/java/io/canonicalize_md.c
2232
2233
2234 // Check the given path to see if it might need normalization. A path
2235 // might need normalization if it contains duplicate slashes, a "."
2236 // segment, or a ".." segment. Return -1 if no further normalization is
2237 // possible, otherwise return the number of segments found.
2238 //
2239 // This method takes a string argument rather than a char array so that
2240 // this test can be performed without invoking path.toCharArray().
2241 //
2242 private static int needsNormalization(String path) {
2243 boolean normal = true;
2244 int ns = 0; // Number of segments
2245 int end = path.length() - 1; // Index of last char in path
2246 int p = 0; // Index of next char in path
2247
2248 // Skip initial slashes
2249 while (p <= end) {
2250 if (path.charAt(p) != '/') break;
2251 p++;
2252 }
2253 if (p > 1) normal = false;
2254
2255 // Scan segments
2256 while (p <= end) {
2257
2258 // Looking at "." or ".." ?
2259 if ((path.charAt(p) == '.')
2260 && ((p == end)
2261 || ((path.charAt(p + 1) == '/')
2262 || ((path.charAt(p + 1) == '.')
2263 && ((p + 1 == end)
2264 || (path.charAt(p + 2) == '/')))))) {
2265 normal = false;
2266 }
2267 ns++;
2268
2269 // Find beginning of next segment
2270 while (p <= end) {
2271 if (path.charAt(p++) != '/')
2272 continue;
2273
2274 // Skip redundant slashes
2275 while (p <= end) {
2276 if (path.charAt(p) != '/') break;
2277 normal = false;
2278 p++;
2279 }
2280
2281 break;
2282 }
2283 }
2284
2285 return normal ? -1 : ns;
2286 }
2287
2288
2289 // Split the given path into segments, replacing slashes with nulls and
2290 // filling in the given segment-index array.
2291 //
2292 // Preconditions:
2293 // segs.length == Number of segments in path
2294 //
2295 // Postconditions:
2296 // All slashes in path replaced by '\0'
2297 // segs[i] == Index of first char in segment i (0 <= i < segs.length)
2298 //
2299 private static void split(char[] path, int[] segs) {
2300 int end = path.length - 1; // Index of last char in path
2301 int p = 0; // Index of next char in path
2302 int i = 0; // Index of current segment
2303
2304 // Skip initial slashes
2305 while (p <= end) {
2306 if (path[p] != '/') break;
2307 path[p] = '\0';
2308 p++;
2309 }
2310
2311 while (p <= end) {
2312
2313 // Note start of segment
2314 segs[i++] = p++;
2315
2316 // Find beginning of next segment
2317 while (p <= end) {
2318 if (path[p++] != '/')
2319 continue;
2320 path[p - 1] = '\0';
2321
2322 // Skip redundant slashes
2323 while (p <= end) {
2324 if (path[p] != '/') break;
2325 path[p++] = '\0';
2326 }
2327 break;
2328 }
2329 }
2330
2331 if (i != segs.length)
2332 throw new InternalError(); // ASSERT
2333 }
2334
2335
2336 // Join the segments in the given path according to the given segment-index
2337 // array, ignoring those segments whose index entries have been set to -1,
2338 // and inserting slashes as needed. Return the length of the resulting
2339 // path.
2340 //
2341 // Preconditions:
2342 // segs[i] == -1 implies segment i is to be ignored
2343 // path computed by split, as above, with '\0' having replaced '/'
2344 //
2345 // Postconditions:
2346 // path[0] .. path[return value] == Resulting path
2347 //
2348 private static int join(char[] path, int[] segs) {
2349 int ns = segs.length; // Number of segments
2350 int end = path.length - 1; // Index of last char in path
2351 int p = 0; // Index of next path char to write
2352
2353 if (path[p] == '\0') {
2354 // Restore initial slash for absolute paths
2355 path[p++] = '/';
2356 }
2357
2358 for (int i = 0; i < ns; i++) {
2359 int q = segs[i]; // Current segment
2360 if (q == -1)
2361 // Ignore this segment
2362 continue;
2363
2364 if (p == q) {
2365 // We're already at this segment, so just skip to its end
2366 while ((p <= end) && (path[p] != '\0'))
2367 p++;
2368 if (p <= end) {
2369 // Preserve trailing slash
2370 path[p++] = '/';
2371 }
2372 } else if (p < q) {
2373 // Copy q down to p
2374 while ((q <= end) && (path[q] != '\0'))
2375 path[p++] = path[q++];
2376 if (q <= end) {
2377 // Preserve trailing slash
2378 path[p++] = '/';
2379 }
2380 } else
2381 throw new InternalError(); // ASSERT false
2382 }
2383
2384 return p;
2385 }
2386
2387
2388 // Remove "." segments from the given path, and remove segment pairs
2389 // consisting of a non-".." segment followed by a ".." segment.
2390 //
2391 private static void removeDots(char[] path, int[] segs) {
2392 int ns = segs.length;
2393 int end = path.length - 1;
2394
2395 for (int i = 0; i < ns; i++) {
2396 int dots = 0; // Number of dots found (0, 1, or 2)
2397
2398 // Find next occurrence of "." or ".."
2399 do {
2400 int p = segs[i];
2401 if (path[p] == '.') {
2402 if (p == end) {
2403 dots = 1;
2404 break;
2405 } else if (path[p + 1] == '\0') {
2406 dots = 1;
2407 break;
2408 } else if ((path[p + 1] == '.')
2409 && ((p + 1 == end)
2410 || (path[p + 2] == '\0'))) {
2411 dots = 2;
2412 break;
2413 }
2414 }
2415 i++;
2416 } while (i < ns);
2417 if ((i > ns) || (dots == 0))
2418 break;
2419
2420 if (dots == 1) {
2421 // Remove this occurrence of "."
2422 segs[i] = -1;
2423 } else {
2424 // If there is a preceding non-".." segment, remove both that
2425 // segment and this occurrence of ".."; otherwise, leave this
2426 // ".." segment as-is.
2427 int j;
2428 for (j = i - 1; j >= 0; j--) {
2429 if (segs[j] != -1) break;
2430 }
2431 if (j >= 0) {
2432 int q = segs[j];
2433 if (!((path[q] == '.')
2434 && (path[q + 1] == '.')
2435 && (path[q + 2] == '\0'))) {
2436 segs[i] = -1;
2437 segs[j] = -1;
2438 }
2439 }
2440 }
2441 }
2442 }
2443
2444
2445 // DEVIATION: If the normalized path is relative, and if the first
2446 // segment could be parsed as a scheme name, then prepend a "." segment
2447 //
2448 private static void maybeAddLeadingDot(char[] path, int[] segs) {
2449
2450 if (path[0] == '\0')
2451 // The path is absolute
2452 return;
2453
2454 int ns = segs.length;
2455 int f = 0; // Index of first segment
2456 while (f < ns) {
2457 if (segs[f] >= 0)
2458 break;
2459 f++;
2460 }
2461 if ((f >= ns) || (f == 0))
2462 // The path is empty, or else the original first segment survived,
2463 // in which case we already know that no leading "." is needed
2464 return;
2465
2466 int p = segs[f];
2467 while ((p < path.length) && (path[p] != ':') && (path[p] != '\0')) p++;
2468 if (p >= path.length || path[p] == '\0')
2469 // No colon in first segment, so no "." needed
2470 return;
2471
2472 // At this point we know that the first segment is unused,
2473 // hence we can insert a "." segment at that position
2474 path[0] = '.';
2475 path[1] = '\0';
2476 segs[0] = 0;
2477 }
2478
2479
2480 // Normalize the given path string. A normal path string has no empty
2481 // segments (i.e., occurrences of "//"), no segments equal to ".", and no
2482 // segments equal to ".." that are preceded by a segment not equal to "..".
2483 // In contrast to Unix-style pathname normalization, for URI paths we
2484 // always retain trailing slashes.
2485 //
2486 private static String normalize(String ps) {
2487
2488 // Does this path need normalization?
2489 int ns = needsNormalization(ps); // Number of segments
2490 if (ns < 0)
2491 // Nope -- just return it
2492 return ps;
2493
2494 char[] path = ps.toCharArray(); // Path in char-array form
2495
2496 // Split path into segments
2497 int[] segs = new int[ns]; // Segment-index array
2498 split(path, segs);
2499
2500 // Remove dots
2501 removeDots(path, segs);
2502
2503 // Prevent scheme-name confusion
2504 maybeAddLeadingDot(path, segs);
2505
2506 // Join the remaining segments and return the result
2507 String s = new String(path, 0, join(path, segs));
2508 if (s.equals(ps)) {
2509 // string was already normalized
2510 return ps;
2511 }
2512 return s;
2513 }
2514
2515
2516
2517 // -- Character classes for parsing --
2518
2519 // RFC2396 precisely specifies which characters in the US-ASCII charset are
2520 // permissible in the various components of a URI reference. We here
2521 // define a set of mask pairs to aid in enforcing these restrictions. Each
2522 // mask pair consists of two longs, a low mask and a high mask. Taken
2523 // together they represent a 128-bit mask, where bit i is set iff the
2524 // character with value i is permitted.
2525 //
2526 // This approach is more efficient than sequentially searching arrays of
2527 // permitted characters. It could be made still more efficient by
2528 // precompiling the mask information so that a character's presence in a
2529 // given mask could be determined by a single table lookup.
2530
2531 // Compute the low-order mask for the characters in the given string
2532 private static long lowMask(String chars) {
2533 int n = chars.length();
2534 long m = 0;
2535 for (int i = 0; i < n; i++) {
2536 char c = chars.charAt(i);
2537 if (c < 64)
2538 m |= (1L << c);
2539 }
2540 return m;
2541 }
2542
2543 // Compute the high-order mask for the characters in the given string
2544 private static long highMask(String chars) {
2545 int n = chars.length();
2546 long m = 0;
2547 for (int i = 0; i < n; i++) {
2548 char c = chars.charAt(i);
2549 if ((c >= 64) && (c < 128))
2550 m |= (1L << (c - 64));
2551 }
2552 return m;
2553 }
2554
2555 // Compute a low-order mask for the characters
2556 // between first and last, inclusive
2557 private static long lowMask(char first, char last) {
2558 long m = 0;
2559 int f = Math.max(Math.min(first, 63), 0);
2560 int l = Math.max(Math.min(last, 63), 0);
2561 for (int i = f; i <= l; i++)
2562 m |= 1L << i;
2563 return m;
2564 }
2565
2566 // Compute a high-order mask for the characters
2567 // between first and last, inclusive
2568 private static long highMask(char first, char last) {
2569 long m = 0;
2570 int f = Math.max(Math.min(first, 127), 64) - 64;
2571 int l = Math.max(Math.min(last, 127), 64) - 64;
2572 for (int i = f; i <= l; i++)
2573 m |= 1L << i;
2574 return m;
2575 }
2576
2577 // Tell whether the given character is permitted by the given mask pair
2578 private static boolean match(char c, long lowMask, long highMask) {
2579 if (c == 0) // 0 doesn't have a slot in the mask. So, it never matches.
2580 return false;
2581 if (c < 64)
2582 return ((1L << c) & lowMask) != 0;
2583 if (c < 128)
2584 return ((1L << (c - 64)) & highMask) != 0;
2585 return false;
2586 }
2587
2588 // Character-class masks, in reverse order from RFC2396 because
2589 // initializers for static fields cannot make forward references.
2590
2591 // digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" |
2592 // "8" | "9"
2593 private static final long L_DIGIT = lowMask('0', '9');
2594 private static final long H_DIGIT = 0L;
2595
2596 // upalpha = "A" | "B" | "C" | "D" | "E" | "F" | "G" | "H" | "I" |
2597 // "J" | "K" | "L" | "M" | "N" | "O" | "P" | "Q" | "R" |
2598 // "S" | "T" | "U" | "V" | "W" | "X" | "Y" | "Z"
2599 private static final long L_UPALPHA = 0L;
2600 private static final long H_UPALPHA = highMask('A', 'Z');
2601
2602 // lowalpha = "a" | "b" | "c" | "d" | "e" | "f" | "g" | "h" | "i" |
2603 // "j" | "k" | "l" | "m" | "n" | "o" | "p" | "q" | "r" |
2604 // "s" | "t" | "u" | "v" | "w" | "x" | "y" | "z"
2605 private static final long L_LOWALPHA = 0L;
2606 private static final long H_LOWALPHA = highMask('a', 'z');
2607
2608 // alpha = lowalpha | upalpha
2609 private static final long L_ALPHA = L_LOWALPHA | L_UPALPHA;
2610 private static final long H_ALPHA = H_LOWALPHA | H_UPALPHA;
2611
2612 // alphanum = alpha | digit
2613 private static final long L_ALPHANUM = L_DIGIT | L_ALPHA;
2614 private static final long H_ALPHANUM = H_DIGIT | H_ALPHA;
2615
2616 // hex = digit | "A" | "B" | "C" | "D" | "E" | "F" |
2617 // "a" | "b" | "c" | "d" | "e" | "f"
2618 private static final long L_HEX = L_DIGIT;
2619 private static final long H_HEX = highMask('A', 'F') | highMask('a', 'f');
2620
2621 // mark = "-" | "_" | "." | "!" | "~" | "*" | "'" |
2622 // "(" | ")"
2623 private static final long L_MARK = lowMask("-_.!~*'()");
2624 private static final long H_MARK = highMask("-_.!~*'()");
2625
2626 // unreserved = alphanum | mark
2627 private static final long L_UNRESERVED = L_ALPHANUM | L_MARK;
2628 private static final long H_UNRESERVED = H_ALPHANUM | H_MARK;
2629
2630 // reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" |
2631 // "$" | "," | "[" | "]"
2632 // Added per RFC2732: "[", "]"
2633 private static final long L_RESERVED = lowMask(";/?:@&=+$,[]");
2634 private static final long H_RESERVED = highMask(";/?:@&=+$,[]");
2635
2636 // The zero'th bit is used to indicate that escape pairs and non-US-ASCII
2637 // characters are allowed; this is handled by the scanEscape method below.
2638 private static final long L_ESCAPED = 1L;
2639 private static final long H_ESCAPED = 0L;
2640
2641 // uric = reserved | unreserved | escaped
2642 private static final long L_URIC = L_RESERVED | L_UNRESERVED | L_ESCAPED;
2643 private static final long H_URIC = H_RESERVED | H_UNRESERVED | H_ESCAPED;
2644
2645 // pchar = unreserved | escaped |
2646 // ":" | "@" | "&" | "=" | "+" | "$" | ","
2647 private static final long L_PCHAR
2648 = L_UNRESERVED | L_ESCAPED | lowMask(":@&=+$,");
2649 private static final long H_PCHAR
2650 = H_UNRESERVED | H_ESCAPED | highMask(":@&=+$,");
2651
2652 // All valid path characters
2653 private static final long L_PATH = L_PCHAR | lowMask(";/");
2654 private static final long H_PATH = H_PCHAR | highMask(";/");
2655
2656 // Dash, for use in domainlabel and toplabel
2657 private static final long L_DASH = lowMask("-");
2658 private static final long H_DASH = highMask("-");
2659
2660 // Dot, for use in hostnames
2661 private static final long L_DOT = lowMask(".");
2662 private static final long H_DOT = highMask(".");
2663
2664 // userinfo = *( unreserved | escaped |
2665 // ";" | ":" | "&" | "=" | "+" | "$" | "," )
2666 private static final long L_USERINFO
2667 = L_UNRESERVED | L_ESCAPED | lowMask(";:&=+$,");
2668 private static final long H_USERINFO
2669 = H_UNRESERVED | H_ESCAPED | highMask(";:&=+$,");
2670
2671 // reg_name = 1*( unreserved | escaped | "$" | "," |
2672 // ";" | ":" | "@" | "&" | "=" | "+" )
2673 private static final long L_REG_NAME
2674 = L_UNRESERVED | L_ESCAPED | lowMask("$,;:@&=+");
2675 private static final long H_REG_NAME
2676 = H_UNRESERVED | H_ESCAPED | highMask("$,;:@&=+");
2677
2678 // All valid characters for server-based authorities
2679 private static final long L_SERVER
2680 = L_USERINFO | L_ALPHANUM | L_DASH | lowMask(".:@[]");
2681 private static final long H_SERVER
2682 = H_USERINFO | H_ALPHANUM | H_DASH | highMask(".:@[]");
2683
2684 // Special case of server authority that represents an IPv6 address
2685 // In this case, a % does not signify an escape sequence
2686 private static final long L_SERVER_PERCENT
2687 = L_SERVER | lowMask("%");
2688 private static final long H_SERVER_PERCENT
2689 = H_SERVER | highMask("%");
2690 private static final long L_LEFT_BRACKET = lowMask("[");
2691 private static final long H_LEFT_BRACKET = highMask("[");
2692
2693 // scheme = alpha *( alpha | digit | "+" | "-" | "." )
2694 private static final long L_SCHEME = L_ALPHA | L_DIGIT | lowMask("+-.");
2695 private static final long H_SCHEME = H_ALPHA | H_DIGIT | highMask("+-.");
2696
2697 // scope_id = alpha | digit | "_" | "."
2698 private static final long L_SCOPE_ID
2699 = L_ALPHANUM | lowMask("_.");
2700 private static final long H_SCOPE_ID
2701 = H_ALPHANUM | highMask("_.");
2702
2703 // -- Escaping and encoding --
2704
2705 private static final char[] hexDigits = {
2706 '0', '1', '2', '3', '4', '5', '6', '7',
2707 '8', '9', 'A', 'B', 'C', 'D', 'E', 'F'
2708 };
2709
2710 private static void appendEscape(StringBuilder sb, byte b) {
2711 sb.append('%');
2712 sb.append(hexDigits[(b >> 4) & 0x0f]);
2713 sb.append(hexDigits[(b >> 0) & 0x0f]);
2714 }
2715
2716 private static void appendEncoded(StringBuilder sb, char c) {
2717 ByteBuffer bb = null;
2718 try {
2719 bb = ThreadLocalCoders.encoderFor("UTF-8")
2720 .encode(CharBuffer.wrap("" + c));
2721 } catch (CharacterCodingException x) {
2722 assert false;
2723 }
2724 while (bb.hasRemaining()) {
2725 int b = bb.get() & 0xff;
2726 if (b >= 0x80)
2727 appendEscape(sb, (byte)b);
2728 else
2729 sb.append((char)b);
2730 }
2731 }
2732
2733 // Quote any characters in s that are not permitted
2734 // by the given mask pair
2735 //
2736 private static String quote(String s, long lowMask, long highMask) {
2737 StringBuilder sb = null;
2738 boolean allowNonASCII = ((lowMask & L_ESCAPED) != 0);
2739 for (int i = 0; i < s.length(); i++) {
2740 char c = s.charAt(i);
2741 if (c < '\u0080') {
2742 if (!match(c, lowMask, highMask)) {
2743 if (sb == null) {
2744 sb = new StringBuilder();
2745 sb.append(s, 0, i);
2746 }
2747 appendEscape(sb, (byte)c);
2748 } else {
2749 if (sb != null)
2750 sb.append(c);
2751 }
2752 } else if (allowNonASCII
2753 && (Character.isSpaceChar(c)
2754 || Character.isISOControl(c))) {
2755 if (sb == null) {
2756 sb = new StringBuilder();
2757 sb.append(s, 0, i);
2758 }
2759 appendEncoded(sb, c);
2760 } else {
2761 if (sb != null)
2762 sb.append(c);
2763 }
2764 }
2765 return (sb == null) ? s : sb.toString();
2766 }
2767
2768 // Encodes all characters >= \u0080 into escaped, normalized UTF-8 octets,
2769 // assuming that s is otherwise legal
2770 //
2771 private static String encode(String s) {
2772 int n = s.length();
2773 if (n == 0)
2774 return s;
2775
2776 // First check whether we actually need to encode
2777 for (int i = 0;;) {
2778 if (s.charAt(i) >= '\u0080')
2779 break;
2780 if (++i >= n)
2781 return s;
2782 }
2783
2784 String ns = Normalizer.normalize(s, Normalizer.Form.NFC);
2785 ByteBuffer bb = null;
2786 try {
2787 bb = ThreadLocalCoders.encoderFor("UTF-8")
2788 .encode(CharBuffer.wrap(ns));
2789 } catch (CharacterCodingException x) {
2790 assert false;
2791 }
2792
2793 StringBuilder sb = new StringBuilder();
2794 while (bb.hasRemaining()) {
2795 int b = bb.get() & 0xff;
2796 if (b >= 0x80)
2797 appendEscape(sb, (byte)b);
2798 else
2799 sb.append((char)b);
2800 }
2801 return sb.toString();
2802 }
2803
2804 private static int decode(char c) {
2805 if ((c >= '0') && (c <= '9'))
2806 return c - '0';
2807 if ((c >= 'a') && (c <= 'f'))
2808 return c - 'a' + 10;
2809 if ((c >= 'A') && (c <= 'F'))
2810 return c - 'A' + 10;
2811 assert false;
2812 return -1;
2813 }
2814
2815 private static byte decode(char c1, char c2) {
2816 return (byte)( ((decode(c1) & 0xf) << 4)
2817 | ((decode(c2) & 0xf) << 0));
2818 }
2819
2820 // Evaluates all escapes in s, applying UTF-8 decoding if needed. Assumes
2821 // that escapes are well-formed syntactically, i.e., of the form %XX. If a
2822 // sequence of escaped octets is not valid UTF-8 then the erroneous octets
2823 // are replaced with '\uFFFD'.
2824 // Exception: any "%" found between "[]" is left alone. It is an IPv6 literal
2825 // with a scope_id
2826 //
2827 private static String decode(String s) {
2828 return decode(s, true);
2829 }
2830
2831 // This method was introduced as a generalization of URI.decode method
2832 // to provide a fix for JDK-8037396
2833 private static String decode(String s, boolean ignorePercentInBrackets) {
2834 if (s == null)
2835 return s;
2836 int n = s.length();
2837 if (n == 0)
2838 return s;
2839 if (s.indexOf('%') < 0)
2840 return s;
2841
2842 StringBuilder sb = new StringBuilder(n);
2843 ByteBuffer bb = ByteBuffer.allocate(n);
2844 CharBuffer cb = CharBuffer.allocate(n);
2845 CharsetDecoder dec = ThreadLocalCoders.decoderFor("UTF-8")
2846 .onMalformedInput(CodingErrorAction.REPLACE)
2847 .onUnmappableCharacter(CodingErrorAction.REPLACE);
2848
2849 // This is not horribly efficient, but it will do for now
2850 char c = s.charAt(0);
2851 boolean betweenBrackets = false;
2852
2853 for (int i = 0; i < n;) {
2854 assert c == s.charAt(i); // Loop invariant
2855 if (c == '[') {
2856 betweenBrackets = true;
2857 } else if (betweenBrackets && c == ']') {
2858 betweenBrackets = false;
2859 }
2860 if (c != '%' || (betweenBrackets && ignorePercentInBrackets)) {
2861 sb.append(c);
2862 if (++i >= n)
2863 break;
2864 c = s.charAt(i);
2865 continue;
2866 }
2867 bb.clear();
2868 int ui = i;
2869 for (;;) {
2870 assert (n - i >= 2);
2871 bb.put(decode(s.charAt(++i), s.charAt(++i)));
2872 if (++i >= n)
2873 break;
2874 c = s.charAt(i);
2875 if (c != '%')
2876 break;
2877 }
2878 bb.flip();
2879 cb.clear();
2880 dec.reset();
2881 CoderResult cr = dec.decode(bb, cb, true);
2882 assert cr.isUnderflow();
2883 cr = dec.flush(cb);
2884 assert cr.isUnderflow();
2885 sb.append(cb.flip().toString());
2886 }
2887
2888 return sb.toString();
2889 }
2890
2891
2892 // -- Parsing --
2893
2894 // For convenience we wrap the input URI string in a new instance of the
2895 // following internal class. This saves always having to pass the input
2896 // string as an argument to each internal scan/parse method.
2897
2898 private class Parser {
2899
2900 private String input; // URI input string
2901 private boolean requireServerAuthority = false;
2902
2903 Parser(String s) {
2904 input = s;
2905 string = s;
2906 }
2907
2908 // -- Methods for throwing URISyntaxException in various ways --
2909
2910 private void fail(String reason) throws URISyntaxException {
2911 throw new URISyntaxException(input, reason);
2912 }
2913
2914 private void fail(String reason, int p) throws URISyntaxException {
2915 throw new URISyntaxException(input, reason, p);
2916 }
2917
2918 private void failExpecting(String expected, int p)
2919 throws URISyntaxException
2920 {
2921 fail("Expected " + expected, p);
2922 }
2923
2924
2925 // -- Simple access to the input string --
2926
2927 // Tells whether start < end and, if so, whether charAt(start) == c
2928 //
2929 private boolean at(int start, int end, char c) {
2930 return (start < end) && (input.charAt(start) == c);
2931 }
2932
2933 // Tells whether start + s.length() < end and, if so,
2934 // whether the chars at the start position match s exactly
2935 //
2936 private boolean at(int start, int end, String s) {
2937 int p = start;
2938 int sn = s.length();
2939 if (sn > end - p)
2940 return false;
2941 int i = 0;
2942 while (i < sn) {
2943 if (input.charAt(p++) != s.charAt(i)) {
2944 break;
2945 }
2946 i++;
2947 }
2948 return (i == sn);
2949 }
2950
2951
2952 // -- Scanning --
2953
2954 // The various scan and parse methods that follow use a uniform
2955 // convention of taking the current start position and end index as
2956 // their first two arguments. The start is inclusive while the end is
2957 // exclusive, just as in the String class, i.e., a start/end pair
2958 // denotes the left-open interval [start, end) of the input string.
2959 //
2960 // These methods never proceed past the end position. They may return
2961 // -1 to indicate outright failure, but more often they simply return
2962 // the position of the first char after the last char scanned. Thus
2963 // a typical idiom is
2964 //
2965 // int p = start;
2966 // int q = scan(p, end, ...);
2967 // if (q > p)
2968 // // We scanned something
2969 // ...;
2970 // else if (q == p)
2971 // // We scanned nothing
2972 // ...;
2973 // else if (q == -1)
2974 // // Something went wrong
2975 // ...;
2976
2977
2978 // Scan a specific char: If the char at the given start position is
2979 // equal to c, return the index of the next char; otherwise, return the
2980 // start position.
2981 //
2982 private int scan(int start, int end, char c) {
2983 if ((start < end) && (input.charAt(start) == c))
2984 return start + 1;
2985 return start;
2986 }
2987
2988 // Scan forward from the given start position. Stop at the first char
2989 // in the err string (in which case -1 is returned), or the first char
2990 // in the stop string (in which case the index of the preceding char is
2991 // returned), or the end of the input string (in which case the length
2992 // of the input string is returned). May return the start position if
2993 // nothing matches.
2994 //
2995 private int scan(int start, int end, String err, String stop) {
2996 int p = start;
2997 while (p < end) {
2998 char c = input.charAt(p);
2999 if (err.indexOf(c) >= 0)
3000 return -1;
3001 if (stop.indexOf(c) >= 0)
3002 break;
3003 p++;
3004 }
3005 return p;
3006 }
3007
3008 // Scan forward from the given start position. Stop at the first char
3009 // in the stop string (in which case the index of the preceding char is
3010 // returned), or the end of the input string (in which case the length
3011 // of the input string is returned). May return the start position if
3012 // nothing matches.
3013 //
3014 private int scan(int start, int end, String stop) {
3015 int p = start;
3016 while (p < end) {
3017 char c = input.charAt(p);
3018 if (stop.indexOf(c) >= 0)
3019 break;
3020 p++;
3021 }
3022 return p;
3023 }
3024
3025 // Scan a potential escape sequence, starting at the given position,
3026 // with the given first char (i.e., charAt(start) == c).
3027 //
3028 // This method assumes that if escapes are allowed then visible
3029 // non-US-ASCII chars are also allowed.
3030 //
3031 private int scanEscape(int start, int n, char first)
3032 throws URISyntaxException
3033 {
3034 int p = start;
3035 char c = first;
3036 if (c == '%') {
3037 // Process escape pair
3038 if ((p + 3 <= n)
3039 && match(input.charAt(p + 1), L_HEX, H_HEX)
3040 && match(input.charAt(p + 2), L_HEX, H_HEX)) {
3041 return p + 3;
3042 }
3043 fail("Malformed escape pair", p);
3044 } else if ((c > 128)
3045 && !Character.isSpaceChar(c)
3046 && !Character.isISOControl(c)) {
3047 // Allow unescaped but visible non-US-ASCII chars
3048 return p + 1;
3049 }
3050 return p;
3051 }
3052
3053 // Scan chars that match the given mask pair
3054 //
3055 private int scan(int start, int n, long lowMask, long highMask)
3056 throws URISyntaxException
3057 {
3058 int p = start;
3059 while (p < n) {
3060 char c = input.charAt(p);
3061 if (match(c, lowMask, highMask)) {
3062 p++;
3063 continue;
3064 }
3065 if ((lowMask & L_ESCAPED) != 0) {
3066 int q = scanEscape(p, n, c);
3067 if (q > p) {
3068 p = q;
3069 continue;
3070 }
3071 }
3072 break;
3073 }
3074 return p;
3075 }
3076
3077 // Check that each of the chars in [start, end) matches the given mask
3078 //
3079 private void checkChars(int start, int end,
3080 long lowMask, long highMask,
3081 String what)
3082 throws URISyntaxException
3083 {
3084 int p = scan(start, end, lowMask, highMask);
3085 if (p < end)
3086 fail("Illegal character in " + what, p);
3087 }
3088
3089 // Check that the char at position p matches the given mask
3090 //
3091 private void checkChar(int p,
3092 long lowMask, long highMask,
3093 String what)
3094 throws URISyntaxException
3095 {
3096 checkChars(p, p + 1, lowMask, highMask, what);
3097 }
3098
3099
3100 // -- Parsing --
3101
3102 // [<scheme>:]<scheme-specific-part>[#<fragment>]
3103 //
3104 void parse(boolean rsa) throws URISyntaxException {
3105 requireServerAuthority = rsa;
3106 int n = input.length();
3107 int p = scan(0, n, "/?#", ":");
3108 if ((p >= 0) && at(p, n, ':')) {
3109 if (p == 0)
3110 failExpecting("scheme name", 0);
3111 checkChar(0, L_ALPHA, H_ALPHA, "scheme name");
3112 checkChars(1, p, L_SCHEME, H_SCHEME, "scheme name");
3113 scheme = input.substring(0, p);
3114 p++; // Skip ':'
3115 if (at(p, n, '/')) {
3116 p = parseHierarchical(p, n);
3117 } else {
3118 // opaque; need to create the schemeSpecificPart
3119 int q = scan(p, n, "#");
3120 if (q <= p)
3121 failExpecting("scheme-specific part", p);
3122 checkChars(p, q, L_URIC, H_URIC, "opaque part");
3123 schemeSpecificPart = input.substring(p, q);
3124 p = q;
3125 }
3126 } else {
3127 p = parseHierarchical(0, n);
3128 }
3129 if (at(p, n, '#')) {
3130 checkChars(p + 1, n, L_URIC, H_URIC, "fragment");
3131 fragment = input.substring(p + 1, n);
3132 p = n;
3133 }
3134 if (p < n)
3135 fail("end of URI", p);
3136 }
3137
3138 // [//authority]<path>[?<query>]
3139 //
3140 // DEVIATION from RFC2396: We allow an empty authority component as
3141 // long as it's followed by a non-empty path, query component, or
3142 // fragment component. This is so that URIs such as "file:///foo/bar"
3143 // will parse. This seems to be the intent of RFC2396, though the
3144 // grammar does not permit it. If the authority is empty then the
3145 // userInfo, host, and port components are undefined.
3146 //
3147 // DEVIATION from RFC2396: We allow empty relative paths. This seems
3148 // to be the intent of RFC2396, but the grammar does not permit it.
3149 // The primary consequence of this deviation is that "#f" parses as a
3150 // relative URI with an empty path.
3151 //
3152 private int parseHierarchical(int start, int n)
3153 throws URISyntaxException
3154 {
3155 int p = start;
3156 if (at(p, n, '/') && at(p + 1, n, '/')) {
3157 p += 2;
3158 int q = scan(p, n, "/?#");
3159 if (q > p) {
3160 p = parseAuthority(p, q);
3161 } else if (q < n) {
3162 // DEVIATION: Allow empty authority prior to non-empty
3163 // path, query component or fragment identifier
3164 } else
3165 failExpecting("authority", p);
3166 }
3167 int q = scan(p, n, "?#"); // DEVIATION: May be empty
3168 checkChars(p, q, L_PATH, H_PATH, "path");
3169 path = input.substring(p, q);
3170 p = q;
3171 if (at(p, n, '?')) {
3172 p++;
3173 q = scan(p, n, "#");
3174 checkChars(p, q, L_URIC, H_URIC, "query");
3175 query = input.substring(p, q);
3176 p = q;
3177 }
3178 return p;
3179 }
3180
3181 // authority = server | reg_name
3182 //
3183 // Ambiguity: An authority that is a registry name rather than a server
3184 // might have a prefix that parses as a server. We use the fact that
3185 // the authority component is always followed by '/' or the end of the
3186 // input string to resolve this: If the complete authority did not
3187 // parse as a server then we try to parse it as a registry name.
3188 //
3189 private int parseAuthority(int start, int n)
3190 throws URISyntaxException
3191 {
3192 int p = start;
3193 int q = p;
3194 URISyntaxException ex = null;
3195
3196 boolean serverChars;
3197 boolean regChars;
3198
3199 if (scan(p, n, "]") > p) {
3200 // contains a literal IPv6 address, therefore % is allowed
3201 serverChars = (scan(p, n, L_SERVER_PERCENT, H_SERVER_PERCENT) == n);
3202 } else {
3203 serverChars = (scan(p, n, L_SERVER, H_SERVER) == n);
3204 }
3205 regChars = (scan(p, n, L_REG_NAME, H_REG_NAME) == n);
3206
3207 if (regChars && !serverChars) {
3208 // Must be a registry-based authority
3209 authority = input.substring(p, n);
3210 return n;
3211 }
3212
3213 if (serverChars) {
3214 // Might be (probably is) a server-based authority, so attempt
3215 // to parse it as such. If the attempt fails, try to treat it
3216 // as a registry-based authority.
3217 try {
3218 q = parseServer(p, n);
3219 if (q < n)
3220 failExpecting("end of authority", q);
3221 authority = input.substring(p, n);
3222 } catch (URISyntaxException x) {
3223 // Undo results of failed parse
3224 userInfo = null;
3225 host = null;
3226 port = -1;
3227 if (requireServerAuthority) {
3228 // If we're insisting upon a server-based authority,
3229 // then just re-throw the exception
3230 throw x;
3231 } else {
3232 // Save the exception in case it doesn't parse as a
3233 // registry either
3234 ex = x;
3235 q = p;
3236 }
3237 }
3238 }
3239
3240 if (q < n) {
3241 if (regChars) {
3242 // Registry-based authority
3243 authority = input.substring(p, n);
3244 } else if (ex != null) {
3245 // Re-throw exception; it was probably due to
3246 // a malformed IPv6 address
3247 throw ex;
3248 } else {
3249 fail("Illegal character in authority", q);
3250 }
3251 }
3252
3253 return n;
3254 }
3255
3256
3257 // [<userinfo>@]<host>[:<port>]
3258 //
3259 private int parseServer(int start, int n)
3260 throws URISyntaxException
3261 {
3262 int p = start;
3263 int q;
3264
3265 // userinfo
3266 q = scan(p, n, "/?#", "@");
3267 if ((q >= p) && at(q, n, '@')) {
3268 checkChars(p, q, L_USERINFO, H_USERINFO, "user info");
3269 userInfo = input.substring(p, q);
3270 p = q + 1; // Skip '@'
3271 }
3272
3273 // hostname, IPv4 address, or IPv6 address
3274 if (at(p, n, '[')) {
3275 // DEVIATION from RFC2396: Support IPv6 addresses, per RFC2732
3276 p++;
3277 q = scan(p, n, "/?#", "]");
3278 if ((q > p) && at(q, n, ']')) {
3279 // look for a "%" scope id
3280 int r = scan (p, q, "%");
3281 if (r > p) {
3282 parseIPv6Reference(p, r);
3283 if (r+1 == q) {
3284 fail ("scope id expected");
3285 }
3286 checkChars (r+1, q, L_SCOPE_ID, H_SCOPE_ID,
3287 "scope id");
3288 } else {
3289 parseIPv6Reference(p, q);
3290 }
3291 host = input.substring(p-1, q+1);
3292 p = q + 1;
3293 } else {
3294 failExpecting("closing bracket for IPv6 address", q);
3295 }
3296 } else {
3297 q = parseIPv4Address(p, n);
3298 if (q <= p)
3299 q = parseHostname(p, n);
3300 p = q;
3301 }
3302
3303 // port
3304 if (at(p, n, ':')) {
3305 p++;
3306 q = scan(p, n, "/");
3307 if (q > p) {
3308 checkChars(p, q, L_DIGIT, H_DIGIT, "port number");
3309 try {
3310 port = Integer.parseInt(input, p, q, 10);
3311 } catch (NumberFormatException x) {
3312 fail("Malformed port number", p);
3313 }
3314 p = q;
3315 }
3316 }
3317 if (p < n)
3318 failExpecting("port number", p);
3319
3320 return p;
3321 }
3322
3323 // Scan a string of decimal digits whose value fits in a byte
3324 //
3325 private int scanByte(int start, int n)
3326 throws URISyntaxException
3327 {
3328 int p = start;
3329 int q = scan(p, n, L_DIGIT, H_DIGIT);
3330 if (q <= p) return q;
3331 if (Integer.parseInt(input, p, q, 10) > 255) return p;
3332 return q;
3333 }
3334
3335 // Scan an IPv4 address.
3336 //
3337 // If the strict argument is true then we require that the given
3338 // interval contain nothing besides an IPv4 address; if it is false
3339 // then we only require that it start with an IPv4 address.
3340 //
3341 // If the interval does not contain or start with (depending upon the
3342 // strict argument) a legal IPv4 address characters then we return -1
3343 // immediately; otherwise we insist that these characters parse as a
3344 // legal IPv4 address and throw an exception on failure.
3345 //
3346 // We assume that any string of decimal digits and dots must be an IPv4
3347 // address. It won't parse as a hostname anyway, so making that
3348 // assumption here allows more meaningful exceptions to be thrown.
3349 //
3350 private int scanIPv4Address(int start, int n, boolean strict)
3351 throws URISyntaxException
3352 {
3353 int p = start;
3354 int q;
3355 int m = scan(p, n, L_DIGIT | L_DOT, H_DIGIT | H_DOT);
3356 if ((m <= p) || (strict && (m != n)))
3357 return -1;
3358 for (;;) {
3359 // Per RFC2732: At most three digits per byte
3360 // Further constraint: Each element fits in a byte
3361 if ((q = scanByte(p, m)) <= p) break; p = q;
3362 if ((q = scan(p, m, '.')) <= p) break; p = q;
3363 if ((q = scanByte(p, m)) <= p) break; p = q;
3364 if ((q = scan(p, m, '.')) <= p) break; p = q;
3365 if ((q = scanByte(p, m)) <= p) break; p = q;
3366 if ((q = scan(p, m, '.')) <= p) break; p = q;
3367 if ((q = scanByte(p, m)) <= p) break; p = q;
3368 if (q < m) break;
3369 return q;
3370 }
3371 fail("Malformed IPv4 address", q);
3372 return -1;
3373 }
3374
3375 // Take an IPv4 address: Throw an exception if the given interval
3376 // contains anything except an IPv4 address
3377 //
3378 private int takeIPv4Address(int start, int n, String expected)
3379 throws URISyntaxException
3380 {
3381 int p = scanIPv4Address(start, n, true);
3382 if (p <= start)
3383 failExpecting(expected, start);
3384 return p;
3385 }
3386
3387 // Attempt to parse an IPv4 address, returning -1 on failure but
3388 // allowing the given interval to contain [:<characters>] after
3389 // the IPv4 address.
3390 //
3391 private int parseIPv4Address(int start, int n) {
3392 int p;
3393
3394 try {
3395 p = scanIPv4Address(start, n, false);
3396 } catch (URISyntaxException x) {
3397 return -1;
3398 } catch (NumberFormatException nfe) {
3399 return -1;
3400 }
3401
3402 if (p > start && p < n) {
3403 // IPv4 address is followed by something - check that
3404 // it's a ":" as this is the only valid character to
3405 // follow an address.
3406 if (input.charAt(p) != ':') {
3407 p = -1;
3408 }
3409 }
3410
3411 if (p > start)
3412 host = input.substring(start, p);
3413
3414 return p;
3415 }
3416
3417 // hostname = domainlabel [ "." ] | 1*( domainlabel "." ) toplabel [ "." ]
3418 // domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum
3419 // toplabel = alpha | alpha *( alphanum | "-" ) alphanum
3420 //
3421 private int parseHostname(int start, int n)
3422 throws URISyntaxException
3423 {
3424 int p = start;
3425 int q;
3426 int l = -1; // Start of last parsed label
3427
3428 do {
3429 // domainlabel = alphanum [ *( alphanum | "-" ) alphanum ]
3430 q = scan(p, n, L_ALPHANUM, H_ALPHANUM);
3431 if (q <= p)
3432 break;
3433 l = p;
3434 if (q > p) {
3435 p = q;
3436 q = scan(p, n, L_ALPHANUM | L_DASH, H_ALPHANUM | H_DASH);
3437 if (q > p) {
3438 if (input.charAt(q - 1) == '-')
3439 fail("Illegal character in hostname", q - 1);
3440 p = q;
3441 }
3442 }
3443 q = scan(p, n, '.');
3444 if (q <= p)
3445 break;
3446 p = q;
3447 } while (p < n);
3448
3449 if ((p < n) && !at(p, n, ':'))
3450 fail("Illegal character in hostname", p);
3451
3452 if (l < 0)
3453 failExpecting("hostname", start);
3454
3455 // for a fully qualified hostname check that the rightmost
3456 // label starts with an alpha character.
3457 if (l > start && !match(input.charAt(l), L_ALPHA, H_ALPHA)) {
3458 fail("Illegal character in hostname", l);
3459 }
3460
3461 host = input.substring(start, p);
3462 return p;
3463 }
3464
3465
3466 // IPv6 address parsing, from RFC2373: IPv6 Addressing Architecture
3467 //
3468 // Bug: The grammar in RFC2373 Appendix B does not allow addresses of
3469 // the form ::12.34.56.78, which are clearly shown in the examples
3470 // earlier in the document. Here is the original grammar:
3471 //
3472 // IPv6address = hexpart [ ":" IPv4address ]
3473 // hexpart = hexseq | hexseq "::" [ hexseq ] | "::" [ hexseq ]
3474 // hexseq = hex4 *( ":" hex4)
3475 // hex4 = 1*4HEXDIG
3476 //
3477 // We therefore use the following revised grammar:
3478 //
3479 // IPv6address = hexseq [ ":" IPv4address ]
3480 // | hexseq [ "::" [ hexpost ] ]
3481 // | "::" [ hexpost ]
3482 // hexpost = hexseq | hexseq ":" IPv4address | IPv4address
3483 // hexseq = hex4 *( ":" hex4)
3484 // hex4 = 1*4HEXDIG
3485 //
3486 // This covers all and only the following cases:
3487 //
3488 // hexseq
3489 // hexseq : IPv4address
3490 // hexseq ::
3491 // hexseq :: hexseq
3492 // hexseq :: hexseq : IPv4address
3493 // hexseq :: IPv4address
3494 // :: hexseq
3495 // :: hexseq : IPv4address
3496 // :: IPv4address
3497 // ::
3498 //
3499 // Additionally we constrain the IPv6 address as follows :-
3500 //
3501 // i. IPv6 addresses without compressed zeros should contain
3502 // exactly 16 bytes.
3503 //
3504 // ii. IPv6 addresses with compressed zeros should contain
3505 // less than 16 bytes.
3506
3507 private int ipv6byteCount = 0;
3508
3509 private int parseIPv6Reference(int start, int n)
3510 throws URISyntaxException
3511 {
3512 int p = start;
3513 int q;
3514 boolean compressedZeros = false;
3515
3516 q = scanHexSeq(p, n);
3517
3518 if (q > p) {
3519 p = q;
3520 if (at(p, n, "::")) {
3521 compressedZeros = true;
3522 p = scanHexPost(p + 2, n);
3523 } else if (at(p, n, ':')) {
3524 p = takeIPv4Address(p + 1, n, "IPv4 address");
3525 ipv6byteCount += 4;
3526 }
3527 } else if (at(p, n, "::")) {
3528 compressedZeros = true;
3529 p = scanHexPost(p + 2, n);
3530 }
3531 if (p < n)
3532 fail("Malformed IPv6 address", start);
3533 if (ipv6byteCount > 16)
3534 fail("IPv6 address too long", start);
3535 if (!compressedZeros && ipv6byteCount < 16)
3536 fail("IPv6 address too short", start);
3537 if (compressedZeros && ipv6byteCount == 16)
3538 fail("Malformed IPv6 address", start);
3539
3540 return p;
3541 }
3542
3543 private int scanHexPost(int start, int n)
3544 throws URISyntaxException
3545 {
3546 int p = start;
3547 int q;
3548
3549 if (p == n)
3550 return p;
3551
3552 q = scanHexSeq(p, n);
3553 if (q > p) {
3554 p = q;
3555 if (at(p, n, ':')) {
3556 p++;
3557 p = takeIPv4Address(p, n, "hex digits or IPv4 address");
3558 ipv6byteCount += 4;
3559 }
3560 } else {
3561 p = takeIPv4Address(p, n, "hex digits or IPv4 address");
3562 ipv6byteCount += 4;
3563 }
3564 return p;
3565 }
3566
3567 // Scan a hex sequence; return -1 if one could not be scanned
3568 //
3569 private int scanHexSeq(int start, int n)
3570 throws URISyntaxException
3571 {
3572 int p = start;
3573 int q;
3574
3575 q = scan(p, n, L_HEX, H_HEX);
3576 if (q <= p)
3577 return -1;
3578 if (at(q, n, '.')) // Beginning of IPv4 address
3579 return -1;
3580 if (q > p + 4)
3581 fail("IPv6 hexadecimal digit sequence too long", p);
3582 ipv6byteCount += 2;
3583 p = q;
3584 while (p < n) {
3585 if (!at(p, n, ':'))
3586 break;
3587 if (at(p + 1, n, ':'))
3588 break; // "::"
3589 p++;
3590 q = scan(p, n, L_HEX, H_HEX);
3591 if (q <= p)
3592 failExpecting("digits for an IPv6 address", p);
3593 if (at(q, n, '.')) { // Beginning of IPv4 address
3594 p--;
3595 break;
3596 }
3597 if (q > p + 4)
3598 fail("IPv6 hexadecimal digit sequence too long", p);
3599 ipv6byteCount += 2;
3600 p = q;
3601 }
3602
3603 return p;
3604 }
3605
3606 }
3607 static {
3608 SharedSecrets.setJavaNetUriAccess(
3609 new JavaNetUriAccess() {
3610 public URI create(String scheme, String path) {
3611 return new URI(scheme, path);
3612 }
3613 }
3614 );
3615 }
3616 }