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