1 /*
2 * Copyright (c) 2000, 2018, 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.access.JavaNetUriAccess;
41 import jdk.internal.access.SharedSecrets;
42 import sun.nio.cs.ThreadLocalCoders;
43
44 import java.lang.Character; // for javadoc
45 import java.lang.NullPointerException; // for javadoc
46
47
48 /**
49 * Represents a Uniform Resource Identifier (URI) reference.
50 *
51 * <p> Aside from some minor deviations noted below, an instance of this
52 * class represents a URI reference as defined by
53 * <a href="http://www.ietf.org/rfc/rfc2396.txt"><i>RFC 2396: Uniform
54 * Resource Identifiers (URI): Generic Syntax</i></a>, amended by <a
55 * href="http://www.ietf.org/rfc/rfc2732.txt"><i>RFC 2732: Format for
56 * Literal IPv6 Addresses in URLs</i></a>. The Literal IPv6 address format
57 * also supports scope_ids. The syntax and usage of scope_ids is described
58 * <a href="Inet6Address.html#scoped">here</a>.
59 * This class provides constructors for creating URI instances from
60 * their components or by parsing their string forms, methods for accessing the
61 * various components of an instance, and methods for normalizing, resolving,
62 * and relativizing URI instances. Instances of this class are immutable.
63 *
64 *
65 * <h3> URI syntax and components </h3>
66 *
67 * At the highest level a URI reference (hereinafter simply "URI") in string
68 * form has the syntax
69 *
70 * <blockquote>
71 * [<i>scheme</i><b>{@code :}</b>]<i>scheme-specific-part</i>[<b>{@code #}</b><i>fragment</i>]
72 * </blockquote>
73 *
74 * where square brackets [...] delineate optional components and the characters
75 * <b>{@code :}</b> and <b>{@code #}</b> stand for themselves.
76 *
77 * <p> An <i>absolute</i> URI specifies a scheme; a URI that is not absolute is
78 * said to be <i>relative</i>. URIs are also classified according to whether
79 * they are <i>opaque</i> or <i>hierarchical</i>.
80 *
81 * <p> An <i>opaque</i> URI is an absolute URI whose scheme-specific part does
82 * not begin with a slash character ({@code '/'}). Opaque URIs are not
83 * subject to further parsing. Some examples of opaque URIs are:
84 *
85 * <blockquote><ul style="list-style-type:none">
86 * <li>{@code mailto:java-net@java.sun.com}</li>
87 * <li>{@code news:comp.lang.java}</li>
88 * <li>{@code urn:isbn:096139210x}</li>
89 * </ul></blockquote>
90 *
91 * <p> A <i>hierarchical</i> URI is either an absolute URI whose
92 * scheme-specific part begins with a slash character, or a relative URI, that
93 * is, a URI that does not specify a scheme. Some examples of hierarchical
94 * URIs are:
95 *
96 * <blockquote>
97 * {@code http://example.com/languages/java/}<br>
98 * {@code sample/a/index.html#28}<br>
99 * {@code ../../demo/b/index.html}<br>
100 * {@code file:///~/calendar}
101 * </blockquote>
102 *
103 * <p> A hierarchical URI is subject to further parsing according to the syntax
104 *
105 * <blockquote>
106 * [<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>]
107 * </blockquote>
108 *
109 * where the characters <b>{@code :}</b>, <b>{@code /}</b>,
110 * <b>{@code ?}</b>, and <b>{@code #}</b> stand for themselves. The
111 * scheme-specific part of a hierarchical URI consists of the characters
112 * between the scheme and fragment components.
113 *
114 * <p> The authority component of a hierarchical URI is, if specified, either
115 * <i>server-based</i> or <i>registry-based</i>. A server-based authority
116 * parses according to the familiar syntax
117 *
118 * <blockquote>
119 * [<i>user-info</i><b>{@code @}</b>]<i>host</i>[<b>{@code :}</b><i>port</i>]
120 * </blockquote>
121 *
122 * where the characters <b>{@code @}</b> and <b>{@code :}</b> stand for
123 * themselves. Nearly all URI schemes currently in use are server-based. An
124 * authority component that does not parse in this way is considered to be
125 * registry-based.
126 *
127 * <p> The path component of a hierarchical URI is itself said to be absolute
128 * if it begins with a slash character ({@code '/'}); otherwise it is
129 * relative. The path of a hierarchical URI that is either absolute or
130 * specifies an authority is always absolute.
131 *
132 * <p> All told, then, a URI instance has the following nine components:
133 *
134 * <table class="striped" style="margin-left:2em">
135 * <caption style="display:none">Describes the components of a URI:scheme,scheme-specific-part,authority,user-info,host,port,path,query,fragment</caption>
136 * <thead>
137 * <tr><th scope="col">Component</th><th scope="col">Type</th></tr>
138 * </thead>
139 * <tbody style="text-align:left">
140 * <tr><th scope="row">scheme</th><td>{@code String}</td></tr>
141 * <tr><th scope="row">scheme-specific-part</th><td>{@code String}</td></tr>
142 * <tr><th scope="row">authority</th><td>{@code String}</td></tr>
143 * <tr><th scope="row">user-info</th><td>{@code String}</td></tr>
144 * <tr><th scope="row">host</th><td>{@code String}</td></tr>
145 * <tr><th scope="row">port</th><td>{@code int}</td></tr>
146 * <tr><th scope="row">path</th><td>{@code String}</td></tr>
147 * <tr><th scope="row">query</th><td>{@code String}</td></tr>
148 * <tr><th scope="row">fragment</th><td>{@code String}</td></tr>
149 * </tbody>
150 * </table>
151 *
152 * In a given instance any particular component is either <i>undefined</i> or
153 * <i>defined</i> with a distinct value. Undefined string components are
154 * represented by {@code null}, while undefined integer components are
155 * represented by {@code -1}. A string component may be defined to have the
156 * empty string as its value; this is not equivalent to that component being
157 * undefined.
158 *
159 * <p> Whether a particular component is or is not defined in an instance
160 * depends upon the type of the URI being represented. An absolute URI has a
161 * scheme component. An opaque URI has a scheme, a scheme-specific part, and
162 * possibly a fragment, but has no other components. A hierarchical URI always
163 * has a path (though it may be empty) and a scheme-specific-part (which at
164 * least contains the path), and may have any of the other components. If the
165 * authority component is present and is server-based then the host component
166 * will be defined and the user-information and port components may be defined.
167 *
168 *
169 * <h4> Operations on URI instances </h4>
170 *
171 * The key operations supported by this class are those of
172 * <i>normalization</i>, <i>resolution</i>, and <i>relativization</i>.
173 *
174 * <p> <i>Normalization</i> is the process of removing unnecessary {@code "."}
175 * and {@code ".."} segments from the path component of a hierarchical URI.
176 * Each {@code "."} segment is simply removed. A {@code ".."} segment is
177 * removed only if it is preceded by a non-{@code ".."} segment.
178 * Normalization has no effect upon opaque URIs.
179 *
180 * <p> <i>Resolution</i> is the process of resolving one URI against another,
181 * <i>base</i> URI. The resulting URI is constructed from components of both
182 * URIs in the manner specified by RFC 2396, taking components from the
183 * base URI for those not specified in the original. For hierarchical URIs,
184 * the path of the original is resolved against the path of the base and then
185 * normalized. The result, for example, of resolving
186 *
187 * <blockquote>
188 * {@code sample/a/index.html#28}
189 *
190 * (1)
191 * </blockquote>
192 *
193 * against the base URI {@code http://example.com/languages/java/} is the result
194 * URI
195 *
196 * <blockquote>
197 * {@code http://example.com/languages/java/sample/a/index.html#28}
198 * </blockquote>
199 *
200 * Resolving the relative URI
201 *
202 * <blockquote>
203 * {@code ../../demo/b/index.html} (2)
204 * </blockquote>
205 *
206 * against this result yields, in turn,
207 *
208 * <blockquote>
209 * {@code http://example.com/languages/java/demo/b/index.html}
210 * </blockquote>
211 *
212 * Resolution of both absolute and relative URIs, and of both absolute and
213 * relative paths in the case of hierarchical URIs, is supported. Resolving
214 * the URI {@code file:///~calendar} against any other URI simply yields the
215 * original URI, since it is absolute. Resolving the relative URI (2) above
216 * against the relative base URI (1) yields the normalized, but still relative,
217 * URI
218 *
219 * <blockquote>
220 * {@code demo/b/index.html}
221 * </blockquote>
222 *
223 * <p> <i>Relativization</i>, finally, is the inverse of resolution: For any
224 * two normalized URIs <i>u</i> and <i>v</i>,
225 *
226 * <blockquote>
227 * <i>u</i>{@code .relativize(}<i>u</i>{@code .resolve(}<i>v</i>{@code )).equals(}<i>v</i>{@code )} and<br>
228 * <i>u</i>{@code .resolve(}<i>u</i>{@code .relativize(}<i>v</i>{@code )).equals(}<i>v</i>{@code )} .<br>
229 * </blockquote>
230 *
231 * This operation is often useful when constructing a document containing URIs
232 * that must be made relative to the base URI of the document wherever
233 * possible. For example, relativizing the URI
234 *
235 * <blockquote>
236 * {@code http://example.com/languages/java/sample/a/index.html#28}
237 * </blockquote>
238 *
239 * against the base URI
240 *
241 * <blockquote>
242 * {@code http://example.com/languages/java/}
243 * </blockquote>
244 *
245 * yields the relative URI {@code sample/a/index.html#28}.
246 *
247 *
248 * <h4> Character categories </h4>
249 *
250 * RFC 2396 specifies precisely which characters are permitted in the
251 * various components of a URI reference. The following categories, most of
252 * which are taken from that specification, are used below to describe these
253 * constraints:
254 *
255 * <table class="striped" style="margin-left:2em">
256 * <caption style="display:none">Describes categories alpha,digit,alphanum,unreserved,punct,reserved,escaped,and other</caption>
257 * <thead>
258 * <tr><th scope="col">Category</th><th scope="col">Description</th></tr>
259 * </thead>
260 * <tbody style="text-align:left">
261 * <tr><th scope="row" style="vertical-align:top">alpha</th>
262 * <td>The US-ASCII alphabetic characters,
263 * {@code 'A'} through {@code 'Z'}
264 * and {@code 'a'} through {@code 'z'}</td></tr>
265 * <tr><th scope="row" style="vertical-align:top">digit</th>
266 * <td>The US-ASCII decimal digit characters,
267 * {@code '0'} through {@code '9'}</td></tr>
268 * <tr><th scope="row" style="vertical-align:top">alphanum</th>
269 * <td>All <i>alpha</i> and <i>digit</i> characters</td></tr>
270 * <tr><th scope="row" style="vertical-align:top">unreserved</th>
271 * <td>All <i>alphanum</i> characters together with those in the string
272 * {@code "_-!.~'()*"}</td></tr>
273 * <tr><th scope="row" style="vertical-align:top">punct</th>
274 * <td>The characters in the string {@code ",;:$&+="}</td></tr>
275 * <tr><th scope="row" style="vertical-align:top">reserved</th>
276 * <td>All <i>punct</i> characters together with those in the string
277 * {@code "?/[]@"}</td></tr>
278 * <tr><th scope="row" style="vertical-align:top">escaped</th>
279 * <td>Escaped octets, that is, triplets consisting of the percent
280 * character ({@code '%'}) followed by two hexadecimal digits
281 * ({@code '0'}-{@code '9'}, {@code 'A'}-{@code 'F'}, and
282 * {@code 'a'}-{@code 'f'})</td></tr>
283 * <tr><th scope="row" style="vertical-align:top">other</th>
284 * <td>The Unicode characters that are not in the US-ASCII character set,
285 * are not control characters (according to the {@link
286 * java.lang.Character#isISOControl(char) Character.isISOControl}
287 * method), and are not space characters (according to the {@link
288 * java.lang.Character#isSpaceChar(char) Character.isSpaceChar}
289 * method) <i>(<b>Deviation from RFC 2396</b>, which is
290 * limited to US-ASCII)</i></td></tr>
291 * </tbody>
292 * </table>
293 *
294 * <p><a id="legal-chars"></a> The set of all legal URI characters consists of
295 * the <i>unreserved</i>, <i>reserved</i>, <i>escaped</i>, and <i>other</i>
296 * characters.
297 *
298 *
299 * <h4> Escaped octets, quotation, encoding, and decoding </h4>
300 *
301 * RFC 2396 allows escaped octets to appear in the user-info, path, query, and
302 * fragment components. Escaping serves two purposes in URIs:
303 *
304 * <ul>
305 *
306 * <li><p> To <i>encode</i> non-US-ASCII characters when a URI is required to
307 * conform strictly to RFC 2396 by not containing any <i>other</i>
308 * characters. </p></li>
309 *
310 * <li><p> To <i>quote</i> characters that are otherwise illegal in a
311 * component. The user-info, path, query, and fragment components differ
312 * slightly in terms of which characters are considered legal and illegal.
313 * </p></li>
314 *
315 * </ul>
316 *
317 * These purposes are served in this class by three related operations:
318 *
319 * <ul>
320 *
321 * <li><p><a id="encode"></a> A character is <i>encoded</i> by replacing it
322 * with the sequence of escaped octets that represent that character in the
323 * UTF-8 character set. The Euro currency symbol ({@code '\u005Cu20AC'}),
324 * for example, is encoded as {@code "%E2%82%AC"}. <i>(<b>Deviation from
325 * RFC 2396</b>, which does not specify any particular character
326 * set.)</i> </p></li>
327 *
328 * <li><p><a id="quote"></a> An illegal character is <i>quoted</i> simply by
329 * encoding it. The space character, for example, is quoted by replacing it
330 * with {@code "%20"}. UTF-8 contains US-ASCII, hence for US-ASCII
331 * characters this transformation has exactly the effect required by
332 * RFC 2396. </p></li>
333 *
334 * <li><p><a id="decode"></a>
335 * A sequence of escaped octets is <i>decoded</i> by
336 * replacing it with the sequence of characters that it represents in the
337 * UTF-8 character set. UTF-8 contains US-ASCII, hence decoding has the
338 * effect of de-quoting any quoted US-ASCII characters as well as that of
339 * decoding any encoded non-US-ASCII characters. If a <a
340 * href="../nio/charset/CharsetDecoder.html#ce">decoding error</a> occurs
341 * when decoding the escaped octets then the erroneous octets are replaced by
342 * {@code '\u005CuFFFD'}, the Unicode replacement character. </p></li>
343 *
344 * </ul>
345 *
346 * These operations are exposed in the constructors and methods of this class
347 * as follows:
348 *
349 * <ul>
350 *
351 * <li><p> The {@linkplain #URI(java.lang.String) single-argument
352 * constructor} requires any illegal characters in its argument to be
353 * quoted and preserves any escaped octets and <i>other</i> characters that
354 * are present. </p></li>
355 *
356 * <li><p> The {@linkplain
357 * #URI(java.lang.String,java.lang.String,java.lang.String,int,java.lang.String,java.lang.String,java.lang.String)
358 * multi-argument constructors} quote illegal characters as
359 * required by the components in which they appear. The percent character
360 * ({@code '%'}) is always quoted by these constructors. Any <i>other</i>
361 * characters are preserved. </p></li>
362 *
363 * <li><p> The {@link #getRawUserInfo() getRawUserInfo}, {@link #getRawPath()
364 * getRawPath}, {@link #getRawQuery() getRawQuery}, {@link #getRawFragment()
365 * getRawFragment}, {@link #getRawAuthority() getRawAuthority}, and {@link
366 * #getRawSchemeSpecificPart() getRawSchemeSpecificPart} methods return the
367 * values of their corresponding components in raw form, without interpreting
368 * any escaped octets. The strings returned by these methods may contain
369 * both escaped octets and <i>other</i> characters, and will not contain any
370 * illegal characters. </p></li>
371 *
372 * <li><p> The {@link #getUserInfo() getUserInfo}, {@link #getPath()
373 * getPath}, {@link #getQuery() getQuery}, {@link #getFragment()
374 * getFragment}, {@link #getAuthority() getAuthority}, and {@link
375 * #getSchemeSpecificPart() getSchemeSpecificPart} methods decode any escaped
376 * octets in their corresponding components. The strings returned by these
377 * methods may contain both <i>other</i> characters and illegal characters,
378 * and will not contain any escaped octets. </p></li>
379 *
380 * <li><p> The {@link #toString() toString} method returns a URI string with
381 * all necessary quotation but which may contain <i>other</i> characters.
382 * </p></li>
383 *
384 * <li><p> The {@link #toASCIIString() toASCIIString} method returns a fully
385 * quoted and encoded URI string that does not contain any <i>other</i>
386 * characters. </p></li>
387 *
388 * </ul>
389 *
390 *
391 * <h4> Identities </h4>
392 *
393 * For any URI <i>u</i>, it is always the case that
394 *
395 * <blockquote>
396 * {@code new URI(}<i>u</i>{@code .toString()).equals(}<i>u</i>{@code )} .
397 * </blockquote>
398 *
399 * For any URI <i>u</i> that does not contain redundant syntax such as two
400 * slashes before an empty authority (as in {@code file:///tmp/} ) or a
401 * colon following a host name but no port (as in
402 * {@code http://java.sun.com:} ), and that does not encode characters
403 * except those that must be quoted, the following identities also hold:
404 * <pre>
405 * new URI(<i>u</i>.getScheme(),
406 * <i>u</i>.getSchemeSpecificPart(),
407 * <i>u</i>.getFragment())
408 * .equals(<i>u</i>)</pre>
409 * in all cases,
410 * <pre>
411 * new URI(<i>u</i>.getScheme(),
412 * <i>u</i>.getAuthority(),
413 * <i>u</i>.getPath(), <i>u</i>.getQuery(),
414 * <i>u</i>.getFragment())
415 * .equals(<i>u</i>)</pre>
416 * if <i>u</i> is hierarchical, and
417 * <pre>
418 * new URI(<i>u</i>.getScheme(),
419 * <i>u</i>.getUserInfo(), <i>u</i>.getHost(), <i>u</i>.getPort(),
420 * <i>u</i>.getPath(), <i>u</i>.getQuery(),
421 * <i>u</i>.getFragment())
422 * .equals(<i>u</i>)</pre>
423 * if <i>u</i> is hierarchical and has either no authority or a server-based
424 * authority.
425 *
426 *
427 * <h4> URIs, URLs, and URNs </h4>
428 *
429 * A URI is a uniform resource <i>identifier</i> while a URL is a uniform
430 * resource <i>locator</i>. Hence every URL is a URI, abstractly speaking, but
431 * not every URI is a URL. This is because there is another subcategory of
432 * URIs, uniform resource <i>names</i> (URNs), which name resources but do not
433 * specify how to locate them. The {@code mailto}, {@code news}, and
434 * {@code isbn} URIs shown above are examples of URNs.
435 *
436 * <p> The conceptual distinction between URIs and URLs is reflected in the
437 * differences between this class and the {@link URL} class.
438 *
439 * <p> An instance of this class represents a URI reference in the syntactic
440 * sense defined by RFC 2396. A URI may be either absolute or relative.
441 * A URI string is parsed according to the generic syntax without regard to the
442 * scheme, if any, that it specifies. No lookup of the host, if any, is
443 * performed, and no scheme-dependent stream handler is constructed. Equality,
444 * hashing, and comparison are defined strictly in terms of the character
445 * content of the instance. In other words, a URI instance is little more than
446 * a structured string that supports the syntactic, scheme-independent
447 * operations of comparison, normalization, resolution, and relativization.
448 *
449 * <p> An instance of the {@link URL} class, by contrast, represents the
450 * syntactic components of a URL together with some of the information required
451 * to access the resource that it describes. A URL must be absolute, that is,
452 * it must always specify a scheme. A URL string is parsed according to its
453 * scheme. A stream handler is always established for a URL, and in fact it is
454 * impossible to create a URL instance for a scheme for which no handler is
455 * available. Equality and hashing depend upon both the scheme and the
456 * Internet address of the host, if any; comparison is not defined. In other
457 * words, a URL is a structured string that supports the syntactic operation of
458 * resolution as well as the network I/O operations of looking up the host and
459 * opening a connection to the specified resource.
460 *
461 *
462 * @author Mark Reinhold
463 * @since 1.4
464 *
465 * @see <a href="http://www.ietf.org/rfc/rfc2279.txt"><i>RFC 2279: UTF-8, a
466 * transformation format of ISO 10646</i></a>, <br><a
467 * href="http://www.ietf.org/rfc/rfc2373.txt"><i>RFC 2373: IPv6 Addressing
468 * Architecture</i></a>, <br><a
469 * href="http://www.ietf.org/rfc/rfc2396.txt"><i>RFC 2396: Uniform
470 * Resource Identifiers (URI): Generic Syntax</i></a>, <br><a
471 * href="http://www.ietf.org/rfc/rfc2732.txt"><i>RFC 2732: Format for
472 * Literal IPv6 Addresses in URLs</i></a>, <br><a
473 * href="URISyntaxException.html">URISyntaxException</a>
474 */
475
476 public final class URI
477 implements Comparable<URI>, Serializable
478 {
479
480 // Note: Comments containing the word "ASSERT" indicate places where a
481 // throw of an InternalError should be replaced by an appropriate assertion
482 // statement once asserts are enabled in the build.
483
484 static final long serialVersionUID = -6052424284110960213L;
485
486
487 // -- Properties and components of this instance --
488
489 // Components of all URIs: [<scheme>:]<scheme-specific-part>[#<fragment>]
490 private transient String scheme; // null ==> relative URI
491 private transient String fragment;
492
493 // Hierarchical URI components: [//<authority>]<path>[?<query>]
494 private transient String authority; // Registry or server
495
496 // Server-based authority: [<userInfo>@]<host>[:<port>]
497 private transient String userInfo;
498 private transient String host; // null ==> registry-based
499 private transient int port = -1; // -1 ==> undefined
500
501 // Remaining components of hierarchical URIs
502 private transient String path; // null ==> opaque
503 private transient String query;
504
505 // The remaining fields may be computed on demand, which is safe even in
506 // the face of multiple threads racing to initialize them
507 private transient String schemeSpecificPart;
508 private transient int hash; // Zero ==> undefined
509
510 private transient String decodedUserInfo;
511 private transient String decodedAuthority;
512 private transient String decodedPath;
513 private transient String decodedQuery;
514 private transient String decodedFragment;
515 private transient String decodedSchemeSpecificPart;
516
517 /**
518 * The string form of this URI.
519 *
520 * @serial
521 */
522 private volatile String string; // The only serializable field
523
524
525
526 // -- Constructors and factories --
527
528 private URI() { } // Used internally
529
530 /**
531 * Constructs a URI by parsing the given string.
532 *
533 * <p> This constructor parses the given string exactly as specified by the
534 * grammar in <a
535 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
536 * Appendix A, <b><i>except for the following deviations:</i></b> </p>
537 *
538 * <ul>
539 *
540 * <li><p> An empty authority component is permitted as long as it is
541 * followed by a non-empty path, a query component, or a fragment
542 * component. This allows the parsing of URIs such as
543 * {@code "file:///foo/bar"}, which seems to be the intent of
544 * RFC 2396 although the grammar does not permit it. If the
545 * authority component is empty then the user-information, host, and port
546 * components are undefined. </p></li>
547 *
548 * <li><p> Empty relative paths are permitted; this seems to be the
549 * intent of RFC 2396 although the grammar does not permit it. The
550 * primary consequence of this deviation is that a standalone fragment
551 * such as {@code "#foo"} parses as a relative URI with an empty path
552 * and the given fragment, and can be usefully <a
553 * href="#resolve-frag">resolved</a> against a base URI.
554 *
555 * <li><p> IPv4 addresses in host components are parsed rigorously, as
556 * specified by <a
557 * href="http://www.ietf.org/rfc/rfc2732.txt">RFC 2732</a>: Each
558 * element of a dotted-quad address must contain no more than three
559 * decimal digits. Each element is further constrained to have a value
560 * no greater than 255. </p></li>
561 *
562 * <li> <p> Hostnames in host components that comprise only a single
563 * domain label are permitted to start with an <i>alphanum</i>
564 * character. This seems to be the intent of <a
565 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>
566 * section 3.2.2 although the grammar does not permit it. The
567 * consequence of this deviation is that the authority component of a
568 * hierarchical URI such as {@code s://123}, will parse as a server-based
569 * authority. </p></li>
570 *
571 * <li><p> IPv6 addresses are permitted for the host component. An IPv6
572 * address must be enclosed in square brackets ({@code '['} and
573 * {@code ']'}) as specified by <a
574 * href="http://www.ietf.org/rfc/rfc2732.txt">RFC 2732</a>. The
575 * IPv6 address itself must parse according to <a
576 * href="http://www.ietf.org/rfc/rfc2373.txt">RFC 2373</a>. IPv6
577 * addresses are further constrained to describe no more than sixteen
578 * bytes of address information, a constraint implicit in RFC 2373
579 * but not expressible in the grammar. </p></li>
580 *
581 * <li><p> Characters in the <i>other</i> category are permitted wherever
582 * RFC 2396 permits <i>escaped</i> octets, that is, in the
583 * user-information, path, query, and fragment components, as well as in
584 * the authority component if the authority is registry-based. This
585 * allows URIs to contain Unicode characters beyond those in the US-ASCII
586 * character set. </p></li>
587 *
588 * </ul>
589 *
590 * @param str The string to be parsed into a URI
591 *
592 * @throws NullPointerException
593 * If {@code str} is {@code null}
594 *
595 * @throws URISyntaxException
596 * If the given string violates RFC 2396, as augmented
597 * by the above deviations
598 */
599 public URI(String str) throws URISyntaxException {
600 new Parser(str).parse(false);
601 }
602
603 /**
604 * Constructs a hierarchical URI from the given components.
605 *
606 * <p> If a scheme is given then the path, if also given, must either be
607 * empty or begin with a slash character ({@code '/'}). Otherwise a
608 * component of the new URI may be left undefined by passing {@code null}
609 * for the corresponding parameter or, in the case of the {@code port}
610 * parameter, by passing {@code -1}.
611 *
612 * <p> This constructor first builds a URI string from the given components
613 * according to the rules specified in <a
614 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
615 * section 5.2, step 7: </p>
616 *
617 * <ol>
618 *
619 * <li><p> Initially, the result string is empty. </p></li>
620 *
621 * <li><p> If a scheme is given then it is appended to the result,
622 * followed by a colon character ({@code ':'}). </p></li>
623 *
624 * <li><p> If user information, a host, or a port are given then the
625 * string {@code "//"} is appended. </p></li>
626 *
627 * <li><p> If user information is given then it is appended, followed by
628 * a commercial-at character ({@code '@'}). Any character not in the
629 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
630 * categories is <a href="#quote">quoted</a>. </p></li>
631 *
632 * <li><p> If a host is given then it is appended. If the host is a
633 * literal IPv6 address but is not enclosed in square brackets
634 * ({@code '['} and {@code ']'}) then the square brackets are added.
635 * </p></li>
636 *
637 * <li><p> If a port number is given then a colon character
638 * ({@code ':'}) is appended, followed by the port number in decimal.
639 * </p></li>
640 *
641 * <li><p> If a path is given then it is appended. Any character not in
642 * the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
643 * categories, and not equal to the slash character ({@code '/'}) or the
644 * commercial-at character ({@code '@'}), is quoted. </p></li>
645 *
646 * <li><p> If a query is given then a question-mark character
647 * ({@code '?'}) is appended, followed by the query. Any character that
648 * is not a <a href="#legal-chars">legal URI character</a> is quoted.
649 * </p></li>
650 *
651 * <li><p> Finally, if a fragment is given then a hash character
652 * ({@code '#'}) is appended, followed by the fragment. Any character
653 * that is not a legal URI character is quoted. </p></li>
654 *
655 * </ol>
656 *
657 * <p> The resulting URI string is then parsed as if by invoking the {@link
658 * #URI(String)} constructor and then invoking the {@link
659 * #parseServerAuthority()} method upon the result; this may cause a {@link
660 * URISyntaxException} to be thrown. </p>
661 *
662 * @param scheme Scheme name
663 * @param userInfo User name and authorization information
664 * @param host Host name
665 * @param port Port number
666 * @param path Path
667 * @param query Query
668 * @param fragment Fragment
669 *
670 * @throws URISyntaxException
671 * If both a scheme and a path are given but the path is relative,
672 * if the URI string constructed from the given components violates
673 * RFC 2396, or if the authority component of the string is
674 * present but cannot be parsed as a server-based authority
675 */
676 public URI(String scheme,
677 String userInfo, String host, int port,
678 String path, String query, String fragment)
679 throws URISyntaxException
680 {
681 String s = toString(scheme, null,
682 null, userInfo, host, port,
683 path, query, fragment);
684 checkPath(s, scheme, path);
685 new Parser(s).parse(true);
686 }
687
688 /**
689 * Constructs a hierarchical URI from the given components.
690 *
691 * <p> If a scheme is given then the path, if also given, must either be
692 * empty or begin with a slash character ({@code '/'}). Otherwise a
693 * component of the new URI may be left undefined by passing {@code null}
694 * for the corresponding parameter.
695 *
696 * <p> This constructor first builds a URI string from the given components
697 * according to the rules specified in <a
698 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
699 * section 5.2, step 7: </p>
700 *
701 * <ol>
702 *
703 * <li><p> Initially, the result string is empty. </p></li>
704 *
705 * <li><p> If a scheme is given then it is appended to the result,
706 * followed by a colon character ({@code ':'}). </p></li>
707 *
708 * <li><p> If an authority is given then the string {@code "//"} is
709 * appended, followed by the authority. If the authority contains a
710 * literal IPv6 address then the address must be enclosed in square
711 * brackets ({@code '['} and {@code ']'}). Any character not in the
712 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
713 * categories, and not equal to the commercial-at character
714 * ({@code '@'}), is <a href="#quote">quoted</a>. </p></li>
715 *
716 * <li><p> If a path is given then it is appended. Any character not in
717 * the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
718 * categories, and not equal to the slash character ({@code '/'}) or the
719 * commercial-at character ({@code '@'}), is quoted. </p></li>
720 *
721 * <li><p> If a query is given then a question-mark character
722 * ({@code '?'}) is appended, followed by the query. Any character that
723 * is not a <a href="#legal-chars">legal URI character</a> is quoted.
724 * </p></li>
725 *
726 * <li><p> Finally, if a fragment is given then a hash character
727 * ({@code '#'}) is appended, followed by the fragment. Any character
728 * that is not a legal URI character is quoted. </p></li>
729 *
730 * </ol>
731 *
732 * <p> The resulting URI string is then parsed as if by invoking the {@link
733 * #URI(String)} constructor and then invoking the {@link
734 * #parseServerAuthority()} method upon the result; this may cause a {@link
735 * URISyntaxException} to be thrown. </p>
736 *
737 * @param scheme Scheme name
738 * @param authority Authority
739 * @param path Path
740 * @param query Query
741 * @param fragment Fragment
742 *
743 * @throws URISyntaxException
744 * If both a scheme and a path are given but the path is relative,
745 * if the URI string constructed from the given components violates
746 * RFC 2396, or if the authority component of the string is
747 * present but cannot be parsed as a server-based authority
748 */
749 public URI(String scheme,
750 String authority,
751 String path, String query, String fragment)
752 throws URISyntaxException
753 {
754 String s = toString(scheme, null,
755 authority, null, null, -1,
756 path, query, fragment);
757 checkPath(s, scheme, path);
758 new Parser(s).parse(false);
759 }
760
761 /**
762 * Constructs a hierarchical URI from the given components.
763 *
764 * <p> A component may be left undefined by passing {@code null}.
765 *
766 * <p> This convenience constructor works as if by invoking the
767 * seven-argument constructor as follows:
768 *
769 * <blockquote>
770 * {@code new} {@link #URI(String, String, String, int, String, String, String)
771 * URI}{@code (scheme, null, host, -1, path, null, fragment);}
772 * </blockquote>
773 *
774 * @param scheme Scheme name
775 * @param host Host name
776 * @param path Path
777 * @param fragment Fragment
778 *
779 * @throws URISyntaxException
780 * If the URI string constructed from the given components
781 * violates RFC 2396
782 */
783 public URI(String scheme, String host, String path, String fragment)
784 throws URISyntaxException
785 {
786 this(scheme, null, host, -1, path, null, fragment);
787 }
788
789 /**
790 * Constructs a URI from the given components.
791 *
792 * <p> A component may be left undefined by passing {@code null}.
793 *
794 * <p> This constructor first builds a URI in string form using the given
795 * components as follows: </p>
796 *
797 * <ol>
798 *
799 * <li><p> Initially, the result string is empty. </p></li>
800 *
801 * <li><p> If a scheme is given then it is appended to the result,
802 * followed by a colon character ({@code ':'}). </p></li>
803 *
804 * <li><p> If a scheme-specific part is given then it is appended. Any
805 * character that is not a <a href="#legal-chars">legal URI character</a>
806 * is <a href="#quote">quoted</a>. </p></li>
807 *
808 * <li><p> Finally, if a fragment is given then a hash character
809 * ({@code '#'}) is appended to the string, followed by the fragment.
810 * Any character that is not a legal URI character is quoted. </p></li>
811 *
812 * </ol>
813 *
814 * <p> The resulting URI string is then parsed in order to create the new
815 * URI instance as if by invoking the {@link #URI(String)} constructor;
816 * this may cause a {@link URISyntaxException} to be thrown. </p>
817 *
818 * @param scheme Scheme name
819 * @param ssp Scheme-specific part
820 * @param fragment Fragment
821 *
822 * @throws URISyntaxException
823 * If the URI string constructed from the given components
824 * violates RFC 2396
825 */
826 public URI(String scheme, String ssp, String fragment)
827 throws URISyntaxException
828 {
829 new Parser(toString(scheme, ssp,
830 null, null, null, -1,
831 null, null, fragment))
832 .parse(false);
833 }
834
835 /**
836 * Constructs a simple URI consisting of only a scheme and a pre-validated
837 * path. Provides a fast-path for some internal cases.
838 */
839 URI(String scheme, String path) {
840 assert validSchemeAndPath(scheme, path);
841 this.scheme = scheme;
842 this.path = path;
843 }
844
845 private static boolean validSchemeAndPath(String scheme, String path) {
846 try {
847 URI u = new URI(scheme + ":" + path);
848 return scheme.equals(u.scheme) && path.equals(u.path);
849 } catch (URISyntaxException e) {
850 return false;
851 }
852 }
853
854 /**
855 * Creates a URI by parsing the given string.
856 *
857 * <p> This convenience factory method works as if by invoking the {@link
858 * #URI(String)} constructor; any {@link URISyntaxException} thrown by the
859 * constructor is caught and wrapped in a new {@link
860 * IllegalArgumentException} object, which is then thrown.
861 *
862 * <p> This method is provided for use in situations where it is known that
863 * the given string is a legal URI, for example for URI constants declared
864 * within in a program, and so it would be considered a programming error
865 * for the string not to parse as such. The constructors, which throw
866 * {@link URISyntaxException} directly, should be used situations where a
867 * URI is being constructed from user input or from some other source that
868 * may be prone to errors. </p>
869 *
870 * @param str The string to be parsed into a URI
871 * @return The new URI
872 *
873 * @throws NullPointerException
874 * If {@code str} is {@code null}
875 *
876 * @throws IllegalArgumentException
877 * If the given string violates RFC 2396
878 */
879 public static URI create(String str) {
880 try {
881 return new URI(str);
882 } catch (URISyntaxException x) {
883 throw new IllegalArgumentException(x.getMessage(), x);
884 }
885 }
886
887
888 // -- Operations --
889
890 /**
891 * Attempts to parse this URI's authority component, if defined, into
892 * user-information, host, and port components.
893 *
894 * <p> If this URI's authority component has already been recognized as
895 * being server-based then it will already have been parsed into
896 * user-information, host, and port components. In this case, or if this
897 * URI has no authority component, this method simply returns this URI.
898 *
899 * <p> Otherwise this method attempts once more to parse the authority
900 * component into user-information, host, and port components, and throws
901 * an exception describing why the authority component could not be parsed
902 * in that way.
903 *
904 * <p> This method is provided because the generic URI syntax specified in
905 * <a href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>
906 * cannot always distinguish a malformed server-based authority from a
907 * legitimate registry-based authority. It must therefore treat some
908 * instances of the former as instances of the latter. The authority
909 * component in the URI string {@code "//foo:bar"}, for example, is not a
910 * legal server-based authority but it is legal as a registry-based
911 * authority.
912 *
913 * <p> In many common situations, for example when working URIs that are
914 * known to be either URNs or URLs, the hierarchical URIs being used will
915 * always be server-based. They therefore must either be parsed as such or
916 * treated as an error. In these cases a statement such as
917 *
918 * <blockquote>
919 * {@code URI }<i>u</i>{@code = new URI(str).parseServerAuthority();}
920 * </blockquote>
921 *
922 * <p> can be used to ensure that <i>u</i> always refers to a URI that, if
923 * it has an authority component, has a server-based authority with proper
924 * user-information, host, and port components. Invoking this method also
925 * ensures that if the authority could not be parsed in that way then an
926 * appropriate diagnostic message can be issued based upon the exception
927 * that is thrown. </p>
928 *
929 * @return A URI whose authority field has been parsed
930 * as a server-based authority
931 *
932 * @throws URISyntaxException
933 * If the authority component of this URI is defined
934 * but cannot be parsed as a server-based authority
935 * according to RFC 2396
936 */
937 public URI parseServerAuthority()
938 throws URISyntaxException
939 {
940 // We could be clever and cache the error message and index from the
941 // exception thrown during the original parse, but that would require
942 // either more fields or a more-obscure representation.
943 if ((host != null) || (authority == null))
944 return this;
945 new Parser(toString()).parse(true);
946 return this;
947 }
948
949 /**
950 * Normalizes this URI's path.
951 *
952 * <p> If this URI is opaque, or if its path is already in normal form,
953 * then this URI is returned. Otherwise a new URI is constructed that is
954 * identical to this URI except that its path is computed by normalizing
955 * this URI's path in a manner consistent with <a
956 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
957 * section 5.2, step 6, sub-steps c through f; that is:
958 * </p>
959 *
960 * <ol>
961 *
962 * <li><p> All {@code "."} segments are removed. </p></li>
963 *
964 * <li><p> If a {@code ".."} segment is preceded by a non-{@code ".."}
965 * segment then both of these segments are removed. This step is
966 * repeated until it is no longer applicable. </p></li>
967 *
968 * <li><p> If the path is relative, and if its first segment contains a
969 * colon character ({@code ':'}), then a {@code "."} segment is
970 * prepended. This prevents a relative URI with a path such as
971 * {@code "a:b/c/d"} from later being re-parsed as an opaque URI with a
972 * scheme of {@code "a"} and a scheme-specific part of {@code "b/c/d"}.
973 * <b><i>(Deviation from RFC 2396)</i></b> </p></li>
974 *
975 * </ol>
976 *
977 * <p> A normalized path will begin with one or more {@code ".."} segments
978 * if there were insufficient non-{@code ".."} segments preceding them to
979 * allow their removal. A normalized path will begin with a {@code "."}
980 * segment if one was inserted by step 3 above. Otherwise, a normalized
981 * path will not contain any {@code "."} or {@code ".."} segments. </p>
982 *
983 * @return A URI equivalent to this URI,
984 * but whose path is in normal form
985 */
986 public URI normalize() {
987 return normalize(this);
988 }
989
990 /**
991 * Resolves the given URI against this URI.
992 *
993 * <p> If the given URI is already absolute, or if this URI is opaque, then
994 * the given URI is returned.
995 *
996 * <p><a id="resolve-frag"></a> If the given URI's fragment component is
997 * defined, its path component is empty, and its scheme, authority, and
998 * query components are undefined, then a URI with the given fragment but
999 * with all other components equal to those of this URI is returned. This
1000 * allows a URI representing a standalone fragment reference, such as
1001 * {@code "#foo"}, to be usefully resolved against a base URI.
1002 *
1003 * <p> Otherwise this method constructs a new hierarchical URI in a manner
1004 * consistent with <a
1005 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
1006 * section 5.2; that is: </p>
1007 *
1008 * <ol>
1009 *
1010 * <li><p> A new URI is constructed with this URI's scheme and the given
1011 * URI's query and fragment components. </p></li>
1012 *
1013 * <li><p> If the given URI has an authority component then the new URI's
1014 * authority and path are taken from the given URI. </p></li>
1015 *
1016 * <li><p> Otherwise the new URI's authority component is copied from
1017 * this URI, and its path is computed as follows: </p>
1018 *
1019 * <ol>
1020 *
1021 * <li><p> If the given URI's path is absolute then the new URI's path
1022 * is taken from the given URI. </p></li>
1023 *
1024 * <li><p> Otherwise the given URI's path is relative, and so the new
1025 * URI's path is computed by resolving the path of the given URI
1026 * against the path of this URI. This is done by concatenating all but
1027 * the last segment of this URI's path, if any, with the given URI's
1028 * path and then normalizing the result as if by invoking the {@link
1029 * #normalize() normalize} method. </p></li>
1030 *
1031 * </ol></li>
1032 *
1033 * </ol>
1034 *
1035 * <p> The result of this method is absolute if, and only if, either this
1036 * URI is absolute or the given URI is absolute. </p>
1037 *
1038 * @param uri The URI to be resolved against this URI
1039 * @return The resulting URI
1040 *
1041 * @throws NullPointerException
1042 * If {@code uri} is {@code null}
1043 */
1044 public URI resolve(URI uri) {
1045 return resolve(this, uri);
1046 }
1047
1048 /**
1049 * Constructs a new URI by parsing the given string and then resolving it
1050 * against this URI.
1051 *
1052 * <p> This convenience method works as if invoking it were equivalent to
1053 * evaluating the expression {@link #resolve(java.net.URI)
1054 * resolve}{@code (URI.}{@link #create(String) create}{@code (str))}. </p>
1055 *
1056 * @param str The string to be parsed into a URI
1057 * @return The resulting URI
1058 *
1059 * @throws NullPointerException
1060 * If {@code str} is {@code null}
1061 *
1062 * @throws IllegalArgumentException
1063 * If the given string violates RFC 2396
1064 */
1065 public URI resolve(String str) {
1066 return resolve(URI.create(str));
1067 }
1068
1069 /**
1070 * Relativizes the given URI against this URI.
1071 *
1072 * <p> The relativization of the given URI against this URI is computed as
1073 * follows: </p>
1074 *
1075 * <ol>
1076 *
1077 * <li><p> If either this URI or the given URI are opaque, or if the
1078 * scheme and authority components of the two URIs are not identical, or
1079 * if the path of this URI is not a prefix of the path of the given URI,
1080 * then the given URI is returned. </p></li>
1081 *
1082 * <li><p> Otherwise a new relative hierarchical URI is constructed with
1083 * query and fragment components taken from the given URI and with a path
1084 * component computed by removing this URI's path from the beginning of
1085 * the given URI's path. </p></li>
1086 *
1087 * </ol>
1088 *
1089 * @param uri The URI to be relativized against this URI
1090 * @return The resulting URI
1091 *
1092 * @throws NullPointerException
1093 * If {@code uri} is {@code null}
1094 */
1095 public URI relativize(URI uri) {
1096 return relativize(this, uri);
1097 }
1098
1099 /**
1100 * Constructs a URL from this URI.
1101 *
1102 * <p> This convenience method works as if invoking it were equivalent to
1103 * evaluating the expression {@code new URL(this.toString())} after
1104 * first checking that this URI is absolute. </p>
1105 *
1106 * @return A URL constructed from this URI
1107 *
1108 * @throws IllegalArgumentException
1109 * If this URL is not absolute
1110 *
1111 * @throws MalformedURLException
1112 * If a protocol handler for the URL could not be found,
1113 * or if some other error occurred while constructing the URL
1114 */
1115 public URL toURL() throws MalformedURLException {
1116 return URL.fromURI(this);
1117 }
1118
1119 // -- Component access methods --
1120
1121 /**
1122 * Returns the scheme component of this URI.
1123 *
1124 * <p> The scheme component of a URI, if defined, only contains characters
1125 * in the <i>alphanum</i> category and in the string {@code "-.+"}. A
1126 * scheme always starts with an <i>alpha</i> character. <p>
1127 *
1128 * The scheme component of a URI cannot contain escaped octets, hence this
1129 * method does not perform any decoding.
1130 *
1131 * @return The scheme component of this URI,
1132 * or {@code null} if the scheme is undefined
1133 */
1134 public String getScheme() {
1135 return scheme;
1136 }
1137
1138 /**
1139 * Tells whether or not this URI is absolute.
1140 *
1141 * <p> A URI is absolute if, and only if, it has a scheme component. </p>
1142 *
1143 * @return {@code true} if, and only if, this URI is absolute
1144 */
1145 public boolean isAbsolute() {
1146 return scheme != null;
1147 }
1148
1149 /**
1150 * Tells whether or not this URI is opaque.
1151 *
1152 * <p> A URI is opaque if, and only if, it is absolute and its
1153 * scheme-specific part does not begin with a slash character ('/').
1154 * An opaque URI has a scheme, a scheme-specific part, and possibly
1155 * a fragment; all other components are undefined. </p>
1156 *
1157 * @return {@code true} if, and only if, this URI is opaque
1158 */
1159 public boolean isOpaque() {
1160 return path == null;
1161 }
1162
1163 /**
1164 * Returns the raw scheme-specific part of this URI. The scheme-specific
1165 * part is never undefined, though it may be empty.
1166 *
1167 * <p> The scheme-specific part of a URI only contains legal URI
1168 * characters. </p>
1169 *
1170 * @return The raw scheme-specific part of this URI
1171 * (never {@code null})
1172 */
1173 public String getRawSchemeSpecificPart() {
1174 String part = schemeSpecificPart;
1175 if (part != null) {
1176 return part;
1177 }
1178
1179 String s = string;
1180 if (s != null) {
1181 // if string is defined, components will have been parsed
1182 int start = 0;
1183 int end = s.length();
1184 if (scheme != null) {
1185 start = scheme.length() + 1;
1186 }
1187 if (fragment != null) {
1188 end -= fragment.length() + 1;
1189 }
1190 if (path != null && path.length() == end - start) {
1191 part = path;
1192 } else {
1193 part = s.substring(start, end);
1194 }
1195 } else {
1196 StringBuilder sb = new StringBuilder();
1197 appendSchemeSpecificPart(sb, null, getAuthority(), getUserInfo(),
1198 host, port, getPath(), getQuery());
1199 part = sb.toString();
1200 }
1201 return schemeSpecificPart = part;
1202 }
1203
1204 /**
1205 * Returns the decoded scheme-specific part of this URI.
1206 *
1207 * <p> The string returned by this method is equal to that returned by the
1208 * {@link #getRawSchemeSpecificPart() getRawSchemeSpecificPart} method
1209 * except that all sequences of escaped octets are <a
1210 * href="#decode">decoded</a>. </p>
1211 *
1212 * @return The decoded scheme-specific part of this URI
1213 * (never {@code null})
1214 */
1215 public String getSchemeSpecificPart() {
1216 String part = decodedSchemeSpecificPart;
1217 if (part == null) {
1218 decodedSchemeSpecificPart = part = decode(getRawSchemeSpecificPart());
1219 }
1220 return part;
1221 }
1222
1223 /**
1224 * Returns the raw authority component of this URI.
1225 *
1226 * <p> The authority component of a URI, if defined, only contains the
1227 * commercial-at character ({@code '@'}) and characters in the
1228 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, and <i>other</i>
1229 * categories. If the authority is server-based then it is further
1230 * constrained to have valid user-information, host, and port
1231 * components. </p>
1232 *
1233 * @return The raw authority component of this URI,
1234 * or {@code null} if the authority is undefined
1235 */
1236 public String getRawAuthority() {
1237 return authority;
1238 }
1239
1240 /**
1241 * Returns the decoded authority component of this URI.
1242 *
1243 * <p> The string returned by this method is equal to that returned by the
1244 * {@link #getRawAuthority() getRawAuthority} method except that all
1245 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1246 *
1247 * @return The decoded authority component of this URI,
1248 * or {@code null} if the authority is undefined
1249 */
1250 public String getAuthority() {
1251 String auth = decodedAuthority;
1252 if ((auth == null) && (authority != null)) {
1253 decodedAuthority = auth = decode(authority);
1254 }
1255 return auth;
1256 }
1257
1258 /**
1259 * Returns the raw user-information component of this URI.
1260 *
1261 * <p> The user-information component of a URI, if defined, only contains
1262 * characters in the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, and
1263 * <i>other</i> categories. </p>
1264 *
1265 * @return The raw user-information component of this URI,
1266 * or {@code null} if the user information is undefined
1267 */
1268 public String getRawUserInfo() {
1269 return userInfo;
1270 }
1271
1272 /**
1273 * Returns the decoded user-information component of this URI.
1274 *
1275 * <p> The string returned by this method is equal to that returned by the
1276 * {@link #getRawUserInfo() getRawUserInfo} method except that all
1277 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1278 *
1279 * @return The decoded user-information component of this URI,
1280 * or {@code null} if the user information is undefined
1281 */
1282 public String getUserInfo() {
1283 String user = decodedUserInfo;
1284 if ((user == null) && (userInfo != null)) {
1285 decodedUserInfo = user = decode(userInfo);
1286 }
1287 return user;
1288 }
1289
1290 /**
1291 * Returns the host component of this URI.
1292 *
1293 * <p> The host component of a URI, if defined, will have one of the
1294 * following forms: </p>
1295 *
1296 * <ul>
1297 *
1298 * <li><p> A domain name consisting of one or more <i>labels</i>
1299 * separated by period characters ({@code '.'}), optionally followed by
1300 * a period character. Each label consists of <i>alphanum</i> characters
1301 * as well as hyphen characters ({@code '-'}), though hyphens never
1302 * occur as the first or last characters in a label. The rightmost
1303 * label of a domain name consisting of two or more labels, begins
1304 * with an <i>alpha</i> character. </li>
1305 *
1306 * <li><p> A dotted-quad IPv4 address of the form
1307 * <i>digit</i>{@code +.}<i>digit</i>{@code +.}<i>digit</i>{@code +.}<i>digit</i>{@code +},
1308 * where no <i>digit</i> sequence is longer than three characters and no
1309 * sequence has a value larger than 255. </p></li>
1310 *
1311 * <li><p> An IPv6 address enclosed in square brackets ({@code '['} and
1312 * {@code ']'}) and consisting of hexadecimal digits, colon characters
1313 * ({@code ':'}), and possibly an embedded IPv4 address. The full
1314 * syntax of IPv6 addresses is specified in <a
1315 * href="http://www.ietf.org/rfc/rfc2373.txt"><i>RFC 2373: IPv6
1316 * Addressing Architecture</i></a>. </p></li>
1317 *
1318 * </ul>
1319 *
1320 * The host component of a URI cannot contain escaped octets, hence this
1321 * method does not perform any decoding.
1322 *
1323 * @return The host component of this URI,
1324 * or {@code null} if the host is undefined
1325 */
1326 public String getHost() {
1327 return host;
1328 }
1329
1330 /**
1331 * Returns the port number of this URI.
1332 *
1333 * <p> The port component of a URI, if defined, is a non-negative
1334 * integer. </p>
1335 *
1336 * @return The port component of this URI,
1337 * or {@code -1} if the port is undefined
1338 */
1339 public int getPort() {
1340 return port;
1341 }
1342
1343 /**
1344 * Returns the raw path component of this URI.
1345 *
1346 * <p> The path component of a URI, if defined, only contains the slash
1347 * character ({@code '/'}), the commercial-at character ({@code '@'}),
1348 * and characters in the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>,
1349 * and <i>other</i> categories. </p>
1350 *
1351 * @return The path component of this URI,
1352 * or {@code null} if the path is undefined
1353 */
1354 public String getRawPath() {
1355 return path;
1356 }
1357
1358 /**
1359 * Returns the decoded path component of this URI.
1360 *
1361 * <p> The string returned by this method is equal to that returned by the
1362 * {@link #getRawPath() getRawPath} method except that all sequences of
1363 * escaped octets are <a href="#decode">decoded</a>. </p>
1364 *
1365 * @return The decoded path component of this URI,
1366 * or {@code null} if the path is undefined
1367 */
1368 public String getPath() {
1369 String decoded = decodedPath;
1370 if ((decoded == null) && (path != null)) {
1371 decodedPath = decoded = decode(path);
1372 }
1373 return decoded;
1374 }
1375
1376 /**
1377 * Returns the raw query component of this URI.
1378 *
1379 * <p> The query component of a URI, if defined, only contains legal URI
1380 * characters. </p>
1381 *
1382 * @return The raw query component of this URI,
1383 * or {@code null} if the query is undefined
1384 */
1385 public String getRawQuery() {
1386 return query;
1387 }
1388
1389 /**
1390 * Returns the decoded query component of this URI.
1391 *
1392 * <p> The string returned by this method is equal to that returned by the
1393 * {@link #getRawQuery() getRawQuery} method except that all sequences of
1394 * escaped octets are <a href="#decode">decoded</a>. </p>
1395 *
1396 * @return The decoded query component of this URI,
1397 * or {@code null} if the query is undefined
1398 */
1399 public String getQuery() {
1400 String decoded = decodedQuery;
1401 if ((decoded == null) && (query != null)) {
1402 decodedQuery = decoded = decode(query, false);
1403 }
1404 return decoded;
1405 }
1406
1407 /**
1408 * Returns the raw fragment component of this URI.
1409 *
1410 * <p> The fragment component of a URI, if defined, only contains legal URI
1411 * characters. </p>
1412 *
1413 * @return The raw fragment component of this URI,
1414 * or {@code null} if the fragment is undefined
1415 */
1416 public String getRawFragment() {
1417 return fragment;
1418 }
1419
1420 /**
1421 * Returns the decoded fragment component of this URI.
1422 *
1423 * <p> The string returned by this method is equal to that returned by the
1424 * {@link #getRawFragment() getRawFragment} method except that all
1425 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1426 *
1427 * @return The decoded fragment component of this URI,
1428 * or {@code null} if the fragment is undefined
1429 */
1430 public String getFragment() {
1431 String decoded = decodedFragment;
1432 if ((decoded == null) && (fragment != null)) {
1433 decodedFragment = decoded = decode(fragment, false);
1434 }
1435 return decoded;
1436 }
1437
1438
1439 // -- Equality, comparison, hash code, toString, and serialization --
1440
1441 /**
1442 * Tests this URI for equality with another object.
1443 *
1444 * <p> If the given object is not a URI then this method immediately
1445 * returns {@code false}.
1446 *
1447 * <p> For two URIs to be considered equal requires that either both are
1448 * opaque or both are hierarchical. Their schemes must either both be
1449 * undefined or else be equal without regard to case. Their fragments
1450 * must either both be undefined or else be equal.
1451 *
1452 * <p> For two opaque URIs to be considered equal, their scheme-specific
1453 * parts must be equal.
1454 *
1455 * <p> For two hierarchical URIs to be considered equal, their paths must
1456 * be equal and their queries must either both be undefined or else be
1457 * equal. Their authorities must either both be undefined, or both be
1458 * registry-based, or both be server-based. If their authorities are
1459 * defined and are registry-based, then they must be equal. If their
1460 * authorities are defined and are server-based, then their hosts must be
1461 * equal without regard to case, their port numbers must be equal, and
1462 * their user-information components must be equal.
1463 *
1464 * <p> When testing the user-information, path, query, fragment, authority,
1465 * or scheme-specific parts of two URIs for equality, the raw forms rather
1466 * than the encoded forms of these components are compared and the
1467 * hexadecimal digits of escaped octets are compared without regard to
1468 * case.
1469 *
1470 * <p> This method satisfies the general contract of the {@link
1471 * java.lang.Object#equals(Object) Object.equals} method. </p>
1472 *
1473 * @param ob The object to which this object is to be compared
1474 *
1475 * @return {@code true} if, and only if, the given object is a URI that
1476 * is identical to this URI
1477 */
1478 public boolean equals(Object ob) {
1479 if (ob == this)
1480 return true;
1481 if (!(ob instanceof URI))
1482 return false;
1483 URI that = (URI)ob;
1484 if (this.isOpaque() != that.isOpaque()) return false;
1485 if (!equalIgnoringCase(this.scheme, that.scheme)) return false;
1486 if (!equal(this.fragment, that.fragment)) return false;
1487
1488 // Opaque
1489 if (this.isOpaque())
1490 return equal(this.schemeSpecificPart, that.schemeSpecificPart);
1491
1492 // Hierarchical
1493 if (!equal(this.path, that.path)) return false;
1494 if (!equal(this.query, that.query)) return false;
1495
1496 // Authorities
1497 if (this.authority == that.authority) return true;
1498 if (this.host != null) {
1499 // Server-based
1500 if (!equal(this.userInfo, that.userInfo)) return false;
1501 if (!equalIgnoringCase(this.host, that.host)) return false;
1502 if (this.port != that.port) return false;
1503 } else if (this.authority != null) {
1504 // Registry-based
1505 if (!equal(this.authority, that.authority)) return false;
1506 } else if (this.authority != that.authority) {
1507 return false;
1508 }
1509
1510 return true;
1511 }
1512
1513 /**
1514 * Returns a hash-code value for this URI. The hash code is based upon all
1515 * of the URI's components, and satisfies the general contract of the
1516 * {@link java.lang.Object#hashCode() Object.hashCode} method.
1517 *
1518 * @return A hash-code value for this URI
1519 */
1520 public int hashCode() {
1521 int h = hash;
1522 if (h == 0) {
1523 h = hashIgnoringCase(0, scheme);
1524 h = hash(h, fragment);
1525 if (isOpaque()) {
1526 h = hash(h, schemeSpecificPart);
1527 } else {
1528 h = hash(h, path);
1529 h = hash(h, query);
1530 if (host != null) {
1531 h = hash(h, userInfo);
1532 h = hashIgnoringCase(h, host);
1533 h += 1949 * port;
1534 } else {
1535 h = hash(h, authority);
1536 }
1537 }
1538 if (h != 0) {
1539 hash = h;
1540 }
1541 }
1542 return h;
1543 }
1544
1545 /**
1546 * Compares this URI to another object, which must be a URI.
1547 *
1548 * <p> When comparing corresponding components of two URIs, if one
1549 * component is undefined but the other is defined then the first is
1550 * considered to be less than the second. Unless otherwise noted, string
1551 * components are ordered according to their natural, case-sensitive
1552 * ordering as defined by the {@link java.lang.String#compareTo(Object)
1553 * String.compareTo} method. String components that are subject to
1554 * encoding are compared by comparing their raw forms rather than their
1555 * encoded forms.
1556 *
1557 * <p> The ordering of URIs is defined as follows: </p>
1558 *
1559 * <ul>
1560 *
1561 * <li><p> Two URIs with different schemes are ordered according the
1562 * ordering of their schemes, without regard to case. </p></li>
1563 *
1564 * <li><p> A hierarchical URI is considered to be less than an opaque URI
1565 * with an identical scheme. </p></li>
1566 *
1567 * <li><p> Two opaque URIs with identical schemes are ordered according
1568 * to the ordering of their scheme-specific parts. </p></li>
1569 *
1570 * <li><p> Two opaque URIs with identical schemes and scheme-specific
1571 * parts are ordered according to the ordering of their
1572 * fragments. </p></li>
1573 *
1574 * <li><p> Two hierarchical URIs with identical schemes are ordered
1575 * according to the ordering of their authority components: </p>
1576 *
1577 * <ul>
1578 *
1579 * <li><p> If both authority components are server-based then the URIs
1580 * are ordered according to their user-information components; if these
1581 * components are identical then the URIs are ordered according to the
1582 * ordering of their hosts, without regard to case; if the hosts are
1583 * identical then the URIs are ordered according to the ordering of
1584 * their ports. </p></li>
1585 *
1586 * <li><p> If one or both authority components are registry-based then
1587 * the URIs are ordered according to the ordering of their authority
1588 * components. </p></li>
1589 *
1590 * </ul></li>
1591 *
1592 * <li><p> Finally, two hierarchical URIs with identical schemes and
1593 * authority components are ordered according to the ordering of their
1594 * paths; if their paths are identical then they are ordered according to
1595 * the ordering of their queries; if the queries are identical then they
1596 * are ordered according to the order of their fragments. </p></li>
1597 *
1598 * </ul>
1599 *
1600 * <p> This method satisfies the general contract of the {@link
1601 * java.lang.Comparable#compareTo(Object) Comparable.compareTo}
1602 * method. </p>
1603 *
1604 * @param that
1605 * The object to which this URI is to be compared
1606 *
1607 * @return A negative integer, zero, or a positive integer as this URI is
1608 * less than, equal to, or greater than the given URI
1609 *
1610 * @throws ClassCastException
1611 * If the given object is not a URI
1612 */
1613 public int compareTo(URI that) {
1614 int c;
1615
1616 if ((c = compareIgnoringCase(this.scheme, that.scheme)) != 0)
1617 return c;
1618
1619 if (this.isOpaque()) {
1620 if (that.isOpaque()) {
1621 // Both opaque
1622 if ((c = compare(this.schemeSpecificPart,
1623 that.schemeSpecificPart)) != 0)
1624 return c;
1625 return compare(this.fragment, that.fragment);
1626 }
1627 return +1; // Opaque > hierarchical
1628 } else if (that.isOpaque()) {
1629 return -1; // Hierarchical < opaque
1630 }
1631
1632 // Hierarchical
1633 if ((this.host != null) && (that.host != null)) {
1634 // Both server-based
1635 if ((c = compare(this.userInfo, that.userInfo)) != 0)
1636 return c;
1637 if ((c = compareIgnoringCase(this.host, that.host)) != 0)
1638 return c;
1639 if ((c = this.port - that.port) != 0)
1640 return c;
1641 } else {
1642 // If one or both authorities are registry-based then we simply
1643 // compare them in the usual, case-sensitive way. If one is
1644 // registry-based and one is server-based then the strings are
1645 // guaranteed to be unequal, hence the comparison will never return
1646 // zero and the compareTo and equals methods will remain
1647 // consistent.
1648 if ((c = compare(this.authority, that.authority)) != 0) return c;
1649 }
1650
1651 if ((c = compare(this.path, that.path)) != 0) return c;
1652 if ((c = compare(this.query, that.query)) != 0) return c;
1653 return compare(this.fragment, that.fragment);
1654 }
1655
1656 /**
1657 * Returns the content of this URI as a string.
1658 *
1659 * <p> If this URI was created by invoking one of the constructors in this
1660 * class then a string equivalent to the original input string, or to the
1661 * string computed from the originally-given components, as appropriate, is
1662 * returned. Otherwise this URI was created by normalization, resolution,
1663 * or relativization, and so a string is constructed from this URI's
1664 * components according to the rules specified in <a
1665 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
1666 * section 5.2, step 7. </p>
1667 *
1668 * @return The string form of this URI
1669 */
1670 public String toString() {
1671 String s = string;
1672 if (s == null) {
1673 s = defineString();
1674 }
1675 return s;
1676 }
1677
1678 private String defineString() {
1679 String s = string;
1680 if (s != null) {
1681 return s;
1682 }
1683
1684 StringBuilder sb = new StringBuilder();
1685 if (scheme != null) {
1686 sb.append(scheme);
1687 sb.append(':');
1688 }
1689 if (isOpaque()) {
1690 sb.append(schemeSpecificPart);
1691 } else {
1692 if (host != null) {
1693 sb.append("//");
1694 if (userInfo != null) {
1695 sb.append(userInfo);
1696 sb.append('@');
1697 }
1698 boolean needBrackets = ((host.indexOf(':') >= 0)
1699 && !host.startsWith("[")
1700 && !host.endsWith("]"));
1701 if (needBrackets) sb.append('[');
1702 sb.append(host);
1703 if (needBrackets) sb.append(']');
1704 if (port != -1) {
1705 sb.append(':');
1706 sb.append(port);
1707 }
1708 } else if (authority != null) {
1709 sb.append("//");
1710 sb.append(authority);
1711 }
1712 if (path != null)
1713 sb.append(path);
1714 if (query != null) {
1715 sb.append('?');
1716 sb.append(query);
1717 }
1718 }
1719 if (fragment != null) {
1720 sb.append('#');
1721 sb.append(fragment);
1722 }
1723 return string = sb.toString();
1724 }
1725
1726 /**
1727 * Returns the content of this URI as a US-ASCII string.
1728 *
1729 * <p> If this URI does not contain any characters in the <i>other</i>
1730 * category then an invocation of this method will return the same value as
1731 * an invocation of the {@link #toString() toString} method. Otherwise
1732 * this method works as if by invoking that method and then <a
1733 * href="#encode">encoding</a> the result. </p>
1734 *
1735 * @return The string form of this URI, encoded as needed
1736 * so that it only contains characters in the US-ASCII
1737 * charset
1738 */
1739 public String toASCIIString() {
1740 return encode(toString());
1741 }
1742
1743
1744 // -- Serialization support --
1745
1746 /**
1747 * Saves the content of this URI to the given serial stream.
1748 *
1749 * <p> The only serializable field of a URI instance is its {@code string}
1750 * field. That field is given a value, if it does not have one already,
1751 * and then the {@link java.io.ObjectOutputStream#defaultWriteObject()}
1752 * method of the given object-output stream is invoked. </p>
1753 *
1754 * @param os The object-output stream to which this object
1755 * is to be written
1756 */
1757 private void writeObject(ObjectOutputStream os)
1758 throws IOException
1759 {
1760 defineString();
1761 os.defaultWriteObject(); // Writes the string field only
1762 }
1763
1764 /**
1765 * Reconstitutes a URI from the given serial stream.
1766 *
1767 * <p> The {@link java.io.ObjectInputStream#defaultReadObject()} method is
1768 * invoked to read the value of the {@code string} field. The result is
1769 * then parsed in the usual way.
1770 *
1771 * @param is The object-input stream from which this object
1772 * is being read
1773 */
1774 private void readObject(ObjectInputStream is)
1775 throws ClassNotFoundException, IOException
1776 {
1777 port = -1; // Argh
1778 is.defaultReadObject();
1779 try {
1780 new Parser(string).parse(false);
1781 } catch (URISyntaxException x) {
1782 IOException y = new InvalidObjectException("Invalid URI");
1783 y.initCause(x);
1784 throw y;
1785 }
1786 }
1787
1788
1789 // -- End of public methods --
1790
1791
1792 // -- Utility methods for string-field comparison and hashing --
1793
1794 // These methods return appropriate values for null string arguments,
1795 // thereby simplifying the equals, hashCode, and compareTo methods.
1796 //
1797 // The case-ignoring methods should only be applied to strings whose
1798 // characters are all known to be US-ASCII. Because of this restriction,
1799 // these methods are faster than the similar methods in the String class.
1800
1801 // US-ASCII only
1802 private static int toLower(char c) {
1803 if ((c >= 'A') && (c <= 'Z'))
1804 return c + ('a' - 'A');
1805 return c;
1806 }
1807
1808 // US-ASCII only
1809 private static int toUpper(char c) {
1810 if ((c >= 'a') && (c <= 'z'))
1811 return c - ('a' - 'A');
1812 return c;
1813 }
1814
1815 private static boolean equal(String s, String t) {
1816 if (s == t) return true;
1817 if ((s != null) && (t != null)) {
1818 if (s.length() != t.length())
1819 return false;
1820 if (s.indexOf('%') < 0)
1821 return s.equals(t);
1822 int n = s.length();
1823 for (int i = 0; i < n;) {
1824 char c = s.charAt(i);
1825 char d = t.charAt(i);
1826 if (c != '%') {
1827 if (c != d)
1828 return false;
1829 i++;
1830 continue;
1831 }
1832 if (d != '%')
1833 return false;
1834 i++;
1835 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1836 return false;
1837 i++;
1838 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1839 return false;
1840 i++;
1841 }
1842 return true;
1843 }
1844 return false;
1845 }
1846
1847 // US-ASCII only
1848 private static boolean equalIgnoringCase(String s, String t) {
1849 if (s == t) return true;
1850 if ((s != null) && (t != null)) {
1851 int n = s.length();
1852 if (t.length() != n)
1853 return false;
1854 for (int i = 0; i < n; i++) {
1855 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1856 return false;
1857 }
1858 return true;
1859 }
1860 return false;
1861 }
1862
1863 private static int hash(int hash, String s) {
1864 if (s == null) return hash;
1865 return s.indexOf('%') < 0 ? hash * 127 + s.hashCode()
1866 : normalizedHash(hash, s);
1867 }
1868
1869
1870 private static int normalizedHash(int hash, String s) {
1871 int h = 0;
1872 for (int index = 0; index < s.length(); index++) {
1873 char ch = s.charAt(index);
1874 h = 31 * h + ch;
1875 if (ch == '%') {
1876 /*
1877 * Process the next two encoded characters
1878 */
1879 for (int i = index + 1; i < index + 3; i++)
1880 h = 31 * h + toUpper(s.charAt(i));
1881 index += 2;
1882 }
1883 }
1884 return hash * 127 + h;
1885 }
1886
1887 // US-ASCII only
1888 private static int hashIgnoringCase(int hash, String s) {
1889 if (s == null) return hash;
1890 int h = hash;
1891 int n = s.length();
1892 for (int i = 0; i < n; i++)
1893 h = 31 * h + toLower(s.charAt(i));
1894 return h;
1895 }
1896
1897 private static int compare(String s, String t) {
1898 if (s == t) return 0;
1899 if (s != null) {
1900 if (t != null)
1901 return s.compareTo(t);
1902 else
1903 return +1;
1904 } else {
1905 return -1;
1906 }
1907 }
1908
1909 // US-ASCII only
1910 private static int compareIgnoringCase(String s, String t) {
1911 if (s == t) return 0;
1912 if (s != null) {
1913 if (t != null) {
1914 int sn = s.length();
1915 int tn = t.length();
1916 int n = sn < tn ? sn : tn;
1917 for (int i = 0; i < n; i++) {
1918 int c = toLower(s.charAt(i)) - toLower(t.charAt(i));
1919 if (c != 0)
1920 return c;
1921 }
1922 return sn - tn;
1923 }
1924 return +1;
1925 } else {
1926 return -1;
1927 }
1928 }
1929
1930
1931 // -- String construction --
1932
1933 // If a scheme is given then the path, if given, must be absolute
1934 //
1935 private static void checkPath(String s, String scheme, String path)
1936 throws URISyntaxException
1937 {
1938 if (scheme != null) {
1939 if (path != null && !path.isEmpty() && path.charAt(0) != '/')
1940 throw new URISyntaxException(s, "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.isEmpty() && 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.isEmpty())
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 // To save startup time, we manually calculate the low-/highMask constants.
2529 // For reference, the following methods were used to calculate the values:
2530
2531 // Compute the low-order mask for the characters in the given string
2532 // private static long lowMask(String chars) {
2533 // int n = chars.length();
2534 // long m = 0;
2535 // for (int i = 0; i < n; i++) {
2536 // char c = chars.charAt(i);
2537 // if (c < 64)
2538 // m |= (1L << c);
2539 // }
2540 // return m;
2541 // }
2542
2543 // Compute the high-order mask for the characters in the given string
2544 // private static long highMask(String chars) {
2545 // int n = chars.length();
2546 // long m = 0;
2547 // for (int i = 0; i < n; i++) {
2548 // char c = chars.charAt(i);
2549 // if ((c >= 64) && (c < 128))
2550 // m |= (1L << (c - 64));
2551 // }
2552 // return m;
2553 // }
2554
2555 // Compute a low-order mask for the characters
2556 // between first and last, inclusive
2557 // private static long lowMask(char first, char last) {
2558 // long m = 0;
2559 // int f = Math.max(Math.min(first, 63), 0);
2560 // int l = Math.max(Math.min(last, 63), 0);
2561 // for (int i = f; i <= l; i++)
2562 // m |= 1L << i;
2563 // return m;
2564 // }
2565
2566 // Compute a high-order mask for the characters
2567 // between first and last, inclusive
2568 // private static long highMask(char first, char last) {
2569 // long m = 0;
2570 // int f = Math.max(Math.min(first, 127), 64) - 64;
2571 // int l = Math.max(Math.min(last, 127), 64) - 64;
2572 // for (int i = f; i <= l; i++)
2573 // m |= 1L << i;
2574 // return m;
2575 // }
2576
2577 // Tell whether the given character is permitted by the given mask pair
2578 private static boolean match(char c, long lowMask, long highMask) {
2579 if (c == 0) // 0 doesn't have a slot in the mask. So, it never matches.
2580 return false;
2581 if (c < 64)
2582 return ((1L << c) & lowMask) != 0;
2583 if (c < 128)
2584 return ((1L << (c - 64)) & highMask) != 0;
2585 return false;
2586 }
2587
2588 // Character-class masks, in reverse order from RFC2396 because
2589 // initializers for static fields cannot make forward references.
2590
2591 // digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" |
2592 // "8" | "9"
2593 private static final long L_DIGIT = 0x3FF000000000000L; // lowMask('0', '9');
2594 private static final long H_DIGIT = 0L;
2595
2596 // upalpha = "A" | "B" | "C" | "D" | "E" | "F" | "G" | "H" | "I" |
2597 // "J" | "K" | "L" | "M" | "N" | "O" | "P" | "Q" | "R" |
2598 // "S" | "T" | "U" | "V" | "W" | "X" | "Y" | "Z"
2599 private static final long L_UPALPHA = 0L;
2600 private static final long H_UPALPHA = 0x7FFFFFEL; // highMask('A', 'Z');
2601
2602 // lowalpha = "a" | "b" | "c" | "d" | "e" | "f" | "g" | "h" | "i" |
2603 // "j" | "k" | "l" | "m" | "n" | "o" | "p" | "q" | "r" |
2604 // "s" | "t" | "u" | "v" | "w" | "x" | "y" | "z"
2605 private static final long L_LOWALPHA = 0L;
2606 private static final long H_LOWALPHA = 0x7FFFFFE00000000L; // highMask('a', 'z');
2607
2608 // alpha = lowalpha | upalpha
2609 private static final long L_ALPHA = L_LOWALPHA | L_UPALPHA;
2610 private static final long H_ALPHA = H_LOWALPHA | H_UPALPHA;
2611
2612 // alphanum = alpha | digit
2613 private static final long L_ALPHANUM = L_DIGIT | L_ALPHA;
2614 private static final long H_ALPHANUM = H_DIGIT | H_ALPHA;
2615
2616 // hex = digit | "A" | "B" | "C" | "D" | "E" | "F" |
2617 // "a" | "b" | "c" | "d" | "e" | "f"
2618 private static final long L_HEX = L_DIGIT;
2619 private static final long H_HEX = 0x7E0000007EL; // highMask('A', 'F') | highMask('a', 'f');
2620
2621 // mark = "-" | "_" | "." | "!" | "~" | "*" | "'" |
2622 // "(" | ")"
2623 private static final long L_MARK = 0x678200000000L; // lowMask("-_.!~*'()");
2624 private static final long H_MARK = 0x4000000080000000L; // highMask("-_.!~*'()");
2625
2626 // unreserved = alphanum | mark
2627 private static final long L_UNRESERVED = L_ALPHANUM | L_MARK;
2628 private static final long H_UNRESERVED = H_ALPHANUM | H_MARK;
2629
2630 // reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" |
2631 // "$" | "," | "[" | "]"
2632 // Added per RFC2732: "[", "]"
2633 private static final long L_RESERVED = 0xAC00985000000000L; // lowMask(";/?:@&=+$,[]");
2634 private static final long H_RESERVED = 0x28000001L; // highMask(";/?:@&=+$,[]");
2635
2636 // The zero'th bit is used to indicate that escape pairs and non-US-ASCII
2637 // characters are allowed; this is handled by the scanEscape method below.
2638 private static final long L_ESCAPED = 1L;
2639 private static final long H_ESCAPED = 0L;
2640
2641 // uric = reserved | unreserved | escaped
2642 private static final long L_URIC = L_RESERVED | L_UNRESERVED | L_ESCAPED;
2643 private static final long H_URIC = H_RESERVED | H_UNRESERVED | H_ESCAPED;
2644
2645 // pchar = unreserved | escaped |
2646 // ":" | "@" | "&" | "=" | "+" | "$" | ","
2647 private static final long L_PCHAR
2648 = L_UNRESERVED | L_ESCAPED | 0x2400185000000000L; // lowMask(":@&=+$,");
2649 private static final long H_PCHAR
2650 = H_UNRESERVED | H_ESCAPED | 0x1L; // highMask(":@&=+$,");
2651
2652 // All valid path characters
2653 private static final long L_PATH = L_PCHAR | 0x800800000000000L; // lowMask(";/");
2654 private static final long H_PATH = H_PCHAR; // highMask(";/") == 0x0L;
2655
2656 // Dash, for use in domainlabel and toplabel
2657 private static final long L_DASH = 0x200000000000L; // lowMask("-");
2658 private static final long H_DASH = 0x0L; // highMask("-");
2659
2660 // Dot, for use in hostnames
2661 private static final long L_DOT = 0x400000000000L; // lowMask(".");
2662 private static final long H_DOT = 0x0L; // highMask(".");
2663
2664 // userinfo = *( unreserved | escaped |
2665 // ";" | ":" | "&" | "=" | "+" | "$" | "," )
2666 private static final long L_USERINFO
2667 = L_UNRESERVED | L_ESCAPED | 0x2C00185000000000L; // lowMask(";:&=+$,");
2668 private static final long H_USERINFO
2669 = H_UNRESERVED | H_ESCAPED; // | highMask(";:&=+$,") == 0L;
2670
2671 // reg_name = 1*( unreserved | escaped | "$" | "," |
2672 // ";" | ":" | "@" | "&" | "=" | "+" )
2673 private static final long L_REG_NAME
2674 = L_UNRESERVED | L_ESCAPED | 0x2C00185000000000L; // lowMask("$,;:@&=+");
2675 private static final long H_REG_NAME
2676 = H_UNRESERVED | H_ESCAPED | 0x1L; // highMask("$,;:@&=+");
2677
2678 // All valid characters for server-based authorities
2679 private static final long L_SERVER
2680 = L_USERINFO | L_ALPHANUM | L_DASH | 0x400400000000000L; // lowMask(".:@[]");
2681 private static final long H_SERVER
2682 = H_USERINFO | H_ALPHANUM | H_DASH | 0x28000001L; // highMask(".:@[]");
2683
2684 // Special case of server authority that represents an IPv6 address
2685 // In this case, a % does not signify an escape sequence
2686 private static final long L_SERVER_PERCENT
2687 = L_SERVER | 0x2000000000L; // lowMask("%");
2688 private static final long H_SERVER_PERCENT
2689 = H_SERVER; // | highMask("%") == 0L;
2690
2691 // scheme = alpha *( alpha | digit | "+" | "-" | "." )
2692 private static final long L_SCHEME = L_ALPHA | L_DIGIT | 0x680000000000L; // lowMask("+-.");
2693 private static final long H_SCHEME = H_ALPHA | H_DIGIT; // | highMask("+-.") == 0L
2694
2695 // scope_id = alpha | digit | "_" | "."
2696 private static final long L_SCOPE_ID
2697 = L_ALPHANUM | 0x400000000000L; // lowMask("_.");
2698 private static final long H_SCOPE_ID
2699 = H_ALPHANUM | 0x80000000L; // highMask("_.");
2700
2701 // -- Escaping and encoding --
2702
2703 private static final char[] hexDigits = {
2704 '0', '1', '2', '3', '4', '5', '6', '7',
2705 '8', '9', 'A', 'B', 'C', 'D', 'E', 'F'
2706 };
2707
2708 private static void appendEscape(StringBuilder sb, byte b) {
2709 sb.append('%');
2710 sb.append(hexDigits[(b >> 4) & 0x0f]);
2711 sb.append(hexDigits[(b >> 0) & 0x0f]);
2712 }
2713
2714 private static void appendEncoded(StringBuilder sb, char c) {
2715 ByteBuffer bb = null;
2716 try {
2717 bb = ThreadLocalCoders.encoderFor("UTF-8")
2718 .encode(CharBuffer.wrap("" + c));
2719 } catch (CharacterCodingException x) {
2720 assert false;
2721 }
2722 while (bb.hasRemaining()) {
2723 int b = bb.get() & 0xff;
2724 if (b >= 0x80)
2725 appendEscape(sb, (byte)b);
2726 else
2727 sb.append((char)b);
2728 }
2729 }
2730
2731 // Quote any characters in s that are not permitted
2732 // by the given mask pair
2733 //
2734 private static String quote(String s, long lowMask, long highMask) {
2735 StringBuilder sb = null;
2736 boolean allowNonASCII = ((lowMask & L_ESCAPED) != 0);
2737 for (int i = 0; i < s.length(); i++) {
2738 char c = s.charAt(i);
2739 if (c < '\u0080') {
2740 if (!match(c, lowMask, highMask)) {
2741 if (sb == null) {
2742 sb = new StringBuilder();
2743 sb.append(s, 0, i);
2744 }
2745 appendEscape(sb, (byte)c);
2746 } else {
2747 if (sb != null)
2748 sb.append(c);
2749 }
2750 } else if (allowNonASCII
2751 && (Character.isSpaceChar(c)
2752 || Character.isISOControl(c))) {
2753 if (sb == null) {
2754 sb = new StringBuilder();
2755 sb.append(s, 0, i);
2756 }
2757 appendEncoded(sb, c);
2758 } else {
2759 if (sb != null)
2760 sb.append(c);
2761 }
2762 }
2763 return (sb == null) ? s : sb.toString();
2764 }
2765
2766 // Encodes all characters >= \u0080 into escaped, normalized UTF-8 octets,
2767 // assuming that s is otherwise legal
2768 //
2769 private static String encode(String s) {
2770 int n = s.length();
2771 if (n == 0)
2772 return s;
2773
2774 // First check whether we actually need to encode
2775 for (int i = 0;;) {
2776 if (s.charAt(i) >= '\u0080')
2777 break;
2778 if (++i >= n)
2779 return s;
2780 }
2781
2782 String ns = Normalizer.normalize(s, Normalizer.Form.NFC);
2783 ByteBuffer bb = null;
2784 try {
2785 bb = ThreadLocalCoders.encoderFor("UTF-8")
2786 .encode(CharBuffer.wrap(ns));
2787 } catch (CharacterCodingException x) {
2788 assert false;
2789 }
2790
2791 StringBuilder sb = new StringBuilder();
2792 while (bb.hasRemaining()) {
2793 int b = bb.get() & 0xff;
2794 if (b >= 0x80)
2795 appendEscape(sb, (byte)b);
2796 else
2797 sb.append((char)b);
2798 }
2799 return sb.toString();
2800 }
2801
2802 private static int decode(char c) {
2803 if ((c >= '0') && (c <= '9'))
2804 return c - '0';
2805 if ((c >= 'a') && (c <= 'f'))
2806 return c - 'a' + 10;
2807 if ((c >= 'A') && (c <= 'F'))
2808 return c - 'A' + 10;
2809 assert false;
2810 return -1;
2811 }
2812
2813 private static byte decode(char c1, char c2) {
2814 return (byte)( ((decode(c1) & 0xf) << 4)
2815 | ((decode(c2) & 0xf) << 0));
2816 }
2817
2818 // Evaluates all escapes in s, applying UTF-8 decoding if needed. Assumes
2819 // that escapes are well-formed syntactically, i.e., of the form %XX. If a
2820 // sequence of escaped octets is not valid UTF-8 then the erroneous octets
2821 // are replaced with '\uFFFD'.
2822 // Exception: any "%" found between "[]" is left alone. It is an IPv6 literal
2823 // with a scope_id
2824 //
2825 private static String decode(String s) {
2826 return decode(s, true);
2827 }
2828
2829 // This method was introduced as a generalization of URI.decode method
2830 // to provide a fix for JDK-8037396
2831 private static String decode(String s, boolean ignorePercentInBrackets) {
2832 if (s == null)
2833 return s;
2834 int n = s.length();
2835 if (n == 0)
2836 return s;
2837 if (s.indexOf('%') < 0)
2838 return s;
2839
2840 StringBuilder sb = new StringBuilder(n);
2841 ByteBuffer bb = ByteBuffer.allocate(n);
2842 CharBuffer cb = CharBuffer.allocate(n);
2843 CharsetDecoder dec = ThreadLocalCoders.decoderFor("UTF-8")
2844 .onMalformedInput(CodingErrorAction.REPLACE)
2845 .onUnmappableCharacter(CodingErrorAction.REPLACE);
2846
2847 // This is not horribly efficient, but it will do for now
2848 char c = s.charAt(0);
2849 boolean betweenBrackets = false;
2850
2851 for (int i = 0; i < n;) {
2852 assert c == s.charAt(i); // Loop invariant
2853 if (c == '[') {
2854 betweenBrackets = true;
2855 } else if (betweenBrackets && c == ']') {
2856 betweenBrackets = false;
2857 }
2858 if (c != '%' || (betweenBrackets && ignorePercentInBrackets)) {
2859 sb.append(c);
2860 if (++i >= n)
2861 break;
2862 c = s.charAt(i);
2863 continue;
2864 }
2865 bb.clear();
2866 int ui = i;
2867 for (;;) {
2868 assert (n - i >= 2);
2869 bb.put(decode(s.charAt(++i), s.charAt(++i)));
2870 if (++i >= n)
2871 break;
2872 c = s.charAt(i);
2873 if (c != '%')
2874 break;
2875 }
2876 bb.flip();
2877 cb.clear();
2878 dec.reset();
2879 CoderResult cr = dec.decode(bb, cb, true);
2880 assert cr.isUnderflow();
2881 cr = dec.flush(cb);
2882 assert cr.isUnderflow();
2883 sb.append(cb.flip().toString());
2884 }
2885
2886 return sb.toString();
2887 }
2888
2889
2890 // -- Parsing --
2891
2892 // For convenience we wrap the input URI string in a new instance of the
2893 // following internal class. This saves always having to pass the input
2894 // string as an argument to each internal scan/parse method.
2895
2896 private class Parser {
2897
2898 private String input; // URI input string
2899 private boolean requireServerAuthority = false;
2900
2901 Parser(String s) {
2902 input = s;
2903 string = s;
2904 }
2905
2906 // -- Methods for throwing URISyntaxException in various ways --
2907
2908 private void fail(String reason) throws URISyntaxException {
2909 throw new URISyntaxException(input, reason);
2910 }
2911
2912 private void fail(String reason, int p) throws URISyntaxException {
2913 throw new URISyntaxException(input, reason, p);
2914 }
2915
2916 private void failExpecting(String expected, int p)
2917 throws URISyntaxException
2918 {
2919 fail("Expected " + expected, p);
2920 }
2921
2922
2923 // -- Simple access to the input string --
2924
2925 // Tells whether start < end and, if so, whether charAt(start) == c
2926 //
2927 private boolean at(int start, int end, char c) {
2928 return (start < end) && (input.charAt(start) == c);
2929 }
2930
2931 // Tells whether start + s.length() < end and, if so,
2932 // whether the chars at the start position match s exactly
2933 //
2934 private boolean at(int start, int end, String s) {
2935 int p = start;
2936 int sn = s.length();
2937 if (sn > end - p)
2938 return false;
2939 int i = 0;
2940 while (i < sn) {
2941 if (input.charAt(p++) != s.charAt(i)) {
2942 break;
2943 }
2944 i++;
2945 }
2946 return (i == sn);
2947 }
2948
2949
2950 // -- Scanning --
2951
2952 // The various scan and parse methods that follow use a uniform
2953 // convention of taking the current start position and end index as
2954 // their first two arguments. The start is inclusive while the end is
2955 // exclusive, just as in the String class, i.e., a start/end pair
2956 // denotes the left-open interval [start, end) of the input string.
2957 //
2958 // These methods never proceed past the end position. They may return
2959 // -1 to indicate outright failure, but more often they simply return
2960 // the position of the first char after the last char scanned. Thus
2961 // a typical idiom is
2962 //
2963 // int p = start;
2964 // int q = scan(p, end, ...);
2965 // if (q > p)
2966 // // We scanned something
2967 // ...;
2968 // else if (q == p)
2969 // // We scanned nothing
2970 // ...;
2971 // else if (q == -1)
2972 // // Something went wrong
2973 // ...;
2974
2975
2976 // Scan a specific char: If the char at the given start position is
2977 // equal to c, return the index of the next char; otherwise, return the
2978 // start position.
2979 //
2980 private int scan(int start, int end, char c) {
2981 if ((start < end) && (input.charAt(start) == c))
2982 return start + 1;
2983 return start;
2984 }
2985
2986 // Scan forward from the given start position. Stop at the first char
2987 // in the err string (in which case -1 is returned), or the first char
2988 // in the stop string (in which case the index of the preceding char is
2989 // returned), or the end of the input string (in which case the length
2990 // of the input string is returned). May return the start position if
2991 // nothing matches.
2992 //
2993 private int scan(int start, int end, String err, String stop) {
2994 int p = start;
2995 while (p < end) {
2996 char c = input.charAt(p);
2997 if (err.indexOf(c) >= 0)
2998 return -1;
2999 if (stop.indexOf(c) >= 0)
3000 break;
3001 p++;
3002 }
3003 return p;
3004 }
3005
3006 // Scan forward from the given start position. Stop at the first char
3007 // in the stop string (in which case the index of the preceding char is
3008 // returned), or the end of the input string (in which case the length
3009 // of the input string is returned). May return the start position if
3010 // nothing matches.
3011 //
3012 private int scan(int start, int end, String stop) {
3013 int p = start;
3014 while (p < end) {
3015 char c = input.charAt(p);
3016 if (stop.indexOf(c) >= 0)
3017 break;
3018 p++;
3019 }
3020 return p;
3021 }
3022
3023 // Scan a potential escape sequence, starting at the given position,
3024 // with the given first char (i.e., charAt(start) == c).
3025 //
3026 // This method assumes that if escapes are allowed then visible
3027 // non-US-ASCII chars are also allowed.
3028 //
3029 private int scanEscape(int start, int n, char first)
3030 throws URISyntaxException
3031 {
3032 int p = start;
3033 char c = first;
3034 if (c == '%') {
3035 // Process escape pair
3036 if ((p + 3 <= n)
3037 && match(input.charAt(p + 1), L_HEX, H_HEX)
3038 && match(input.charAt(p + 2), L_HEX, H_HEX)) {
3039 return p + 3;
3040 }
3041 fail("Malformed escape pair", p);
3042 } else if ((c > 128)
3043 && !Character.isSpaceChar(c)
3044 && !Character.isISOControl(c)) {
3045 // Allow unescaped but visible non-US-ASCII chars
3046 return p + 1;
3047 }
3048 return p;
3049 }
3050
3051 // Scan chars that match the given mask pair
3052 //
3053 private int scan(int start, int n, long lowMask, long highMask)
3054 throws URISyntaxException
3055 {
3056 int p = start;
3057 while (p < n) {
3058 char c = input.charAt(p);
3059 if (match(c, lowMask, highMask)) {
3060 p++;
3061 continue;
3062 }
3063 if ((lowMask & L_ESCAPED) != 0) {
3064 int q = scanEscape(p, n, c);
3065 if (q > p) {
3066 p = q;
3067 continue;
3068 }
3069 }
3070 break;
3071 }
3072 return p;
3073 }
3074
3075 // Check that each of the chars in [start, end) matches the given mask
3076 //
3077 private void checkChars(int start, int end,
3078 long lowMask, long highMask,
3079 String what)
3080 throws URISyntaxException
3081 {
3082 int p = scan(start, end, lowMask, highMask);
3083 if (p < end)
3084 fail("Illegal character in " + what, p);
3085 }
3086
3087 // Check that the char at position p matches the given mask
3088 //
3089 private void checkChar(int p,
3090 long lowMask, long highMask,
3091 String what)
3092 throws URISyntaxException
3093 {
3094 checkChars(p, p + 1, lowMask, highMask, what);
3095 }
3096
3097
3098 // -- Parsing --
3099
3100 // [<scheme>:]<scheme-specific-part>[#<fragment>]
3101 //
3102 void parse(boolean rsa) throws URISyntaxException {
3103 requireServerAuthority = rsa;
3104 int n = input.length();
3105 int p = scan(0, n, "/?#", ":");
3106 if ((p >= 0) && at(p, n, ':')) {
3107 if (p == 0)
3108 failExpecting("scheme name", 0);
3109 checkChar(0, L_ALPHA, H_ALPHA, "scheme name");
3110 checkChars(1, p, L_SCHEME, H_SCHEME, "scheme name");
3111 scheme = input.substring(0, p);
3112 p++; // Skip ':'
3113 if (at(p, n, '/')) {
3114 p = parseHierarchical(p, n);
3115 } else {
3116 // opaque; need to create the schemeSpecificPart
3117 int q = scan(p, n, "#");
3118 if (q <= p)
3119 failExpecting("scheme-specific part", p);
3120 checkChars(p, q, L_URIC, H_URIC, "opaque part");
3121 schemeSpecificPart = input.substring(p, q);
3122 p = q;
3123 }
3124 } else {
3125 p = parseHierarchical(0, n);
3126 }
3127 if (at(p, n, '#')) {
3128 checkChars(p + 1, n, L_URIC, H_URIC, "fragment");
3129 fragment = input.substring(p + 1, n);
3130 p = n;
3131 }
3132 if (p < n)
3133 fail("end of URI", p);
3134 }
3135
3136 // [//authority]<path>[?<query>]
3137 //
3138 // DEVIATION from RFC2396: We allow an empty authority component as
3139 // long as it's followed by a non-empty path, query component, or
3140 // fragment component. This is so that URIs such as "file:///foo/bar"
3141 // will parse. This seems to be the intent of RFC2396, though the
3142 // grammar does not permit it. If the authority is empty then the
3143 // userInfo, host, and port components are undefined.
3144 //
3145 // DEVIATION from RFC2396: We allow empty relative paths. This seems
3146 // to be the intent of RFC2396, but the grammar does not permit it.
3147 // The primary consequence of this deviation is that "#f" parses as a
3148 // relative URI with an empty path.
3149 //
3150 private int parseHierarchical(int start, int n)
3151 throws URISyntaxException
3152 {
3153 int p = start;
3154 if (at(p, n, '/') && at(p + 1, n, '/')) {
3155 p += 2;
3156 int q = scan(p, n, "/?#");
3157 if (q > p) {
3158 p = parseAuthority(p, q);
3159 } else if (q < n) {
3160 // DEVIATION: Allow empty authority prior to non-empty
3161 // path, query component or fragment identifier
3162 } else
3163 failExpecting("authority", p);
3164 }
3165 int q = scan(p, n, "?#"); // DEVIATION: May be empty
3166 checkChars(p, q, L_PATH, H_PATH, "path");
3167 path = input.substring(p, q);
3168 p = q;
3169 if (at(p, n, '?')) {
3170 p++;
3171 q = scan(p, n, "#");
3172 checkChars(p, q, L_URIC, H_URIC, "query");
3173 query = input.substring(p, q);
3174 p = q;
3175 }
3176 return p;
3177 }
3178
3179 // authority = server | reg_name
3180 //
3181 // Ambiguity: An authority that is a registry name rather than a server
3182 // might have a prefix that parses as a server. We use the fact that
3183 // the authority component is always followed by '/' or the end of the
3184 // input string to resolve this: If the complete authority did not
3185 // parse as a server then we try to parse it as a registry name.
3186 //
3187 private int parseAuthority(int start, int n)
3188 throws URISyntaxException
3189 {
3190 int p = start;
3191 int q = p;
3192 URISyntaxException ex = null;
3193
3194 boolean serverChars;
3195 boolean regChars;
3196
3197 if (scan(p, n, "]") > p) {
3198 // contains a literal IPv6 address, therefore % is allowed
3199 serverChars = (scan(p, n, L_SERVER_PERCENT, H_SERVER_PERCENT) == n);
3200 } else {
3201 serverChars = (scan(p, n, L_SERVER, H_SERVER) == n);
3202 }
3203 regChars = (scan(p, n, L_REG_NAME, H_REG_NAME) == n);
3204
3205 if (regChars && !serverChars) {
3206 // Must be a registry-based authority
3207 authority = input.substring(p, n);
3208 return n;
3209 }
3210
3211 if (serverChars) {
3212 // Might be (probably is) a server-based authority, so attempt
3213 // to parse it as such. If the attempt fails, try to treat it
3214 // as a registry-based authority.
3215 try {
3216 q = parseServer(p, n);
3217 if (q < n)
3218 failExpecting("end of authority", q);
3219 authority = input.substring(p, n);
3220 } catch (URISyntaxException x) {
3221 // Undo results of failed parse
3222 userInfo = null;
3223 host = null;
3224 port = -1;
3225 if (requireServerAuthority) {
3226 // If we're insisting upon a server-based authority,
3227 // then just re-throw the exception
3228 throw x;
3229 } else {
3230 // Save the exception in case it doesn't parse as a
3231 // registry either
3232 ex = x;
3233 q = p;
3234 }
3235 }
3236 }
3237
3238 if (q < n) {
3239 if (regChars) {
3240 // Registry-based authority
3241 authority = input.substring(p, n);
3242 } else if (ex != null) {
3243 // Re-throw exception; it was probably due to
3244 // a malformed IPv6 address
3245 throw ex;
3246 } else {
3247 fail("Illegal character in authority", q);
3248 }
3249 }
3250
3251 return n;
3252 }
3253
3254
3255 // [<userinfo>@]<host>[:<port>]
3256 //
3257 private int parseServer(int start, int n)
3258 throws URISyntaxException
3259 {
3260 int p = start;
3261 int q;
3262
3263 // userinfo
3264 q = scan(p, n, "/?#", "@");
3265 if ((q >= p) && at(q, n, '@')) {
3266 checkChars(p, q, L_USERINFO, H_USERINFO, "user info");
3267 userInfo = input.substring(p, q);
3268 p = q + 1; // Skip '@'
3269 }
3270
3271 // hostname, IPv4 address, or IPv6 address
3272 if (at(p, n, '[')) {
3273 // DEVIATION from RFC2396: Support IPv6 addresses, per RFC2732
3274 p++;
3275 q = scan(p, n, "/?#", "]");
3276 if ((q > p) && at(q, n, ']')) {
3277 // look for a "%" scope id
3278 int r = scan (p, q, "%");
3279 if (r > p) {
3280 parseIPv6Reference(p, r);
3281 if (r+1 == q) {
3282 fail ("scope id expected");
3283 }
3284 checkChars (r+1, q, L_SCOPE_ID, H_SCOPE_ID,
3285 "scope id");
3286 } else {
3287 parseIPv6Reference(p, q);
3288 }
3289 host = input.substring(p-1, q+1);
3290 p = q + 1;
3291 } else {
3292 failExpecting("closing bracket for IPv6 address", q);
3293 }
3294 } else {
3295 q = parseIPv4Address(p, n);
3296 if (q <= p)
3297 q = parseHostname(p, n);
3298 p = q;
3299 }
3300
3301 // port
3302 if (at(p, n, ':')) {
3303 p++;
3304 q = scan(p, n, "/");
3305 if (q > p) {
3306 checkChars(p, q, L_DIGIT, H_DIGIT, "port number");
3307 try {
3308 port = Integer.parseInt(input, p, q, 10);
3309 } catch (NumberFormatException x) {
3310 fail("Malformed port number", p);
3311 }
3312 p = q;
3313 }
3314 }
3315 if (p < n)
3316 failExpecting("port number", p);
3317
3318 return p;
3319 }
3320
3321 // Scan a string of decimal digits whose value fits in a byte
3322 //
3323 private int scanByte(int start, int n)
3324 throws URISyntaxException
3325 {
3326 int p = start;
3327 int q = scan(p, n, L_DIGIT, H_DIGIT);
3328 if (q <= p) return q;
3329 if (Integer.parseInt(input, p, q, 10) > 255) return p;
3330 return q;
3331 }
3332
3333 // Scan an IPv4 address.
3334 //
3335 // If the strict argument is true then we require that the given
3336 // interval contain nothing besides an IPv4 address; if it is false
3337 // then we only require that it start with an IPv4 address.
3338 //
3339 // If the interval does not contain or start with (depending upon the
3340 // strict argument) a legal IPv4 address characters then we return -1
3341 // immediately; otherwise we insist that these characters parse as a
3342 // legal IPv4 address and throw an exception on failure.
3343 //
3344 // We assume that any string of decimal digits and dots must be an IPv4
3345 // address. It won't parse as a hostname anyway, so making that
3346 // assumption here allows more meaningful exceptions to be thrown.
3347 //
3348 private int scanIPv4Address(int start, int n, boolean strict)
3349 throws URISyntaxException
3350 {
3351 int p = start;
3352 int q;
3353 int m = scan(p, n, L_DIGIT | L_DOT, H_DIGIT | H_DOT);
3354 if ((m <= p) || (strict && (m != n)))
3355 return -1;
3356 for (;;) {
3357 // Per RFC2732: At most three digits per byte
3358 // Further constraint: Each element fits in a byte
3359 if ((q = scanByte(p, m)) <= p) break; p = q;
3360 if ((q = scan(p, m, '.')) <= p) break; p = q;
3361 if ((q = scanByte(p, m)) <= p) break; p = q;
3362 if ((q = scan(p, m, '.')) <= p) break; p = q;
3363 if ((q = scanByte(p, m)) <= p) break; p = q;
3364 if ((q = scan(p, m, '.')) <= p) break; p = q;
3365 if ((q = scanByte(p, m)) <= p) break; p = q;
3366 if (q < m) break;
3367 return q;
3368 }
3369 fail("Malformed IPv4 address", q);
3370 return -1;
3371 }
3372
3373 // Take an IPv4 address: Throw an exception if the given interval
3374 // contains anything except an IPv4 address
3375 //
3376 private int takeIPv4Address(int start, int n, String expected)
3377 throws URISyntaxException
3378 {
3379 int p = scanIPv4Address(start, n, true);
3380 if (p <= start)
3381 failExpecting(expected, start);
3382 return p;
3383 }
3384
3385 // Attempt to parse an IPv4 address, returning -1 on failure but
3386 // allowing the given interval to contain [:<characters>] after
3387 // the IPv4 address.
3388 //
3389 private int parseIPv4Address(int start, int n) {
3390 int p;
3391
3392 try {
3393 p = scanIPv4Address(start, n, false);
3394 } catch (URISyntaxException x) {
3395 return -1;
3396 } catch (NumberFormatException nfe) {
3397 return -1;
3398 }
3399
3400 if (p > start && p < n) {
3401 // IPv4 address is followed by something - check that
3402 // it's a ":" as this is the only valid character to
3403 // follow an address.
3404 if (input.charAt(p) != ':') {
3405 p = -1;
3406 }
3407 }
3408
3409 if (p > start)
3410 host = input.substring(start, p);
3411
3412 return p;
3413 }
3414
3415 // hostname = domainlabel [ "." ] | 1*( domainlabel "." ) toplabel [ "." ]
3416 // domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum
3417 // toplabel = alpha | alpha *( alphanum | "-" ) alphanum
3418 //
3419 private int parseHostname(int start, int n)
3420 throws URISyntaxException
3421 {
3422 int p = start;
3423 int q;
3424 int l = -1; // Start of last parsed label
3425
3426 do {
3427 // domainlabel = alphanum [ *( alphanum | "-" ) alphanum ]
3428 q = scan(p, n, L_ALPHANUM, H_ALPHANUM);
3429 if (q <= p)
3430 break;
3431 l = p;
3432 if (q > p) {
3433 p = q;
3434 q = scan(p, n, L_ALPHANUM | L_DASH, H_ALPHANUM | H_DASH);
3435 if (q > p) {
3436 if (input.charAt(q - 1) == '-')
3437 fail("Illegal character in hostname", q - 1);
3438 p = q;
3439 }
3440 }
3441 q = scan(p, n, '.');
3442 if (q <= p)
3443 break;
3444 p = q;
3445 } while (p < n);
3446
3447 if ((p < n) && !at(p, n, ':'))
3448 fail("Illegal character in hostname", p);
3449
3450 if (l < 0)
3451 failExpecting("hostname", start);
3452
3453 // for a fully qualified hostname check that the rightmost
3454 // label starts with an alpha character.
3455 if (l > start && !match(input.charAt(l), L_ALPHA, H_ALPHA)) {
3456 fail("Illegal character in hostname", l);
3457 }
3458
3459 host = input.substring(start, p);
3460 return p;
3461 }
3462
3463
3464 // IPv6 address parsing, from RFC2373: IPv6 Addressing Architecture
3465 //
3466 // Bug: The grammar in RFC2373 Appendix B does not allow addresses of
3467 // the form ::12.34.56.78, which are clearly shown in the examples
3468 // earlier in the document. Here is the original grammar:
3469 //
3470 // IPv6address = hexpart [ ":" IPv4address ]
3471 // hexpart = hexseq | hexseq "::" [ hexseq ] | "::" [ hexseq ]
3472 // hexseq = hex4 *( ":" hex4)
3473 // hex4 = 1*4HEXDIG
3474 //
3475 // We therefore use the following revised grammar:
3476 //
3477 // IPv6address = hexseq [ ":" IPv4address ]
3478 // | hexseq [ "::" [ hexpost ] ]
3479 // | "::" [ hexpost ]
3480 // hexpost = hexseq | hexseq ":" IPv4address | IPv4address
3481 // hexseq = hex4 *( ":" hex4)
3482 // hex4 = 1*4HEXDIG
3483 //
3484 // This covers all and only the following cases:
3485 //
3486 // hexseq
3487 // hexseq : IPv4address
3488 // hexseq ::
3489 // hexseq :: hexseq
3490 // hexseq :: hexseq : IPv4address
3491 // hexseq :: IPv4address
3492 // :: hexseq
3493 // :: hexseq : IPv4address
3494 // :: IPv4address
3495 // ::
3496 //
3497 // Additionally we constrain the IPv6 address as follows :-
3498 //
3499 // i. IPv6 addresses without compressed zeros should contain
3500 // exactly 16 bytes.
3501 //
3502 // ii. IPv6 addresses with compressed zeros should contain
3503 // less than 16 bytes.
3504
3505 private int ipv6byteCount = 0;
3506
3507 private int parseIPv6Reference(int start, int n)
3508 throws URISyntaxException
3509 {
3510 int p = start;
3511 int q;
3512 boolean compressedZeros = false;
3513
3514 q = scanHexSeq(p, n);
3515
3516 if (q > p) {
3517 p = q;
3518 if (at(p, n, "::")) {
3519 compressedZeros = true;
3520 p = scanHexPost(p + 2, n);
3521 } else if (at(p, n, ':')) {
3522 p = takeIPv4Address(p + 1, n, "IPv4 address");
3523 ipv6byteCount += 4;
3524 }
3525 } else if (at(p, n, "::")) {
3526 compressedZeros = true;
3527 p = scanHexPost(p + 2, n);
3528 }
3529 if (p < n)
3530 fail("Malformed IPv6 address", start);
3531 if (ipv6byteCount > 16)
3532 fail("IPv6 address too long", start);
3533 if (!compressedZeros && ipv6byteCount < 16)
3534 fail("IPv6 address too short", start);
3535 if (compressedZeros && ipv6byteCount == 16)
3536 fail("Malformed IPv6 address", start);
3537
3538 return p;
3539 }
3540
3541 private int scanHexPost(int start, int n)
3542 throws URISyntaxException
3543 {
3544 int p = start;
3545 int q;
3546
3547 if (p == n)
3548 return p;
3549
3550 q = scanHexSeq(p, n);
3551 if (q > p) {
3552 p = q;
3553 if (at(p, n, ':')) {
3554 p++;
3555 p = takeIPv4Address(p, n, "hex digits or IPv4 address");
3556 ipv6byteCount += 4;
3557 }
3558 } else {
3559 p = takeIPv4Address(p, n, "hex digits or IPv4 address");
3560 ipv6byteCount += 4;
3561 }
3562 return p;
3563 }
3564
3565 // Scan a hex sequence; return -1 if one could not be scanned
3566 //
3567 private int scanHexSeq(int start, int n)
3568 throws URISyntaxException
3569 {
3570 int p = start;
3571 int q;
3572
3573 q = scan(p, n, L_HEX, H_HEX);
3574 if (q <= p)
3575 return -1;
3576 if (at(q, n, '.')) // Beginning of IPv4 address
3577 return -1;
3578 if (q > p + 4)
3579 fail("IPv6 hexadecimal digit sequence too long", p);
3580 ipv6byteCount += 2;
3581 p = q;
3582 while (p < n) {
3583 if (!at(p, n, ':'))
3584 break;
3585 if (at(p + 1, n, ':'))
3586 break; // "::"
3587 p++;
3588 q = scan(p, n, L_HEX, H_HEX);
3589 if (q <= p)
3590 failExpecting("digits for an IPv6 address", p);
3591 if (at(q, n, '.')) { // Beginning of IPv4 address
3592 p--;
3593 break;
3594 }
3595 if (q > p + 4)
3596 fail("IPv6 hexadecimal digit sequence too long", p);
3597 ipv6byteCount += 2;
3598 p = q;
3599 }
3600
3601 return p;
3602 }
3603
3604 }
3605 static {
3606 SharedSecrets.setJavaNetUriAccess(
3607 new JavaNetUriAccess() {
3608 public URI create(String scheme, String path) {
3609 return new URI(scheme, path);
3610 }
3611 }
3612 );
3613 }
3614 }