1 /*
2 * Copyright (c) 2000, 2025, 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.File;
29 import java.io.IOException;
30 import java.io.InvalidObjectException;
31 import java.io.ObjectInputStream;
32 import java.io.ObjectOutputStream;
33 import java.io.Serializable;
34 import java.nio.ByteBuffer;
35 import java.nio.CharBuffer;
36 import java.nio.charset.CharsetDecoder;
37 import java.nio.charset.CharsetEncoder;
38 import java.nio.charset.CoderResult;
39 import java.nio.charset.CodingErrorAction;
40 import java.nio.charset.CharacterCodingException;
41 import java.nio.file.Path;
42 import java.text.Normalizer;
43 import jdk.internal.access.JavaNetUriAccess;
44 import jdk.internal.access.SharedSecrets;
45 import jdk.internal.util.Exceptions;
46 import sun.nio.cs.UTF_8;
47
48 import static jdk.internal.util.Exceptions.filterNonSocketInfo;
49 import static jdk.internal.util.Exceptions.formatMsg;
50
51 /**
52 * Represents a Uniform Resource Identifier (URI) reference.
53 *
54 * <p> Aside from some minor deviations noted below, an instance of this
55 * class represents a URI reference as defined by
56 * <a href="http://www.ietf.org/rfc/rfc2396.txt"><i>RFC 2396: Uniform
57 * Resource Identifiers (URI): Generic Syntax</i></a>, amended by <a
58 * href="http://www.ietf.org/rfc/rfc2732.txt"><i>RFC 2732: Format for
59 * Literal IPv6 Addresses in URLs</i></a>. The Literal IPv6 address format
60 * also supports scope_ids. The syntax and usage of scope_ids is described
61 * <a href="Inet6Address.html#scoped">here</a>.
62 * This class provides constructors for creating URI instances from
63 * their components or by parsing their string forms, methods for accessing the
64 * various components of an instance, and methods for normalizing, resolving,
65 * and relativizing URI instances. Instances of this class are immutable.
66 *
67 *
68 * <h2> URI syntax and components </h2>
69 *
70 * At the highest level a URI reference (hereinafter simply "URI") in string
71 * form has the syntax
72 *
73 * <blockquote>
74 * [<i>scheme</i><b>{@code :}</b>]<i>scheme-specific-part</i>[<b>{@code #}</b><i>fragment</i>]
75 * </blockquote>
76 *
77 * where square brackets [...] delineate optional components and the characters
78 * <b>{@code :}</b> and <b>{@code #}</b> stand for themselves.
79 *
80 * <p> An <i>absolute</i> URI specifies a scheme; a URI that is not absolute is
81 * said to be <i>relative</i>. URIs are also classified according to whether
82 * they are <i>opaque</i> or <i>hierarchical</i>.
83 *
84 * <p> An <i>opaque</i> URI is an absolute URI whose scheme-specific part does
85 * not begin with a slash character ({@code '/'}). Opaque URIs are not
86 * subject to further parsing. Some examples of opaque URIs are:
87 *
88 * <blockquote><ul style="list-style-type:none">
89 * <li>{@code mailto:java-net@www.example.com}</li>
90 * <li>{@code news:comp.lang.java}</li>
91 * <li>{@code urn:isbn:096139210x}</li>
92 * </ul></blockquote>
93 *
94 * <p> A <i>hierarchical</i> URI is either an absolute URI whose
95 * scheme-specific part begins with a slash character, or a relative URI, that
96 * is, a URI that does not specify a scheme. Some examples of hierarchical
97 * URIs are:
98 *
99 * <blockquote>
100 * {@code http://example.com/languages/java/}<br>
101 * {@code sample/a/index.html#28}<br>
102 * {@code ../../demo/b/index.html}<br>
103 * {@code file:///~/calendar}
104 * </blockquote>
105 *
106 * <p> A hierarchical URI is subject to further parsing according to the syntax
107 *
108 * <blockquote>
109 * [<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>]
110 * </blockquote>
111 *
112 * where the characters <b>{@code :}</b>, <b>{@code /}</b>,
113 * <b>{@code ?}</b>, and <b>{@code #}</b> stand for themselves. The
114 * scheme-specific part of a hierarchical URI consists of the characters
115 * between the scheme and fragment components.
116 *
117 * <p> The authority component of a hierarchical URI is, if specified, either
118 * <i>server-based</i> or <i>registry-based</i>. A server-based authority
119 * parses according to the familiar syntax
120 *
121 * <blockquote>
122 * [<i>user-info</i><b>{@code @}</b>]<i>host</i>[<b>{@code :}</b><i>port</i>]
123 * </blockquote>
124 *
125 * where the characters <b>{@code @}</b> and <b>{@code :}</b> stand for
126 * themselves. Nearly all URI schemes currently in use are server-based. An
127 * authority component that does not parse in this way is considered to be
128 * registry-based.
129 *
130 * <p> The path component of a hierarchical URI is itself said to be absolute
131 * if it begins with a slash character ({@code '/'}); otherwise it is
132 * relative. The path of a hierarchical URI that is either absolute or
133 * specifies an authority is always absolute.
134 *
135 * <p> All told, then, a URI instance has the following nine components:
136 *
137 * <table class="striped" style="margin-left:2em">
138 * <caption style="display:none">Describes the components of a URI:scheme,scheme-specific-part,authority,user-info,host,port,path,query,fragment</caption>
139 * <thead>
140 * <tr><th scope="col">Component</th><th scope="col">Type</th></tr>
141 * </thead>
142 * <tbody style="text-align:left">
143 * <tr><th scope="row">scheme</th><td>{@code String}</td></tr>
144 * <tr><th scope="row">scheme-specific-part</th><td>{@code String}</td></tr>
145 * <tr><th scope="row">authority</th><td>{@code String}</td></tr>
146 * <tr><th scope="row">user-info</th><td>{@code String}</td></tr>
147 * <tr><th scope="row">host</th><td>{@code String}</td></tr>
148 * <tr><th scope="row">port</th><td>{@code int}</td></tr>
149 * <tr><th scope="row">path</th><td>{@code String}</td></tr>
150 * <tr><th scope="row">query</th><td>{@code String}</td></tr>
151 * <tr><th scope="row">fragment</th><td>{@code String}</td></tr>
152 * </tbody>
153 * </table>
154 *
155 * In a given instance any particular component is either <i>undefined</i> or
156 * <i>defined</i> with a distinct value. Undefined string components are
157 * represented by {@code null}, while undefined integer components are
158 * represented by {@code -1}. A string component may be defined to have the
159 * empty string as its value; this is not equivalent to that component being
160 * undefined.
161 *
162 * <p> Whether a particular component is or is not defined in an instance
163 * depends upon the type of the URI being represented. An absolute URI has a
164 * scheme component. An opaque URI has a scheme, a scheme-specific part, and
165 * possibly a fragment, but has no other components. A hierarchical URI always
166 * has a path (though it may be empty) and a scheme-specific-part (which at
167 * least contains the path), and may have any of the other components. If the
168 * authority component is present and is server-based then the host component
169 * will be defined and the user-information and port components may be defined.
170 *
171 *
172 * <h3> Operations on URI instances </h3>
173 *
174 * The key operations supported by this class are those of
175 * <i>normalization</i>, <i>resolution</i>, and <i>relativization</i>.
176 *
177 * <p> <i>Normalization</i> is the process of removing unnecessary {@code "."}
178 * and {@code ".."} segments from the path component of a hierarchical URI.
179 * Each {@code "."} segment is simply removed. A {@code ".."} segment is
180 * removed only if it is preceded by a non-{@code ".."} segment.
181 * Normalization has no effect upon opaque URIs.
182 *
183 * <p> <i>Resolution</i> is the process of resolving one URI against another,
184 * <i>base</i> URI. The resulting URI is constructed from components of both
185 * URIs in the manner specified by RFC 2396, taking components from the
186 * base URI for those not specified in the original. For hierarchical URIs,
187 * the path of the original is resolved against the path of the base and then
188 * normalized. The result, for example, of resolving
189 *
190 * <blockquote>
191 * {@code sample/a/index.html#28}
192 *
193 * (1)
194 * </blockquote>
195 *
196 * against the base URI {@code http://example.com/languages/java/} is the result
197 * URI
198 *
199 * <blockquote>
200 * {@code http://example.com/languages/java/sample/a/index.html#28}
201 * </blockquote>
202 *
203 * Resolving the relative URI
204 *
205 * <blockquote>
206 * {@code ../../demo/b/index.html} (2)
207 * </blockquote>
208 *
209 * against this result yields, in turn,
210 *
211 * <blockquote>
212 * {@code http://example.com/languages/java/demo/b/index.html}
213 * </blockquote>
214 *
215 * Resolution of both absolute and relative URIs, and of both absolute and
216 * relative paths in the case of hierarchical URIs, is supported. Resolving
217 * the URI {@code file:///~calendar} against any other URI simply yields the
218 * original URI, since it is absolute. Resolving the relative URI (2) above
219 * against the relative base URI (1) yields the normalized, but still relative,
220 * URI
221 *
222 * <blockquote>
223 * {@code demo/b/index.html}
224 * </blockquote>
225 *
226 * <p> <i>Relativization</i>, finally, can be regarded as the inverse of resolution.
227 * Let <i>u</i> be any normalized absolute URI ending with a slash character ({@code '/'})
228 * and <i>v</i> be any normalized relative URI not beginning with a period character ({@code '.'})
229 * or slash character ({@code '/'}). Then, the following statement is true:
230 *
231 * <blockquote>
232 * <i>u</i>{@code .relativize(}<i>u</i>{@code .resolve(}<i>v</i>{@code )).equals(}<i>v</i>{@code )}
233 * </blockquote>
234 *
235 * Let <i>u</i> be any normalized absolute URI ending with a slash character ({@code '/'})
236 * and <i>v</i> be any normalized absolute URI. Then, the following statement is true:
237 *
238 * <blockquote>
239 * <i>u</i>{@code .resolve(}<i>u</i>{@code .relativize(}<i>v</i>{@code )).equals(}<i>v</i>{@code )}
240 * </blockquote>
241 *
242 * This operation is often useful when constructing a document containing URIs
243 * that must be made relative to the base URI of the document wherever
244 * possible. For example, relativizing the URI
245 *
246 * <blockquote>
247 * {@code http://example.com/languages/java/sample/a/index.html#28}
248 * </blockquote>
249 *
250 * against the base URI
251 *
252 * <blockquote>
253 * {@code http://example.com/languages/java/}
254 * </blockquote>
255 *
256 * yields the relative URI {@code sample/a/index.html#28}.
257 *
258 *
259 * <h3> Character categories </h3>
260 *
261 * RFC 2396 specifies precisely which characters are permitted in the
262 * various components of a URI reference. The following categories, most of
263 * which are taken from that specification, are used below to describe these
264 * constraints:
265 *
266 * <table class="striped" style="margin-left:2em">
267 * <caption style="display:none">Describes categories alpha,digit,alphanum,unreserved,punct,reserved,escaped,and other</caption>
268 * <thead>
269 * <tr><th scope="col">Category</th><th scope="col">Description</th></tr>
270 * </thead>
271 * <tbody style="text-align:left">
272 * <tr><th scope="row" style="vertical-align:top">alpha</th>
273 * <td>The US-ASCII alphabetic characters,
274 * {@code 'A'} through {@code 'Z'}
275 * and {@code 'a'} through {@code 'z'}</td></tr>
276 * <tr><th scope="row" style="vertical-align:top">digit</th>
277 * <td>The US-ASCII decimal digit characters,
278 * {@code '0'} through {@code '9'}</td></tr>
279 * <tr><th scope="row" style="vertical-align:top">alphanum</th>
280 * <td>All <i>alpha</i> and <i>digit</i> characters</td></tr>
281 * <tr><th scope="row" style="vertical-align:top">unreserved</th>
282 * <td>All <i>alphanum</i> characters together with those in the string
283 * {@code "_-!.~'()*"}</td></tr>
284 * <tr><th scope="row" style="vertical-align:top">punct</th>
285 * <td>The characters in the string {@code ",;:$&+="}</td></tr>
286 * <tr><th scope="row" style="vertical-align:top">reserved</th>
287 * <td>All <i>punct</i> characters together with those in the string
288 * {@code "?/[]@"}</td></tr>
289 * <tr><th scope="row" style="vertical-align:top">escaped</th>
290 * <td>Escaped octets, that is, triplets consisting of the percent
291 * character ({@code '%'}) followed by two hexadecimal digits
292 * ({@code '0'}-{@code '9'}, {@code 'A'}-{@code 'F'}, and
293 * {@code 'a'}-{@code 'f'})</td></tr>
294 * <tr><th scope="row" style="vertical-align:top">other</th>
295 * <td>The Unicode characters that are not in the US-ASCII character set,
296 * are not control characters (according to the {@link
297 * java.lang.Character#isISOControl(char) Character.isISOControl}
298 * method), and are not space characters (according to the {@link
299 * java.lang.Character#isSpaceChar(char) Character.isSpaceChar}
300 * method) <i>(<b>Deviation from RFC 2396</b>, which is
301 * limited to US-ASCII)</i></td></tr>
302 * </tbody>
303 * </table>
304 *
305 * <p><a id="legal-chars"></a> The set of all legal URI characters consists of
306 * the <i>unreserved</i>, <i>reserved</i>, <i>escaped</i>, and <i>other</i>
307 * characters.
308 *
309 *
310 * <h3> Escaped octets, quotation, encoding, and decoding </h3>
311 *
312 * RFC 2396 allows escaped octets to appear in the user-info, path, query, and
313 * fragment components. Escaping serves two purposes in URIs:
314 *
315 * <ul>
316 *
317 * <li><p> To <i>encode</i> non-US-ASCII characters when a URI is required to
318 * conform strictly to RFC 2396 by not containing any <i>other</i>
319 * characters. </p></li>
320 *
321 * <li><p> To <i>quote</i> characters that are otherwise illegal in a
322 * component. The user-info, path, query, and fragment components differ
323 * slightly in terms of which characters are considered legal and illegal.
324 * </p></li>
325 *
326 * </ul>
327 *
328 * These purposes are served in this class by three related operations:
329 *
330 * <ul>
331 *
332 * <li><p><a id="encode"></a> A character is <i>encoded</i> by replacing it
333 * with the sequence of escaped octets that represent that character in the
334 * UTF-8 character set. The Euro currency symbol ({@code '\u005Cu20AC'}),
335 * for example, is encoded as {@code "%E2%82%AC"}. <i>(<b>Deviation from
336 * RFC 2396</b>, which does not specify any particular character
337 * set.)</i> </p></li>
338 *
339 * <li><p><a id="quote"></a> An illegal character is <i>quoted</i> simply by
340 * encoding it. The space character, for example, is quoted by replacing it
341 * with {@code "%20"}. UTF-8 contains US-ASCII, hence for US-ASCII
342 * characters this transformation has exactly the effect required by
343 * RFC 2396. </p></li>
344 *
345 * <li><p><a id="decode"></a>
346 * A sequence of escaped octets is <i>decoded</i> by
347 * replacing it with the sequence of characters that it represents in the
348 * UTF-8 character set. UTF-8 contains US-ASCII, hence decoding has the
349 * effect of de-quoting any quoted US-ASCII characters as well as that of
350 * decoding any encoded non-US-ASCII characters. If a <a
351 * href="../nio/charset/CharsetDecoder.html#ce">decoding error</a> occurs
352 * when decoding the escaped octets then the erroneous octets are replaced by
353 * {@code '\u005CuFFFD'}, the Unicode replacement character. </p></li>
354 *
355 * </ul>
356 *
357 * These operations are exposed in the constructors and methods of this class
358 * as follows:
359 *
360 * <ul>
361 *
362 * <li><p> The {@linkplain #URI(java.lang.String) single-argument
363 * constructor} requires any illegal characters in its argument to be
364 * quoted and preserves any escaped octets and <i>other</i> characters that
365 * are present. </p></li>
366 *
367 * <li><p> The {@linkplain
368 * #URI(java.lang.String,java.lang.String,java.lang.String,int,java.lang.String,java.lang.String,java.lang.String)
369 * multi-argument constructors} quote illegal characters as
370 * required by the components in which they appear. The percent character
371 * ({@code '%'}) is always quoted by these constructors. Any <i>other</i>
372 * characters are preserved. </p></li>
373 *
374 * <li><p> The {@link #getRawUserInfo() getRawUserInfo}, {@link #getRawPath()
375 * getRawPath}, {@link #getRawQuery() getRawQuery}, {@link #getRawFragment()
376 * getRawFragment}, {@link #getRawAuthority() getRawAuthority}, and {@link
377 * #getRawSchemeSpecificPart() getRawSchemeSpecificPart} methods return the
378 * values of their corresponding components in raw form, without interpreting
379 * any escaped octets. The strings returned by these methods may contain
380 * both escaped octets and <i>other</i> characters, and will not contain any
381 * illegal characters. </p></li>
382 *
383 * <li><p> The {@link #getUserInfo() getUserInfo}, {@link #getPath()
384 * getPath}, {@link #getQuery() getQuery}, {@link #getFragment()
385 * getFragment}, {@link #getAuthority() getAuthority}, and {@link
386 * #getSchemeSpecificPart() getSchemeSpecificPart} methods decode any escaped
387 * octets in their corresponding components. The strings returned by these
388 * methods may contain both <i>other</i> characters and illegal characters,
389 * and will not contain any escaped octets. </p></li>
390 *
391 * <li><p> The {@link #toString() toString} method returns a URI string with
392 * all necessary quotation but which may contain <i>other</i> characters.
393 * </p></li>
394 *
395 * <li><p> The {@link #toASCIIString() toASCIIString} method returns a fully
396 * quoted and encoded URI string that does not contain any <i>other</i>
397 * characters. </p></li>
398 *
399 * </ul>
400 *
401 *
402 * <h3> Identities </h3>
403 *
404 * For any URI <i>u</i>, it is always the case that
405 *
406 * <blockquote>
407 * {@code new URI(}<i>u</i>{@code .toString()).equals(}<i>u</i>{@code )} .
408 * </blockquote>
409 *
410 * For any URI <i>u</i> that does not contain redundant syntax such as two
411 * slashes before an empty authority (as in {@code file:///tmp/} ) or a
412 * colon following a host name but no port (as in
413 * {@code http://www.example.com:} ), and that does not encode characters
414 * except those that must be quoted, the following identities also hold:
415 * <pre>
416 * new URI(<i>u</i>.getScheme(),
417 * <i>u</i>.getSchemeSpecificPart(),
418 * <i>u</i>.getFragment())
419 * .equals(<i>u</i>)</pre>
420 * in all cases,
421 * <pre>
422 * new URI(<i>u</i>.getScheme(),
423 * <i>u</i>.getAuthority(),
424 * <i>u</i>.getPath(), <i>u</i>.getQuery(),
425 * <i>u</i>.getFragment())
426 * .equals(<i>u</i>)</pre>
427 * if <i>u</i> is hierarchical, and
428 * <pre>
429 * new URI(<i>u</i>.getScheme(),
430 * <i>u</i>.getUserInfo(), <i>u</i>.getHost(), <i>u</i>.getPort(),
431 * <i>u</i>.getPath(), <i>u</i>.getQuery(),
432 * <i>u</i>.getFragment())
433 * .equals(<i>u</i>)</pre>
434 * if <i>u</i> is hierarchical and has either no authority or a server-based
435 * authority.
436 *
437 *
438 * <h3> URIs, URLs, and URNs </h3>
439 *
440 * A URI is a uniform resource <i>identifier</i> while a URL is a uniform
441 * resource <i>locator</i>. Hence every URL is a URI, abstractly speaking, but
442 * not every URI is a URL. This is because there is another subcategory of
443 * URIs, uniform resource <i>names</i> (URNs), which name resources but do not
444 * specify how to locate them. The {@code mailto}, {@code news}, and
445 * {@code isbn} URIs shown above are examples of URNs.
446 *
447 * <p> The conceptual distinction between URIs and URLs is reflected in the
448 * differences between this class and the {@link URL} class.
449 *
450 * <p> An instance of this class represents a URI reference in the syntactic
451 * sense defined by RFC 2396. A URI may be either absolute or relative.
452 * A URI string is parsed according to the generic syntax without regard to the
453 * scheme, if any, that it specifies. No lookup of the host, if any, is
454 * performed, and no scheme-dependent stream handler is constructed. Equality,
455 * hashing, and comparison are defined strictly in terms of the character
456 * content of the instance. In other words, a URI instance is little more than
457 * a structured string that supports the syntactic, scheme-independent
458 * operations of comparison, normalization, resolution, and relativization.
459 *
460 * <p> An instance of the {@link URL} class, by contrast, represents the
461 * syntactic components of a URL together with some of the information required
462 * to access the resource that it describes. A URL must be absolute, that is,
463 * it must always specify a scheme. A URL string is parsed according to its
464 * scheme. A stream handler is always established for a URL, and in fact it is
465 * impossible to create a URL instance for a scheme for which no handler is
466 * available. Equality and hashing depend upon both the scheme and the
467 * Internet address of the host, if any; comparison is not defined. In other
468 * words, a URL is a structured string that supports the syntactic operation of
469 * resolution as well as the network I/O operations of looking up the host and
470 * opening a connection to the specified resource.
471 *
472 * @apiNote
473 *
474 * Applications working with file paths and file URIs should take great
475 * care to use the appropriate methods to convert between the two.
476 * The {@link Path#of(URI)} factory method and the {@link File#File(URI)}
477 * constructor can be used to create {@link Path} or {@link File}
478 * objects from a file URI. {@link Path#toUri()} and {@link File#toURI()}
479 * can be used to create a {@link URI} from a file path.
480 * Applications should never try to {@linkplain
481 * #URI(String, String, String, int, String, String, String)
482 * construct}, {@linkplain #URI(String) parse}, or
483 * {@linkplain #resolve(String) resolve} a {@code URI}
484 * from the direct string representation of a {@code File} or {@code Path}
485 * instance.
486 * <p>
487 * Some components of a URL or URI, such as <i>userinfo</i>, may
488 * be abused to construct misleading URLs or URIs. Applications
489 * that deal with URLs or URIs should take into account
490 * the recommendations advised in <a
491 * href="https://tools.ietf.org/html/rfc3986#section-7">RFC3986,
492 * Section 7, Security Considerations</a>.
493 *
494 * @author Mark Reinhold
495 * @since 1.4
496 *
497 * @spec https://www.rfc-editor.org/info/rfc2279
498 * RFC 2279: UTF-8, a transformation format of ISO 10646
499 * @spec https://www.rfc-editor.org/info/rfc2373
500 * RFC 2373: IP Version 6 Addressing Architecture
501 * @spec https://www.rfc-editor.org/info/rfc2396
502 * RFC 2396: Uniform Resource Identifiers (URI): Generic Syntax
503 * @spec https://www.rfc-editor.org/info/rfc2732
504 * RFC 2732: Format for Literal IPv6 Addresses in URL's
505 * @spec https://www.rfc-editor.org/info/rfc3986
506 * RFC 3986: Uniform Resource Identifier (URI): Generic Syntax
507 *
508 * @see <a href="http://www.ietf.org/rfc/rfc2279.txt"><i>RFC 2279: UTF-8, a
509 * transformation format of ISO 10646</i></a>
510 * @see <a href="http://www.ietf.org/rfc/rfc2373.txt"><i>RFC 2373: IPv6 Addressing
511 * Architecture</i></a>
512 * @see <a href="http://www.ietf.org/rfc/rfc2396.txt"><i>RFC 2396: Uniform
513 * Resource Identifiers (URI): Generic Syntax</i></a>
514 * @see <a href="http://www.ietf.org/rfc/rfc2732.txt"><i>RFC 2732: Format for
515 * Literal IPv6 Addresses in URLs</i></a>
516 * @see <a href="URISyntaxException.html">URISyntaxException</a>
517 */
518
519 public final class URI
520 implements Comparable<URI>, Serializable
521 {
522
523 // Note: Comments containing the word "ASSERT" indicate places where a
524 // throw of an InternalError should be replaced by an appropriate assertion
525 // statement once asserts are enabled in the build.
526 @java.io.Serial
527 static final long serialVersionUID = -6052424284110960213L;
528
529
530 // -- Properties and components of this instance --
531
532 // Components of all URIs: [<scheme>:]<scheme-specific-part>[#<fragment>]
533 private transient String scheme; // null ==> relative URI
534 private transient String fragment;
535
536 // Hierarchical URI components: [//<authority>]<path>[?<query>]
537 private transient String authority; // Registry or server
538
539 // Server-based authority: [<userInfo>@]<host>[:<port>]
540 private transient String userInfo;
541 private transient String host; // null ==> registry-based
542 private transient int port = -1; // -1 ==> undefined
543
544 // Remaining components of hierarchical URIs
545 private transient String path; // null ==> opaque
546 private transient String query;
547
548 // The remaining fields may be computed on demand, which is safe even in
549 // the face of multiple threads racing to initialize them
550 private transient String schemeSpecificPart;
551 private transient int hash; // Zero ==> undefined
552
553 private transient String decodedUserInfo;
554 private transient String decodedAuthority;
555 private transient String decodedPath;
556 private transient String decodedQuery;
557 private transient String decodedFragment;
558 private transient String decodedSchemeSpecificPart;
559
560 /**
561 * The string form of this URI.
562 *
563 * @serial
564 */
565 private volatile String string; // The only serializable field
566
567
568
569 // -- Constructors and factories --
570
571 private URI() { } // Used internally
572
573 /**
574 * Constructs a URI by parsing the given string.
575 *
576 * <p> This constructor parses the given string exactly as specified by the
577 * grammar in <a
578 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
579 * Appendix A, <b><i>except for the following deviations:</i></b> </p>
580 *
581 * <ul>
582 *
583 * <li><p> An empty authority component is permitted as long as it is
584 * followed by a non-empty path, a query component, or a fragment
585 * component. This allows the parsing of URIs such as
586 * {@code "file:///foo/bar"}, which seems to be the intent of
587 * RFC 2396 although the grammar does not permit it. If the
588 * authority component is empty then the user-information, host, and port
589 * components are undefined. </p></li>
590 *
591 * <li><p> Empty relative paths are permitted; this seems to be the
592 * intent of RFC 2396 although the grammar does not permit it. The
593 * primary consequence of this deviation is that a standalone fragment
594 * such as {@code "#foo"} parses as a relative URI with an empty path
595 * and the given fragment, and can be usefully <a
596 * href="#resolve-frag">resolved</a> against a base URI.
597 *
598 * <li><p> IPv4 addresses in host components are parsed rigorously, as
599 * specified by <a
600 * href="http://www.ietf.org/rfc/rfc2732.txt">RFC 2732</a>: Each
601 * element of a dotted-quad address must contain no more than three
602 * decimal digits. Each element is further constrained to have a value
603 * no greater than 255. </p></li>
604 *
605 * <li> <p> Hostnames in host components that comprise only a single
606 * domain label are permitted to start with an <i>alphanum</i>
607 * character. This seems to be the intent of <a
608 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>
609 * section 3.2.2 although the grammar does not permit it. The
610 * consequence of this deviation is that the authority component of a
611 * hierarchical URI such as {@code s://123}, will parse as a server-based
612 * authority. </p></li>
613 *
614 * <li><p> IPv6 addresses are permitted for the host component. An IPv6
615 * address must be enclosed in square brackets ({@code '['} and
616 * {@code ']'}) as specified by <a
617 * href="http://www.ietf.org/rfc/rfc2732.txt">RFC 2732</a>. The
618 * IPv6 address itself must parse according to <a
619 * href="http://www.ietf.org/rfc/rfc2373.txt">RFC 2373</a>. IPv6
620 * addresses are further constrained to describe no more than sixteen
621 * bytes of address information, a constraint implicit in RFC 2373
622 * but not expressible in the grammar. </p></li>
623 *
624 * <li><p> Characters in the <i>other</i> category are permitted wherever
625 * RFC 2396 permits <i>escaped</i> octets, that is, in the
626 * user-information, path, query, and fragment components, as well as in
627 * the authority component if the authority is registry-based. This
628 * allows URIs to contain Unicode characters beyond those in the US-ASCII
629 * character set. </p></li>
630 *
631 * </ul>
632 *
633 * @param str The string to be parsed into a URI
634 *
635 * @throws NullPointerException
636 * If {@code str} is {@code null}
637 *
638 * @throws URISyntaxException
639 * If the given string violates RFC 2396, as augmented
640 * by the above deviations
641 * @spec https://www.rfc-editor.org/info/rfc2373
642 * RFC 2373: IP Version 6 Addressing Architecture
643 * @spec https://www.rfc-editor.org/info/rfc2396
644 * RFC 2396: Uniform Resource Identifiers (URI): Generic Syntax
645 * @spec https://www.rfc-editor.org/info/rfc2732
646 * RFC 2732: Format for Literal IPv6 Addresses in URL's
647 */
648 public URI(String str) throws URISyntaxException {
649 new Parser(str).parse(false);
650 }
651
652 /**
653 * Constructs a hierarchical URI from the given components.
654 *
655 * <p> If a scheme is given then the path, if also given, must either be
656 * empty or begin with a slash character ({@code '/'}). Otherwise a
657 * component of the new URI may be left undefined by passing {@code null}
658 * for the corresponding parameter or, in the case of the {@code port}
659 * parameter, by passing {@code -1}.
660 *
661 * <p> This constructor first builds a URI string from the given components
662 * according to the rules specified in <a
663 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
664 * section 5.2, step 7: </p>
665 *
666 * <ol>
667 *
668 * <li><p> Initially, the result string is empty. </p></li>
669 *
670 * <li><p> If a scheme is given then it is appended to the result,
671 * followed by a colon character ({@code ':'}). </p></li>
672 *
673 * <li><p> If user information, a host, or a port are given then the
674 * string {@code "//"} is appended. </p></li>
675 *
676 * <li><p> If user information is given then it is appended, followed by
677 * a commercial-at character ({@code '@'}). Any character not in the
678 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
679 * categories is <a href="#quote">quoted</a>. </p></li>
680 *
681 * <li><p> If a host is given then it is appended. If the host is a
682 * literal IPv6 address but is not enclosed in square brackets
683 * ({@code '['} and {@code ']'}) then the square brackets are added.
684 * </p></li>
685 *
686 * <li><p> If a port number is given then a colon character
687 * ({@code ':'}) is appended, followed by the port number in decimal.
688 * </p></li>
689 *
690 * <li><p> If a path is given then it is appended. Any character not in
691 * the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
692 * categories, and not equal to the slash character ({@code '/'}) or the
693 * commercial-at character ({@code '@'}), is quoted. </p></li>
694 *
695 * <li><p> If a query is given then a question-mark character
696 * ({@code '?'}) is appended, followed by the query. Any character that
697 * is not a <a href="#legal-chars">legal URI character</a> is quoted.
698 * </p></li>
699 *
700 * <li><p> Finally, if a fragment is given then a hash character
701 * ({@code '#'}) is appended, followed by the fragment. Any character
702 * that is not a legal URI character is quoted. </p></li>
703 *
704 * </ol>
705 *
706 * <p> The resulting URI string is then parsed as if by invoking the {@link
707 * #URI(String)} constructor and then invoking the {@link
708 * #parseServerAuthority()} method upon the result; this may cause a {@link
709 * URISyntaxException} to be thrown. </p>
710 *
711 * @param scheme Scheme name
712 * @param userInfo User name and authorization information
713 * @param host Host name
714 * @param port Port number
715 * @param path Path
716 * @param query Query
717 * @param fragment Fragment
718 *
719 * @throws URISyntaxException
720 * If both a scheme and a path are given but the path is relative,
721 * if the URI string constructed from the given components violates
722 * RFC 2396, or if the authority component of the string is
723 * present but cannot be parsed as a server-based authority
724 * @spec https://www.rfc-editor.org/info/rfc2396
725 * RFC 2396: Uniform Resource Identifiers (URI): Generic Syntax
726 */
727 public URI(String scheme,
728 String userInfo, String host, int port,
729 String path, String query, String fragment)
730 throws URISyntaxException
731 {
732 String s = toString(scheme, null,
733 null, userInfo, host, port,
734 path, query, fragment);
735 checkPath(s, scheme, path);
736 new Parser(s).parse(true);
737 }
738
739 /**
740 * Constructs a hierarchical URI from the given components.
741 *
742 * <p> If a scheme is given then the path, if also given, must either be
743 * empty or begin with a slash character ({@code '/'}). Otherwise a
744 * component of the new URI may be left undefined by passing {@code null}
745 * for the corresponding parameter.
746 *
747 * <p> This constructor first builds a URI string from the given components
748 * according to the rules specified in <a
749 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
750 * section 5.2, step 7: </p>
751 *
752 * <ol>
753 *
754 * <li><p> Initially, the result string is empty. </p></li>
755 *
756 * <li><p> If a scheme is given then it is appended to the result,
757 * followed by a colon character ({@code ':'}). </p></li>
758 *
759 * <li><p> If an authority is given then the string {@code "//"} is
760 * appended, followed by the authority. If the authority contains a
761 * literal IPv6 address then the address must be enclosed in square
762 * brackets ({@code '['} and {@code ']'}). Any character not in the
763 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
764 * categories, and not equal to the commercial-at character
765 * ({@code '@'}), is <a href="#quote">quoted</a>. </p></li>
766 *
767 * <li><p> If a path is given then it is appended. Any character not in
768 * the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
769 * categories, and not equal to the slash character ({@code '/'}) or the
770 * commercial-at character ({@code '@'}), is quoted. </p></li>
771 *
772 * <li><p> If a query is given then a question-mark character
773 * ({@code '?'}) is appended, followed by the query. Any character that
774 * is not a <a href="#legal-chars">legal URI character</a> is quoted.
775 * </p></li>
776 *
777 * <li><p> Finally, if a fragment is given then a hash character
778 * ({@code '#'}) is appended, followed by the fragment. Any character
779 * that is not a legal URI character is quoted. </p></li>
780 *
781 * </ol>
782 *
783 * <p> The resulting URI string is then parsed as if by invoking the {@link
784 * #URI(String)} constructor and then invoking the {@link
785 * #parseServerAuthority()} method upon the result; this may cause a {@link
786 * URISyntaxException} to be thrown. </p>
787 *
788 * @param scheme Scheme name
789 * @param authority Authority
790 * @param path Path
791 * @param query Query
792 * @param fragment Fragment
793 *
794 * @throws URISyntaxException
795 * If both a scheme and a path are given but the path is relative,
796 * if the URI string constructed from the given components violates
797 * RFC 2396, or if the authority component of the string is
798 * present but cannot be parsed as a server-based authority
799 * @spec https://www.rfc-editor.org/info/rfc2396
800 * RFC 2396: Uniform Resource Identifiers (URI): Generic Syntax
801 */
802 public URI(String scheme,
803 String authority,
804 String path, String query, String fragment)
805 throws URISyntaxException
806 {
807 String s = toString(scheme, null,
808 authority, null, null, -1,
809 path, query, fragment);
810 checkPath(s, scheme, path);
811 new Parser(s).parse(false);
812 }
813
814 /**
815 * Constructs a hierarchical URI from the given components.
816 *
817 * <p> A component may be left undefined by passing {@code null}.
818 *
819 * <p> This convenience constructor works as if by invoking the
820 * seven-argument constructor as follows:
821 *
822 * <blockquote>
823 * {@code new} {@link #URI(String, String, String, int, String, String, String)
824 * URI}{@code (scheme, null, host, -1, path, null, fragment);}
825 * </blockquote>
826 *
827 * @param scheme Scheme name
828 * @param host Host name
829 * @param path Path
830 * @param fragment Fragment
831 *
832 * @throws URISyntaxException
833 * If the URI string constructed from the given components
834 * violates RFC 2396
835 */
836 public URI(String scheme, String host, String path, String fragment)
837 throws URISyntaxException
838 {
839 this(scheme, null, host, -1, path, null, fragment);
840 }
841
842 /**
843 * Constructs a URI from the given components.
844 *
845 * <p> A component may be left undefined by passing {@code null}.
846 *
847 * <p> This constructor first builds a URI in string form using the given
848 * components as follows: </p>
849 *
850 * <ol>
851 *
852 * <li><p> Initially, the result string is empty. </p></li>
853 *
854 * <li><p> If a scheme is given then it is appended to the result,
855 * followed by a colon character ({@code ':'}). </p></li>
856 *
857 * <li><p> If a scheme-specific part is given then it is appended. Any
858 * character that is not a <a href="#legal-chars">legal URI character</a>
859 * is <a href="#quote">quoted</a>. </p></li>
860 *
861 * <li><p> Finally, if a fragment is given then a hash character
862 * ({@code '#'}) is appended to the string, followed by the fragment.
863 * Any character that is not a legal URI character is quoted. </p></li>
864 *
865 * </ol>
866 *
867 * <p> The resulting URI string is then parsed in order to create the new
868 * URI instance as if by invoking the {@link #URI(String)} constructor;
869 * this may cause a {@link URISyntaxException} to be thrown. </p>
870 *
871 * @param scheme Scheme name
872 * @param ssp Scheme-specific part
873 * @param fragment Fragment
874 *
875 * @throws URISyntaxException
876 * If the URI string constructed from the given components
877 * violates RFC 2396
878 */
879 public URI(String scheme, String ssp, String fragment)
880 throws URISyntaxException
881 {
882 new Parser(toString(scheme, ssp,
883 null, null, null, -1,
884 null, null, fragment))
885 .parse(false);
886 }
887
888 /**
889 * Constructs a simple URI consisting of only a scheme and a pre-validated
890 * path. Provides a fast-path for some internal cases.
891 */
892 URI(String scheme, String path) {
893 assert validSchemeAndPath(scheme, path);
894 this.scheme = scheme;
895 this.path = path;
896 }
897
898 private static boolean validSchemeAndPath(String scheme, String path) {
899 try {
900 URI u = new URI(scheme + ':' + path);
901 return scheme.equals(u.scheme) && path.equals(u.path);
902 } catch (URISyntaxException e) {
903 return false;
904 }
905 }
906
907 /**
908 * Creates a URI by parsing the given string.
909 *
910 * <p> This convenience factory method works as if by invoking the {@link
911 * #URI(String)} constructor; any {@link URISyntaxException} thrown by the
912 * constructor is caught and wrapped in a new {@link
913 * IllegalArgumentException} object, which is then thrown.
914 *
915 * <p> This method is provided for use in situations where it is known that
916 * the given string is a legal URI, for example for URI constants declared
917 * within a program, and so it would be considered a programming error
918 * for the string not to parse as such. The constructors, which throw
919 * {@link URISyntaxException} directly, should be used in situations where a
920 * URI is being constructed from user input or from some other source that
921 * may be prone to errors. </p>
922 *
923 * @param str The string to be parsed into a URI
924 * @return The new URI
925 *
926 * @throws NullPointerException
927 * If {@code str} is {@code null}
928 *
929 * @throws IllegalArgumentException
930 * If the given string violates RFC 2396
931 */
932 public static URI create(String str) {
933 try {
934 return new URI(str);
935 } catch (URISyntaxException x) {
936 throw new IllegalArgumentException(x.getMessage(), x);
937 }
938 }
939
940
941 // -- Operations --
942
943 /**
944 * Attempts to parse this URI's authority component, if defined, into
945 * user-information, host, and port components.
946 *
947 * <p> If this URI's authority component has already been recognized as
948 * being server-based then it will already have been parsed into
949 * user-information, host, and port components. In this case, or if this
950 * URI has no authority component, this method simply returns this URI.
951 *
952 * <p> Otherwise this method attempts once more to parse the authority
953 * component into user-information, host, and port components, and throws
954 * an exception describing why the authority component could not be parsed
955 * in that way.
956 *
957 * <p> This method is provided because the generic URI syntax specified in
958 * <a href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>
959 * cannot always distinguish a malformed server-based authority from a
960 * legitimate registry-based authority. It must therefore treat some
961 * instances of the former as instances of the latter. The authority
962 * component in the URI string {@code "//foo:bar"}, for example, is not a
963 * legal server-based authority but it is legal as a registry-based
964 * authority.
965 *
966 * <p> In many common situations, for example when working URIs that are
967 * known to be either URNs or URLs, the hierarchical URIs being used will
968 * always be server-based. They therefore must either be parsed as such or
969 * treated as an error. In these cases a statement such as
970 *
971 * <blockquote>
972 * {@code URI }<i>u</i>{@code = new URI(str).parseServerAuthority();}
973 * </blockquote>
974 *
975 * <p> can be used to ensure that <i>u</i> always refers to a URI that, if
976 * it has an authority component, has a server-based authority with proper
977 * user-information, host, and port components. Invoking this method also
978 * ensures that if the authority could not be parsed in that way then an
979 * appropriate diagnostic message can be issued based upon the exception
980 * that is thrown. </p>
981 *
982 * @return A URI whose authority field has been parsed
983 * as a server-based authority
984 *
985 * @throws URISyntaxException
986 * If the authority component of this URI is defined
987 * but cannot be parsed as a server-based authority
988 * according to RFC 2396
989 *
990 * @spec https://www.rfc-editor.org/info/rfc2396
991 * RFC 2396: Uniform Resource Identifiers (URI): Generic Syntax
992 */
993 public URI parseServerAuthority()
994 throws URISyntaxException
995 {
996 // We could be clever and cache the error message and index from the
997 // exception thrown during the original parse, but that would require
998 // either more fields or a more-obscure representation.
999 if ((host != null) || (authority == null))
1000 return this;
1001 new Parser(toString()).parse(true);
1002 return this;
1003 }
1004
1005 /**
1006 * Normalizes this URI's path.
1007 *
1008 * <p> If this URI is opaque, or if its path is already in normal form,
1009 * then this URI is returned. Otherwise a new URI is constructed that is
1010 * identical to this URI except that its path is computed by normalizing
1011 * this URI's path in a manner consistent with <a
1012 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
1013 * section 5.2, step 6, sub-steps c through f; that is:
1014 * </p>
1015 *
1016 * <ol>
1017 *
1018 * <li><p> All {@code "."} segments are removed. </p></li>
1019 *
1020 * <li><p> If a {@code ".."} segment is preceded by a non-{@code ".."}
1021 * segment then both of these segments are removed. This step is
1022 * repeated until it is no longer applicable. </p></li>
1023 *
1024 * <li><p> If the path is relative, and if its first segment contains a
1025 * colon character ({@code ':'}), then a {@code "."} segment is
1026 * prepended. This prevents a relative URI with a path such as
1027 * {@code "a:b/c/d"} from later being re-parsed as an opaque URI with a
1028 * scheme of {@code "a"} and a scheme-specific part of {@code "b/c/d"}.
1029 * <b><i>(Deviation from RFC 2396)</i></b> </p></li>
1030 *
1031 * </ol>
1032 *
1033 * <p> A normalized path will begin with one or more {@code ".."} segments
1034 * if there were insufficient non-{@code ".."} segments preceding them to
1035 * allow their removal. A normalized path will begin with a {@code "."}
1036 * segment if one was inserted by step 3 above. Otherwise, a normalized
1037 * path will not contain any {@code "."} or {@code ".."} segments. </p>
1038 *
1039 * @return A URI equivalent to this URI,
1040 * but whose path is in normal form
1041 * @spec https://www.rfc-editor.org/info/rfc2396
1042 * RFC 2396: Uniform Resource Identifiers (URI): Generic Syntax
1043 */
1044 public URI normalize() {
1045 return normalize(this);
1046 }
1047
1048 /**
1049 * Resolves the given URI against this URI.
1050 *
1051 * <p> If the given URI is already absolute, or if this URI is opaque, then
1052 * the given URI is returned.
1053 *
1054 * <p><a id="resolve-frag"></a> If the given URI's fragment component is
1055 * defined, its path component is empty, and its scheme, authority, and
1056 * query components are undefined, then a URI with the given fragment but
1057 * with all other components equal to those of this URI is returned. This
1058 * allows a URI representing a standalone fragment reference, such as
1059 * {@code "#foo"}, to be usefully resolved against a base URI.
1060 *
1061 * <p> Otherwise this method constructs a new hierarchical URI in a manner
1062 * consistent with <a
1063 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
1064 * section 5.2; that is: </p>
1065 *
1066 * <ol>
1067 *
1068 * <li><p> A new URI is constructed with this URI's scheme and the given
1069 * URI's query and fragment components. </p></li>
1070 *
1071 * <li><p> If the given URI has an authority component then the new URI's
1072 * authority and path are taken from the given URI. </p></li>
1073 *
1074 * <li><p> Otherwise the new URI's authority component is copied from
1075 * this URI, and its path is computed as follows: </p>
1076 *
1077 * <ol>
1078 *
1079 * <li><p> If the given URI's path is absolute then the new URI's path
1080 * is taken from the given URI. </p></li>
1081 *
1082 * <li><p> Otherwise the given URI's path is relative, and so the new
1083 * URI's path is computed by resolving the path of the given URI
1084 * against the path of this URI. This is done by concatenating all but
1085 * the last segment of this URI's path, if any, with the given URI's
1086 * path and then normalizing the result as if by invoking the {@link
1087 * #normalize() normalize} method. </p></li>
1088 *
1089 * </ol></li>
1090 *
1091 * </ol>
1092 *
1093 * <p> The result of this method is absolute if, and only if, either this
1094 * URI is absolute or the given URI is absolute. </p>
1095 *
1096 * @param uri The URI to be resolved against this URI
1097 * @return The resulting URI
1098 *
1099 * @throws NullPointerException
1100 * If {@code uri} is {@code null}
1101 * @spec https://www.rfc-editor.org/info/rfc2396
1102 * RFC 2396: Uniform Resource Identifiers (URI): Generic Syntax
1103 */
1104 public URI resolve(URI uri) {
1105 return resolve(this, uri);
1106 }
1107
1108 /**
1109 * Constructs a new URI by parsing the given string and then resolving it
1110 * against this URI.
1111 *
1112 * <p> This convenience method works as if invoking it were equivalent to
1113 * evaluating the expression {@link #resolve(java.net.URI)
1114 * resolve}{@code (URI.}{@link #create(String) create}{@code (str))}. </p>
1115 *
1116 * @param str The string to be parsed into a URI
1117 * @return The resulting URI
1118 *
1119 * @throws NullPointerException
1120 * If {@code str} is {@code null}
1121 *
1122 * @throws IllegalArgumentException
1123 * If the given string violates RFC 2396
1124 */
1125 public URI resolve(String str) {
1126 return resolve(URI.create(str));
1127 }
1128
1129 /**
1130 * Relativizes the given URI against this URI.
1131 *
1132 * <p> The relativization of the given URI against this URI is computed as
1133 * follows: </p>
1134 *
1135 * <ol>
1136 *
1137 * <li><p> If either this URI or the given URI are opaque, or if the
1138 * scheme and authority components of the two URIs are not identical, or
1139 * if the path of this URI is not a prefix of the path of the given URI,
1140 * then the given URI is returned. </p></li>
1141 *
1142 * <li><p> Otherwise a new relative hierarchical URI is constructed with
1143 * query and fragment components taken from the given URI and with a path
1144 * component computed by removing this URI's path from the beginning of
1145 * the given URI's path. </p></li>
1146 *
1147 * </ol>
1148 *
1149 * @param uri The URI to be relativized against this URI
1150 * @return The resulting URI
1151 *
1152 * @throws NullPointerException
1153 * If {@code uri} is {@code null}
1154 */
1155 public URI relativize(URI uri) {
1156 return relativize(this, uri);
1157 }
1158
1159 /**
1160 * Constructs a URL from this URI.
1161 *
1162 * <p> This convenience method works as if invoking it were equivalent to
1163 * evaluating the expression {@code new URL(this.toString())} after
1164 * first checking that this URI is absolute. </p>
1165 *
1166 * @return A URL constructed from this URI
1167 *
1168 * @throws IllegalArgumentException
1169 * If this URL is not absolute
1170 *
1171 * @throws MalformedURLException
1172 * If a protocol handler for the URL could not be found,
1173 * or if some other error occurred while constructing the URL
1174 */
1175 public URL toURL() throws MalformedURLException {
1176 return URL.of(this, null);
1177 }
1178
1179 // -- Component access methods --
1180
1181 /**
1182 * Returns the scheme component of this URI.
1183 *
1184 * <p> The scheme component of a URI, if defined, only contains characters
1185 * in the <i>alphanum</i> category and in the string {@code "-.+"}. A
1186 * scheme always starts with an <i>alpha</i> character. <p>
1187 *
1188 * The scheme component of a URI cannot contain escaped octets, hence this
1189 * method does not perform any decoding.
1190 *
1191 * @return The scheme component of this URI,
1192 * or {@code null} if the scheme is undefined
1193 */
1194 public String getScheme() {
1195 return scheme;
1196 }
1197
1198 /**
1199 * Tells whether or not this URI is absolute.
1200 *
1201 * <p> A URI is absolute if, and only if, it has a scheme component. </p>
1202 *
1203 * @return {@code true} if, and only if, this URI is absolute
1204 */
1205 public boolean isAbsolute() {
1206 return scheme != null;
1207 }
1208
1209 /**
1210 * Tells whether or not this URI is opaque.
1211 *
1212 * <p> A URI is opaque if, and only if, it is absolute and its
1213 * scheme-specific part does not begin with a slash character ('/').
1214 * An opaque URI has a scheme, a scheme-specific part, and possibly
1215 * a fragment; all other components are undefined. </p>
1216 *
1217 * @return {@code true} if, and only if, this URI is opaque
1218 */
1219 public boolean isOpaque() {
1220 return path == null;
1221 }
1222
1223 /**
1224 * Returns the raw scheme-specific part of this URI. The scheme-specific
1225 * part is never undefined, though it may be empty.
1226 *
1227 * <p> The scheme-specific part of a URI only contains legal URI
1228 * characters. </p>
1229 *
1230 * @return The raw scheme-specific part of this URI
1231 * (never {@code null})
1232 */
1233 public String getRawSchemeSpecificPart() {
1234 String part = schemeSpecificPart;
1235 if (part != null) {
1236 return part;
1237 }
1238
1239 String s = string;
1240 if (s != null) {
1241 // if string is defined, components will have been parsed
1242 int start = 0;
1243 int end = s.length();
1244 if (scheme != null) {
1245 start = scheme.length() + 1;
1246 }
1247 if (fragment != null) {
1248 end -= fragment.length() + 1;
1249 }
1250 if (path != null && path.length() == end - start) {
1251 part = path;
1252 } else {
1253 part = s.substring(start, end);
1254 }
1255 } else {
1256 StringBuilder sb = new StringBuilder();
1257 appendSchemeSpecificPart(sb, null, getAuthority(), getUserInfo(),
1258 host, port, getPath(), getQuery());
1259 part = sb.toString();
1260 }
1261 return schemeSpecificPart = part;
1262 }
1263
1264 /**
1265 * Returns the decoded scheme-specific part of this URI.
1266 *
1267 * <p> The string returned by this method is equal to that returned by the
1268 * {@link #getRawSchemeSpecificPart() getRawSchemeSpecificPart} method
1269 * except that all sequences of escaped octets are <a
1270 * href="#decode">decoded</a>. </p>
1271 *
1272 * @return The decoded scheme-specific part of this URI
1273 * (never {@code null})
1274 */
1275 public String getSchemeSpecificPart() {
1276 String part = decodedSchemeSpecificPart;
1277 if (part == null) {
1278 decodedSchemeSpecificPart = part = decode(getRawSchemeSpecificPart());
1279 }
1280 return part;
1281 }
1282
1283 /**
1284 * Returns the raw authority component of this URI.
1285 *
1286 * <p> The authority component of a URI, if defined, only contains the
1287 * commercial-at character ({@code '@'}) and characters in the
1288 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, and <i>other</i>
1289 * categories. If the authority is server-based then it is further
1290 * constrained to have valid user-information, host, and port
1291 * components. </p>
1292 *
1293 * @return The raw authority component of this URI,
1294 * or {@code null} if the authority is undefined
1295 */
1296 public String getRawAuthority() {
1297 return authority;
1298 }
1299
1300 /**
1301 * Returns the decoded authority component of this URI.
1302 *
1303 * <p> The string returned by this method is equal to that returned by the
1304 * {@link #getRawAuthority() getRawAuthority} method except that all
1305 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1306 *
1307 * @return The decoded authority component of this URI,
1308 * or {@code null} if the authority is undefined
1309 */
1310 public String getAuthority() {
1311 String auth = decodedAuthority;
1312 if ((auth == null) && (authority != null)) {
1313 decodedAuthority = auth = decode(authority);
1314 }
1315 return auth;
1316 }
1317
1318 /**
1319 * Returns the raw user-information component of this URI.
1320 *
1321 * <p> The user-information component of a URI, if defined, only contains
1322 * characters in the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, and
1323 * <i>other</i> categories. </p>
1324 *
1325 * @return The raw user-information component of this URI,
1326 * or {@code null} if the user information is undefined
1327 */
1328 public String getRawUserInfo() {
1329 return userInfo;
1330 }
1331
1332 /**
1333 * Returns the decoded user-information component of this URI.
1334 *
1335 * <p> The string returned by this method is equal to that returned by the
1336 * {@link #getRawUserInfo() getRawUserInfo} method except that all
1337 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1338 *
1339 * @return The decoded user-information component of this URI,
1340 * or {@code null} if the user information is undefined
1341 */
1342 public String getUserInfo() {
1343 String user = decodedUserInfo;
1344 if ((user == null) && (userInfo != null)) {
1345 decodedUserInfo = user = decode(userInfo);
1346 }
1347 return user;
1348 }
1349
1350 /**
1351 * Returns the host component of this URI.
1352 *
1353 * <p> The host component of a URI, if defined, will have one of the
1354 * following forms: </p>
1355 *
1356 * <ul>
1357 *
1358 * <li><p> A domain name consisting of one or more <i>labels</i>
1359 * separated by period characters ({@code '.'}), optionally followed by
1360 * a period character. Each label consists of <i>alphanum</i> characters
1361 * as well as hyphen characters ({@code '-'}), though hyphens never
1362 * occur as the first or last characters in a label. The rightmost
1363 * label of a domain name consisting of two or more labels, begins
1364 * with an <i>alpha</i> character. </li>
1365 *
1366 * <li><p> A dotted-quad IPv4 address of the form
1367 * <i>digit</i>{@code +.}<i>digit</i>{@code +.}<i>digit</i>{@code +.}<i>digit</i>{@code +},
1368 * where no <i>digit</i> sequence is longer than three characters and no
1369 * sequence has a value larger than 255. </p></li>
1370 *
1371 * <li><p> An IPv6 address enclosed in square brackets ({@code '['} and
1372 * {@code ']'}) and consisting of hexadecimal digits, colon characters
1373 * ({@code ':'}), and possibly an embedded IPv4 address. The full
1374 * syntax of IPv6 addresses is specified in <a
1375 * href="http://www.ietf.org/rfc/rfc2373.txt"><i>RFC 2373: IPv6
1376 * Addressing Architecture</i></a>. </p></li>
1377 *
1378 * </ul>
1379 *
1380 * The host component of a URI cannot contain escaped octets, hence this
1381 * method does not perform any decoding.
1382 *
1383 * @return The host component of this URI,
1384 * or {@code null} if the host is undefined
1385 * @spec https://www.rfc-editor.org/info/rfc2373
1386 * RFC 2373: IP Version 6 Addressing Architecture
1387 */
1388 public String getHost() {
1389 return host;
1390 }
1391
1392 /**
1393 * Returns the port number of this URI.
1394 *
1395 * <p> The port component of a URI, if defined, is a non-negative
1396 * integer. </p>
1397 *
1398 * @return The port component of this URI,
1399 * or {@code -1} if the port is undefined
1400 */
1401 public int getPort() {
1402 return port;
1403 }
1404
1405 /**
1406 * Returns the raw path component of this URI.
1407 *
1408 * <p> The path component of a URI, if defined, only contains the slash
1409 * character ({@code '/'}), the commercial-at character ({@code '@'}),
1410 * and characters in the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>,
1411 * and <i>other</i> categories. </p>
1412 *
1413 * @return The path component of this URI,
1414 * or {@code null} if the path is undefined
1415 */
1416 public String getRawPath() {
1417 return path;
1418 }
1419
1420 /**
1421 * Returns the decoded path component of this URI.
1422 *
1423 * <p> The string returned by this method is equal to that returned by the
1424 * {@link #getRawPath() getRawPath} method except that all sequences of
1425 * escaped octets are <a href="#decode">decoded</a>. </p>
1426 *
1427 * @return The decoded path component of this URI,
1428 * or {@code null} if the path is undefined
1429 */
1430 public String getPath() {
1431 String decoded = decodedPath;
1432 if ((decoded == null) && (path != null)) {
1433 decodedPath = decoded = decode(path);
1434 }
1435 return decoded;
1436 }
1437
1438 /**
1439 * Returns the raw query component of this URI.
1440 *
1441 * <p> The query component of a URI, if defined, only contains legal URI
1442 * characters. </p>
1443 *
1444 * @return The raw query component of this URI,
1445 * or {@code null} if the query is undefined
1446 */
1447 public String getRawQuery() {
1448 return query;
1449 }
1450
1451 /**
1452 * Returns the decoded query component of this URI.
1453 *
1454 * <p> The string returned by this method is equal to that returned by the
1455 * {@link #getRawQuery() getRawQuery} method except that all sequences of
1456 * escaped octets are <a href="#decode">decoded</a>. </p>
1457 *
1458 * @return The decoded query component of this URI,
1459 * or {@code null} if the query is undefined
1460 */
1461 public String getQuery() {
1462 String decoded = decodedQuery;
1463 if ((decoded == null) && (query != null)) {
1464 decodedQuery = decoded = decode(query, false);
1465 }
1466 return decoded;
1467 }
1468
1469 /**
1470 * Returns the raw fragment component of this URI.
1471 *
1472 * <p> The fragment component of a URI, if defined, only contains legal URI
1473 * characters. </p>
1474 *
1475 * @return The raw fragment component of this URI,
1476 * or {@code null} if the fragment is undefined
1477 */
1478 public String getRawFragment() {
1479 return fragment;
1480 }
1481
1482 /**
1483 * Returns the decoded fragment component of this URI.
1484 *
1485 * <p> The string returned by this method is equal to that returned by the
1486 * {@link #getRawFragment() getRawFragment} method except that all
1487 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1488 *
1489 * @return The decoded fragment component of this URI,
1490 * or {@code null} if the fragment is undefined
1491 */
1492 public String getFragment() {
1493 String decoded = decodedFragment;
1494 if ((decoded == null) && (fragment != null)) {
1495 decodedFragment = decoded = decode(fragment, false);
1496 }
1497 return decoded;
1498 }
1499
1500
1501 // -- Equality, comparison, hash code, toString, and serialization --
1502
1503 /**
1504 * Tests this URI for equality with another object.
1505 *
1506 * <p> If the given object is not a URI then this method immediately
1507 * returns {@code false}.
1508 *
1509 * <p> For two URIs to be considered equal requires that either both are
1510 * opaque or both are hierarchical. Their schemes must either both be
1511 * undefined or else be equal without regard to case. Their fragments
1512 * must either both be undefined or else be equal.
1513 *
1514 * <p> For two opaque URIs to be considered equal, their scheme-specific
1515 * parts must be equal.
1516 *
1517 * <p> For two hierarchical URIs to be considered equal, their paths must
1518 * be equal and their queries must either both be undefined or else be
1519 * equal. Their authorities must either both be undefined, or both be
1520 * registry-based, or both be server-based. If their authorities are
1521 * defined and are registry-based, then they must be equal. If their
1522 * authorities are defined and are server-based, then their hosts must be
1523 * equal without regard to case, their port numbers must be equal, and
1524 * their user-information components must be equal.
1525 *
1526 * <p> When testing the user-information, path, query, fragment, authority,
1527 * or scheme-specific parts of two URIs for equality, the raw forms rather
1528 * than the encoded forms of these components are compared and the
1529 * hexadecimal digits of escaped octets are compared without regard to
1530 * case.
1531 *
1532 * <p> This method satisfies the general contract of the {@link
1533 * java.lang.Object#equals(Object) Object.equals} method. </p>
1534 *
1535 * @param ob The object to which this object is to be compared
1536 *
1537 * @return {@code true} if, and only if, the given object is a URI that
1538 * is identical to this URI
1539 */
1540 public boolean equals(Object ob) {
1541 if (ob == this)
1542 return true;
1543 if (!(ob instanceof URI that))
1544 return false;
1545 if (this.isOpaque() != that.isOpaque()) return false;
1546 if (!equalIgnoringCase(this.scheme, that.scheme)) return false;
1547 if (!equal(this.fragment, that.fragment)) return false;
1548
1549 // Opaque
1550 if (this.isOpaque())
1551 return equal(this.schemeSpecificPart, that.schemeSpecificPart);
1552
1553 // Hierarchical
1554 if (!equal(this.path, that.path)) return false;
1555 if (!equal(this.query, that.query)) return false;
1556
1557 // Authorities
1558 if (this.authority == that.authority) return true;
1559 if (this.host != null) {
1560 // Server-based
1561 if (!equal(this.userInfo, that.userInfo)) return false;
1562 if (!equalIgnoringCase(this.host, that.host)) return false;
1563 if (this.port != that.port) return false;
1564 } else if (this.authority != null) {
1565 // Registry-based
1566 if (!equal(this.authority, that.authority)) return false;
1567 } else if (this.authority != that.authority) {
1568 return false;
1569 }
1570
1571 return true;
1572 }
1573
1574 /**
1575 * Returns a hash-code value for this URI. The hash code is based upon all
1576 * of the URI's components, and satisfies the general contract of the
1577 * {@link java.lang.Object#hashCode() Object.hashCode} method.
1578 *
1579 * @return A hash-code value for this URI
1580 */
1581 public int hashCode() {
1582 int h = hash;
1583 if (h == 0) {
1584 h = hashIgnoringCase(0, scheme);
1585 h = hash(h, fragment);
1586 if (isOpaque()) {
1587 h = hash(h, schemeSpecificPart);
1588 } else {
1589 h = hash(h, path);
1590 h = hash(h, query);
1591 if (host != null) {
1592 h = hash(h, userInfo);
1593 h = hashIgnoringCase(h, host);
1594 h += 1949 * port;
1595 } else {
1596 h = hash(h, authority);
1597 }
1598 }
1599 if (h != 0) {
1600 hash = h;
1601 }
1602 }
1603 return h;
1604 }
1605
1606 /**
1607 * Compares this URI to another object, which must be a URI.
1608 *
1609 * <p> When comparing corresponding components of two URIs, if one
1610 * component is undefined but the other is defined then the first is
1611 * considered to be less than the second. Unless otherwise noted, string
1612 * components are ordered according to their natural, case-sensitive
1613 * ordering as defined by the {@link java.lang.String#compareTo(String)
1614 * String.compareTo} method. String components that are subject to
1615 * encoding are compared by comparing their raw forms rather than their
1616 * encoded forms and the hexadecimal digits of escaped octets are compared
1617 * without regard to case.
1618 *
1619 * <p> The ordering of URIs is defined as follows: </p>
1620 *
1621 * <ul>
1622 *
1623 * <li><p> Two URIs with different schemes are ordered according the
1624 * ordering of their schemes, without regard to case. </p></li>
1625 *
1626 * <li><p> A hierarchical URI is considered to be less than an opaque URI
1627 * with an identical scheme. </p></li>
1628 *
1629 * <li><p> Two opaque URIs with identical schemes are ordered according
1630 * to the ordering of their scheme-specific parts. </p></li>
1631 *
1632 * <li><p> Two opaque URIs with identical schemes and scheme-specific
1633 * parts are ordered according to the ordering of their
1634 * fragments. </p></li>
1635 *
1636 * <li><p> Two hierarchical URIs with identical schemes are ordered
1637 * according to the ordering of their authority components: </p>
1638 *
1639 * <ul>
1640 *
1641 * <li><p> If both authority components are server-based then the URIs
1642 * are ordered according to their user-information components; if these
1643 * components are identical then the URIs are ordered according to the
1644 * ordering of their hosts, without regard to case; if the hosts are
1645 * identical then the URIs are ordered according to the ordering of
1646 * their ports. </p></li>
1647 *
1648 * <li><p> If one or both authority components are registry-based then
1649 * the URIs are ordered according to the ordering of their authority
1650 * components. </p></li>
1651 *
1652 * </ul></li>
1653 *
1654 * <li><p> Finally, two hierarchical URIs with identical schemes and
1655 * authority components are ordered according to the ordering of their
1656 * paths; if their paths are identical then they are ordered according to
1657 * the ordering of their queries; if the queries are identical then they
1658 * are ordered according to the order of their fragments. </p></li>
1659 *
1660 * </ul>
1661 *
1662 * <p> This method satisfies the general contract of the {@link
1663 * java.lang.Comparable#compareTo(Object) Comparable.compareTo}
1664 * method. </p>
1665 *
1666 * @param that
1667 * The object to which this URI is to be compared
1668 *
1669 * @return A negative integer, zero, or a positive integer as this URI is
1670 * less than, equal to, or greater than the given URI
1671 *
1672 * @throws ClassCastException
1673 * If the given object is not a URI
1674 */
1675 public int compareTo(URI that) {
1676 int c;
1677
1678 if ((c = compareIgnoringCase(this.scheme, that.scheme)) != 0)
1679 return c;
1680
1681 if (this.isOpaque()) {
1682 if (that.isOpaque()) {
1683 // Both opaque
1684 if ((c = compare(this.schemeSpecificPart,
1685 that.schemeSpecificPart)) != 0)
1686 return c;
1687 return compare(this.fragment, that.fragment);
1688 }
1689 return +1; // Opaque > hierarchical
1690 } else if (that.isOpaque()) {
1691 return -1; // Hierarchical < opaque
1692 }
1693
1694 // Hierarchical
1695 if ((this.host != null) && (that.host != null)) {
1696 // Both server-based
1697 if ((c = compare(this.userInfo, that.userInfo)) != 0)
1698 return c;
1699 if ((c = compareIgnoringCase(this.host, that.host)) != 0)
1700 return c;
1701 if ((c = this.port - that.port) != 0)
1702 return c;
1703 } else {
1704 // If one or both authorities are registry-based then we simply
1705 // compare them in the usual, case-sensitive way. If one is
1706 // registry-based and one is server-based then the strings are
1707 // guaranteed to be unequal, hence the comparison will never return
1708 // zero and the compareTo and equals methods will remain
1709 // consistent.
1710 if ((c = compare(this.authority, that.authority)) != 0) return c;
1711 }
1712
1713 if ((c = compare(this.path, that.path)) != 0) return c;
1714 if ((c = compare(this.query, that.query)) != 0) return c;
1715 return compare(this.fragment, that.fragment);
1716 }
1717
1718 /**
1719 * Returns the content of this URI as a string.
1720 *
1721 * <p> If this URI was created by invoking one of the constructors in this
1722 * class then a string equivalent to the original input string, or to the
1723 * string computed from the originally-given components, as appropriate, is
1724 * returned. Otherwise this URI was created by normalization, resolution,
1725 * or relativization, and so a string is constructed from this URI's
1726 * components according to the rules specified in <a
1727 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
1728 * section 5.2, step 7. </p>
1729 *
1730 * @return The string form of this URI
1731 * @spec https://www.rfc-editor.org/info/rfc2396
1732 * RFC 2396: Uniform Resource Identifiers (URI): Generic Syntax
1733 */
1734 public String toString() {
1735 String s = string;
1736 if (s == null) {
1737 s = defineString();
1738 }
1739 return s;
1740 }
1741
1742 private String defineString() {
1743 String s = string;
1744 if (s != null) {
1745 return s;
1746 }
1747
1748 StringBuilder sb = new StringBuilder();
1749 if (scheme != null) {
1750 sb.append(scheme);
1751 sb.append(':');
1752 }
1753 if (isOpaque()) {
1754 sb.append(schemeSpecificPart);
1755 } else {
1756 if (host != null) {
1757 sb.append("//");
1758 if (userInfo != null) {
1759 sb.append(userInfo);
1760 sb.append('@');
1761 }
1762 boolean needBrackets = ((host.indexOf(':') >= 0)
1763 && !host.startsWith("[")
1764 && !host.endsWith("]"));
1765 if (needBrackets) sb.append('[');
1766 sb.append(host);
1767 if (needBrackets) sb.append(']');
1768 if (port != -1) {
1769 sb.append(':');
1770 sb.append(port);
1771 }
1772 } else if (authority != null) {
1773 sb.append("//");
1774 sb.append(authority);
1775 }
1776 if (path != null)
1777 sb.append(path);
1778 if (query != null) {
1779 sb.append('?');
1780 sb.append(query);
1781 }
1782 }
1783 if (fragment != null) {
1784 sb.append('#');
1785 sb.append(fragment);
1786 }
1787 return string = sb.toString();
1788 }
1789
1790 /**
1791 * Returns the content of this URI as a US-ASCII string.
1792 *
1793 * <p> If this URI does not contain any characters in the <i>other</i>
1794 * category then an invocation of this method will return the same value as
1795 * an invocation of the {@link #toString() toString} method. Otherwise
1796 * this method works as if by invoking that method and then <a
1797 * href="#encode">encoding</a> the result. </p>
1798 *
1799 * @return The string form of this URI, encoded as needed
1800 * so that it only contains characters in the US-ASCII
1801 * charset
1802 */
1803 public String toASCIIString() {
1804 return encode(toString());
1805 }
1806
1807
1808 // -- Serialization support --
1809
1810 /**
1811 * Saves the content of this URI to the given serial stream.
1812 *
1813 * <p> The only serializable field of a URI instance is its {@code string}
1814 * field. That field is given a value, if it does not have one already,
1815 * and then the {@link java.io.ObjectOutputStream#defaultWriteObject()}
1816 * method of the given object-output stream is invoked. </p>
1817 *
1818 * @param os The object-output stream to which this object
1819 * is to be written
1820 *
1821 * @throws IOException
1822 * If an I/O error occurs
1823 */
1824 @java.io.Serial
1825 private void writeObject(ObjectOutputStream os)
1826 throws IOException
1827 {
1828 defineString();
1829 os.defaultWriteObject(); // Writes the string field only
1830 }
1831
1832 /**
1833 * Reconstitutes a URI from the given serial stream.
1834 *
1835 * <p> The {@link java.io.ObjectInputStream#defaultReadObject()} method is
1836 * invoked to read the value of the {@code string} field. The result is
1837 * then parsed in the usual way.
1838 *
1839 * @param is The object-input stream from which this object
1840 * is being read
1841 *
1842 * @throws IOException
1843 * If an I/O error occurs
1844 *
1845 * @throws ClassNotFoundException
1846 * If a serialized class cannot be loaded
1847 */
1848 @java.io.Serial
1849 private void readObject(ObjectInputStream is)
1850 throws ClassNotFoundException, IOException
1851 {
1852 port = -1; // Argh
1853 is.defaultReadObject();
1854 try {
1855 new Parser(string).parse(false);
1856 } catch (URISyntaxException x) {
1857 IOException y = new InvalidObjectException("Invalid URI");
1858 y.initCause(x);
1859 throw y;
1860 }
1861 }
1862
1863
1864 // -- End of public methods --
1865
1866
1867 // -- Utility methods for string-field comparison and hashing --
1868
1869 // These methods return appropriate values for null string arguments,
1870 // thereby simplifying the equals, hashCode, and compareTo methods.
1871 //
1872 // The case-ignoring methods should only be applied to strings whose
1873 // characters are all known to be US-ASCII. Because of this restriction,
1874 // these methods are faster than the similar methods in the String class.
1875
1876 // US-ASCII only
1877 private static int toLower(char c) {
1878 if ((c >= 'A') && (c <= 'Z'))
1879 return c + ('a' - 'A');
1880 return c;
1881 }
1882
1883 // US-ASCII only
1884 private static int toUpper(char c) {
1885 if ((c >= 'a') && (c <= 'z'))
1886 return c - ('a' - 'A');
1887 return c;
1888 }
1889
1890 private static boolean equal(String s, String t) {
1891 boolean testForEquality = true;
1892 int result = percentNormalizedComparison(s, t, testForEquality);
1893 return result == 0;
1894 }
1895
1896 // US-ASCII only
1897 private static boolean equalIgnoringCase(String s, String t) {
1898 if (s == t) return true;
1899 if ((s != null) && (t != null)) {
1900 int n = s.length();
1901 if (t.length() != n)
1902 return false;
1903 for (int i = 0; i < n; i++) {
1904 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1905 return false;
1906 }
1907 return true;
1908 }
1909 return false;
1910 }
1911
1912 private static int hash(int hash, String s) {
1913 if (s == null) return hash;
1914 return s.indexOf('%') < 0 ? hash * 127 + s.hashCode()
1915 : normalizedHash(hash, s);
1916 }
1917
1918
1919 private static int normalizedHash(int hash, String s) {
1920 int h = 0;
1921 for (int index = 0; index < s.length(); index++) {
1922 char ch = s.charAt(index);
1923 h = 31 * h + ch;
1924 if (ch == '%') {
1925 /*
1926 * Process the next two encoded characters
1927 */
1928 for (int i = index + 1; i < index + 3; i++)
1929 h = 31 * h + toUpper(s.charAt(i));
1930 index += 2;
1931 }
1932 }
1933 return hash * 127 + h;
1934 }
1935
1936 // US-ASCII only
1937 private static int hashIgnoringCase(int hash, String s) {
1938 if (s == null) return hash;
1939 int h = hash;
1940 int n = s.length();
1941 for (int i = 0; i < n; i++)
1942 h = 31 * h + toLower(s.charAt(i));
1943 return h;
1944 }
1945
1946 private static int compare(String s, String t) {
1947 boolean testForEquality = false;
1948 int result = percentNormalizedComparison(s, t, testForEquality);
1949 return result;
1950 }
1951
1952 // The percentNormalizedComparison method does not verify two
1953 // characters that follow the % sign are hexadecimal digits.
1954 // Reason being:
1955 // 1) percentNormalizedComparison method is not called with
1956 // 'decoded' strings
1957 // 2) The only place where a percent can be followed by anything
1958 // other than hexadecimal digits is in the authority component
1959 // (for a IPv6 scope) and the whole authority component is case
1960 // insensitive.
1961 private static int percentNormalizedComparison(String s, String t,
1962 boolean testForEquality) {
1963
1964 if (s == t) return 0;
1965 if (s != null) {
1966 if (t != null) {
1967 if (s.indexOf('%') < 0) {
1968 return s.compareTo(t);
1969 }
1970 int sn = s.length();
1971 int tn = t.length();
1972 if ((sn != tn) && testForEquality)
1973 return sn - tn;
1974 int val = 0;
1975 int n = Math.min(sn, tn);
1976 for (int i = 0; i < n; ) {
1977 char c = s.charAt(i);
1978 char d = t.charAt(i);
1979 val = c - d;
1980 if (c != '%') {
1981 if (val != 0)
1982 return val;
1983 i++;
1984 continue;
1985 }
1986 if (d != '%') {
1987 if (val != 0)
1988 return val;
1989 }
1990 i++;
1991 val = toLower(s.charAt(i)) - toLower(t.charAt(i));
1992 if (val != 0)
1993 return val;
1994 i++;
1995 val = toLower(s.charAt(i)) - toLower(t.charAt(i));
1996 if (val != 0)
1997 return val;
1998 i++;
1999 }
2000 return sn - tn;
2001 } else
2002 return +1;
2003 } else {
2004 return -1;
2005 }
2006 }
2007
2008 // US-ASCII only
2009 private static int compareIgnoringCase(String s, String t) {
2010 if (s == t) return 0;
2011 if (s != null) {
2012 if (t != null) {
2013 int sn = s.length();
2014 int tn = t.length();
2015 int n = sn < tn ? sn : tn;
2016 for (int i = 0; i < n; i++) {
2017 int c = toLower(s.charAt(i)) - toLower(t.charAt(i));
2018 if (c != 0)
2019 return c;
2020 }
2021 return sn - tn;
2022 }
2023 return +1;
2024 } else {
2025 return -1;
2026 }
2027 }
2028
2029
2030 // -- String construction --
2031
2032 // If a scheme is given then the path, if given, must be absolute
2033 //
2034 private static void checkPath(String s, String scheme, String path)
2035 throws URISyntaxException
2036 {
2037 if (scheme != null) {
2038 if (path != null && !path.isEmpty() && path.charAt(0) != '/')
2039 throw new URISyntaxException(formatMsg("%s", filterNonSocketInfo(s)),
2040 "Relative path in absolute URI");
2041 }
2042 }
2043
2044 private void appendAuthority(StringBuilder sb,
2045 String authority,
2046 String userInfo,
2047 String host,
2048 int port)
2049 {
2050 if (host != null) {
2051 sb.append("//");
2052 if (userInfo != null) {
2053 sb.append(quote(userInfo, L_USERINFO, H_USERINFO));
2054 sb.append('@');
2055 }
2056 boolean needBrackets = ((host.indexOf(':') >= 0)
2057 && !host.startsWith("[")
2058 && !host.endsWith("]"));
2059 if (needBrackets) sb.append('[');
2060 sb.append(host);
2061 if (needBrackets) sb.append(']');
2062 if (port != -1) {
2063 sb.append(':');
2064 sb.append(port);
2065 }
2066 } else if (authority != null) {
2067 sb.append("//");
2068 if (authority.startsWith("[")) {
2069 // authority should (but may not) contain an embedded IPv6 address
2070 int end = authority.indexOf(']');
2071 String doquote = authority;
2072 if (end != -1 && authority.indexOf(':') != -1) {
2073 // the authority contains an IPv6 address
2074 sb.append(authority, 0, end + 1);
2075 doquote = authority.substring(end + 1);
2076 }
2077 sb.append(quote(doquote,
2078 L_REG_NAME | L_SERVER,
2079 H_REG_NAME | H_SERVER));
2080 } else {
2081 sb.append(quote(authority,
2082 L_REG_NAME | L_SERVER,
2083 H_REG_NAME | H_SERVER));
2084 }
2085 }
2086 }
2087
2088 private void appendSchemeSpecificPart(StringBuilder sb,
2089 String opaquePart,
2090 String authority,
2091 String userInfo,
2092 String host,
2093 int port,
2094 String path,
2095 String query)
2096 {
2097 if (opaquePart != null) {
2098 /* check if SSP begins with an IPv6 address
2099 * because we must not quote a literal IPv6 address
2100 */
2101 if (opaquePart.startsWith("//[")) {
2102 int end = opaquePart.indexOf(']');
2103 if (end != -1 && opaquePart.indexOf(':')!=-1) {
2104 String doquote = opaquePart.substring(end + 1);
2105 sb.append(opaquePart, 0, end + 1);
2106 sb.append(quote(doquote, L_URIC, H_URIC));
2107 }
2108 } else {
2109 sb.append(quote(opaquePart, L_URIC, H_URIC));
2110 }
2111 } else {
2112 appendAuthority(sb, authority, userInfo, host, port);
2113 if (path != null)
2114 sb.append(quote(path, L_PATH, H_PATH));
2115 if (query != null) {
2116 sb.append('?');
2117 sb.append(quote(query, L_URIC, H_URIC));
2118 }
2119 }
2120 }
2121
2122 private void appendFragment(StringBuilder sb, String fragment) {
2123 if (fragment != null) {
2124 sb.append('#');
2125 sb.append(quote(fragment, L_URIC, H_URIC));
2126 }
2127 }
2128
2129 private String toString(String scheme,
2130 String opaquePart,
2131 String authority,
2132 String userInfo,
2133 String host,
2134 int port,
2135 String path,
2136 String query,
2137 String fragment)
2138 {
2139 StringBuilder sb = new StringBuilder();
2140 if (scheme != null) {
2141 sb.append(scheme);
2142 sb.append(':');
2143 }
2144 appendSchemeSpecificPart(sb, opaquePart,
2145 authority, userInfo, host, port,
2146 path, query);
2147 appendFragment(sb, fragment);
2148 return sb.toString();
2149 }
2150
2151 // -- Normalization, resolution, and relativization --
2152
2153 // RFC2396 5.2 (6)
2154 private static String resolvePath(String base, String child, boolean absolute)
2155 {
2156 int i = base.lastIndexOf('/');
2157 int cn = child.length();
2158 String path = "";
2159
2160 if (cn == 0) {
2161 // 5.2 (6a)
2162 if (i >= 0)
2163 path = base.substring(0, i + 1);
2164 } else {
2165 // 5.2 (6a-b)
2166 if (i >= 0 || !absolute) {
2167 path = base.substring(0, i + 1).concat(child);
2168 } else {
2169 path = "/".concat(child);
2170 }
2171
2172 }
2173
2174 // 5.2 (6c-f)
2175 String np = normalize(path);
2176
2177 // 5.2 (6g): If the result is absolute but the path begins with "../",
2178 // then we simply leave the path as-is
2179
2180 return np;
2181 }
2182
2183 // RFC2396 5.2
2184 private static URI resolve(URI base, URI child) {
2185 // check if child if opaque first so that NPE is thrown
2186 // if child is null.
2187 if (child.isOpaque() || base.isOpaque())
2188 return child;
2189
2190 // 5.2 (2): Reference to current document (lone fragment)
2191 if ((child.scheme == null) && (child.authority == null)
2192 && child.path.isEmpty() && (child.fragment != null)
2193 && (child.query == null)) {
2194 if ((base.fragment != null)
2195 && child.fragment.equals(base.fragment)) {
2196 return base;
2197 }
2198 URI ru = new URI();
2199 ru.scheme = base.scheme;
2200 ru.authority = base.authority;
2201 ru.userInfo = base.userInfo;
2202 ru.host = base.host;
2203 ru.port = base.port;
2204 ru.path = base.path;
2205 ru.fragment = child.fragment;
2206 ru.query = base.query;
2207 return ru;
2208 }
2209
2210 // 5.2 (3): Child is absolute
2211 if (child.scheme != null)
2212 return child;
2213
2214 URI ru = new URI(); // Resolved URI
2215 ru.scheme = base.scheme;
2216 ru.query = child.query;
2217 ru.fragment = child.fragment;
2218
2219 // 5.2 (4): Authority
2220 if (child.authority == null) {
2221 ru.authority = base.authority;
2222 ru.host = base.host;
2223 ru.userInfo = base.userInfo;
2224 ru.port = base.port;
2225
2226 String cp = child.path;
2227 if (!cp.isEmpty() && cp.charAt(0) == '/') {
2228 // 5.2 (5): Child path is absolute
2229 ru.path = child.path;
2230 } else {
2231 // 5.2 (6): Resolve relative path
2232 ru.path = resolvePath(base.path, cp, base.isAbsolute());
2233 }
2234 } else {
2235 ru.authority = child.authority;
2236 ru.host = child.host;
2237 ru.userInfo = child.userInfo;
2238 ru.port = child.port;
2239 ru.path = child.path;
2240 }
2241
2242 // 5.2 (7): Recombine (nothing to do here)
2243 return ru;
2244 }
2245
2246 // If the given URI's path is normal then return the URI;
2247 // o.w., return a new URI containing the normalized path.
2248 //
2249 private static URI normalize(URI u) {
2250 if (u.isOpaque() || u.path == null || u.path.isEmpty())
2251 return u;
2252
2253 String np = normalize(u.path);
2254 if (np == u.path)
2255 return u;
2256
2257 URI v = new URI();
2258 v.scheme = u.scheme;
2259 v.fragment = u.fragment;
2260 v.authority = u.authority;
2261 v.userInfo = u.userInfo;
2262 v.host = u.host;
2263 v.port = u.port;
2264 v.path = np;
2265 v.query = u.query;
2266 return v;
2267 }
2268
2269 // If both URIs are hierarchical, their scheme and authority components are
2270 // identical, and the base path is a prefix of the child's path, then
2271 // return a relative URI that, when resolved against the base, yields the
2272 // child; otherwise, return the child.
2273 //
2274 private static URI relativize(URI base, URI child) {
2275 // check if child if opaque first so that NPE is thrown
2276 // if child is null.
2277 if (child.isOpaque() || base.isOpaque())
2278 return child;
2279 if (!equalIgnoringCase(base.scheme, child.scheme)
2280 || !equal(base.authority, child.authority))
2281 return child;
2282
2283 String bp = normalize(base.path);
2284 String cp = normalize(child.path);
2285 if (!bp.equals(cp)) {
2286 if (!bp.endsWith("/"))
2287 bp = bp + "/";
2288 if (!cp.startsWith(bp))
2289 return child;
2290 }
2291
2292 URI v = new URI();
2293 v.path = cp.substring(bp.length());
2294 v.query = child.query;
2295 v.fragment = child.fragment;
2296 return v;
2297 }
2298
2299
2300
2301 // -- Path normalization --
2302
2303 // The following algorithm for path normalization avoids the creation of a
2304 // string object for each segment, as well as the use of a string buffer to
2305 // compute the final result, by using a single char array and editing it in
2306 // place. The array is first split into segments, replacing each slash
2307 // with '\0' and creating a segment-index array, each element of which is
2308 // the index of the first char in the corresponding segment. We then walk
2309 // through both arrays, removing ".", "..", and other segments as necessary
2310 // by setting their entries in the index array to -1. Finally, the two
2311 // arrays are used to rejoin the segments and compute the final result.
2312 //
2313 // This code is based upon src/solaris/native/java/io/canonicalize_md.c
2314
2315
2316 // Check the given path to see if it might need normalization. A path
2317 // might need normalization if it contains duplicate slashes, a "."
2318 // segment, or a ".." segment. Return -1 if no further normalization is
2319 // possible, otherwise return the number of segments found.
2320 //
2321 // This method takes a string argument rather than a char array so that
2322 // this test can be performed without invoking path.toCharArray().
2323 //
2324 private static int needsNormalization(String path) {
2325 boolean normal = true;
2326 int ns = 0; // Number of segments
2327 int end = path.length() - 1; // Index of last char in path
2328 int p = 0; // Index of next char in path
2329
2330 // Skip initial slashes
2331 while (p <= end) {
2332 if (path.charAt(p) != '/') break;
2333 p++;
2334 }
2335 if (p > 1) normal = false;
2336
2337 // Scan segments
2338 while (p <= end) {
2339
2340 // Looking at "." or ".." ?
2341 if ((path.charAt(p) == '.')
2342 && ((p == end)
2343 || ((path.charAt(p + 1) == '/')
2344 || ((path.charAt(p + 1) == '.')
2345 && ((p + 1 == end)
2346 || (path.charAt(p + 2) == '/')))))) {
2347 normal = false;
2348 }
2349 ns++;
2350
2351 // Find beginning of next segment
2352 while (p <= end) {
2353 if (path.charAt(p++) != '/')
2354 continue;
2355
2356 // Skip redundant slashes
2357 while (p <= end) {
2358 if (path.charAt(p) != '/') break;
2359 normal = false;
2360 p++;
2361 }
2362
2363 break;
2364 }
2365 }
2366
2367 return normal ? -1 : ns;
2368 }
2369
2370
2371 // Split the given path into segments, replacing slashes with nulls and
2372 // filling in the given segment-index array.
2373 //
2374 // Preconditions:
2375 // segs.length == Number of segments in path
2376 //
2377 // Postconditions:
2378 // All slashes in path replaced by '\0'
2379 // segs[i] == Index of first char in segment i (0 <= i < segs.length)
2380 //
2381 private static void split(char[] path, int[] segs) {
2382 int end = path.length - 1; // Index of last char in path
2383 int p = 0; // Index of next char in path
2384 int i = 0; // Index of current segment
2385
2386 // Skip initial slashes
2387 while (p <= end) {
2388 if (path[p] != '/') break;
2389 path[p] = '\0';
2390 p++;
2391 }
2392
2393 while (p <= end) {
2394
2395 // Note start of segment
2396 segs[i++] = p++;
2397
2398 // Find beginning of next segment
2399 while (p <= end) {
2400 if (path[p++] != '/')
2401 continue;
2402 path[p - 1] = '\0';
2403
2404 // Skip redundant slashes
2405 while (p <= end) {
2406 if (path[p] != '/') break;
2407 path[p++] = '\0';
2408 }
2409 break;
2410 }
2411 }
2412
2413 if (i != segs.length)
2414 throw new InternalError(); // ASSERT
2415 }
2416
2417
2418 // Join the segments in the given path according to the given segment-index
2419 // array, ignoring those segments whose index entries have been set to -1,
2420 // and inserting slashes as needed. Return the length of the resulting
2421 // path.
2422 //
2423 // Preconditions:
2424 // segs[i] == -1 implies segment i is to be ignored
2425 // path computed by split, as above, with '\0' having replaced '/'
2426 //
2427 // Postconditions:
2428 // path[0] .. path[return value] == Resulting path
2429 //
2430 private static int join(char[] path, int[] segs) {
2431 int ns = segs.length; // Number of segments
2432 int end = path.length - 1; // Index of last char in path
2433 int p = 0; // Index of next path char to write
2434
2435 if (path[p] == '\0') {
2436 // Restore initial slash for absolute paths
2437 path[p++] = '/';
2438 }
2439
2440 for (int i = 0; i < ns; i++) {
2441 int q = segs[i]; // Current segment
2442 if (q == -1)
2443 // Ignore this segment
2444 continue;
2445
2446 if (p == q) {
2447 // We're already at this segment, so just skip to its end
2448 while ((p <= end) && (path[p] != '\0'))
2449 p++;
2450 if (p <= end) {
2451 // Preserve trailing slash
2452 path[p++] = '/';
2453 }
2454 } else if (p < q) {
2455 // Copy q down to p
2456 while ((q <= end) && (path[q] != '\0'))
2457 path[p++] = path[q++];
2458 if (q <= end) {
2459 // Preserve trailing slash
2460 path[p++] = '/';
2461 }
2462 } else
2463 throw new InternalError(); // ASSERT false
2464 }
2465
2466 return p;
2467 }
2468
2469
2470 // Remove "." segments from the given path, and remove segment pairs
2471 // consisting of a non-".." segment followed by a ".." segment.
2472 //
2473 private static void removeDots(char[] path, int[] segs) {
2474 int ns = segs.length;
2475 int end = path.length - 1;
2476
2477 for (int i = 0; i < ns; i++) {
2478 int dots = 0; // Number of dots found (0, 1, or 2)
2479
2480 // Find next occurrence of "." or ".."
2481 do {
2482 int p = segs[i];
2483 if (path[p] == '.') {
2484 if (p == end) {
2485 dots = 1;
2486 break;
2487 } else if (path[p + 1] == '\0') {
2488 dots = 1;
2489 break;
2490 } else if ((path[p + 1] == '.')
2491 && ((p + 1 == end)
2492 || (path[p + 2] == '\0'))) {
2493 dots = 2;
2494 break;
2495 }
2496 }
2497 i++;
2498 } while (i < ns);
2499 if ((i > ns) || (dots == 0))
2500 break;
2501
2502 if (dots == 1) {
2503 // Remove this occurrence of "."
2504 segs[i] = -1;
2505 } else {
2506 // If there is a preceding non-".." segment, remove both that
2507 // segment and this occurrence of ".."; otherwise, leave this
2508 // ".." segment as-is.
2509 int j;
2510 for (j = i - 1; j >= 0; j--) {
2511 if (segs[j] != -1) break;
2512 }
2513 if (j >= 0) {
2514 int q = segs[j];
2515 if (!((path[q] == '.')
2516 && (path[q + 1] == '.')
2517 && (path[q + 2] == '\0'))) {
2518 segs[i] = -1;
2519 segs[j] = -1;
2520 }
2521 }
2522 }
2523 }
2524 }
2525
2526
2527 // DEVIATION: If the normalized path is relative, and if the first
2528 // segment could be parsed as a scheme name, then prepend a "." segment
2529 //
2530 private static void maybeAddLeadingDot(char[] path, int[] segs) {
2531
2532 if (path[0] == '\0')
2533 // The path is absolute
2534 return;
2535
2536 int ns = segs.length;
2537 int f = 0; // Index of first segment
2538 while (f < ns) {
2539 if (segs[f] >= 0)
2540 break;
2541 f++;
2542 }
2543 if ((f >= ns) || (f == 0))
2544 // The path is empty, or else the original first segment survived,
2545 // in which case we already know that no leading "." is needed
2546 return;
2547
2548 int p = segs[f];
2549 while ((p < path.length) && (path[p] != ':') && (path[p] != '\0')) p++;
2550 if (p >= path.length || path[p] == '\0')
2551 // No colon in first segment, so no "." needed
2552 return;
2553
2554 // At this point we know that the first segment is unused,
2555 // hence we can insert a "." segment at that position
2556 path[0] = '.';
2557 path[1] = '\0';
2558 segs[0] = 0;
2559 }
2560
2561
2562 // Normalize the given path string. A normal path string has no empty
2563 // segments (i.e., occurrences of "//"), no segments equal to ".", and no
2564 // segments equal to ".." that are preceded by a segment not equal to "..".
2565 // In contrast to Unix-style pathname normalization, for URI paths we
2566 // always retain trailing slashes.
2567 //
2568 private static String normalize(String ps) {
2569
2570 // Does this path need normalization?
2571 int ns = needsNormalization(ps); // Number of segments
2572 if (ns < 0)
2573 // Nope -- just return it
2574 return ps;
2575
2576 char[] path = ps.toCharArray(); // Path in char-array form
2577
2578 // Split path into segments
2579 int[] segs = new int[ns]; // Segment-index array
2580 split(path, segs);
2581
2582 // Remove dots
2583 removeDots(path, segs);
2584
2585 // Prevent scheme-name confusion
2586 maybeAddLeadingDot(path, segs);
2587
2588 // Join the remaining segments and return the result
2589 String s = new String(path, 0, join(path, segs));
2590 if (s.equals(ps)) {
2591 // string was already normalized
2592 return ps;
2593 }
2594 return s;
2595 }
2596
2597
2598
2599 // -- Character classes for parsing --
2600
2601 // RFC2396 precisely specifies which characters in the US-ASCII charset are
2602 // permissible in the various components of a URI reference. We here
2603 // define a set of mask pairs to aid in enforcing these restrictions. Each
2604 // mask pair consists of two longs, a low mask and a high mask. Taken
2605 // together they represent a 128-bit mask, where bit i is set iff the
2606 // character with value i is permitted.
2607 //
2608 // This approach is more efficient than sequentially searching arrays of
2609 // permitted characters. It could be made still more efficient by
2610 // precompiling the mask information so that a character's presence in a
2611 // given mask could be determined by a single table lookup.
2612
2613 // To save startup time, we manually calculate the low-/highMask constants.
2614 // For reference, the following methods were used to calculate the values:
2615
2616 // Compute the low-order mask for the characters in the given string
2617 // private static long lowMask(String chars) {
2618 // int n = chars.length();
2619 // long m = 0;
2620 // for (int i = 0; i < n; i++) {
2621 // char c = chars.charAt(i);
2622 // if (c < 64)
2623 // m |= (1L << c);
2624 // }
2625 // return m;
2626 // }
2627
2628 // Compute the high-order mask for the characters in the given string
2629 // private static long highMask(String chars) {
2630 // int n = chars.length();
2631 // long m = 0;
2632 // for (int i = 0; i < n; i++) {
2633 // char c = chars.charAt(i);
2634 // if ((c >= 64) && (c < 128))
2635 // m |= (1L << (c - 64));
2636 // }
2637 // return m;
2638 // }
2639
2640 // Compute a low-order mask for the characters
2641 // between first and last, inclusive
2642 // private static long lowMask(char first, char last) {
2643 // long m = 0;
2644 // int f = Math.max(Math.min(first, 63), 0);
2645 // int l = Math.max(Math.min(last, 63), 0);
2646 // for (int i = f; i <= l; i++)
2647 // m |= 1L << i;
2648 // return m;
2649 // }
2650
2651 // Compute a high-order mask for the characters
2652 // between first and last, inclusive
2653 // private static long highMask(char first, char last) {
2654 // long m = 0;
2655 // int f = Math.max(Math.min(first, 127), 64) - 64;
2656 // int l = Math.max(Math.min(last, 127), 64) - 64;
2657 // for (int i = f; i <= l; i++)
2658 // m |= 1L << i;
2659 // return m;
2660 // }
2661
2662 // Tell whether the given character is permitted by the given mask pair
2663 private static boolean match(char c, long lowMask, long highMask) {
2664 if (c == 0) // 0 doesn't have a slot in the mask. So, it never matches.
2665 return false;
2666 if (c < 64)
2667 return ((1L << c) & lowMask) != 0;
2668 if (c < 128)
2669 return ((1L << (c - 64)) & highMask) != 0;
2670 return false;
2671 }
2672
2673 // Character-class masks, in reverse order from RFC2396 because
2674 // initializers for static fields cannot make forward references.
2675
2676 // digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" |
2677 // "8" | "9"
2678 private static final long L_DIGIT = 0x3FF000000000000L; // lowMask('0', '9');
2679 private static final long H_DIGIT = 0L;
2680
2681 // upalpha = "A" | "B" | "C" | "D" | "E" | "F" | "G" | "H" | "I" |
2682 // "J" | "K" | "L" | "M" | "N" | "O" | "P" | "Q" | "R" |
2683 // "S" | "T" | "U" | "V" | "W" | "X" | "Y" | "Z"
2684 private static final long L_UPALPHA = 0L;
2685 private static final long H_UPALPHA = 0x7FFFFFEL; // highMask('A', 'Z');
2686
2687 // lowalpha = "a" | "b" | "c" | "d" | "e" | "f" | "g" | "h" | "i" |
2688 // "j" | "k" | "l" | "m" | "n" | "o" | "p" | "q" | "r" |
2689 // "s" | "t" | "u" | "v" | "w" | "x" | "y" | "z"
2690 private static final long L_LOWALPHA = 0L;
2691 private static final long H_LOWALPHA = 0x7FFFFFE00000000L; // highMask('a', 'z');
2692
2693 // alpha = lowalpha | upalpha
2694 private static final long L_ALPHA = L_LOWALPHA | L_UPALPHA;
2695 private static final long H_ALPHA = H_LOWALPHA | H_UPALPHA;
2696
2697 // alphanum = alpha | digit
2698 private static final long L_ALPHANUM = L_DIGIT | L_ALPHA;
2699 private static final long H_ALPHANUM = H_DIGIT | H_ALPHA;
2700
2701 // hex = digit | "A" | "B" | "C" | "D" | "E" | "F" |
2702 // "a" | "b" | "c" | "d" | "e" | "f"
2703 private static final long L_HEX = L_DIGIT;
2704 private static final long H_HEX = 0x7E0000007EL; // highMask('A', 'F') | highMask('a', 'f');
2705
2706 // mark = "-" | "_" | "." | "!" | "~" | "*" | "'" |
2707 // "(" | ")"
2708 private static final long L_MARK = 0x678200000000L; // lowMask("-_.!~*'()");
2709 private static final long H_MARK = 0x4000000080000000L; // highMask("-_.!~*'()");
2710
2711 // unreserved = alphanum | mark
2712 private static final long L_UNRESERVED = L_ALPHANUM | L_MARK;
2713 private static final long H_UNRESERVED = H_ALPHANUM | H_MARK;
2714
2715 // reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" |
2716 // "$" | "," | "[" | "]"
2717 // Added per RFC2732: "[", "]"
2718 private static final long L_RESERVED = 0xAC00985000000000L; // lowMask(";/?:@&=+$,[]");
2719 private static final long H_RESERVED = 0x28000001L; // highMask(";/?:@&=+$,[]");
2720
2721 // The zero'th bit is used to indicate that escape pairs and non-US-ASCII
2722 // characters are allowed; this is handled by the scanEscape method below.
2723 private static final long L_ESCAPED = 1L;
2724 private static final long H_ESCAPED = 0L;
2725
2726 // uric = reserved | unreserved | escaped
2727 private static final long L_URIC = L_RESERVED | L_UNRESERVED | L_ESCAPED;
2728 private static final long H_URIC = H_RESERVED | H_UNRESERVED | H_ESCAPED;
2729
2730 // pchar = unreserved | escaped |
2731 // ":" | "@" | "&" | "=" | "+" | "$" | ","
2732 private static final long L_PCHAR
2733 = L_UNRESERVED | L_ESCAPED | 0x2400185000000000L; // lowMask(":@&=+$,");
2734 private static final long H_PCHAR
2735 = H_UNRESERVED | H_ESCAPED | 0x1L; // highMask(":@&=+$,");
2736
2737 // All valid path characters
2738 private static final long L_PATH = L_PCHAR | 0x800800000000000L; // lowMask(";/");
2739 private static final long H_PATH = H_PCHAR; // highMask(";/") == 0x0L;
2740
2741 // Dash, for use in domainlabel and toplabel
2742 private static final long L_DASH = 0x200000000000L; // lowMask("-");
2743 private static final long H_DASH = 0x0L; // highMask("-");
2744
2745 // Dot, for use in hostnames
2746 private static final long L_DOT = 0x400000000000L; // lowMask(".");
2747 private static final long H_DOT = 0x0L; // highMask(".");
2748
2749 // userinfo = *( unreserved | escaped |
2750 // ";" | ":" | "&" | "=" | "+" | "$" | "," )
2751 private static final long L_USERINFO
2752 = L_UNRESERVED | L_ESCAPED | 0x2C00185000000000L; // lowMask(";:&=+$,");
2753 private static final long H_USERINFO
2754 = H_UNRESERVED | H_ESCAPED; // | highMask(";:&=+$,") == 0L;
2755
2756 // reg_name = 1*( unreserved | escaped | "$" | "," |
2757 // ";" | ":" | "@" | "&" | "=" | "+" )
2758 private static final long L_REG_NAME
2759 = L_UNRESERVED | L_ESCAPED | 0x2C00185000000000L; // lowMask("$,;:@&=+");
2760 private static final long H_REG_NAME
2761 = H_UNRESERVED | H_ESCAPED | 0x1L; // highMask("$,;:@&=+");
2762
2763 // All valid characters for server-based authorities
2764 private static final long L_SERVER
2765 = L_USERINFO | L_ALPHANUM | L_DASH | 0x400400000000000L; // lowMask(".:@[]");
2766 private static final long H_SERVER
2767 = H_USERINFO | H_ALPHANUM | H_DASH | 0x28000001L; // highMask(".:@[]");
2768
2769 // Special case of server authority that represents an IPv6 address
2770 // In this case, a % does not signify an escape sequence
2771 private static final long L_SERVER_PERCENT
2772 = L_SERVER | 0x2000000000L; // lowMask("%");
2773 private static final long H_SERVER_PERCENT
2774 = H_SERVER; // | highMask("%") == 0L;
2775
2776 // scheme = alpha *( alpha | digit | "+" | "-" | "." )
2777 private static final long L_SCHEME = L_ALPHA | L_DIGIT | 0x680000000000L; // lowMask("+-.");
2778 private static final long H_SCHEME = H_ALPHA | H_DIGIT; // | highMask("+-.") == 0L
2779
2780 // scope_id = alpha | digit | "_" | "."
2781 private static final long L_SCOPE_ID
2782 = L_ALPHANUM | 0x400000000000L; // lowMask("_.");
2783 private static final long H_SCOPE_ID
2784 = H_ALPHANUM | 0x80000000L; // highMask("_.");
2785
2786 // -- Escaping and encoding --
2787
2788 private static final char[] hexDigits = {
2789 '0', '1', '2', '3', '4', '5', '6', '7',
2790 '8', '9', 'A', 'B', 'C', 'D', 'E', 'F'
2791 };
2792
2793 private static void appendEscape(StringBuilder sb, byte b) {
2794 sb.append('%');
2795 sb.append(hexDigits[(b >> 4) & 0x0f]);
2796 sb.append(hexDigits[(b >> 0) & 0x0f]);
2797 }
2798
2799 private static void appendEncoded(CharsetEncoder encoder, StringBuilder sb, char c) {
2800 ByteBuffer bb = null;
2801 try {
2802 bb = encoder.encode(CharBuffer.wrap(new char[]{c}));
2803 } catch (CharacterCodingException x) {
2804 assert false;
2805 }
2806 while (bb.hasRemaining()) {
2807 int b = bb.get() & 0xff;
2808 if (b >= 0x80)
2809 appendEscape(sb, (byte)b);
2810 else
2811 sb.append((char)b);
2812 }
2813 }
2814
2815 // Quote any characters in s that are not permitted
2816 // by the given mask pair
2817 //
2818 private static String quote(String s, long lowMask, long highMask) {
2819 StringBuilder sb = null;
2820 CharsetEncoder encoder = null;
2821 boolean allowNonASCII = ((lowMask & L_ESCAPED) != 0);
2822 for (int i = 0; i < s.length(); i++) {
2823 char c = s.charAt(i);
2824 if (c < '\u0080') {
2825 if (!match(c, lowMask, highMask)) {
2826 if (sb == null) {
2827 sb = new StringBuilder();
2828 sb.append(s, 0, i);
2829 }
2830 appendEscape(sb, (byte)c);
2831 } else {
2832 if (sb != null)
2833 sb.append(c);
2834 }
2835 } else if (allowNonASCII
2836 && (Character.isSpaceChar(c)
2837 || Character.isISOControl(c))) {
2838 if (encoder == null)
2839 encoder = UTF_8.INSTANCE.newEncoder();
2840 if (sb == null) {
2841 sb = new StringBuilder();
2842 sb.append(s, 0, i);
2843 }
2844 appendEncoded(encoder, sb, c);
2845 } else {
2846 if (sb != null)
2847 sb.append(c);
2848 }
2849 }
2850 return (sb == null) ? s : sb.toString();
2851 }
2852
2853 // Encodes all characters >= \u0080 into escaped, normalized UTF-8 octets,
2854 // assuming that s is otherwise legal
2855 //
2856 private static String encode(String s) {
2857 int n = s.length();
2858 if (n == 0)
2859 return s;
2860
2861 // First check whether we actually need to encode
2862 for (int i = 0;;) {
2863 if (s.charAt(i) >= '\u0080')
2864 break;
2865 if (++i >= n)
2866 return s;
2867 }
2868
2869 String ns = Normalizer.normalize(s, Normalizer.Form.NFC);
2870 ByteBuffer bb = null;
2871 try {
2872 bb = UTF_8.INSTANCE.newEncoder()
2873 .encode(CharBuffer.wrap(ns));
2874
2875 } catch (CharacterCodingException x) {
2876 assert false;
2877 }
2878
2879 StringBuilder sb = new StringBuilder();
2880 while (bb.hasRemaining()) {
2881 int b = bb.get() & 0xff;
2882 if (b >= 0x80)
2883 appendEscape(sb, (byte)b);
2884 else
2885 sb.append((char)b);
2886 }
2887 return sb.toString();
2888 }
2889
2890 private static int decode(char c) {
2891 if ((c >= '0') && (c <= '9'))
2892 return c - '0';
2893 if ((c >= 'a') && (c <= 'f'))
2894 return c - 'a' + 10;
2895 if ((c >= 'A') && (c <= 'F'))
2896 return c - 'A' + 10;
2897 assert false;
2898 return -1;
2899 }
2900
2901 private static byte decode(char c1, char c2) {
2902 return (byte)( ((decode(c1) & 0xf) << 4)
2903 | ((decode(c2) & 0xf) << 0));
2904 }
2905
2906 // Evaluates all escapes in s, applying UTF-8 decoding if needed. Assumes
2907 // that escapes are well-formed syntactically, i.e., of the form %XX. If a
2908 // sequence of escaped octets is not valid UTF-8 then the erroneous octets
2909 // are replaced with '\uFFFD'.
2910 // Exception: any "%" found between "[]" is left alone. It is an IPv6 literal
2911 // with a scope_id
2912 //
2913 private static String decode(String s) {
2914 return decode(s, true);
2915 }
2916
2917 // This method was introduced as a generalization of URI.decode method
2918 // to provide a fix for JDK-8037396
2919 private static String decode(String s, boolean ignorePercentInBrackets) {
2920 if (s == null)
2921 return s;
2922 int n = s.length();
2923 if (n == 0)
2924 return s;
2925 if (s.indexOf('%') < 0)
2926 return s;
2927
2928 StringBuilder sb = new StringBuilder(n);
2929 ByteBuffer bb = ByteBuffer.allocate(n);
2930 CharBuffer cb = CharBuffer.allocate(n);
2931 CharsetDecoder dec = UTF_8.INSTANCE.newDecoder()
2932 .onMalformedInput(CodingErrorAction.REPLACE)
2933 .onUnmappableCharacter(CodingErrorAction.REPLACE);
2934
2935 // This is not horribly efficient, but it will do for now
2936 char c = s.charAt(0);
2937 boolean betweenBrackets = false;
2938
2939 for (int i = 0; i < n;) {
2940 assert c == s.charAt(i); // Loop invariant
2941 if (c == '[') {
2942 betweenBrackets = true;
2943 } else if (betweenBrackets && c == ']') {
2944 betweenBrackets = false;
2945 }
2946 if (c != '%' || (betweenBrackets && ignorePercentInBrackets)) {
2947 sb.append(c);
2948 if (++i >= n)
2949 break;
2950 c = s.charAt(i);
2951 continue;
2952 }
2953 bb.clear();
2954 for (;;) {
2955 assert (n - i >= 2);
2956 bb.put(decode(s.charAt(++i), s.charAt(++i)));
2957 if (++i >= n)
2958 break;
2959 c = s.charAt(i);
2960 if (c != '%')
2961 break;
2962 }
2963 bb.flip();
2964 cb.clear();
2965 dec.reset();
2966 CoderResult cr = dec.decode(bb, cb, true);
2967 assert cr.isUnderflow();
2968 cr = dec.flush(cb);
2969 assert cr.isUnderflow();
2970 sb.append(cb.flip().toString());
2971 }
2972
2973 return sb.toString();
2974 }
2975
2976
2977 // -- Parsing --
2978
2979 // For convenience we wrap the input URI string in a new instance of the
2980 // following internal class. This saves always having to pass the input
2981 // string as an argument to each internal scan/parse method.
2982
2983 private class Parser {
2984
2985 private final String input; // URI input string
2986 private boolean requireServerAuthority = false;
2987
2988 Parser(String s) {
2989 input = s;
2990 string = s;
2991 }
2992
2993 // -- Methods for throwing URISyntaxException in various ways --
2994
2995 private void fail(String reason) throws URISyntaxException {
2996 throw new URISyntaxException(formatMsg("%s", filterNonSocketInfo(input)), reason);
2997 }
2998
2999 private void fail(String reason, int p) throws URISyntaxException {
3000 if (!Exceptions.enhancedNonSocketExceptions()) {
3001 p = -1;
3002 }
3003 throw new URISyntaxException(formatMsg("%s", filterNonSocketInfo(input)), reason, p);
3004 }
3005
3006 private void failExpecting(String expected, int p)
3007 throws URISyntaxException
3008 {
3009 fail("Expected " + expected, p);
3010 }
3011
3012
3013 // -- Simple access to the input string --
3014
3015 // Tells whether start < end and, if so, whether charAt(start) == c
3016 //
3017 private boolean at(int start, int end, char c) {
3018 return (start < end) && (input.charAt(start) == c);
3019 }
3020
3021 // Tells whether start + s.length() < end and, if so,
3022 // whether the chars at the start position match s exactly
3023 //
3024 private boolean at(int start, int end, String s) {
3025 int p = start;
3026 int sn = s.length();
3027 if (sn > end - p)
3028 return false;
3029 int i = 0;
3030 while (i < sn) {
3031 if (input.charAt(p++) != s.charAt(i)) {
3032 break;
3033 }
3034 i++;
3035 }
3036 return (i == sn);
3037 }
3038
3039
3040 // -- Scanning --
3041
3042 // The various scan and parse methods that follow use a uniform
3043 // convention of taking the current start position and end index as
3044 // their first two arguments. The start is inclusive while the end is
3045 // exclusive, just as in the String class, i.e., a start/end pair
3046 // denotes the left-open interval [start, end) of the input string.
3047 //
3048 // These methods never proceed past the end position. They may return
3049 // -1 to indicate outright failure, but more often they simply return
3050 // the position of the first char after the last char scanned. Thus
3051 // a typical idiom is
3052 //
3053 // int p = start;
3054 // int q = scan(p, end, ...);
3055 // if (q > p)
3056 // // We scanned something
3057 // ...;
3058 // else if (q == p)
3059 // // We scanned nothing
3060 // ...;
3061 // else if (q == -1)
3062 // // Something went wrong
3063 // ...;
3064
3065
3066 // Scan a specific char: If the char at the given start position is
3067 // equal to c, return the index of the next char; otherwise, return the
3068 // start position.
3069 //
3070 private int scan(int start, int end, char c) {
3071 if ((start < end) && (input.charAt(start) == c))
3072 return start + 1;
3073 return start;
3074 }
3075
3076 // Scan forward from the given start position. Stop at the first char
3077 // in the err string (in which case -1 is returned), or the first char
3078 // in the stop string (in which case the index of the preceding char is
3079 // returned), or the end of the input string (in which case the length
3080 // of the input string is returned). May return the start position if
3081 // nothing matches.
3082 //
3083 private int scan(int start, int end, String err, String stop) {
3084 int p = start;
3085 while (p < end) {
3086 char c = input.charAt(p);
3087 if (err.indexOf(c) >= 0)
3088 return -1;
3089 if (stop.indexOf(c) >= 0)
3090 break;
3091 p++;
3092 }
3093 return p;
3094 }
3095
3096 // Scan forward from the given start position. Stop at the first char
3097 // in the stop string (in which case the index of the preceding char is
3098 // returned), or the end of the input string (in which case the length
3099 // of the input string is returned). May return the start position if
3100 // nothing matches.
3101 //
3102 private int scan(int start, int end, String stop) {
3103 int p = start;
3104 while (p < end) {
3105 char c = input.charAt(p);
3106 if (stop.indexOf(c) >= 0)
3107 break;
3108 p++;
3109 }
3110 return p;
3111 }
3112
3113 // Scan a potential escape sequence, starting at the given position,
3114 // with the given first char (i.e., charAt(start) == c).
3115 //
3116 // This method assumes that if escapes are allowed then visible
3117 // non-US-ASCII chars are also allowed.
3118 //
3119 private int scanEscape(int start, int n, char first)
3120 throws URISyntaxException
3121 {
3122 int p = start;
3123 char c = first;
3124 if (c == '%') {
3125 // Process escape pair
3126 if ((p + 3 <= n)
3127 && match(input.charAt(p + 1), L_HEX, H_HEX)
3128 && match(input.charAt(p + 2), L_HEX, H_HEX)) {
3129 return p + 3;
3130 }
3131 fail("Malformed escape pair", p);
3132 } else if ((c > 128)
3133 && !Character.isSpaceChar(c)
3134 && !Character.isISOControl(c)) {
3135 // Allow unescaped but visible non-US-ASCII chars
3136 return p + 1;
3137 }
3138 return p;
3139 }
3140
3141 // Scan chars that match the given mask pair
3142 //
3143 private int scan(int start, int n, long lowMask, long highMask)
3144 throws URISyntaxException
3145 {
3146 int p = start;
3147 while (p < n) {
3148 char c = input.charAt(p);
3149 if (match(c, lowMask, highMask)) {
3150 p++;
3151 continue;
3152 }
3153 if ((lowMask & L_ESCAPED) != 0) {
3154 int q = scanEscape(p, n, c);
3155 if (q > p) {
3156 p = q;
3157 continue;
3158 }
3159 }
3160 break;
3161 }
3162 return p;
3163 }
3164
3165 // Check that each of the chars in [start, end) matches the given mask
3166 //
3167 private void checkChars(int start, int end,
3168 long lowMask, long highMask,
3169 String what)
3170 throws URISyntaxException
3171 {
3172 int p = scan(start, end, lowMask, highMask);
3173 if (p < end)
3174 fail("Illegal character in " + what, p);
3175 }
3176
3177 // Check that the char at position p matches the given mask
3178 //
3179 private void checkChar(int p,
3180 long lowMask, long highMask,
3181 String what)
3182 throws URISyntaxException
3183 {
3184 checkChars(p, p + 1, lowMask, highMask, what);
3185 }
3186
3187
3188 // -- Parsing --
3189
3190 // [<scheme>:]<scheme-specific-part>[#<fragment>]
3191 //
3192 void parse(boolean rsa) throws URISyntaxException {
3193 requireServerAuthority = rsa;
3194 int n = input.length();
3195 int p = scan(0, n, "/?#", ":");
3196 if ((p >= 0) && at(p, n, ':')) {
3197 if (p == 0)
3198 failExpecting("scheme name", 0);
3199 checkChar(0, L_ALPHA, H_ALPHA, "scheme name");
3200 checkChars(1, p, L_SCHEME, H_SCHEME, "scheme name");
3201 scheme = input.substring(0, p);
3202 p++; // Skip ':'
3203 if (at(p, n, '/')) {
3204 p = parseHierarchical(p, n);
3205 } else {
3206 // opaque; need to create the schemeSpecificPart
3207 int q = scan(p, n, "#");
3208 if (q <= p)
3209 failExpecting("scheme-specific part", p);
3210 checkChars(p, q, L_URIC, H_URIC, "opaque part");
3211 schemeSpecificPart = input.substring(p, q);
3212 p = q;
3213 }
3214 } else {
3215 p = parseHierarchical(0, n);
3216 }
3217 if (at(p, n, '#')) {
3218 checkChars(p + 1, n, L_URIC, H_URIC, "fragment");
3219 fragment = input.substring(p + 1, n);
3220 p = n;
3221 }
3222 if (p < n)
3223 fail("end of URI", p);
3224 }
3225
3226 // [//authority]<path>[?<query>]
3227 //
3228 // DEVIATION from RFC2396: We allow an empty authority component as
3229 // long as it's followed by a non-empty path, query component, or
3230 // fragment component. This is so that URIs such as "file:///foo/bar"
3231 // will parse. This seems to be the intent of RFC2396, though the
3232 // grammar does not permit it. If the authority is empty then the
3233 // userInfo, host, and port components are undefined.
3234 //
3235 // DEVIATION from RFC2396: We allow empty relative paths. This seems
3236 // to be the intent of RFC2396, but the grammar does not permit it.
3237 // The primary consequence of this deviation is that "#f" parses as a
3238 // relative URI with an empty path.
3239 //
3240 private int parseHierarchical(int start, int n)
3241 throws URISyntaxException
3242 {
3243 int p = start;
3244 if (at(p, n, '/') && at(p + 1, n, '/')) {
3245 p += 2;
3246 int q = scan(p, n, "/?#");
3247 if (q > p) {
3248 p = parseAuthority(p, q);
3249 } else if (q < n) {
3250 // DEVIATION: Allow empty authority prior to non-empty
3251 // path, query component or fragment identifier
3252 } else
3253 failExpecting("authority", p);
3254 }
3255 int q = scan(p, n, "?#"); // DEVIATION: May be empty
3256 checkChars(p, q, L_PATH, H_PATH, "path");
3257 path = input.substring(p, q);
3258 p = q;
3259 if (at(p, n, '?')) {
3260 p++;
3261 q = scan(p, n, "#");
3262 checkChars(p, q, L_URIC, H_URIC, "query");
3263 query = input.substring(p, q);
3264 p = q;
3265 }
3266 return p;
3267 }
3268
3269 // authority = server | reg_name
3270 //
3271 // Ambiguity: An authority that is a registry name rather than a server
3272 // might have a prefix that parses as a server. We use the fact that
3273 // the authority component is always followed by '/' or the end of the
3274 // input string to resolve this: If the complete authority did not
3275 // parse as a server then we try to parse it as a registry name.
3276 //
3277 private int parseAuthority(int start, int n)
3278 throws URISyntaxException
3279 {
3280 int p = start;
3281 int q = p;
3282 int qreg = p;
3283 URISyntaxException ex = null;
3284
3285 boolean serverChars;
3286 boolean regChars;
3287 boolean skipParseException;
3288
3289 if (scan(p, n, "]") > p) {
3290 // contains a literal IPv6 address, therefore % is allowed
3291 serverChars = (scan(p, n, L_SERVER_PERCENT, H_SERVER_PERCENT) == n);
3292 } else {
3293 serverChars = (scan(p, n, L_SERVER, H_SERVER) == n);
3294 }
3295 regChars = ((qreg = scan(p, n, L_REG_NAME, H_REG_NAME)) == n);
3296
3297 if (regChars && !serverChars) {
3298 // Must be a registry-based authority
3299 authority = input.substring(p, n);
3300 return n;
3301 }
3302
3303 // When parsing a URI, skip creating exception objects if the server-based
3304 // authority is not required and the registry parse is successful.
3305 //
3306 skipParseException = (!requireServerAuthority && regChars);
3307 if (serverChars) {
3308 // Might be (probably is) a server-based authority, so attempt
3309 // to parse it as such. If the attempt fails, try to treat it
3310 // as a registry-based authority.
3311 try {
3312 q = parseServer(p, n, skipParseException);
3313 if (q < n) {
3314 if (skipParseException) {
3315 userInfo = null;
3316 host = null;
3317 port = -1;
3318 q = p;
3319 } else {
3320 failExpecting("end of authority", q);
3321 }
3322 } else {
3323 authority = input.substring(p, n);
3324 }
3325 } catch (URISyntaxException x) {
3326 // Undo results of failed parse
3327 userInfo = null;
3328 host = null;
3329 port = -1;
3330 if (requireServerAuthority) {
3331 // If we're insisting upon a server-based authority,
3332 // then just re-throw the exception
3333 throw x;
3334 } else {
3335 // Save the exception in case it doesn't parse as a
3336 // registry either
3337 ex = x;
3338 q = p;
3339 }
3340 }
3341 }
3342
3343 if (q < n) {
3344 if (regChars) {
3345 // Registry-based authority
3346 authority = input.substring(p, n);
3347 } else if (ex != null) {
3348 // Re-throw exception; it was probably due to
3349 // a malformed IPv6 address
3350 throw ex;
3351 } else {
3352 fail("Illegal character in authority", serverChars ? q : qreg);
3353 }
3354 }
3355
3356 return n;
3357 }
3358
3359
3360 // [<userinfo>@]<host>[:<port>]
3361 //
3362 private int parseServer(int start, int n, boolean skipParseException)
3363 throws URISyntaxException
3364 {
3365 int p = start;
3366 int q;
3367
3368 // userinfo
3369 q = scan(p, n, "/?#", "@");
3370 if ((q >= p) && at(q, n, '@')) {
3371 checkChars(p, q, L_USERINFO, H_USERINFO, "user info");
3372 userInfo = input.substring(p, q);
3373 p = q + 1; // Skip '@'
3374 }
3375
3376 // hostname, IPv4 address, or IPv6 address
3377 if (at(p, n, '[')) {
3378 // DEVIATION from RFC2396: Support IPv6 addresses, per RFC2732
3379 p++;
3380 q = scan(p, n, "/?#", "]");
3381 if ((q > p) && at(q, n, ']')) {
3382 // look for a "%" scope id
3383 int r = scan (p, q, "%");
3384 if (r > p) {
3385 parseIPv6Reference(p, r);
3386 if (r+1 == q) {
3387 fail ("scope id expected");
3388 }
3389 checkChars (r+1, q, L_SCOPE_ID, H_SCOPE_ID,
3390 "scope id");
3391 } else {
3392 parseIPv6Reference(p, q);
3393 }
3394 host = input.substring(p-1, q+1);
3395 p = q + 1;
3396 } else {
3397 failExpecting("closing bracket for IPv6 address", q);
3398 }
3399 } else {
3400 q = parseIPv4Address(p, n);
3401 if (q <= p)
3402 q = parseHostname(p, n, skipParseException);
3403 p = q;
3404 }
3405
3406 // port
3407 if (at(p, n, ':')) {
3408 p++;
3409 q = scan(p, n, "/");
3410 if (q > p) {
3411 checkChars(p, q, L_DIGIT, H_DIGIT, "port number");
3412 try {
3413 port = Integer.parseInt(input, p, q, 10);
3414 } catch (NumberFormatException x) {
3415 fail("Malformed port number", p);
3416 }
3417 p = q;
3418 }
3419 } else if (p < n && skipParseException) {
3420 return p;
3421 }
3422
3423 if (p < n)
3424 failExpecting("port number", p);
3425
3426 return p;
3427 }
3428
3429 // Scan a string of decimal digits whose value fits in a byte
3430 //
3431 private int scanByte(int start, int n)
3432 throws URISyntaxException
3433 {
3434 int p = start;
3435 int q = scan(p, n, L_DIGIT, H_DIGIT);
3436 if (q <= p) return q;
3437
3438 // Handle leading zeros
3439 int i = p, j;
3440 while ((j = scan(i, q, '0')) > i) i = j;
3441
3442 // Calculate the number of significant digits (after leading zeros)
3443 int significantDigitsNum = q - i;
3444
3445 if (significantDigitsNum < 3) return q; // definitely < 255
3446
3447 // If more than 3 significant digits, it's definitely > 255
3448 if (significantDigitsNum > 3) return p;
3449
3450 if (Integer.parseInt(input, p, q, 10) > 255) return p;
3451 return q;
3452 }
3453
3454 // Scan an IPv4 address.
3455 //
3456 // If the strict argument is true then we require that the given
3457 // interval contain nothing besides an IPv4 address; if it is false
3458 // then we only require that it start with an IPv4 address.
3459 //
3460 // If the interval does not contain or start with (depending upon the
3461 // strict argument) a legal IPv4 address characters then we return -1
3462 // immediately; otherwise we insist that these characters parse as a
3463 // legal IPv4 address and throw an exception on failure.
3464 //
3465 // We assume that any string of decimal digits and dots must be an IPv4
3466 // address. It won't parse as a hostname anyway, so making that
3467 // assumption here allows more meaningful exceptions to be thrown.
3468 //
3469 private int scanIPv4Address(int start, int n, boolean strict)
3470 throws URISyntaxException
3471 {
3472 int p = start;
3473 int q;
3474 int m = scan(p, n, L_DIGIT | L_DOT, H_DIGIT | H_DOT);
3475 if ((m <= p) || (strict && (m != n)))
3476 return -1;
3477 for (;;) {
3478 // Per RFC2732: At most three digits per byte
3479 // Further constraint: Each element fits in a byte
3480 if ((q = scanByte(p, m)) <= p) break; p = q;
3481 if ((q = scan(p, m, '.')) <= p) break; p = q;
3482 if ((q = scanByte(p, m)) <= p) break; p = q;
3483 if ((q = scan(p, m, '.')) <= p) break; p = q;
3484 if ((q = scanByte(p, m)) <= p) break; p = q;
3485 if ((q = scan(p, m, '.')) <= p) break; p = q;
3486 if ((q = scanByte(p, m)) <= p) break; p = q;
3487 if (q < m) break;
3488 return q;
3489 }
3490 if (strict) fail("Malformed IPv4 address", q);
3491 return -1;
3492 }
3493
3494 // Take an IPv4 address: Throw an exception if the given interval
3495 // contains anything except an IPv4 address
3496 //
3497 private int takeIPv4Address(int start, int n, String expected)
3498 throws URISyntaxException
3499 {
3500 int p = scanIPv4Address(start, n, true);
3501 if (p <= start)
3502 failExpecting(expected, start);
3503 return p;
3504 }
3505
3506 // Attempt to parse an IPv4 address, returning -1 on failure but
3507 // allowing the given interval to contain [:<characters>] after
3508 // the IPv4 address.
3509 //
3510 private int parseIPv4Address(int start, int n) {
3511 int p;
3512
3513 try {
3514 p = scanIPv4Address(start, n, false);
3515 } catch (URISyntaxException | NumberFormatException x) {
3516 return -1;
3517 }
3518
3519 if (p == -1) {
3520 return p;
3521 }
3522
3523 if (p > start && p < n) {
3524 // IPv4 address is followed by something - check that
3525 // it's a ":" as this is the only valid character to
3526 // follow an address.
3527 if (input.charAt(p) != ':') {
3528 return -1;
3529 }
3530 }
3531
3532 if (p > start)
3533 host = input.substring(start, p);
3534
3535 return p;
3536 }
3537
3538 // hostname = domainlabel [ "." ] | 1*( domainlabel "." ) toplabel [ "." ]
3539 // domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum
3540 // toplabel = alpha | alpha *( alphanum | "-" ) alphanum
3541 //
3542 private int parseHostname(int start, int n, boolean skipParseException)
3543 throws URISyntaxException
3544 {
3545 int p = start;
3546 int q;
3547 int l = -1; // Start of last parsed label
3548
3549 do {
3550 // domainlabel = alphanum [ *( alphanum | "-" ) alphanum ]
3551 q = scan(p, n, L_ALPHANUM, H_ALPHANUM);
3552 if (q <= p)
3553 break;
3554 l = p;
3555 p = q;
3556 q = scan(p, n, L_ALPHANUM | L_DASH, H_ALPHANUM | H_DASH);
3557 if (q > p) {
3558 if (input.charAt(q - 1) == '-')
3559 fail("Illegal character in hostname", q - 1);
3560 p = q;
3561 }
3562 q = scan(p, n, '.');
3563 if (q <= p)
3564 break;
3565 p = q;
3566 } while (p < n);
3567
3568 if ((p < n) && !at(p, n, ':')) {
3569 if (skipParseException) {
3570 return p;
3571 }
3572 fail("Illegal character in hostname", p);
3573 }
3574 if (l < 0)
3575 failExpecting("hostname", start);
3576
3577 // for a fully qualified hostname check that the rightmost
3578 // label starts with an alpha character.
3579 if (l > start && !match(input.charAt(l), L_ALPHA, H_ALPHA)) {
3580 fail("Illegal character in hostname", l);
3581 }
3582
3583 host = input.substring(start, p);
3584 return p;
3585 }
3586
3587
3588 // IPv6 address parsing, from RFC2373: IPv6 Addressing Architecture
3589 //
3590 // Bug: The grammar in RFC2373 Appendix B does not allow addresses of
3591 // the form ::12.34.56.78, which are clearly shown in the examples
3592 // earlier in the document. Here is the original grammar:
3593 //
3594 // IPv6address = hexpart [ ":" IPv4address ]
3595 // hexpart = hexseq | hexseq "::" [ hexseq ] | "::" [ hexseq ]
3596 // hexseq = hex4 *( ":" hex4)
3597 // hex4 = 1*4HEXDIG
3598 //
3599 // We therefore use the following revised grammar:
3600 //
3601 // IPv6address = hexseq [ ":" IPv4address ]
3602 // | hexseq [ "::" [ hexpost ] ]
3603 // | "::" [ hexpost ]
3604 // hexpost = hexseq | hexseq ":" IPv4address | IPv4address
3605 // hexseq = hex4 *( ":" hex4)
3606 // hex4 = 1*4HEXDIG
3607 //
3608 // This covers all and only the following cases:
3609 //
3610 // hexseq
3611 // hexseq : IPv4address
3612 // hexseq ::
3613 // hexseq :: hexseq
3614 // hexseq :: hexseq : IPv4address
3615 // hexseq :: IPv4address
3616 // :: hexseq
3617 // :: hexseq : IPv4address
3618 // :: IPv4address
3619 // ::
3620 //
3621 // Additionally we constrain the IPv6 address as follows :-
3622 //
3623 // i. IPv6 addresses without compressed zeros should contain
3624 // exactly 16 bytes.
3625 //
3626 // ii. IPv6 addresses with compressed zeros should contain
3627 // less than 16 bytes.
3628
3629 private int ipv6byteCount = 0;
3630
3631 private int parseIPv6Reference(int start, int n)
3632 throws URISyntaxException
3633 {
3634 int p = start;
3635 int q;
3636 boolean compressedZeros = false;
3637
3638 q = scanHexSeq(p, n);
3639
3640 if (q > p) {
3641 p = q;
3642 if (at(p, n, "::")) {
3643 compressedZeros = true;
3644 p = scanHexPost(p + 2, n);
3645 } else if (at(p, n, ':')) {
3646 p = takeIPv4Address(p + 1, n, "IPv4 address");
3647 ipv6byteCount += 4;
3648 }
3649 } else if (at(p, n, "::")) {
3650 compressedZeros = true;
3651 p = scanHexPost(p + 2, n);
3652 }
3653 if (p < n)
3654 fail("Malformed IPv6 address", start);
3655 if (ipv6byteCount > 16)
3656 fail("IPv6 address too long", start);
3657 if (!compressedZeros && ipv6byteCount < 16)
3658 fail("IPv6 address too short", start);
3659 if (compressedZeros && ipv6byteCount == 16)
3660 fail("Malformed IPv6 address", start);
3661
3662 return p;
3663 }
3664
3665 private int scanHexPost(int start, int n)
3666 throws URISyntaxException
3667 {
3668 int p = start;
3669 int q;
3670
3671 if (p == n)
3672 return p;
3673
3674 q = scanHexSeq(p, n);
3675 if (q > p) {
3676 p = q;
3677 if (at(p, n, ':')) {
3678 p++;
3679 p = takeIPv4Address(p, n, "hex digits or IPv4 address");
3680 ipv6byteCount += 4;
3681 }
3682 } else {
3683 p = takeIPv4Address(p, n, "hex digits or IPv4 address");
3684 ipv6byteCount += 4;
3685 }
3686 return p;
3687 }
3688
3689 // Scan a hex sequence; return -1 if one could not be scanned
3690 //
3691 private int scanHexSeq(int start, int n)
3692 throws URISyntaxException
3693 {
3694 int p = start;
3695 int q;
3696
3697 q = scan(p, n, L_HEX, H_HEX);
3698 if (q <= p)
3699 return -1;
3700 if (at(q, n, '.')) // Beginning of IPv4 address
3701 return -1;
3702 if (q > p + 4)
3703 fail("IPv6 hexadecimal digit sequence too long", p);
3704 ipv6byteCount += 2;
3705 p = q;
3706 while (p < n) {
3707 if (!at(p, n, ':'))
3708 break;
3709 if (at(p + 1, n, ':'))
3710 break; // "::"
3711 p++;
3712 q = scan(p, n, L_HEX, H_HEX);
3713 if (q <= p)
3714 failExpecting("digits for an IPv6 address", p);
3715 if (at(q, n, '.')) { // Beginning of IPv4 address
3716 p--;
3717 break;
3718 }
3719 if (q > p + 4)
3720 fail("IPv6 hexadecimal digit sequence too long", p);
3721 ipv6byteCount += 2;
3722 p = q;
3723 }
3724
3725 return p;
3726 }
3727
3728 }
3729
3730 static {
3731 runtimeSetup();
3732 }
3733
3734 // Called from JVM when loading an AOT cache
3735 private static void runtimeSetup() {
3736 SharedSecrets.setJavaNetUriAccess(
3737 new JavaNetUriAccess() {
3738 public URI create(String scheme, String path) {
3739 return new URI(scheme, path);
3740 }
3741 }
3742 );
3743 }
3744 }