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