2 * Copyright (c) 2000, 2011, Oracle and/or its affiliates. All rights reserved.
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
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.
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).
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.
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
28 import java.io.IOException;
29 import java.io.InvalidObjectException;
30 import java.io.ObjectInputStream;
31 import java.io.ObjectOutputStream;
32 import java.io.Serializable;
33 import java.nio.ByteBuffer;
34 import java.nio.CharBuffer;
35 import java.nio.charset.CharsetDecoder;
36 import java.nio.charset.CharsetEncoder;
37 import java.nio.charset.CoderResult;
38 import java.nio.charset.CodingErrorAction;
39 import java.nio.charset.CharacterCodingException;
40 import java.text.Normalizer;
41 import sun.nio.cs.ThreadLocalCoders;
43 import java.lang.Character; // for javadoc
44 import java.lang.NullPointerException; // for javadoc
48 * Represents a Uniform Resource Identifier (URI) reference.
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.
64 * <h4> URI syntax and components </h4>
66 * At the highest level a URI reference (hereinafter simply "URI") in string
70 * [<i>scheme</i><tt><b>:</b></tt><i></i>]<i>scheme-specific-part</i>[<tt><b>#</b></tt><i>fragment</i>]
73 * where square brackets [...] delineate optional components and the characters
74 * <tt><b>:</b></tt> and <tt><b>#</b></tt> stand for themselves.
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>.
80 * <p> An <i>opaque</i> URI is an absolute URI whose scheme-specific part does
81 * not begin with a slash character (<tt>'/'</tt>). Opaque URIs are not
82 * subject to further parsing. Some examples of opaque URIs are:
84 * <blockquote><table cellpadding=0 cellspacing=0 summary="layout">
85 * <tr><td><tt>mailto:java-net@java.sun.com</tt><td></tr>
86 * <tr><td><tt>news:comp.lang.java</tt><td></tr>
87 * <tr><td><tt>urn:isbn:096139210x</tt></td></tr>
88 * </table></blockquote>
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
96 * <tt>http://java.sun.com/j2se/1.3/</tt><br>
97 * <tt>docs/guide/collections/designfaq.html#28</tt><br>
98 * <tt>../../../demo/jfc/SwingSet2/src/SwingSet2.java</tt><br>
99 * <tt>file:///~/calendar</tt>
102 * <p> A hierarchical URI is subject to further parsing according to the syntax
105 * [<i>scheme</i><tt><b>:</b></tt>][<tt><b>//</b></tt><i>authority</i>][<i>path</i>][<tt><b>?</b></tt><i>query</i>][<tt><b>#</b></tt><i>fragment</i>]
108 * where the characters <tt><b>:</b></tt>, <tt><b>/</b></tt>,
109 * <tt><b>?</b></tt>, and <tt><b>#</b></tt> stand for themselves. The
110 * scheme-specific part of a hierarchical URI consists of the characters
111 * between the scheme and fragment components.
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
118 * [<i>user-info</i><tt><b>@</b></tt>]<i>host</i>[<tt><b>:</b></tt><i>port</i>]
121 * where the characters <tt><b>@</b></tt> and <tt><b>:</b></tt> 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
126 * <p> The path component of a hierarchical URI is itself said to be absolute
127 * if it begins with a slash character (<tt>'/'</tt>); otherwise it is
128 * relative. The path of a hierarchical URI that is either absolute or
129 * specifies an authority is always absolute.
131 * <p> All told, then, a URI instance has the following nine components:
133 * <blockquote><table summary="Describes the components of a URI:scheme,scheme-specific-part,authority,user-info,host,port,path,query,fragment">
134 * <tr><th><i>Component</i></th><th><i>Type</i></th></tr>
135 * <tr><td>scheme</td><td><tt>String</tt></td></tr>
136 * <tr><td>scheme-specific-part </td><td><tt>String</tt></td></tr>
137 * <tr><td>authority</td><td><tt>String</tt></td></tr>
138 * <tr><td>user-info</td><td><tt>String</tt></td></tr>
139 * <tr><td>host</td><td><tt>String</tt></td></tr>
140 * <tr><td>port</td><td><tt>int</tt></td></tr>
141 * <tr><td>path</td><td><tt>String</tt></td></tr>
142 * <tr><td>query</td><td><tt>String</tt></td></tr>
143 * <tr><td>fragment</td><td><tt>String</tt></td></tr>
144 * </table></blockquote>
146 * In a given instance any particular component is either <i>undefined</i> or
147 * <i>defined</i> with a distinct value. Undefined string components are
148 * represented by <tt>null</tt>, while undefined integer components are
149 * represented by <tt>-1</tt>. A string component may be defined to have the
150 * empty string as its value; this is not equivalent to that component being
153 * <p> Whether a particular component is or is not defined in an instance
154 * depends upon the type of the URI being represented. An absolute URI has a
155 * scheme component. An opaque URI has a scheme, a scheme-specific part, and
156 * possibly a fragment, but has no other components. A hierarchical URI always
157 * has a path (though it may be empty) and a scheme-specific-part (which at
158 * least contains the path), and may have any of the other components. If the
159 * authority component is present and is server-based then the host component
160 * will be defined and the user-information and port components may be defined.
163 * <h4> Operations on URI instances </h4>
165 * The key operations supported by this class are those of
166 * <i>normalization</i>, <i>resolution</i>, and <i>relativization</i>.
168 * <p> <i>Normalization</i> is the process of removing unnecessary <tt>"."</tt>
169 * and <tt>".."</tt> segments from the path component of a hierarchical URI.
170 * Each <tt>"."</tt> segment is simply removed. A <tt>".."</tt> segment is
171 * removed only if it is preceded by a non-<tt>".."</tt> segment.
172 * Normalization has no effect upon opaque URIs.
174 * <p> <i>Resolution</i> is the process of resolving one URI against another,
175 * <i>base</i> URI. The resulting URI is constructed from components of both
176 * URIs in the manner specified by RFC 2396, taking components from the
177 * base URI for those not specified in the original. For hierarchical URIs,
178 * the path of the original is resolved against the path of the base and then
179 * normalized. The result, for example, of resolving
182 * <tt>docs/guide/collections/designfaq.html#28 </tt>(1)
185 * against the base URI <tt>http://java.sun.com/j2se/1.3/</tt> is the result
189 * <tt>http://java.sun.com/j2se/1.3/docs/guide/collections/designfaq.html#28</tt>
192 * Resolving the relative URI
195 * <tt>../../../demo/jfc/SwingSet2/src/SwingSet2.java </tt>(2)
198 * against this result yields, in turn,
201 * <tt>http://java.sun.com/j2se/1.3/demo/jfc/SwingSet2/src/SwingSet2.java</tt>
204 * Resolution of both absolute and relative URIs, and of both absolute and
205 * relative paths in the case of hierarchical URIs, is supported. Resolving
206 * the URI <tt>file:///~calendar</tt> against any other URI simply yields the
207 * original URI, since it is absolute. Resolving the relative URI (2) above
208 * against the relative base URI (1) yields the normalized, but still relative,
212 * <tt>demo/jfc/SwingSet2/src/SwingSet2.java</tt>
215 * <p> <i>Relativization</i>, finally, is the inverse of resolution: For any
216 * two normalized URIs <i>u</i> and <i>v</i>,
219 * <i>u</i><tt>.relativize(</tt><i>u</i><tt>.resolve(</tt><i>v</i><tt>)).equals(</tt><i>v</i><tt>)</tt> and<br>
220 * <i>u</i><tt>.resolve(</tt><i>u</i><tt>.relativize(</tt><i>v</i><tt>)).equals(</tt><i>v</i><tt>)</tt> .<br>
223 * This operation is often useful when constructing a document containing URIs
224 * that must be made relative to the base URI of the document wherever
225 * possible. For example, relativizing the URI
228 * <tt>http://java.sun.com/j2se/1.3/docs/guide/index.html</tt>
231 * against the base URI
234 * <tt>http://java.sun.com/j2se/1.3</tt>
237 * yields the relative URI <tt>docs/guide/index.html</tt>.
240 * <h4> Character categories </h4>
242 * RFC 2396 specifies precisely which characters are permitted in the
243 * various components of a URI reference. The following categories, most of
244 * which are taken from that specification, are used below to describe these
247 * <blockquote><table cellspacing=2 summary="Describes categories alpha,digit,alphanum,unreserved,punct,reserved,escaped,and other">
248 * <tr><th valign=top><i>alpha</i></th>
249 * <td>The US-ASCII alphabetic characters,
250 * <tt>'A'</tt> through <tt>'Z'</tt>
251 * and <tt>'a'</tt> through <tt>'z'</tt></td></tr>
252 * <tr><th valign=top><i>digit</i></th>
253 * <td>The US-ASCII decimal digit characters,
254 * <tt>'0'</tt> through <tt>'9'</tt></td></tr>
255 * <tr><th valign=top><i>alphanum</i></th>
256 * <td>All <i>alpha</i> and <i>digit</i> characters</td></tr>
257 * <tr><th valign=top><i>unreserved</i> </th>
258 * <td>All <i>alphanum</i> characters together with those in the string
259 * <tt>"_-!.~'()*"</tt></td></tr>
260 * <tr><th valign=top><i>punct</i></th>
261 * <td>The characters in the string <tt>",;:$&+="</tt></td></tr>
262 * <tr><th valign=top><i>reserved</i></th>
263 * <td>All <i>punct</i> characters together with those in the string
264 * <tt>"?/[]@"</tt></td></tr>
265 * <tr><th valign=top><i>escaped</i></th>
266 * <td>Escaped octets, that is, triplets consisting of the percent
267 * character (<tt>'%'</tt>) followed by two hexadecimal digits
268 * (<tt>'0'</tt>-<tt>'9'</tt>, <tt>'A'</tt>-<tt>'F'</tt>, and
269 * <tt>'a'</tt>-<tt>'f'</tt>)</td></tr>
270 * <tr><th valign=top><i>other</i></th>
271 * <td>The Unicode characters that are not in the US-ASCII character set,
272 * are not control characters (according to the {@link
273 * java.lang.Character#isISOControl(char) Character.isISOControl}
274 * method), and are not space characters (according to the {@link
275 * java.lang.Character#isSpaceChar(char) Character.isSpaceChar}
276 * method) <i>(<b>Deviation from RFC 2396</b>, which is
277 * limited to US-ASCII)</i></td></tr>
278 * </table></blockquote>
280 * <p><a name="legal-chars"></a> The set of all legal URI characters consists of
281 * the <i>unreserved</i>, <i>reserved</i>, <i>escaped</i>, and <i>other</i>
285 * <h4> Escaped octets, quotation, encoding, and decoding </h4>
287 * RFC 2396 allows escaped octets to appear in the user-info, path, query, and
288 * fragment components. Escaping serves two purposes in URIs:
292 * <li><p> To <i>encode</i> non-US-ASCII characters when a URI is required to
293 * conform strictly to RFC 2396 by not containing any <i>other</i>
294 * characters. </p></li>
296 * <li><p> To <i>quote</i> characters that are otherwise illegal in a
297 * component. The user-info, path, query, and fragment components differ
298 * slightly in terms of which characters are considered legal and illegal.
303 * These purposes are served in this class by three related operations:
307 * <li><p><a name="encode"></a> A character is <i>encoded</i> by replacing it
308 * with the sequence of escaped octets that represent that character in the
309 * UTF-8 character set. The Euro currency symbol (<tt>'\u20AC'</tt>),
310 * for example, is encoded as <tt>"%E2%82%AC"</tt>. <i>(<b>Deviation from
311 * RFC 2396</b>, which does not specify any particular character
312 * set.)</i> </p></li>
314 * <li><p><a name="quote"></a> An illegal character is <i>quoted</i> simply by
315 * encoding it. The space character, for example, is quoted by replacing it
316 * with <tt>"%20"</tt>. UTF-8 contains US-ASCII, hence for US-ASCII
317 * characters this transformation has exactly the effect required by
318 * RFC 2396. </p></li>
320 * <li><p><a name="decode"></a>
321 * A sequence of escaped octets is <i>decoded</i> by
322 * replacing it with the sequence of characters that it represents in the
323 * UTF-8 character set. UTF-8 contains US-ASCII, hence decoding has the
324 * effect of de-quoting any quoted US-ASCII characters as well as that of
325 * decoding any encoded non-US-ASCII characters. If a <a
326 * href="../nio/charset/CharsetDecoder.html#ce">decoding error</a> occurs
327 * when decoding the escaped octets then the erroneous octets are replaced by
328 * <tt>'\uFFFD'</tt>, the Unicode replacement character. </p></li>
332 * These operations are exposed in the constructors and methods of this class
337 * <li><p> The {@link #URI(java.lang.String) <code>single-argument
338 * constructor</code>} requires any illegal characters in its argument to be
339 * quoted and preserves any escaped octets and <i>other</i> characters that
340 * are present. </p></li>
343 * #URI(java.lang.String,java.lang.String,java.lang.String,int,java.lang.String,java.lang.String,java.lang.String)
344 * <code>multi-argument constructors</code>} quote illegal characters as
345 * required by the components in which they appear. The percent character
346 * (<tt>'%'</tt>) is always quoted by these constructors. Any <i>other</i>
347 * characters are preserved. </p></li>
349 * <li><p> The {@link #getRawUserInfo() getRawUserInfo}, {@link #getRawPath()
350 * getRawPath}, {@link #getRawQuery() getRawQuery}, {@link #getRawFragment()
351 * getRawFragment}, {@link #getRawAuthority() getRawAuthority}, and {@link
352 * #getRawSchemeSpecificPart() getRawSchemeSpecificPart} methods return the
353 * values of their corresponding components in raw form, without interpreting
354 * any escaped octets. The strings returned by these methods may contain
355 * both escaped octets and <i>other</i> characters, and will not contain any
356 * illegal characters. </p></li>
358 * <li><p> The {@link #getUserInfo() getUserInfo}, {@link #getPath()
359 * getPath}, {@link #getQuery() getQuery}, {@link #getFragment()
360 * getFragment}, {@link #getAuthority() getAuthority}, and {@link
361 * #getSchemeSpecificPart() getSchemeSpecificPart} methods decode any escaped
362 * octets in their corresponding components. The strings returned by these
363 * methods may contain both <i>other</i> characters and illegal characters,
364 * and will not contain any escaped octets. </p></li>
366 * <li><p> The {@link #toString() toString} method returns a URI string with
367 * all necessary quotation but which may contain <i>other</i> characters.
370 * <li><p> The {@link #toASCIIString() toASCIIString} method returns a fully
371 * quoted and encoded URI string that does not contain any <i>other</i>
372 * characters. </p></li>
377 * <h4> Identities </h4>
379 * For any URI <i>u</i>, it is always the case that
382 * <tt>new URI(</tt><i>u</i><tt>.toString()).equals(</tt><i>u</i><tt>)</tt> .
385 * For any URI <i>u</i> that does not contain redundant syntax such as two
386 * slashes before an empty authority (as in <tt>file:///tmp/</tt> ) or a
387 * colon following a host name but no port (as in
388 * <tt>http://java.sun.com:</tt> ), and that does not encode characters
389 * except those that must be quoted, the following identities also hold:
392 * <tt>new URI(</tt><i>u</i><tt>.getScheme(),<br>
393 * </tt><i>u</i><tt>.getSchemeSpecificPart(),<br>
394 * </tt><i>u</i><tt>.getFragment())<br>
395 * .equals(</tt><i>u</i><tt>)</tt>
401 * <tt>new URI(</tt><i>u</i><tt>.getScheme(),<br>
402 * </tt><i>u</i><tt>.getUserInfo(), </tt><i>u</i><tt>.getAuthority(),<br>
403 * </tt><i>u</i><tt>.getPath(), </tt><i>u</i><tt>.getQuery(),<br>
404 * </tt><i>u</i><tt>.getFragment())<br>
405 * .equals(</tt><i>u</i><tt>)</tt>
408 * if <i>u</i> is hierarchical, and
411 * <tt>new URI(</tt><i>u</i><tt>.getScheme(),<br>
412 * </tt><i>u</i><tt>.getUserInfo(), </tt><i>u</i><tt>.getHost(), </tt><i>u</i><tt>.getPort(),<br>
413 * </tt><i>u</i><tt>.getPath(), </tt><i>u</i><tt>.getQuery(),<br>
414 * </tt><i>u</i><tt>.getFragment())<br>
415 * .equals(</tt><i>u</i><tt>)</tt>
418 * if <i>u</i> is hierarchical and has either no authority or a server-based
422 * <h4> URIs, URLs, and URNs </h4>
424 * A URI is a uniform resource <i>identifier</i> while a URL is a uniform
425 * resource <i>locator</i>. Hence every URL is a URI, abstractly speaking, but
426 * not every URI is a URL. This is because there is another subcategory of
427 * URIs, uniform resource <i>names</i> (URNs), which name resources but do not
428 * specify how to locate them. The <tt>mailto</tt>, <tt>news</tt>, and
429 * <tt>isbn</tt> URIs shown above are examples of URNs.
431 * <p> The conceptual distinction between URIs and URLs is reflected in the
432 * differences between this class and the {@link URL} class.
434 * <p> An instance of this class represents a URI reference in the syntactic
435 * sense defined by RFC 2396. A URI may be either absolute or relative.
436 * A URI string is parsed according to the generic syntax without regard to the
437 * scheme, if any, that it specifies. No lookup of the host, if any, is
438 * performed, and no scheme-dependent stream handler is constructed. Equality,
439 * hashing, and comparison are defined strictly in terms of the character
440 * content of the instance. In other words, a URI instance is little more than
441 * a structured string that supports the syntactic, scheme-independent
442 * operations of comparison, normalization, resolution, and relativization.
444 * <p> An instance of the {@link URL} class, by contrast, represents the
445 * syntactic components of a URL together with some of the information required
446 * to access the resource that it describes. A URL must be absolute, that is,
447 * it must always specify a scheme. A URL string is parsed according to its
448 * scheme. A stream handler is always established for a URL, and in fact it is
449 * impossible to create a URL instance for a scheme for which no handler is
450 * available. Equality and hashing depend upon both the scheme and the
451 * Internet address of the host, if any; comparison is not defined. In other
452 * words, a URL is a structured string that supports the syntactic operation of
453 * resolution as well as the network I/O operations of looking up the host and
454 * opening a connection to the specified resource.
457 * @author Mark Reinhold
460 * @see <a href="http://www.ietf.org/rfc/rfc2279.txt"><i>RFC 2279: UTF-8, a
461 * transformation format of ISO 10646</i></a>, <br><a
462 * href="http://www.ietf.org/rfc/rfc2373.txt"><i>RFC 2373: IPv6 Addressing
463 * Architecture</i></a>, <br><a
464 * href="http://www.ietf.org/rfc/rfc2396.txt"><i>RFC 2396: Uniform
465 * Resource Identifiers (URI): Generic Syntax</i></a>, <br><a
466 * href="http://www.ietf.org/rfc/rfc2732.txt"><i>RFC 2732: Format for
467 * Literal IPv6 Addresses in URLs</i></a>, <br><a
468 * href="URISyntaxException.html">URISyntaxException</a>
471 public final class URI
472 implements Comparable<URI>, Serializable
475 // Note: Comments containing the word "ASSERT" indicate places where a
476 // throw of an InternalError should be replaced by an appropriate assertion
477 // statement once asserts are enabled in the build.
479 static final long serialVersionUID = -6052424284110960213L;
482 // -- Properties and components of this instance --
484 // Components of all URIs: [<scheme>:]<scheme-specific-part>[#<fragment>]
485 private transient String scheme; // null ==> relative URI
486 private transient String fragment;
488 // Hierarchical URI components: [//<authority>]<path>[?<query>]
489 private transient String authority; // Registry or server
491 // Server-based authority: [<userInfo>@]<host>[:<port>]
492 private transient String userInfo;
493 private transient String host; // null ==> registry-based
494 private transient int port = -1; // -1 ==> undefined
496 // Remaining components of hierarchical URIs
497 private transient String path; // null ==> opaque
498 private transient String query;
500 // The remaining fields may be computed on demand
502 private volatile transient String schemeSpecificPart;
503 private volatile transient int hash; // Zero ==> undefined
505 private volatile transient String decodedUserInfo = null;
506 private volatile transient String decodedAuthority = null;
507 private volatile transient String decodedPath = null;
508 private volatile transient String decodedQuery = null;
509 private volatile transient String decodedFragment = null;
510 private volatile transient String decodedSchemeSpecificPart = null;
513 * The string form of this URI.
517 private volatile String string; // The only serializable field
521 // -- Constructors and factories --
523 private URI() { } // Used internally
526 * Constructs a URI by parsing the given string.
528 * <p> This constructor parses the given string exactly as specified by the
530 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
531 * Appendix A, <b><i>except for the following deviations:</i></b> </p>
535 * <li><p> An empty authority component is permitted as long as it is
536 * followed by a non-empty path, a query component, or a fragment
537 * component. This allows the parsing of URIs such as
538 * <tt>"file:///foo/bar"</tt>, which seems to be the intent of
539 * RFC 2396 although the grammar does not permit it. If the
540 * authority component is empty then the user-information, host, and port
541 * components are undefined. </p></li>
543 * <li><p> Empty relative paths are permitted; this seems to be the
544 * intent of RFC 2396 although the grammar does not permit it. The
545 * primary consequence of this deviation is that a standalone fragment
546 * such as <tt>"#foo"</tt> parses as a relative URI with an empty path
547 * and the given fragment, and can be usefully <a
548 * href="#resolve-frag">resolved</a> against a base URI.
550 * <li><p> IPv4 addresses in host components are parsed rigorously, as
552 * href="http://www.ietf.org/rfc/rfc2732.txt">RFC 2732</a>: Each
553 * element of a dotted-quad address must contain no more than three
554 * decimal digits. Each element is further constrained to have a value
555 * no greater than 255. </p></li>
557 * <li> <p> Hostnames in host components that comprise only a single
558 * domain label are permitted to start with an <i>alphanum</i>
559 * character. This seems to be the intent of <a
560 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>
561 * section 3.2.2 although the grammar does not permit it. The
562 * consequence of this deviation is that the authority component of a
563 * hierarchical URI such as <tt>s://123</tt>, will parse as a server-based
564 * authority. </p></li>
566 * <li><p> IPv6 addresses are permitted for the host component. An IPv6
567 * address must be enclosed in square brackets (<tt>'['</tt> and
568 * <tt>']'</tt>) as specified by <a
569 * href="http://www.ietf.org/rfc/rfc2732.txt">RFC 2732</a>. The
570 * IPv6 address itself must parse according to <a
571 * href="http://www.ietf.org/rfc/rfc2373.txt">RFC 2373</a>. IPv6
572 * addresses are further constrained to describe no more than sixteen
573 * bytes of address information, a constraint implicit in RFC 2373
574 * but not expressible in the grammar. </p></li>
576 * <li><p> Characters in the <i>other</i> category are permitted wherever
577 * RFC 2396 permits <i>escaped</i> octets, that is, in the
578 * user-information, path, query, and fragment components, as well as in
579 * the authority component if the authority is registry-based. This
580 * allows URIs to contain Unicode characters beyond those in the US-ASCII
581 * character set. </p></li>
585 * @param str The string to be parsed into a URI
587 * @throws NullPointerException
588 * If <tt>str</tt> is <tt>null</tt>
590 * @throws URISyntaxException
591 * If the given string violates RFC 2396, as augmented
592 * by the above deviations
594 public URI(String str) throws URISyntaxException {
595 new Parser(str).parse(false);
599 * Constructs a hierarchical URI from the given components.
601 * <p> If a scheme is given then the path, if also given, must either be
602 * empty or begin with a slash character (<tt>'/'</tt>). Otherwise a
603 * component of the new URI may be left undefined by passing <tt>null</tt>
604 * for the corresponding parameter or, in the case of the <tt>port</tt>
605 * parameter, by passing <tt>-1</tt>.
607 * <p> This constructor first builds a URI string from the given components
608 * according to the rules specified in <a
609 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
610 * section 5.2, step 7: </p>
614 * <li><p> Initially, the result string is empty. </p></li>
616 * <li><p> If a scheme is given then it is appended to the result,
617 * followed by a colon character (<tt>':'</tt>). </p></li>
619 * <li><p> If user information, a host, or a port are given then the
620 * string <tt>"//"</tt> is appended. </p></li>
622 * <li><p> If user information is given then it is appended, followed by
623 * a commercial-at character (<tt>'@'</tt>). Any character not in the
624 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
625 * categories is <a href="#quote">quoted</a>. </p></li>
627 * <li><p> If a host is given then it is appended. If the host is a
628 * literal IPv6 address but is not enclosed in square brackets
629 * (<tt>'['</tt> and <tt>']'</tt>) then the square brackets are added.
632 * <li><p> If a port number is given then a colon character
633 * (<tt>':'</tt>) is appended, followed by the port number in decimal.
636 * <li><p> If a path is given then it is appended. Any character not in
637 * the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
638 * categories, and not equal to the slash character (<tt>'/'</tt>) or the
639 * commercial-at character (<tt>'@'</tt>), is quoted. </p></li>
641 * <li><p> If a query is given then a question-mark character
642 * (<tt>'?'</tt>) is appended, followed by the query. Any character that
643 * is not a <a href="#legal-chars">legal URI character</a> is quoted.
646 * <li><p> Finally, if a fragment is given then a hash character
647 * (<tt>'#'</tt>) is appended, followed by the fragment. Any character
648 * that is not a legal URI character is quoted. </p></li>
652 * <p> The resulting URI string is then parsed as if by invoking the {@link
653 * #URI(String)} constructor and then invoking the {@link
654 * #parseServerAuthority()} method upon the result; this may cause a {@link
655 * URISyntaxException} to be thrown. </p>
657 * @param scheme Scheme name
658 * @param userInfo User name and authorization information
659 * @param host Host name
660 * @param port Port number
663 * @param fragment Fragment
665 * @throws URISyntaxException
666 * If both a scheme and a path are given but the path is relative,
667 * if the URI string constructed from the given components violates
668 * RFC 2396, or if the authority component of the string is
669 * present but cannot be parsed as a server-based authority
671 public URI(String scheme,
672 String userInfo, String host, int port,
673 String path, String query, String fragment)
674 throws URISyntaxException
676 String s = toString(scheme, null,
677 null, userInfo, host, port,
678 path, query, fragment);
679 checkPath(s, scheme, path);
680 new Parser(s).parse(true);
684 * Constructs a hierarchical URI from the given components.
686 * <p> If a scheme is given then the path, if also given, must either be
687 * empty or begin with a slash character (<tt>'/'</tt>). Otherwise a
688 * component of the new URI may be left undefined by passing <tt>null</tt>
689 * for the corresponding parameter.
691 * <p> This constructor first builds a URI string from the given components
692 * according to the rules specified in <a
693 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
694 * section 5.2, step 7: </p>
698 * <li><p> Initially, the result string is empty. </p></li>
700 * <li><p> If a scheme is given then it is appended to the result,
701 * followed by a colon character (<tt>':'</tt>). </p></li>
703 * <li><p> If an authority is given then the string <tt>"//"</tt> is
704 * appended, followed by the authority. If the authority contains a
705 * literal IPv6 address then the address must be enclosed in square
706 * brackets (<tt>'['</tt> and <tt>']'</tt>). Any character not in the
707 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
708 * categories, and not equal to the commercial-at character
709 * (<tt>'@'</tt>), is <a href="#quote">quoted</a>. </p></li>
711 * <li><p> If a path is given then it is appended. Any character not in
712 * the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, or <i>other</i>
713 * categories, and not equal to the slash character (<tt>'/'</tt>) or the
714 * commercial-at character (<tt>'@'</tt>), is quoted. </p></li>
716 * <li><p> If a query is given then a question-mark character
717 * (<tt>'?'</tt>) is appended, followed by the query. Any character that
718 * is not a <a href="#legal-chars">legal URI character</a> is quoted.
721 * <li><p> Finally, if a fragment is given then a hash character
722 * (<tt>'#'</tt>) is appended, followed by the fragment. Any character
723 * that is not a legal URI character is quoted. </p></li>
727 * <p> The resulting URI string is then parsed as if by invoking the {@link
728 * #URI(String)} constructor and then invoking the {@link
729 * #parseServerAuthority()} method upon the result; this may cause a {@link
730 * URISyntaxException} to be thrown. </p>
732 * @param scheme Scheme name
733 * @param authority Authority
736 * @param fragment Fragment
738 * @throws URISyntaxException
739 * If both a scheme and a path are given but the path is relative,
740 * if the URI string constructed from the given components violates
741 * RFC 2396, or if the authority component of the string is
742 * present but cannot be parsed as a server-based authority
744 public URI(String scheme,
746 String path, String query, String fragment)
747 throws URISyntaxException
749 String s = toString(scheme, null,
750 authority, null, null, -1,
751 path, query, fragment);
752 checkPath(s, scheme, path);
753 new Parser(s).parse(false);
757 * Constructs a hierarchical URI from the given components.
759 * <p> A component may be left undefined by passing <tt>null</tt>.
761 * <p> This convenience constructor works as if by invoking the
762 * seven-argument constructor as follows:
765 * new {@link #URI(String, String, String, int, String, String, String)
766 * URI}(scheme, null, host, -1, path, null, fragment);
769 * @param scheme Scheme name
770 * @param host Host name
772 * @param fragment Fragment
774 * @throws URISyntaxException
775 * If the URI string constructed from the given components
776 * violates RFC 2396
778 public URI(String scheme, String host, String path, String fragment)
779 throws URISyntaxException
781 this(scheme, null, host, -1, path, null, fragment);
785 * Constructs a URI from the given components.
787 * <p> A component may be left undefined by passing <tt>null</tt>.
789 * <p> This constructor first builds a URI in string form using the given
790 * components as follows: </p>
794 * <li><p> Initially, the result string is empty. </p></li>
796 * <li><p> If a scheme is given then it is appended to the result,
797 * followed by a colon character (<tt>':'</tt>). </p></li>
799 * <li><p> If a scheme-specific part is given then it is appended. Any
800 * character that is not a <a href="#legal-chars">legal URI character</a>
801 * is <a href="#quote">quoted</a>. </p></li>
803 * <li><p> Finally, if a fragment is given then a hash character
804 * (<tt>'#'</tt>) is appended to the string, followed by the fragment.
805 * Any character that is not a legal URI character is quoted. </p></li>
809 * <p> The resulting URI string is then parsed in order to create the new
810 * URI instance as if by invoking the {@link #URI(String)} constructor;
811 * this may cause a {@link URISyntaxException} to be thrown. </p>
813 * @param scheme Scheme name
814 * @param ssp Scheme-specific part
815 * @param fragment Fragment
817 * @throws URISyntaxException
818 * If the URI string constructed from the given components
819 * violates RFC 2396
821 public URI(String scheme, String ssp, String fragment)
822 throws URISyntaxException
824 new Parser(toString(scheme, ssp,
825 null, null, null, -1,
826 null, null, fragment))
831 * Creates a URI by parsing the given string.
833 * <p> This convenience factory method works as if by invoking the {@link
834 * #URI(String)} constructor; any {@link URISyntaxException} thrown by the
835 * constructor is caught and wrapped in a new {@link
836 * IllegalArgumentException} object, which is then thrown.
838 * <p> This method is provided for use in situations where it is known that
839 * the given string is a legal URI, for example for URI constants declared
840 * within in a program, and so it would be considered a programming error
841 * for the string not to parse as such. The constructors, which throw
842 * {@link URISyntaxException} directly, should be used situations where a
843 * URI is being constructed from user input or from some other source that
844 * may be prone to errors. </p>
846 * @param str The string to be parsed into a URI
847 * @return The new URI
849 * @throws NullPointerException
850 * If <tt>str</tt> is <tt>null</tt>
852 * @throws IllegalArgumentException
853 * If the given string violates RFC 2396
855 public static URI create(String str) {
858 } catch (URISyntaxException x) {
859 throw new IllegalArgumentException(x.getMessage(), x);
867 * Attempts to parse this URI's authority component, if defined, into
868 * user-information, host, and port components.
870 * <p> If this URI's authority component has already been recognized as
871 * being server-based then it will already have been parsed into
872 * user-information, host, and port components. In this case, or if this
873 * URI has no authority component, this method simply returns this URI.
875 * <p> Otherwise this method attempts once more to parse the authority
876 * component into user-information, host, and port components, and throws
877 * an exception describing why the authority component could not be parsed
880 * <p> This method is provided because the generic URI syntax specified in
881 * <a href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>
882 * cannot always distinguish a malformed server-based authority from a
883 * legitimate registry-based authority. It must therefore treat some
884 * instances of the former as instances of the latter. The authority
885 * component in the URI string <tt>"//foo:bar"</tt>, for example, is not a
886 * legal server-based authority but it is legal as a registry-based
889 * <p> In many common situations, for example when working URIs that are
890 * known to be either URNs or URLs, the hierarchical URIs being used will
891 * always be server-based. They therefore must either be parsed as such or
892 * treated as an error. In these cases a statement such as
895 * <tt>URI </tt><i>u</i><tt> = new URI(str).parseServerAuthority();</tt>
898 * <p> can be used to ensure that <i>u</i> always refers to a URI that, if
899 * it has an authority component, has a server-based authority with proper
900 * user-information, host, and port components. Invoking this method also
901 * ensures that if the authority could not be parsed in that way then an
902 * appropriate diagnostic message can be issued based upon the exception
903 * that is thrown. </p>
905 * @return A URI whose authority field has been parsed
906 * as a server-based authority
908 * @throws URISyntaxException
909 * If the authority component of this URI is defined
910 * but cannot be parsed as a server-based authority
911 * according to RFC 2396
913 public URI parseServerAuthority()
914 throws URISyntaxException
916 // We could be clever and cache the error message and index from the
917 // exception thrown during the original parse, but that would require
918 // either more fields or a more-obscure representation.
919 if ((host != null) || (authority == null))
922 new Parser(string).parse(true);
927 * Normalizes this URI's path.
929 * <p> If this URI is opaque, or if its path is already in normal form,
930 * then this URI is returned. Otherwise a new URI is constructed that is
931 * identical to this URI except that its path is computed by normalizing
932 * this URI's path in a manner consistent with <a
933 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
934 * section 5.2, step 6, sub-steps c through f; that is:
939 * <li><p> All <tt>"."</tt> segments are removed. </p></li>
941 * <li><p> If a <tt>".."</tt> segment is preceded by a non-<tt>".."</tt>
942 * segment then both of these segments are removed. This step is
943 * repeated until it is no longer applicable. </p></li>
945 * <li><p> If the path is relative, and if its first segment contains a
946 * colon character (<tt>':'</tt>), then a <tt>"."</tt> segment is
947 * prepended. This prevents a relative URI with a path such as
948 * <tt>"a:b/c/d"</tt> from later being re-parsed as an opaque URI with a
949 * scheme of <tt>"a"</tt> and a scheme-specific part of <tt>"b/c/d"</tt>.
950 * <b><i>(Deviation from RFC 2396)</i></b> </p></li>
954 * <p> A normalized path will begin with one or more <tt>".."</tt> segments
955 * if there were insufficient non-<tt>".."</tt> segments preceding them to
956 * allow their removal. A normalized path will begin with a <tt>"."</tt>
957 * segment if one was inserted by step 3 above. Otherwise, a normalized
958 * path will not contain any <tt>"."</tt> or <tt>".."</tt> segments. </p>
960 * @return A URI equivalent to this URI,
961 * but whose path is in normal form
963 public URI normalize() {
964 return normalize(this);
968 * Resolves the given URI against this URI.
970 * <p> If the given URI is already absolute, or if this URI is opaque, then
971 * the given URI is returned.
973 * <p><a name="resolve-frag"></a> If the given URI's fragment component is
974 * defined, its path component is empty, and its scheme, authority, and
975 * query components are undefined, then a URI with the given fragment but
976 * with all other components equal to those of this URI is returned. This
977 * allows a URI representing a standalone fragment reference, such as
978 * <tt>"#foo"</tt>, to be usefully resolved against a base URI.
980 * <p> Otherwise this method constructs a new hierarchical URI in a manner
982 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
983 * section 5.2; that is: </p>
987 * <li><p> A new URI is constructed with this URI's scheme and the given
988 * URI's query and fragment components. </p></li>
990 * <li><p> If the given URI has an authority component then the new URI's
991 * authority and path are taken from the given URI. </p></li>
993 * <li><p> Otherwise the new URI's authority component is copied from
994 * this URI, and its path is computed as follows: </p>
998 * <li><p> If the given URI's path is absolute then the new URI's path
999 * is taken from the given URI. </p></li>
1001 * <li><p> Otherwise the given URI's path is relative, and so the new
1002 * URI's path is computed by resolving the path of the given URI
1003 * against the path of this URI. This is done by concatenating all but
1004 * the last segment of this URI's path, if any, with the given URI's
1005 * path and then normalizing the result as if by invoking the {@link
1006 * #normalize() normalize} method. </p></li>
1012 * <p> The result of this method is absolute if, and only if, either this
1013 * URI is absolute or the given URI is absolute. </p>
1015 * @param uri The URI to be resolved against this URI
1016 * @return The resulting URI
1018 * @throws NullPointerException
1019 * If <tt>uri</tt> is <tt>null</tt>
1021 public URI resolve(URI uri) {
1022 return resolve(this, uri);
1026 * Constructs a new URI by parsing the given string and then resolving it
1029 * <p> This convenience method works as if invoking it were equivalent to
1030 * evaluating the expression <tt>{@link #resolve(java.net.URI)
1031 * resolve}(URI.{@link #create(String) create}(str))</tt>. </p>
1033 * @param str The string to be parsed into a URI
1034 * @return The resulting URI
1036 * @throws NullPointerException
1037 * If <tt>str</tt> is <tt>null</tt>
1039 * @throws IllegalArgumentException
1040 * If the given string violates RFC 2396
1042 public URI resolve(String str) {
1043 return resolve(URI.create(str));
1047 * Relativizes the given URI against this URI.
1049 * <p> The relativization of the given URI against this URI is computed as
1054 * <li><p> If either this URI or the given URI are opaque, or if the
1055 * scheme and authority components of the two URIs are not identical, or
1056 * if the path of this URI is not a prefix of the path of the given URI,
1057 * then the given URI is returned. </p></li>
1059 * <li><p> Otherwise a new relative hierarchical URI is constructed with
1060 * query and fragment components taken from the given URI and with a path
1061 * component computed by removing this URI's path from the beginning of
1062 * the given URI's path. </p></li>
1066 * @param uri The URI to be relativized against this URI
1067 * @return The resulting URI
1069 * @throws NullPointerException
1070 * If <tt>uri</tt> is <tt>null</tt>
1072 public URI relativize(URI uri) {
1073 return relativize(this, uri);
1077 * Constructs a URL from this URI.
1079 * <p> This convenience method works as if invoking it were equivalent to
1080 * evaluating the expression <tt>new URL(this.toString())</tt> after
1081 * first checking that this URI is absolute. </p>
1083 * @return A URL constructed from this URI
1085 * @throws IllegalArgumentException
1086 * If this URL is not absolute
1088 * @throws MalformedURLException
1089 * If a protocol handler for the URL could not be found,
1090 * or if some other error occurred while constructing the URL
1093 throws MalformedURLException {
1095 throw new IllegalArgumentException("URI is not absolute");
1096 return new URL(toString());
1099 // -- Component access methods --
1102 * Returns the scheme component of this URI.
1104 * <p> The scheme component of a URI, if defined, only contains characters
1105 * in the <i>alphanum</i> category and in the string <tt>"-.+"</tt>. A
1106 * scheme always starts with an <i>alpha</i> character. <p>
1108 * The scheme component of a URI cannot contain escaped octets, hence this
1109 * method does not perform any decoding.
1111 * @return The scheme component of this URI,
1112 * or <tt>null</tt> if the scheme is undefined
1114 public String getScheme() {
1119 * Tells whether or not this URI is absolute.
1121 * <p> A URI is absolute if, and only if, it has a scheme component. </p>
1123 * @return <tt>true</tt> if, and only if, this URI is absolute
1125 public boolean isAbsolute() {
1126 return scheme != null;
1130 * Tells whether or not this URI is opaque.
1132 * <p> A URI is opaque if, and only if, it is absolute and its
1133 * scheme-specific part does not begin with a slash character ('/').
1134 * An opaque URI has a scheme, a scheme-specific part, and possibly
1135 * a fragment; all other components are undefined. </p>
1137 * @return <tt>true</tt> if, and only if, this URI is opaque
1139 public boolean isOpaque() {
1140 return path == null;
1144 * Returns the raw scheme-specific part of this URI. The scheme-specific
1145 * part is never undefined, though it may be empty.
1147 * <p> The scheme-specific part of a URI only contains legal URI
1150 * @return The raw scheme-specific part of this URI
1151 * (never <tt>null</tt>)
1153 public String getRawSchemeSpecificPart() {
1154 defineSchemeSpecificPart();
1155 return schemeSpecificPart;
1159 * Returns the decoded scheme-specific part of this URI.
1161 * <p> The string returned by this method is equal to that returned by the
1162 * {@link #getRawSchemeSpecificPart() getRawSchemeSpecificPart} method
1163 * except that all sequences of escaped octets are <a
1164 * href="#decode">decoded</a>. </p>
1166 * @return The decoded scheme-specific part of this URI
1167 * (never <tt>null</tt>)
1169 public String getSchemeSpecificPart() {
1170 if (decodedSchemeSpecificPart == null)
1171 decodedSchemeSpecificPart = decode(getRawSchemeSpecificPart());
1172 return decodedSchemeSpecificPart;
1176 * Returns the raw authority component of this URI.
1178 * <p> The authority component of a URI, if defined, only contains the
1179 * commercial-at character (<tt>'@'</tt>) and characters in the
1180 * <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, and <i>other</i>
1181 * categories. If the authority is server-based then it is further
1182 * constrained to have valid user-information, host, and port
1185 * @return The raw authority component of this URI,
1186 * or <tt>null</tt> if the authority is undefined
1188 public String getRawAuthority() {
1193 * Returns the decoded authority component of this URI.
1195 * <p> The string returned by this method is equal to that returned by the
1196 * {@link #getRawAuthority() getRawAuthority} method except that all
1197 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1199 * @return The decoded authority component of this URI,
1200 * or <tt>null</tt> if the authority is undefined
1202 public String getAuthority() {
1203 if (decodedAuthority == null)
1204 decodedAuthority = decode(authority);
1205 return decodedAuthority;
1209 * Returns the raw user-information component of this URI.
1211 * <p> The user-information component of a URI, if defined, only contains
1212 * characters in the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>, and
1213 * <i>other</i> categories. </p>
1215 * @return The raw user-information component of this URI,
1216 * or <tt>null</tt> if the user information is undefined
1218 public String getRawUserInfo() {
1223 * Returns the decoded user-information component of this URI.
1225 * <p> The string returned by this method is equal to that returned by the
1226 * {@link #getRawUserInfo() getRawUserInfo} method except that all
1227 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1229 * @return The decoded user-information component of this URI,
1230 * or <tt>null</tt> if the user information is undefined
1232 public String getUserInfo() {
1233 if ((decodedUserInfo == null) && (userInfo != null))
1234 decodedUserInfo = decode(userInfo);
1235 return decodedUserInfo;
1239 * Returns the host component of this URI.
1241 * <p> The host component of a URI, if defined, will have one of the
1242 * following forms: </p>
1246 * <li><p> A domain name consisting of one or more <i>labels</i>
1247 * separated by period characters (<tt>'.'</tt>), optionally followed by
1248 * a period character. Each label consists of <i>alphanum</i> characters
1249 * as well as hyphen characters (<tt>'-'</tt>), though hyphens never
1250 * occur as the first or last characters in a label. The rightmost
1251 * label of a domain name consisting of two or more labels, begins
1252 * with an <i>alpha</i> character. </li>
1254 * <li><p> A dotted-quad IPv4 address of the form
1255 * <i>digit</i><tt>+.</tt><i>digit</i><tt>+.</tt><i>digit</i><tt>+.</tt><i>digit</i><tt>+</tt>,
1256 * where no <i>digit</i> sequence is longer than three characters and no
1257 * sequence has a value larger than 255. </p></li>
1259 * <li><p> An IPv6 address enclosed in square brackets (<tt>'['</tt> and
1260 * <tt>']'</tt>) and consisting of hexadecimal digits, colon characters
1261 * (<tt>':'</tt>), and possibly an embedded IPv4 address. The full
1262 * syntax of IPv6 addresses is specified in <a
1263 * href="http://www.ietf.org/rfc/rfc2373.txt"><i>RFC 2373: IPv6
1264 * Addressing Architecture</i></a>. </p></li>
1268 * The host component of a URI cannot contain escaped octets, hence this
1269 * method does not perform any decoding.
1271 * @return The host component of this URI,
1272 * or <tt>null</tt> if the host is undefined
1274 public String getHost() {
1279 * Returns the port number of this URI.
1281 * <p> The port component of a URI, if defined, is a non-negative
1284 * @return The port component of this URI,
1285 * or <tt>-1</tt> if the port is undefined
1287 public int getPort() {
1292 * Returns the raw path component of this URI.
1294 * <p> The path component of a URI, if defined, only contains the slash
1295 * character (<tt>'/'</tt>), the commercial-at character (<tt>'@'</tt>),
1296 * and characters in the <i>unreserved</i>, <i>punct</i>, <i>escaped</i>,
1297 * and <i>other</i> categories. </p>
1299 * @return The path component of this URI,
1300 * or <tt>null</tt> if the path is undefined
1302 public String getRawPath() {
1307 * Returns the decoded path component of this URI.
1309 * <p> The string returned by this method is equal to that returned by the
1310 * {@link #getRawPath() getRawPath} method except that all sequences of
1311 * escaped octets are <a href="#decode">decoded</a>. </p>
1313 * @return The decoded path component of this URI,
1314 * or <tt>null</tt> if the path is undefined
1316 public String getPath() {
1317 if ((decodedPath == null) && (path != null))
1318 decodedPath = decode(path);
1323 * Returns the raw query component of this URI.
1325 * <p> The query component of a URI, if defined, only contains legal URI
1328 * @return The raw query component of this URI,
1329 * or <tt>null</tt> if the query is undefined
1331 public String getRawQuery() {
1336 * Returns the decoded query component of this URI.
1338 * <p> The string returned by this method is equal to that returned by the
1339 * {@link #getRawQuery() getRawQuery} method except that all sequences of
1340 * escaped octets are <a href="#decode">decoded</a>. </p>
1342 * @return The decoded query component of this URI,
1343 * or <tt>null</tt> if the query is undefined
1345 public String getQuery() {
1346 if ((decodedQuery == null) && (query != null))
1347 decodedQuery = decode(query);
1348 return decodedQuery;
1352 * Returns the raw fragment component of this URI.
1354 * <p> The fragment component of a URI, if defined, only contains legal URI
1357 * @return The raw fragment component of this URI,
1358 * or <tt>null</tt> if the fragment is undefined
1360 public String getRawFragment() {
1365 * Returns the decoded fragment component of this URI.
1367 * <p> The string returned by this method is equal to that returned by the
1368 * {@link #getRawFragment() getRawFragment} method except that all
1369 * sequences of escaped octets are <a href="#decode">decoded</a>. </p>
1371 * @return The decoded fragment component of this URI,
1372 * or <tt>null</tt> if the fragment is undefined
1374 public String getFragment() {
1375 if ((decodedFragment == null) && (fragment != null))
1376 decodedFragment = decode(fragment);
1377 return decodedFragment;
1381 // -- Equality, comparison, hash code, toString, and serialization --
1384 * Tests this URI for equality with another object.
1386 * <p> If the given object is not a URI then this method immediately
1387 * returns <tt>false</tt>.
1389 * <p> For two URIs to be considered equal requires that either both are
1390 * opaque or both are hierarchical. Their schemes must either both be
1391 * undefined or else be equal without regard to case. Their fragments
1392 * must either both be undefined or else be equal.
1394 * <p> For two opaque URIs to be considered equal, their scheme-specific
1395 * parts must be equal.
1397 * <p> For two hierarchical URIs to be considered equal, their paths must
1398 * be equal and their queries must either both be undefined or else be
1399 * equal. Their authorities must either both be undefined, or both be
1400 * registry-based, or both be server-based. If their authorities are
1401 * defined and are registry-based, then they must be equal. If their
1402 * authorities are defined and are server-based, then their hosts must be
1403 * equal without regard to case, their port numbers must be equal, and
1404 * their user-information components must be equal.
1406 * <p> When testing the user-information, path, query, fragment, authority,
1407 * or scheme-specific parts of two URIs for equality, the raw forms rather
1408 * than the encoded forms of these components are compared and the
1409 * hexadecimal digits of escaped octets are compared without regard to
1412 * <p> This method satisfies the general contract of the {@link
1413 * java.lang.Object#equals(Object) Object.equals} method. </p>
1415 * @param ob The object to which this object is to be compared
1417 * @return <tt>true</tt> if, and only if, the given object is a URI that
1418 * is identical to this URI
1420 public boolean equals(Object ob) {
1423 if (!(ob instanceof URI))
1426 if (this.isOpaque() != that.isOpaque()) return false;
1427 if (!equalIgnoringCase(this.scheme, that.scheme)) return false;
1428 if (!equal(this.fragment, that.fragment)) return false;
1431 if (this.isOpaque())
1432 return equal(this.schemeSpecificPart, that.schemeSpecificPart);
1435 if (!equal(this.path, that.path)) return false;
1436 if (!equal(this.query, that.query)) return false;
1439 if (this.authority == that.authority) return true;
1440 if (this.host != null) {
1442 if (!equal(this.userInfo, that.userInfo)) return false;
1443 if (!equalIgnoringCase(this.host, that.host)) return false;
1444 if (this.port != that.port) return false;
1445 } else if (this.authority != null) {
1447 if (!equal(this.authority, that.authority)) return false;
1448 } else if (this.authority != that.authority) {
1456 * Returns a hash-code value for this URI. The hash code is based upon all
1457 * of the URI's components, and satisfies the general contract of the
1458 * {@link java.lang.Object#hashCode() Object.hashCode} method.
1460 * @return A hash-code value for this URI
1462 public int hashCode() {
1465 int h = hashIgnoringCase(0, scheme);
1466 h = hash(h, fragment);
1468 h = hash(h, schemeSpecificPart);
1473 h = hash(h, userInfo);
1474 h = hashIgnoringCase(h, host);
1477 h = hash(h, authority);
1485 * Compares this URI to another object, which must be a URI.
1487 * <p> When comparing corresponding components of two URIs, if one
1488 * component is undefined but the other is defined then the first is
1489 * considered to be less than the second. Unless otherwise noted, string
1490 * components are ordered according to their natural, case-sensitive
1491 * ordering as defined by the {@link java.lang.String#compareTo(Object)
1492 * String.compareTo} method. String components that are subject to
1493 * encoding are compared by comparing their raw forms rather than their
1496 * <p> The ordering of URIs is defined as follows: </p>
1500 * <li><p> Two URIs with different schemes are ordered according the
1501 * ordering of their schemes, without regard to case. </p></li>
1503 * <li><p> A hierarchical URI is considered to be less than an opaque URI
1504 * with an identical scheme. </p></li>
1506 * <li><p> Two opaque URIs with identical schemes are ordered according
1507 * to the ordering of their scheme-specific parts. </p></li>
1509 * <li><p> Two opaque URIs with identical schemes and scheme-specific
1510 * parts are ordered according to the ordering of their
1511 * fragments. </p></li>
1513 * <li><p> Two hierarchical URIs with identical schemes are ordered
1514 * according to the ordering of their authority components: </p>
1518 * <li><p> If both authority components are server-based then the URIs
1519 * are ordered according to their user-information components; if these
1520 * components are identical then the URIs are ordered according to the
1521 * ordering of their hosts, without regard to case; if the hosts are
1522 * identical then the URIs are ordered according to the ordering of
1523 * their ports. </p></li>
1525 * <li><p> If one or both authority components are registry-based then
1526 * the URIs are ordered according to the ordering of their authority
1527 * components. </p></li>
1531 * <li><p> Finally, two hierarchical URIs with identical schemes and
1532 * authority components are ordered according to the ordering of their
1533 * paths; if their paths are identical then they are ordered according to
1534 * the ordering of their queries; if the queries are identical then they
1535 * are ordered according to the order of their fragments. </p></li>
1539 * <p> This method satisfies the general contract of the {@link
1540 * java.lang.Comparable#compareTo(Object) Comparable.compareTo}
1544 * The object to which this URI is to be compared
1546 * @return A negative integer, zero, or a positive integer as this URI is
1547 * less than, equal to, or greater than the given URI
1549 * @throws ClassCastException
1550 * If the given object is not a URI
1552 public int compareTo(URI that) {
1555 if ((c = compareIgnoringCase(this.scheme, that.scheme)) != 0)
1558 if (this.isOpaque()) {
1559 if (that.isOpaque()) {
1561 if ((c = compare(this.schemeSpecificPart,
1562 that.schemeSpecificPart)) != 0)
1564 return compare(this.fragment, that.fragment);
1566 return +1; // Opaque > hierarchical
1567 } else if (that.isOpaque()) {
1568 return -1; // Hierarchical < opaque
1572 if ((this.host != null) && (that.host != null)) {
1573 // Both server-based
1574 if ((c = compare(this.userInfo, that.userInfo)) != 0)
1576 if ((c = compareIgnoringCase(this.host, that.host)) != 0)
1578 if ((c = this.port - that.port) != 0)
1581 // If one or both authorities are registry-based then we simply
1582 // compare them in the usual, case-sensitive way. If one is
1583 // registry-based and one is server-based then the strings are
1584 // guaranteed to be unequal, hence the comparison will never return
1585 // zero and the compareTo and equals methods will remain
1587 if ((c = compare(this.authority, that.authority)) != 0) return c;
1590 if ((c = compare(this.path, that.path)) != 0) return c;
1591 if ((c = compare(this.query, that.query)) != 0) return c;
1592 return compare(this.fragment, that.fragment);
1596 * Returns the content of this URI as a string.
1598 * <p> If this URI was created by invoking one of the constructors in this
1599 * class then a string equivalent to the original input string, or to the
1600 * string computed from the originally-given components, as appropriate, is
1601 * returned. Otherwise this URI was created by normalization, resolution,
1602 * or relativization, and so a string is constructed from this URI's
1603 * components according to the rules specified in <a
1604 * href="http://www.ietf.org/rfc/rfc2396.txt">RFC 2396</a>,
1605 * section 5.2, step 7. </p>
1607 * @return The string form of this URI
1609 public String toString() {
1615 * Returns the content of this URI as a US-ASCII string.
1617 * <p> If this URI does not contain any characters in the <i>other</i>
1618 * category then an invocation of this method will return the same value as
1619 * an invocation of the {@link #toString() toString} method. Otherwise
1620 * this method works as if by invoking that method and then <a
1621 * href="#encode">encoding</a> the result. </p>
1623 * @return The string form of this URI, encoded as needed
1624 * so that it only contains characters in the US-ASCII
1627 public String toASCIIString() {
1629 return encode(string);
1633 // -- Serialization support --
1636 * Saves the content of this URI to the given serial stream.
1638 * <p> The only serializable field of a URI instance is its <tt>string</tt>
1639 * field. That field is given a value, if it does not have one already,
1640 * and then the {@link java.io.ObjectOutputStream#defaultWriteObject()}
1641 * method of the given object-output stream is invoked. </p>
1643 * @param os The object-output stream to which this object
1646 private void writeObject(ObjectOutputStream os)
1650 os.defaultWriteObject(); // Writes the string field only
1654 * Reconstitutes a URI from the given serial stream.
1656 * <p> The {@link java.io.ObjectInputStream#defaultReadObject()} method is
1657 * invoked to read the value of the <tt>string</tt> field. The result is
1658 * then parsed in the usual way.
1660 * @param is The object-input stream from which this object
1663 private void readObject(ObjectInputStream is)
1664 throws ClassNotFoundException, IOException
1667 is.defaultReadObject();
1669 new Parser(string).parse(false);
1670 } catch (URISyntaxException x) {
1671 IOException y = new InvalidObjectException("Invalid URI");
1678 // -- End of public methods --
1681 // -- Utility methods for string-field comparison and hashing --
1683 // These methods return appropriate values for null string arguments,
1684 // thereby simplifying the equals, hashCode, and compareTo methods.
1686 // The case-ignoring methods should only be applied to strings whose
1687 // characters are all known to be US-ASCII. Because of this restriction,
1688 // these methods are faster than the similar methods in the String class.
1691 private static int toLower(char c) {
1692 if ((c >= 'A') && (c <= 'Z'))
1693 return c + ('a' - 'A');
1697 private static boolean equal(String s, String t) {
1698 if (s == t) return true;
1699 if ((s != null) && (t != null)) {
1700 if (s.length() != t.length())
1702 if (s.indexOf('%') < 0)
1705 for (int i = 0; i < n;) {
1706 char c = s.charAt(i);
1707 char d = t.charAt(i);
1715 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1718 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1728 private static boolean equalIgnoringCase(String s, String t) {
1729 if (s == t) return true;
1730 if ((s != null) && (t != null)) {
1732 if (t.length() != n)
1734 for (int i = 0; i < n; i++) {
1735 if (toLower(s.charAt(i)) != toLower(t.charAt(i)))
1743 private static int hash(int hash, String s) {
1744 if (s == null) return hash;
1745 return hash * 127 + s.hashCode();
1749 private static int hashIgnoringCase(int hash, String s) {
1750 if (s == null) return hash;
1753 for (int i = 0; i < n; i++)
1754 h = 31 * h + toLower(s.charAt(i));
1758 private static int compare(String s, String t) {
1759 if (s == t) return 0;
1762 return s.compareTo(t);
1771 private static int compareIgnoringCase(String s, String t) {
1772 if (s == t) return 0;
1775 int sn = s.length();
1776 int tn = t.length();
1777 int n = sn < tn ? sn : tn;
1778 for (int i = 0; i < n; i++) {
1779 int c = toLower(s.charAt(i)) - toLower(t.charAt(i));
1792 // -- String construction --
1794 // If a scheme is given then the path, if given, must be absolute
1796 private static void checkPath(String s, String scheme, String path)
1797 throws URISyntaxException
1799 if (scheme != null) {
1801 && ((path.length() > 0) && (path.charAt(0) != '/')))
1802 throw new URISyntaxException(s,
1803 "Relative path in absolute URI");
1807 private void appendAuthority(StringBuffer sb,
1815 if (userInfo != null) {
1816 sb.append(quote(userInfo, L_USERINFO, H_USERINFO));
1819 boolean needBrackets = ((host.indexOf(':') >= 0)
1820 && !host.startsWith("[")
1821 && !host.endsWith("]"));
1822 if (needBrackets) sb.append('[');
1824 if (needBrackets) sb.append(']');
1829 } else if (authority != null) {
1831 if (authority.startsWith("[")) {
1832 // authority should (but may not) contain an embedded IPv6 address
1833 int end = authority.indexOf("]");
1834 String doquote = authority, dontquote = "";
1835 if (end != -1 && authority.indexOf(":") != -1) {
1836 // the authority contains an IPv6 address
1837 if (end == authority.length()) {
1838 dontquote = authority;
1841 dontquote = authority.substring(0 , end + 1);
1842 doquote = authority.substring(end + 1);
1845 sb.append(dontquote);
1846 sb.append(quote(doquote,
1847 L_REG_NAME | L_SERVER,
1848 H_REG_NAME | H_SERVER));
1850 sb.append(quote(authority,
1851 L_REG_NAME | L_SERVER,
1852 H_REG_NAME | H_SERVER));
1857 private void appendSchemeSpecificPart(StringBuffer sb,
1866 if (opaquePart != null) {
1867 /* check if SSP begins with an IPv6 address
1868 * because we must not quote a literal IPv6 address
1870 if (opaquePart.startsWith("//[")) {
1871 int end = opaquePart.indexOf("]");
1872 if (end != -1 && opaquePart.indexOf(":")!=-1) {
1873 String doquote, dontquote;
1874 if (end == opaquePart.length()) {
1875 dontquote = opaquePart;
1878 dontquote = opaquePart.substring(0,end+1);
1879 doquote = opaquePart.substring(end+1);
1881 sb.append (dontquote);
1882 sb.append(quote(doquote, L_URIC, H_URIC));
1885 sb.append(quote(opaquePart, L_URIC, H_URIC));
1888 appendAuthority(sb, authority, userInfo, host, port);
1890 sb.append(quote(path, L_PATH, H_PATH));
1891 if (query != null) {
1893 sb.append(quote(query, L_URIC, H_URIC));
1898 private void appendFragment(StringBuffer sb, String fragment) {
1899 if (fragment != null) {
1901 sb.append(quote(fragment, L_URIC, H_URIC));
1905 private String toString(String scheme,
1915 StringBuffer sb = new StringBuffer();
1916 if (scheme != null) {
1920 appendSchemeSpecificPart(sb, opaquePart,
1921 authority, userInfo, host, port,
1923 appendFragment(sb, fragment);
1924 return sb.toString();
1927 private void defineSchemeSpecificPart() {
1928 if (schemeSpecificPart != null) return;
1929 StringBuffer sb = new StringBuffer();
1930 appendSchemeSpecificPart(sb, null, getAuthority(), getUserInfo(),
1931 host, port, getPath(), getQuery());
1932 if (sb.length() == 0) return;
1933 schemeSpecificPart = sb.toString();
1936 private void defineString() {
1937 if (string != null) return;
1939 StringBuffer sb = new StringBuffer();
1940 if (scheme != null) {
1945 sb.append(schemeSpecificPart);
1949 if (userInfo != null) {
1950 sb.append(userInfo);
1953 boolean needBrackets = ((host.indexOf(':') >= 0)
1954 && !host.startsWith("[")
1955 && !host.endsWith("]"));
1956 if (needBrackets) sb.append('[');
1958 if (needBrackets) sb.append(']');
1963 } else if (authority != null) {
1965 sb.append(authority);
1969 if (query != null) {
1974 if (fragment != null) {
1976 sb.append(fragment);
1978 string = sb.toString();
1982 // -- Normalization, resolution, and relativization --
1985 private static String resolvePath(String base, String child,
1988 int i = base.lastIndexOf('/');
1989 int cn = child.length();
1995 path = base.substring(0, i + 1);
1997 StringBuffer sb = new StringBuffer(base.length() + cn);
2000 sb.append(base.substring(0, i + 1));
2003 path = sb.toString();
2007 String np = normalize(path);
2009 // 5.2 (6g): If the result is absolute but the path begins with "../",
2010 // then we simply leave the path as-is
2016 private static URI resolve(URI base, URI child) {
2017 // check if child if opaque first so that NPE is thrown
2018 // if child is null.
2019 if (child.isOpaque() || base.isOpaque())
2022 // 5.2 (2): Reference to current document (lone fragment)
2023 if ((child.scheme == null) && (child.authority == null)
2024 && child.path.equals("") && (child.fragment != null)
2025 && (child.query == null)) {
2026 if ((base.fragment != null)
2027 && child.fragment.equals(base.fragment)) {
2031 ru.scheme = base.scheme;
2032 ru.authority = base.authority;
2033 ru.userInfo = base.userInfo;
2034 ru.host = base.host;
2035 ru.port = base.port;
2036 ru.path = base.path;
2037 ru.fragment = child.fragment;
2038 ru.query = base.query;
2042 // 5.2 (3): Child is absolute
2043 if (child.scheme != null)
2046 URI ru = new URI(); // Resolved URI
2047 ru.scheme = base.scheme;
2048 ru.query = child.query;
2049 ru.fragment = child.fragment;
2051 // 5.2 (4): Authority
2052 if (child.authority == null) {
2053 ru.authority = base.authority;
2054 ru.host = base.host;
2055 ru.userInfo = base.userInfo;
2056 ru.port = base.port;
2058 String cp = (child.path == null) ? "" : child.path;
2059 if ((cp.length() > 0) && (cp.charAt(0) == '/')) {
2060 // 5.2 (5): Child path is absolute
2061 ru.path = child.path;
2063 // 5.2 (6): Resolve relative path
2064 ru.path = resolvePath(base.path, cp, base.isAbsolute());
2067 ru.authority = child.authority;
2068 ru.host = child.host;
2069 ru.userInfo = child.userInfo;
2070 ru.host = child.host;
2071 ru.port = child.port;
2072 ru.path = child.path;
2075 // 5.2 (7): Recombine (nothing to do here)
2079 // If the given URI's path is normal then return the URI;
2080 // o.w., return a new URI containing the normalized path.
2082 private static URI normalize(URI u) {
2083 if (u.isOpaque() || (u.path == null) || (u.path.length() == 0))
2086 String np = normalize(u.path);
2091 v.scheme = u.scheme;
2092 v.fragment = u.fragment;
2093 v.authority = u.authority;
2094 v.userInfo = u.userInfo;
2102 // If both URIs are hierarchical, their scheme and authority components are
2103 // identical, and the base path is a prefix of the child's path, then
2104 // return a relative URI that, when resolved against the base, yields the
2105 // child; otherwise, return the child.
2107 private static URI relativize(URI base, URI child) {
2108 // check if child if opaque first so that NPE is thrown
2109 // if child is null.
2110 if (child.isOpaque() || base.isOpaque())
2112 if (!equalIgnoringCase(base.scheme, child.scheme)
2113 || !equal(base.authority, child.authority))
2116 String bp = normalize(base.path);
2117 String cp = normalize(child.path);
2118 if (!bp.equals(cp)) {
2119 if (!bp.endsWith("/"))
2121 if (!cp.startsWith(bp))
2126 v.path = cp.substring(bp.length());
2127 v.query = child.query;
2128 v.fragment = child.fragment;
2134 // -- Path normalization --
2136 // The following algorithm for path normalization avoids the creation of a
2137 // string object for each segment, as well as the use of a string buffer to
2138 // compute the final result, by using a single char array and editing it in
2139 // place. The array is first split into segments, replacing each slash
2140 // with '\0' and creating a segment-index array, each element of which is
2141 // the index of the first char in the corresponding segment. We then walk
2142 // through both arrays, removing ".", "..", and other segments as necessary
2143 // by setting their entries in the index array to -1. Finally, the two
2144 // arrays are used to rejoin the segments and compute the final result.
2146 // This code is based upon src/solaris/native/java/io/canonicalize_md.c
2149 // Check the given path to see if it might need normalization. A path
2150 // might need normalization if it contains duplicate slashes, a "."
2151 // segment, or a ".." segment. Return -1 if no further normalization is
2152 // possible, otherwise return the number of segments found.
2154 // This method takes a string argument rather than a char array so that
2155 // this test can be performed without invoking path.toCharArray().
2157 static private int needsNormalization(String path) {
2158 boolean normal = true;
2159 int ns = 0; // Number of segments
2160 int end = path.length() - 1; // Index of last char in path
2161 int p = 0; // Index of next char in path
2163 // Skip initial slashes
2165 if (path.charAt(p) != '/') break;
2168 if (p > 1) normal = false;
2173 // Looking at "." or ".." ?
2174 if ((path.charAt(p) == '.')
2176 || ((path.charAt(p + 1) == '/')
2177 || ((path.charAt(p + 1) == '.')
2179 || (path.charAt(p + 2) == '/')))))) {
2184 // Find beginning of next segment
2186 if (path.charAt(p++) != '/')
2189 // Skip redundant slashes
2191 if (path.charAt(p) != '/') break;
2200 return normal ? -1 : ns;
2204 // Split the given path into segments, replacing slashes with nulls and
2205 // filling in the given segment-index array.
2208 // segs.length == Number of segments in path
2211 // All slashes in path replaced by '\0'
2212 // segs[i] == Index of first char in segment i (0 <= i < segs.length)
2214 static private void split(char[] path, int[] segs) {
2215 int end = path.length - 1; // Index of last char in path
2216 int p = 0; // Index of next char in path
2217 int i = 0; // Index of current segment
2219 // Skip initial slashes
2221 if (path[p] != '/') break;
2228 // Note start of segment
2231 // Find beginning of next segment
2233 if (path[p++] != '/')
2237 // Skip redundant slashes
2239 if (path[p] != '/') break;
2246 if (i != segs.length)
2247 throw new InternalError(); // ASSERT
2251 // Join the segments in the given path according to the given segment-index
2252 // array, ignoring those segments whose index entries have been set to -1,
2253 // and inserting slashes as needed. Return the length of the resulting
2257 // segs[i] == -1 implies segment i is to be ignored
2258 // path computed by split, as above, with '\0' having replaced '/'
2261 // path[0] .. path[return value] == Resulting path
2263 static private int join(char[] path, int[] segs) {
2264 int ns = segs.length; // Number of segments
2265 int end = path.length - 1; // Index of last char in path
2266 int p = 0; // Index of next path char to write
2268 if (path[p] == '\0') {
2269 // Restore initial slash for absolute paths
2273 for (int i = 0; i < ns; i++) {
2274 int q = segs[i]; // Current segment
2276 // Ignore this segment
2280 // We're already at this segment, so just skip to its end
2281 while ((p <= end) && (path[p] != '\0'))
2284 // Preserve trailing slash
2289 while ((q <= end) && (path[q] != '\0'))
2290 path[p++] = path[q++];
2292 // Preserve trailing slash
2296 throw new InternalError(); // ASSERT false
2303 // Remove "." segments from the given path, and remove segment pairs
2304 // consisting of a non-".." segment followed by a ".." segment.
2306 private static void removeDots(char[] path, int[] segs) {
2307 int ns = segs.length;
2308 int end = path.length - 1;
2310 for (int i = 0; i < ns; i++) {
2311 int dots = 0; // Number of dots found (0, 1, or 2)
2313 // Find next occurrence of "." or ".."
2316 if (path[p] == '.') {
2320 } else if (path[p + 1] == '\0') {
2323 } else if ((path[p + 1] == '.')
2325 || (path[p + 2] == '\0'))) {
2332 if ((i > ns) || (dots == 0))
2336 // Remove this occurrence of "."
2339 // If there is a preceding non-".." segment, remove both that
2340 // segment and this occurrence of ".."; otherwise, leave this
2341 // ".." segment as-is.
2343 for (j = i - 1; j >= 0; j--) {
2344 if (segs[j] != -1) break;
2348 if (!((path[q] == '.')
2349 && (path[q + 1] == '.')
2350 && (path[q + 2] == '\0'))) {
2360 // DEVIATION: If the normalized path is relative, and if the first
2361 // segment could be parsed as a scheme name, then prepend a "." segment
2363 private static void maybeAddLeadingDot(char[] path, int[] segs) {
2365 if (path[0] == '\0')
2366 // The path is absolute
2369 int ns = segs.length;
2370 int f = 0; // Index of first segment
2376 if ((f >= ns) || (f == 0))
2377 // The path is empty, or else the original first segment survived,
2378 // in which case we already know that no leading "." is needed
2382 while ((p < path.length) && (path[p] != ':') && (path[p] != '\0')) p++;
2383 if (p >= path.length || path[p] == '\0')
2384 // No colon in first segment, so no "." needed
2387 // At this point we know that the first segment is unused,
2388 // hence we can insert a "." segment at that position
2395 // Normalize the given path string. A normal path string has no empty
2396 // segments (i.e., occurrences of "//"), no segments equal to ".", and no
2397 // segments equal to ".." that are preceded by a segment not equal to "..".
2398 // In contrast to Unix-style pathname normalization, for URI paths we
2399 // always retain trailing slashes.
2401 private static String normalize(String ps) {
2403 // Does this path need normalization?
2404 int ns = needsNormalization(ps); // Number of segments
2406 // Nope -- just return it
2409 char[] path = ps.toCharArray(); // Path in char-array form
2411 // Split path into segments
2412 int[] segs = new int[ns]; // Segment-index array
2416 removeDots(path, segs);
2418 // Prevent scheme-name confusion
2419 maybeAddLeadingDot(path, segs);
2421 // Join the remaining segments and return the result
2422 String s = new String(path, 0, join(path, segs));
2424 // string was already normalized
2432 // -- Character classes for parsing --
2434 // RFC2396 precisely specifies which characters in the US-ASCII charset are
2435 // permissible in the various components of a URI reference. We here
2436 // define a set of mask pairs to aid in enforcing these restrictions. Each
2437 // mask pair consists of two longs, a low mask and a high mask. Taken
2438 // together they represent a 128-bit mask, where bit i is set iff the
2439 // character with value i is permitted.
2441 // This approach is more efficient than sequentially searching arrays of
2442 // permitted characters. It could be made still more efficient by
2443 // precompiling the mask information so that a character's presence in a
2444 // given mask could be determined by a single table lookup.
2446 // Compute the low-order mask for the characters in the given string
2447 private static long lowMask(String chars) {
2448 int n = chars.length();
2450 for (int i = 0; i < n; i++) {
2451 char c = chars.charAt(i);
2458 // Compute the high-order mask for the characters in the given string
2459 private static long highMask(String chars) {
2460 int n = chars.length();
2462 for (int i = 0; i < n; i++) {
2463 char c = chars.charAt(i);
2464 if ((c >= 64) && (c < 128))
2465 m |= (1L << (c - 64));
2470 // Compute a low-order mask for the characters
2471 // between first and last, inclusive
2472 private static long lowMask(char first, char last) {
2474 int f = Math.max(Math.min(first, 63), 0);
2475 int l = Math.max(Math.min(last, 63), 0);
2476 for (int i = f; i <= l; i++)
2481 // Compute a high-order mask for the characters
2482 // between first and last, inclusive
2483 private static long highMask(char first, char last) {
2485 int f = Math.max(Math.min(first, 127), 64) - 64;
2486 int l = Math.max(Math.min(last, 127), 64) - 64;
2487 for (int i = f; i <= l; i++)
2492 // Tell whether the given character is permitted by the given mask pair
2493 private static boolean match(char c, long lowMask, long highMask) {
2494 if (c == 0) // 0 doesn't have a slot in the mask. So, it never matches.
2497 return ((1L << c) & lowMask) != 0;
2499 return ((1L << (c - 64)) & highMask) != 0;
2503 // Character-class masks, in reverse order from RFC2396 because
2504 // initializers for static fields cannot make forward references.
2506 // digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" |
2508 private static final long L_DIGIT = lowMask('0', '9');
2509 private static final long H_DIGIT = 0L;
2511 // upalpha = "A" | "B" | "C" | "D" | "E" | "F" | "G" | "H" | "I" |
2512 // "J" | "K" | "L" | "M" | "N" | "O" | "P" | "Q" | "R" |
2513 // "S" | "T" | "U" | "V" | "W" | "X" | "Y" | "Z"
2514 private static final long L_UPALPHA = 0L;
2515 private static final long H_UPALPHA = highMask('A', 'Z');
2517 // lowalpha = "a" | "b" | "c" | "d" | "e" | "f" | "g" | "h" | "i" |
2518 // "j" | "k" | "l" | "m" | "n" | "o" | "p" | "q" | "r" |
2519 // "s" | "t" | "u" | "v" | "w" | "x" | "y" | "z"
2520 private static final long L_LOWALPHA = 0L;
2521 private static final long H_LOWALPHA = highMask('a', 'z');
2523 // alpha = lowalpha | upalpha
2524 private static final long L_ALPHA = L_LOWALPHA | L_UPALPHA;
2525 private static final long H_ALPHA = H_LOWALPHA | H_UPALPHA;
2527 // alphanum = alpha | digit
2528 private static final long L_ALPHANUM = L_DIGIT | L_ALPHA;
2529 private static final long H_ALPHANUM = H_DIGIT | H_ALPHA;
2531 // hex = digit | "A" | "B" | "C" | "D" | "E" | "F" |
2532 // "a" | "b" | "c" | "d" | "e" | "f"
2533 private static final long L_HEX = L_DIGIT;
2534 private static final long H_HEX = highMask('A', 'F') | highMask('a', 'f');
2536 // mark = "-" | "_" | "." | "!" | "~" | "*" | "'" |
2538 private static final long L_MARK = lowMask("-_.!~*'()");
2539 private static final long H_MARK = highMask("-_.!~*'()");
2541 // unreserved = alphanum | mark
2542 private static final long L_UNRESERVED = L_ALPHANUM | L_MARK;
2543 private static final long H_UNRESERVED = H_ALPHANUM | H_MARK;
2545 // reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" |
2546 // "$" | "," | "[" | "]"
2547 // Added per RFC2732: "[", "]"
2548 private static final long L_RESERVED = lowMask(";/?:@&=+$,[]");
2549 private static final long H_RESERVED = highMask(";/?:@&=+$,[]");
2551 // The zero'th bit is used to indicate that escape pairs and non-US-ASCII
2552 // characters are allowed; this is handled by the scanEscape method below.
2553 private static final long L_ESCAPED = 1L;
2554 private static final long H_ESCAPED = 0L;
2556 // uric = reserved | unreserved | escaped
2557 private static final long L_URIC = L_RESERVED | L_UNRESERVED | L_ESCAPED;
2558 private static final long H_URIC = H_RESERVED | H_UNRESERVED | H_ESCAPED;
2560 // pchar = unreserved | escaped |
2561 // ":" | "@" | "&" | "=" | "+" | "$" | ","
2562 private static final long L_PCHAR
2563 = L_UNRESERVED | L_ESCAPED | lowMask(":@&=+$,");
2564 private static final long H_PCHAR
2565 = H_UNRESERVED | H_ESCAPED | highMask(":@&=+$,");
2567 // All valid path characters
2568 private static final long L_PATH = L_PCHAR | lowMask(";/");
2569 private static final long H_PATH = H_PCHAR | highMask(";/");
2571 // Dash, for use in domainlabel and toplabel
2572 private static final long L_DASH = lowMask("-");
2573 private static final long H_DASH = highMask("-");
2575 // Dot, for use in hostnames
2576 private static final long L_DOT = lowMask(".");
2577 private static final long H_DOT = highMask(".");
2579 // userinfo = *( unreserved | escaped |
2580 // ";" | ":" | "&" | "=" | "+" | "$" | "," )
2581 private static final long L_USERINFO
2582 = L_UNRESERVED | L_ESCAPED | lowMask(";:&=+$,");
2583 private static final long H_USERINFO
2584 = H_UNRESERVED | H_ESCAPED | highMask(";:&=+$,");
2586 // reg_name = 1*( unreserved | escaped | "$" | "," |
2587 // ";" | ":" | "@" | "&" | "=" | "+" )
2588 private static final long L_REG_NAME
2589 = L_UNRESERVED | L_ESCAPED | lowMask("$,;:@&=+");
2590 private static final long H_REG_NAME
2591 = H_UNRESERVED | H_ESCAPED | highMask("$,;:@&=+");
2593 // All valid characters for server-based authorities
2594 private static final long L_SERVER
2595 = L_USERINFO | L_ALPHANUM | L_DASH | lowMask(".:@[]");
2596 private static final long H_SERVER
2597 = H_USERINFO | H_ALPHANUM | H_DASH | highMask(".:@[]");
2599 // Special case of server authority that represents an IPv6 address
2600 // In this case, a % does not signify an escape sequence
2601 private static final long L_SERVER_PERCENT
2602 = L_SERVER | lowMask("%");
2603 private static final long H_SERVER_PERCENT
2604 = H_SERVER | highMask("%");
2605 private static final long L_LEFT_BRACKET = lowMask("[");
2606 private static final long H_LEFT_BRACKET = highMask("[");
2608 // scheme = alpha *( alpha | digit | "+" | "-" | "." )
2609 private static final long L_SCHEME = L_ALPHA | L_DIGIT | lowMask("+-.");
2610 private static final long H_SCHEME = H_ALPHA | H_DIGIT | highMask("+-.");
2612 // uric_no_slash = unreserved | escaped | ";" | "?" | ":" | "@" |
2613 // "&" | "=" | "+" | "$" | ","
2614 private static final long L_URIC_NO_SLASH
2615 = L_UNRESERVED | L_ESCAPED | lowMask(";?:@&=+$,");
2616 private static final long H_URIC_NO_SLASH
2617 = H_UNRESERVED | H_ESCAPED | highMask(";?:@&=+$,");
2620 // -- Escaping and encoding --
2622 private final static char[] hexDigits = {
2623 '0', '1', '2', '3', '4', '5', '6', '7',
2624 '8', '9', 'A', 'B', 'C', 'D', 'E', 'F'
2627 private static void appendEscape(StringBuffer sb, byte b) {
2629 sb.append(hexDigits[(b >> 4) & 0x0f]);
2630 sb.append(hexDigits[(b >> 0) & 0x0f]);
2633 private static void appendEncoded(StringBuffer sb, char c) {
2634 ByteBuffer bb = null;
2636 bb = ThreadLocalCoders.encoderFor("UTF-8")
2637 .encode(CharBuffer.wrap("" + c));
2638 } catch (CharacterCodingException x) {
2641 while (bb.hasRemaining()) {
2642 int b = bb.get() & 0xff;
2644 appendEscape(sb, (byte)b);
2650 // Quote any characters in s that are not permitted
2651 // by the given mask pair
2653 private static String quote(String s, long lowMask, long highMask) {
2655 StringBuffer sb = null;
2656 boolean allowNonASCII = ((lowMask & L_ESCAPED) != 0);
2657 for (int i = 0; i < s.length(); i++) {
2658 char c = s.charAt(i);
2660 if (!match(c, lowMask, highMask)) {
2662 sb = new StringBuffer();
2663 sb.append(s.substring(0, i));
2665 appendEscape(sb, (byte)c);
2670 } else if (allowNonASCII
2671 && (Character.isSpaceChar(c)
2672 || Character.isISOControl(c))) {
2674 sb = new StringBuffer();
2675 sb.append(s.substring(0, i));
2677 appendEncoded(sb, c);
2683 return (sb == null) ? s : sb.toString();
2686 // Encodes all characters >= \u0080 into escaped, normalized UTF-8 octets,
2687 // assuming that s is otherwise legal
2689 private static String encode(String s) {
2694 // First check whether we actually need to encode
2696 if (s.charAt(i) >= '\u0080')
2702 String ns = Normalizer.normalize(s, Normalizer.Form.NFC);
2703 ByteBuffer bb = null;
2705 bb = ThreadLocalCoders.encoderFor("UTF-8")
2706 .encode(CharBuffer.wrap(ns));
2707 } catch (CharacterCodingException x) {
2711 StringBuffer sb = new StringBuffer();
2712 while (bb.hasRemaining()) {
2713 int b = bb.get() & 0xff;
2715 appendEscape(sb, (byte)b);
2719 return sb.toString();
2722 private static int decode(char c) {
2723 if ((c >= '0') && (c <= '9'))
2725 if ((c >= 'a') && (c <= 'f'))
2726 return c - 'a' + 10;
2727 if ((c >= 'A') && (c <= 'F'))
2728 return c - 'A' + 10;
2733 private static byte decode(char c1, char c2) {
2734 return (byte)( ((decode(c1) & 0xf) << 4)
2735 | ((decode(c2) & 0xf) << 0));
2738 // Evaluates all escapes in s, applying UTF-8 decoding if needed. Assumes
2739 // that escapes are well-formed syntactically, i.e., of the form %XX. If a
2740 // sequence of escaped octets is not valid UTF-8 then the erroneous octets
2741 // are replaced with '\uFFFD'.
2742 // Exception: any "%" found between "[]" is left alone. It is an IPv6 literal
2745 private static String decode(String s) {
2751 if (s.indexOf('%') < 0)
2754 StringBuffer sb = new StringBuffer(n);
2755 ByteBuffer bb = ByteBuffer.allocate(n);
2756 CharBuffer cb = CharBuffer.allocate(n);
2757 CharsetDecoder dec = ThreadLocalCoders.decoderFor("UTF-8")
2758 .onMalformedInput(CodingErrorAction.REPLACE)
2759 .onUnmappableCharacter(CodingErrorAction.REPLACE);
2761 // This is not horribly efficient, but it will do for now
2762 char c = s.charAt(0);
2763 boolean betweenBrackets = false;
2765 for (int i = 0; i < n;) {
2766 assert c == s.charAt(i); // Loop invariant
2768 betweenBrackets = true;
2769 } else if (betweenBrackets && c == ']') {
2770 betweenBrackets = false;
2772 if (c != '%' || betweenBrackets) {
2782 assert (n - i >= 2);
2783 bb.put(decode(s.charAt(++i), s.charAt(++i)));
2793 CoderResult cr = dec.decode(bb, cb, true);
2794 assert cr.isUnderflow();
2796 assert cr.isUnderflow();
2797 sb.append(cb.flip().toString());
2800 return sb.toString();
2806 // For convenience we wrap the input URI string in a new instance of the
2807 // following internal class. This saves always having to pass the input
2808 // string as an argument to each internal scan/parse method.
2810 private class Parser {
2812 private String input; // URI input string
2813 private boolean requireServerAuthority = false;
2820 // -- Methods for throwing URISyntaxException in various ways --
2822 private void fail(String reason) throws URISyntaxException {
2823 throw new URISyntaxException(input, reason);
2826 private void fail(String reason, int p) throws URISyntaxException {
2827 throw new URISyntaxException(input, reason, p);
2830 private void failExpecting(String expected, int p)
2831 throws URISyntaxException
2833 fail("Expected " + expected, p);
2836 private void failExpecting(String expected, String prior, int p)
2837 throws URISyntaxException
2839 fail("Expected " + expected + " following " + prior, p);
2843 // -- Simple access to the input string --
2845 // Return a substring of the input string
2847 private String substring(int start, int end) {
2848 return input.substring(start, end);
2851 // Return the char at position p,
2852 // assuming that p < input.length()
2854 private char charAt(int p) {
2855 return input.charAt(p);
2858 // Tells whether start < end and, if so, whether charAt(start) == c
2860 private boolean at(int start, int end, char c) {
2861 return (start < end) && (charAt(start) == c);
2864 // Tells whether start + s.length() < end and, if so,
2865 // whether the chars at the start position match s exactly
2867 private boolean at(int start, int end, String s) {
2869 int sn = s.length();
2874 if (charAt(p++) != s.charAt(i)) {
2885 // The various scan and parse methods that follow use a uniform
2886 // convention of taking the current start position and end index as
2887 // their first two arguments. The start is inclusive while the end is
2888 // exclusive, just as in the String class, i.e., a start/end pair
2889 // denotes the left-open interval [start, end) of the input string.
2891 // These methods never proceed past the end position. They may return
2892 // -1 to indicate outright failure, but more often they simply return
2893 // the position of the first char after the last char scanned. Thus
2894 // a typical idiom is
2897 // int q = scan(p, end, ...);
2899 // // We scanned something
2902 // // We scanned nothing
2904 // else if (q == -1)
2905 // // Something went wrong
2909 // Scan a specific char: If the char at the given start position is
2910 // equal to c, return the index of the next char; otherwise, return the
2913 private int scan(int start, int end, char c) {
2914 if ((start < end) && (charAt(start) == c))
2919 // Scan forward from the given start position. Stop at the first char
2920 // in the err string (in which case -1 is returned), or the first char
2921 // in the stop string (in which case the index of the preceding char is
2922 // returned), or the end of the input string (in which case the length
2923 // of the input string is returned). May return the start position if
2926 private int scan(int start, int end, String err, String stop) {
2930 if (err.indexOf(c) >= 0)
2932 if (stop.indexOf(c) >= 0)
2939 // Scan a potential escape sequence, starting at the given position,
2940 // with the given first char (i.e., charAt(start) == c).
2942 // This method assumes that if escapes are allowed then visible
2943 // non-US-ASCII chars are also allowed.
2945 private int scanEscape(int start, int n, char first)
2946 throws URISyntaxException
2951 // Process escape pair
2953 && match(charAt(p + 1), L_HEX, H_HEX)
2954 && match(charAt(p + 2), L_HEX, H_HEX)) {
2957 fail("Malformed escape pair", p);
2958 } else if ((c > 128)
2959 && !Character.isSpaceChar(c)
2960 && !Character.isISOControl(c)) {
2961 // Allow unescaped but visible non-US-ASCII chars
2967 // Scan chars that match the given mask pair
2969 private int scan(int start, int n, long lowMask, long highMask)
2970 throws URISyntaxException
2975 if (match(c, lowMask, highMask)) {
2979 if ((lowMask & L_ESCAPED) != 0) {
2980 int q = scanEscape(p, n, c);
2991 // Check that each of the chars in [start, end) matches the given mask
2993 private void checkChars(int start, int end,
2994 long lowMask, long highMask,
2996 throws URISyntaxException
2998 int p = scan(start, end, lowMask, highMask);
3000 fail("Illegal character in " + what, p);
3003 // Check that the char at position p matches the given mask
3005 private void checkChar(int p,
3006 long lowMask, long highMask,
3008 throws URISyntaxException
3010 checkChars(p, p + 1, lowMask, highMask, what);
3016 // [<scheme>:]<scheme-specific-part>[#<fragment>]
3018 void parse(boolean rsa) throws URISyntaxException {
3019 requireServerAuthority = rsa;
3020 int ssp; // Start of scheme-specific part
3021 int n = input.length();
3022 int p = scan(0, n, "/?#", ":");
3023 if ((p >= 0) && at(p, n, ':')) {
3025 failExpecting("scheme name", 0);
3026 checkChar(0, L_ALPHA, H_ALPHA, "scheme name");
3027 checkChars(1, p, L_SCHEME, H_SCHEME, "scheme name");
3028 scheme = substring(0, p);
3031 if (at(p, n, '/')) {
3032 p = parseHierarchical(p, n);
3034 int q = scan(p, n, "", "#");
3036 failExpecting("scheme-specific part", p);
3037 checkChars(p, q, L_URIC, H_URIC, "opaque part");
3042 p = parseHierarchical(0, n);
3044 schemeSpecificPart = substring(ssp, p);
3045 if (at(p, n, '#')) {
3046 checkChars(p + 1, n, L_URIC, H_URIC, "fragment");
3047 fragment = substring(p + 1, n);
3051 fail("end of URI", p);
3054 // [//authority]<path>[?<query>]
3056 // DEVIATION from RFC2396: We allow an empty authority component as
3057 // long as it's followed by a non-empty path, query component, or
3058 // fragment component. This is so that URIs such as "file:///foo/bar"
3059 // will parse. This seems to be the intent of RFC2396, though the
3060 // grammar does not permit it. If the authority is empty then the
3061 // userInfo, host, and port components are undefined.
3063 // DEVIATION from RFC2396: We allow empty relative paths. This seems
3064 // to be the intent of RFC2396, but the grammar does not permit it.
3065 // The primary consequence of this deviation is that "#f" parses as a
3066 // relative URI with an empty path.
3068 private int parseHierarchical(int start, int n)
3069 throws URISyntaxException
3072 if (at(p, n, '/') && at(p + 1, n, '/')) {
3074 int q = scan(p, n, "", "/?#");
3076 p = parseAuthority(p, q);
3078 // DEVIATION: Allow empty authority prior to non-empty
3079 // path, query component or fragment identifier
3081 failExpecting("authority", p);
3083 int q = scan(p, n, "", "?#"); // DEVIATION: May be empty
3084 checkChars(p, q, L_PATH, H_PATH, "path");
3085 path = substring(p, q);
3087 if (at(p, n, '?')) {
3089 q = scan(p, n, "", "#");
3090 checkChars(p, q, L_URIC, H_URIC, "query");
3091 query = substring(p, q);
3097 // authority = server | reg_name
3099 // Ambiguity: An authority that is a registry name rather than a server
3100 // might have a prefix that parses as a server. We use the fact that
3101 // the authority component is always followed by '/' or the end of the
3102 // input string to resolve this: If the complete authority did not
3103 // parse as a server then we try to parse it as a registry name.
3105 private int parseAuthority(int start, int n)
3106 throws URISyntaxException
3110 URISyntaxException ex = null;
3112 boolean serverChars;
3115 if (scan(p, n, "", "]") > p) {
3116 // contains a literal IPv6 address, therefore % is allowed
3117 serverChars = (scan(p, n, L_SERVER_PERCENT, H_SERVER_PERCENT) == n);
3119 serverChars = (scan(p, n, L_SERVER, H_SERVER) == n);
3121 regChars = (scan(p, n, L_REG_NAME, H_REG_NAME) == n);
3123 if (regChars && !serverChars) {
3124 // Must be a registry-based authority
3125 authority = substring(p, n);
3130 // Might be (probably is) a server-based authority, so attempt
3131 // to parse it as such. If the attempt fails, try to treat it
3132 // as a registry-based authority.
3134 q = parseServer(p, n);
3136 failExpecting("end of authority", q);
3137 authority = substring(p, n);
3138 } catch (URISyntaxException x) {
3139 // Undo results of failed parse
3143 if (requireServerAuthority) {
3144 // If we're insisting upon a server-based authority,
3145 // then just re-throw the exception
3148 // Save the exception in case it doesn't parse as a
3158 // Registry-based authority
3159 authority = substring(p, n);
3160 } else if (ex != null) {
3161 // Re-throw exception; it was probably due to
3162 // a malformed IPv6 address
3165 fail("Illegal character in authority", q);
3173 // [<userinfo>@]<host>[:<port>]
3175 private int parseServer(int start, int n)
3176 throws URISyntaxException
3182 q = scan(p, n, "/?#", "@");
3183 if ((q >= p) && at(q, n, '@')) {
3184 checkChars(p, q, L_USERINFO, H_USERINFO, "user info");
3185 userInfo = substring(p, q);
3186 p = q + 1; // Skip '@'
3189 // hostname, IPv4 address, or IPv6 address
3190 if (at(p, n, '[')) {
3191 // DEVIATION from RFC2396: Support IPv6 addresses, per RFC2732
3193 q = scan(p, n, "/?#", "]");
3194 if ((q > p) && at(q, n, ']')) {
3195 // look for a "%" scope id
3196 int r = scan (p, q, "", "%");
3198 parseIPv6Reference(p, r);
3200 fail ("scope id expected");
3202 checkChars (r+1, q, L_ALPHANUM, H_ALPHANUM,
3205 parseIPv6Reference(p, q);
3207 host = substring(p-1, q+1);
3210 failExpecting("closing bracket for IPv6 address", q);
3213 q = parseIPv4Address(p, n);
3215 q = parseHostname(p, n);
3220 if (at(p, n, ':')) {
3222 q = scan(p, n, "", "/");
3224 checkChars(p, q, L_DIGIT, H_DIGIT, "port number");
3226 port = Integer.parseInt(substring(p, q));
3227 } catch (NumberFormatException x) {
3228 fail("Malformed port number", p);
3234 failExpecting("port number", p);
3239 // Scan a string of decimal digits whose value fits in a byte
3241 private int scanByte(int start, int n)
3242 throws URISyntaxException
3245 int q = scan(p, n, L_DIGIT, H_DIGIT);
3246 if (q <= p) return q;
3247 if (Integer.parseInt(substring(p, q)) > 255) return p;
3251 // Scan an IPv4 address.
3253 // If the strict argument is true then we require that the given
3254 // interval contain nothing besides an IPv4 address; if it is false
3255 // then we only require that it start with an IPv4 address.
3257 // If the interval does not contain or start with (depending upon the
3258 // strict argument) a legal IPv4 address characters then we return -1
3259 // immediately; otherwise we insist that these characters parse as a
3260 // legal IPv4 address and throw an exception on failure.
3262 // We assume that any string of decimal digits and dots must be an IPv4
3263 // address. It won't parse as a hostname anyway, so making that
3264 // assumption here allows more meaningful exceptions to be thrown.
3266 private int scanIPv4Address(int start, int n, boolean strict)
3267 throws URISyntaxException
3271 int m = scan(p, n, L_DIGIT | L_DOT, H_DIGIT | H_DOT);
3272 if ((m <= p) || (strict && (m != n)))
3275 // Per RFC2732: At most three digits per byte
3276 // Further constraint: Each element fits in a byte
3277 if ((q = scanByte(p, m)) <= p) break; p = q;
3278 if ((q = scan(p, m, '.')) <= p) break; p = q;
3279 if ((q = scanByte(p, m)) <= p) break; p = q;
3280 if ((q = scan(p, m, '.')) <= p) break; p = q;
3281 if ((q = scanByte(p, m)) <= p) break; p = q;
3282 if ((q = scan(p, m, '.')) <= p) break; p = q;
3283 if ((q = scanByte(p, m)) <= p) break; p = q;
3287 fail("Malformed IPv4 address", q);
3291 // Take an IPv4 address: Throw an exception if the given interval
3292 // contains anything except an IPv4 address
3294 private int takeIPv4Address(int start, int n, String expected)
3295 throws URISyntaxException
3297 int p = scanIPv4Address(start, n, true);
3299 failExpecting(expected, start);
3303 // Attempt to parse an IPv4 address, returning -1 on failure but
3304 // allowing the given interval to contain [:<characters>] after
3305 // the IPv4 address.
3307 private int parseIPv4Address(int start, int n) {
3311 p = scanIPv4Address(start, n, false);
3312 } catch (URISyntaxException x) {
3314 } catch (NumberFormatException nfe) {
3318 if (p > start && p < n) {
3319 // IPv4 address is followed by something - check that
3320 // it's a ":" as this is the only valid character to
3321 // follow an address.
3322 if (charAt(p) != ':') {
3328 host = substring(start, p);
3333 // hostname = domainlabel [ "." ] | 1*( domainlabel "." ) toplabel [ "." ]
3334 // domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum
3335 // toplabel = alpha | alpha *( alphanum | "-" ) alphanum
3337 private int parseHostname(int start, int n)
3338 throws URISyntaxException
3342 int l = -1; // Start of last parsed label
3345 // domainlabel = alphanum [ *( alphanum | "-" ) alphanum ]
3346 q = scan(p, n, L_ALPHANUM, H_ALPHANUM);
3352 q = scan(p, n, L_ALPHANUM | L_DASH, H_ALPHANUM | H_DASH);
3354 if (charAt(q - 1) == '-')
3355 fail("Illegal character in hostname", q - 1);
3359 q = scan(p, n, '.');
3365 if ((p < n) && !at(p, n, ':'))
3366 fail("Illegal character in hostname", p);
3369 failExpecting("hostname", start);
3371 // for a fully qualified hostname check that the rightmost
3372 // label starts with an alpha character.
3373 if (l > start && !match(charAt(l), L_ALPHA, H_ALPHA)) {
3374 fail("Illegal character in hostname", l);
3377 host = substring(start, p);
3382 // IPv6 address parsing, from RFC2373: IPv6 Addressing Architecture
3384 // Bug: The grammar in RFC2373 Appendix B does not allow addresses of
3385 // the form ::12.34.56.78, which are clearly shown in the examples
3386 // earlier in the document. Here is the original grammar:
3388 // IPv6address = hexpart [ ":" IPv4address ]
3389 // hexpart = hexseq | hexseq "::" [ hexseq ] | "::" [ hexseq ]
3390 // hexseq = hex4 *( ":" hex4)
3393 // We therefore use the following revised grammar:
3395 // IPv6address = hexseq [ ":" IPv4address ]
3396 // | hexseq [ "::" [ hexpost ] ]
3397 // | "::" [ hexpost ]
3398 // hexpost = hexseq | hexseq ":" IPv4address | IPv4address
3399 // hexseq = hex4 *( ":" hex4)
3402 // This covers all and only the following cases:
3405 // hexseq : IPv4address
3408 // hexseq :: hexseq : IPv4address
3409 // hexseq :: IPv4address
3411 // :: hexseq : IPv4address
3415 // Additionally we constrain the IPv6 address as follows :-
3417 // i. IPv6 addresses without compressed zeros should contain
3418 // exactly 16 bytes.
3420 // ii. IPv6 addresses with compressed zeros should contain
3421 // less than 16 bytes.
3423 private int ipv6byteCount = 0;
3425 private int parseIPv6Reference(int start, int n)
3426 throws URISyntaxException
3430 boolean compressedZeros = false;
3432 q = scanHexSeq(p, n);
3436 if (at(p, n, "::")) {
3437 compressedZeros = true;
3438 p = scanHexPost(p + 2, n);
3439 } else if (at(p, n, ':')) {
3440 p = takeIPv4Address(p + 1, n, "IPv4 address");
3443 } else if (at(p, n, "::")) {
3444 compressedZeros = true;
3445 p = scanHexPost(p + 2, n);
3448 fail("Malformed IPv6 address", start);
3449 if (ipv6byteCount > 16)
3450 fail("IPv6 address too long", start);
3451 if (!compressedZeros && ipv6byteCount < 16)
3452 fail("IPv6 address too short", start);
3453 if (compressedZeros && ipv6byteCount == 16)
3454 fail("Malformed IPv6 address", start);
3459 private int scanHexPost(int start, int n)
3460 throws URISyntaxException
3468 q = scanHexSeq(p, n);
3471 if (at(p, n, ':')) {
3473 p = takeIPv4Address(p, n, "hex digits or IPv4 address");
3477 p = takeIPv4Address(p, n, "hex digits or IPv4 address");
3483 // Scan a hex sequence; return -1 if one could not be scanned
3485 private int scanHexSeq(int start, int n)
3486 throws URISyntaxException
3491 q = scan(p, n, L_HEX, H_HEX);
3494 if (at(q, n, '.')) // Beginning of IPv4 address
3497 fail("IPv6 hexadecimal digit sequence too long", p);
3503 if (at(p + 1, n, ':'))
3506 q = scan(p, n, L_HEX, H_HEX);
3508 failExpecting("digits for an IPv6 address", p);
3509 if (at(q, n, '.')) { // Beginning of IPv4 address
3514 fail("IPv6 hexadecimal digit sequence too long", p);