|
jpayne@69
|
1 // © 2016 and later: Unicode, Inc. and others.
|
|
jpayne@69
|
2 // License & terms of use: http://www.unicode.org/copyright.html
|
|
jpayne@69
|
3 /*
|
|
jpayne@69
|
4 ***************************************************************************
|
|
jpayne@69
|
5 * Copyright (C) 1999-2016, International Business Machines Corporation
|
|
jpayne@69
|
6 * and others. All Rights Reserved.
|
|
jpayne@69
|
7 ***************************************************************************
|
|
jpayne@69
|
8 * Date Name Description
|
|
jpayne@69
|
9 * 10/20/99 alan Creation.
|
|
jpayne@69
|
10 ***************************************************************************
|
|
jpayne@69
|
11 */
|
|
jpayne@69
|
12
|
|
jpayne@69
|
13 #ifndef UNICODESET_H
|
|
jpayne@69
|
14 #define UNICODESET_H
|
|
jpayne@69
|
15
|
|
jpayne@69
|
16 #include "unicode/utypes.h"
|
|
jpayne@69
|
17
|
|
jpayne@69
|
18 #if U_SHOW_CPLUSPLUS_API
|
|
jpayne@69
|
19
|
|
jpayne@69
|
20 #include "unicode/ucpmap.h"
|
|
jpayne@69
|
21 #include "unicode/unifilt.h"
|
|
jpayne@69
|
22 #include "unicode/unistr.h"
|
|
jpayne@69
|
23 #include "unicode/uset.h"
|
|
jpayne@69
|
24
|
|
jpayne@69
|
25 /**
|
|
jpayne@69
|
26 * \file
|
|
jpayne@69
|
27 * \brief C++ API: Unicode Set
|
|
jpayne@69
|
28 */
|
|
jpayne@69
|
29
|
|
jpayne@69
|
30 U_NAMESPACE_BEGIN
|
|
jpayne@69
|
31
|
|
jpayne@69
|
32 // Forward Declarations.
|
|
jpayne@69
|
33 class BMPSet;
|
|
jpayne@69
|
34 class ParsePosition;
|
|
jpayne@69
|
35 class RBBIRuleScanner;
|
|
jpayne@69
|
36 class SymbolTable;
|
|
jpayne@69
|
37 class UnicodeSetStringSpan;
|
|
jpayne@69
|
38 class UVector;
|
|
jpayne@69
|
39 class RuleCharacterIterator;
|
|
jpayne@69
|
40
|
|
jpayne@69
|
41 /**
|
|
jpayne@69
|
42 * A mutable set of Unicode characters and multicharacter strings. Objects of this class
|
|
jpayne@69
|
43 * represent <em>character classes</em> used in regular expressions.
|
|
jpayne@69
|
44 * A character specifies a subset of Unicode code points. Legal
|
|
jpayne@69
|
45 * code points are U+0000 to U+10FFFF, inclusive.
|
|
jpayne@69
|
46 *
|
|
jpayne@69
|
47 * <p>The UnicodeSet class is not designed to be subclassed.
|
|
jpayne@69
|
48 *
|
|
jpayne@69
|
49 * <p><code>UnicodeSet</code> supports two APIs. The first is the
|
|
jpayne@69
|
50 * <em>operand</em> API that allows the caller to modify the value of
|
|
jpayne@69
|
51 * a <code>UnicodeSet</code> object. It conforms to Java 2's
|
|
jpayne@69
|
52 * <code>java.util.Set</code> interface, although
|
|
jpayne@69
|
53 * <code>UnicodeSet</code> does not actually implement that
|
|
jpayne@69
|
54 * interface. All methods of <code>Set</code> are supported, with the
|
|
jpayne@69
|
55 * modification that they take a character range or single character
|
|
jpayne@69
|
56 * instead of an <code>Object</code>, and they take a
|
|
jpayne@69
|
57 * <code>UnicodeSet</code> instead of a <code>Collection</code>. The
|
|
jpayne@69
|
58 * operand API may be thought of in terms of boolean logic: a boolean
|
|
jpayne@69
|
59 * OR is implemented by <code>add</code>, a boolean AND is implemented
|
|
jpayne@69
|
60 * by <code>retain</code>, a boolean XOR is implemented by
|
|
jpayne@69
|
61 * <code>complement</code> taking an argument, and a boolean NOT is
|
|
jpayne@69
|
62 * implemented by <code>complement</code> with no argument. In terms
|
|
jpayne@69
|
63 * of traditional set theory function names, <code>add</code> is a
|
|
jpayne@69
|
64 * union, <code>retain</code> is an intersection, <code>remove</code>
|
|
jpayne@69
|
65 * is an asymmetric difference, and <code>complement</code> with no
|
|
jpayne@69
|
66 * argument is a set complement with respect to the superset range
|
|
jpayne@69
|
67 * <code>MIN_VALUE-MAX_VALUE</code>
|
|
jpayne@69
|
68 *
|
|
jpayne@69
|
69 * <p>The second API is the
|
|
jpayne@69
|
70 * <code>applyPattern()</code>/<code>toPattern()</code> API from the
|
|
jpayne@69
|
71 * <code>java.text.Format</code>-derived classes. Unlike the
|
|
jpayne@69
|
72 * methods that add characters, add categories, and control the logic
|
|
jpayne@69
|
73 * of the set, the method <code>applyPattern()</code> sets all
|
|
jpayne@69
|
74 * attributes of a <code>UnicodeSet</code> at once, based on a
|
|
jpayne@69
|
75 * string pattern.
|
|
jpayne@69
|
76 *
|
|
jpayne@69
|
77 * <p><b>Pattern syntax</b></p>
|
|
jpayne@69
|
78 *
|
|
jpayne@69
|
79 * Patterns are accepted by the constructors and the
|
|
jpayne@69
|
80 * <code>applyPattern()</code> methods and returned by the
|
|
jpayne@69
|
81 * <code>toPattern()</code> method. These patterns follow a syntax
|
|
jpayne@69
|
82 * similar to that employed by version 8 regular expression character
|
|
jpayne@69
|
83 * classes. Here are some simple examples:
|
|
jpayne@69
|
84 *
|
|
jpayne@69
|
85 * \htmlonly<blockquote>\endhtmlonly
|
|
jpayne@69
|
86 * <table>
|
|
jpayne@69
|
87 * <tr align="top">
|
|
jpayne@69
|
88 * <td nowrap valign="top" align="left"><code>[]</code></td>
|
|
jpayne@69
|
89 * <td valign="top">No characters</td>
|
|
jpayne@69
|
90 * </tr><tr align="top">
|
|
jpayne@69
|
91 * <td nowrap valign="top" align="left"><code>[a]</code></td>
|
|
jpayne@69
|
92 * <td valign="top">The character 'a'</td>
|
|
jpayne@69
|
93 * </tr><tr align="top">
|
|
jpayne@69
|
94 * <td nowrap valign="top" align="left"><code>[ae]</code></td>
|
|
jpayne@69
|
95 * <td valign="top">The characters 'a' and 'e'</td>
|
|
jpayne@69
|
96 * </tr>
|
|
jpayne@69
|
97 * <tr>
|
|
jpayne@69
|
98 * <td nowrap valign="top" align="left"><code>[a-e]</code></td>
|
|
jpayne@69
|
99 * <td valign="top">The characters 'a' through 'e' inclusive, in Unicode code
|
|
jpayne@69
|
100 * point order</td>
|
|
jpayne@69
|
101 * </tr>
|
|
jpayne@69
|
102 * <tr>
|
|
jpayne@69
|
103 * <td nowrap valign="top" align="left"><code>[\\u4E01]</code></td>
|
|
jpayne@69
|
104 * <td valign="top">The character U+4E01</td>
|
|
jpayne@69
|
105 * </tr>
|
|
jpayne@69
|
106 * <tr>
|
|
jpayne@69
|
107 * <td nowrap valign="top" align="left"><code>[a{ab}{ac}]</code></td>
|
|
jpayne@69
|
108 * <td valign="top">The character 'a' and the multicharacter strings "ab" and
|
|
jpayne@69
|
109 * "ac"</td>
|
|
jpayne@69
|
110 * </tr>
|
|
jpayne@69
|
111 * <tr>
|
|
jpayne@69
|
112 * <td nowrap valign="top" align="left"><code>[\\p{Lu}]</code></td>
|
|
jpayne@69
|
113 * <td valign="top">All characters in the general category Uppercase Letter</td>
|
|
jpayne@69
|
114 * </tr>
|
|
jpayne@69
|
115 * </table>
|
|
jpayne@69
|
116 * \htmlonly</blockquote>\endhtmlonly
|
|
jpayne@69
|
117 *
|
|
jpayne@69
|
118 * Any character may be preceded by a backslash in order to remove any special
|
|
jpayne@69
|
119 * meaning. White space characters, as defined by UCharacter.isWhitespace(), are
|
|
jpayne@69
|
120 * ignored, unless they are escaped.
|
|
jpayne@69
|
121 *
|
|
jpayne@69
|
122 * <p>Property patterns specify a set of characters having a certain
|
|
jpayne@69
|
123 * property as defined by the Unicode standard. Both the POSIX-like
|
|
jpayne@69
|
124 * "[:Lu:]" and the Perl-like syntax "\\p{Lu}" are recognized. For a
|
|
jpayne@69
|
125 * complete list of supported property patterns, see the User's Guide
|
|
jpayne@69
|
126 * for UnicodeSet at
|
|
jpayne@69
|
127 * <a href="http://icu-project.org/userguide/unicodeSet.html">
|
|
jpayne@69
|
128 * http://icu-project.org/userguide/unicodeSet.html</a>.
|
|
jpayne@69
|
129 * Actual determination of property data is defined by the underlying
|
|
jpayne@69
|
130 * Unicode database as implemented by UCharacter.
|
|
jpayne@69
|
131 *
|
|
jpayne@69
|
132 * <p>Patterns specify individual characters, ranges of characters, and
|
|
jpayne@69
|
133 * Unicode property sets. When elements are concatenated, they
|
|
jpayne@69
|
134 * specify their union. To complement a set, place a '^' immediately
|
|
jpayne@69
|
135 * after the opening '['. Property patterns are inverted by modifying
|
|
jpayne@69
|
136 * their delimiters; "[:^foo]" and "\\P{foo}". In any other location,
|
|
jpayne@69
|
137 * '^' has no special meaning.
|
|
jpayne@69
|
138 *
|
|
jpayne@69
|
139 * <p>Ranges are indicated by placing two a '-' between two
|
|
jpayne@69
|
140 * characters, as in "a-z". This specifies the range of all
|
|
jpayne@69
|
141 * characters from the left to the right, in Unicode order. If the
|
|
jpayne@69
|
142 * left character is greater than or equal to the
|
|
jpayne@69
|
143 * right character it is a syntax error. If a '-' occurs as the first
|
|
jpayne@69
|
144 * character after the opening '[' or '[^', or if it occurs as the
|
|
jpayne@69
|
145 * last character before the closing ']', then it is taken as a
|
|
jpayne@69
|
146 * literal. Thus "[a\-b]", "[-ab]", and "[ab-]" all indicate the same
|
|
jpayne@69
|
147 * set of three characters, 'a', 'b', and '-'.
|
|
jpayne@69
|
148 *
|
|
jpayne@69
|
149 * <p>Sets may be intersected using the '&' operator or the asymmetric
|
|
jpayne@69
|
150 * set difference may be taken using the '-' operator, for example,
|
|
jpayne@69
|
151 * "[[:L:]&[\\u0000-\\u0FFF]]" indicates the set of all Unicode letters
|
|
jpayne@69
|
152 * with values less than 4096. Operators ('&' and '|') have equal
|
|
jpayne@69
|
153 * precedence and bind left-to-right. Thus
|
|
jpayne@69
|
154 * "[[:L:]-[a-z]-[\\u0100-\\u01FF]]" is equivalent to
|
|
jpayne@69
|
155 * "[[[:L:]-[a-z]]-[\\u0100-\\u01FF]]". This only really matters for
|
|
jpayne@69
|
156 * difference; intersection is commutative.
|
|
jpayne@69
|
157 *
|
|
jpayne@69
|
158 * <table>
|
|
jpayne@69
|
159 * <tr valign=top><td nowrap><code>[a]</code><td>The set containing 'a'
|
|
jpayne@69
|
160 * <tr valign=top><td nowrap><code>[a-z]</code><td>The set containing 'a'
|
|
jpayne@69
|
161 * through 'z' and all letters in between, in Unicode order
|
|
jpayne@69
|
162 * <tr valign=top><td nowrap><code>[^a-z]</code><td>The set containing
|
|
jpayne@69
|
163 * all characters but 'a' through 'z',
|
|
jpayne@69
|
164 * that is, U+0000 through 'a'-1 and 'z'+1 through U+10FFFF
|
|
jpayne@69
|
165 * <tr valign=top><td nowrap><code>[[<em>pat1</em>][<em>pat2</em>]]</code>
|
|
jpayne@69
|
166 * <td>The union of sets specified by <em>pat1</em> and <em>pat2</em>
|
|
jpayne@69
|
167 * <tr valign=top><td nowrap><code>[[<em>pat1</em>]&[<em>pat2</em>]]</code>
|
|
jpayne@69
|
168 * <td>The intersection of sets specified by <em>pat1</em> and <em>pat2</em>
|
|
jpayne@69
|
169 * <tr valign=top><td nowrap><code>[[<em>pat1</em>]-[<em>pat2</em>]]</code>
|
|
jpayne@69
|
170 * <td>The asymmetric difference of sets specified by <em>pat1</em> and
|
|
jpayne@69
|
171 * <em>pat2</em>
|
|
jpayne@69
|
172 * <tr valign=top><td nowrap><code>[:Lu:] or \\p{Lu}</code>
|
|
jpayne@69
|
173 * <td>The set of characters having the specified
|
|
jpayne@69
|
174 * Unicode property; in
|
|
jpayne@69
|
175 * this case, Unicode uppercase letters
|
|
jpayne@69
|
176 * <tr valign=top><td nowrap><code>[:^Lu:] or \\P{Lu}</code>
|
|
jpayne@69
|
177 * <td>The set of characters <em>not</em> having the given
|
|
jpayne@69
|
178 * Unicode property
|
|
jpayne@69
|
179 * </table>
|
|
jpayne@69
|
180 *
|
|
jpayne@69
|
181 * <p><b>Warning</b>: you cannot add an empty string ("") to a UnicodeSet.</p>
|
|
jpayne@69
|
182 *
|
|
jpayne@69
|
183 * <p><b>Formal syntax</b></p>
|
|
jpayne@69
|
184 *
|
|
jpayne@69
|
185 * \htmlonly<blockquote>\endhtmlonly
|
|
jpayne@69
|
186 * <table>
|
|
jpayne@69
|
187 * <tr align="top">
|
|
jpayne@69
|
188 * <td nowrap valign="top" align="right"><code>pattern := </code></td>
|
|
jpayne@69
|
189 * <td valign="top"><code>('[' '^'? item* ']') |
|
|
jpayne@69
|
190 * property</code></td>
|
|
jpayne@69
|
191 * </tr>
|
|
jpayne@69
|
192 * <tr align="top">
|
|
jpayne@69
|
193 * <td nowrap valign="top" align="right"><code>item := </code></td>
|
|
jpayne@69
|
194 * <td valign="top"><code>char | (char '-' char) | pattern-expr<br>
|
|
jpayne@69
|
195 * </code></td>
|
|
jpayne@69
|
196 * </tr>
|
|
jpayne@69
|
197 * <tr align="top">
|
|
jpayne@69
|
198 * <td nowrap valign="top" align="right"><code>pattern-expr := </code></td>
|
|
jpayne@69
|
199 * <td valign="top"><code>pattern | pattern-expr pattern |
|
|
jpayne@69
|
200 * pattern-expr op pattern<br>
|
|
jpayne@69
|
201 * </code></td>
|
|
jpayne@69
|
202 * </tr>
|
|
jpayne@69
|
203 * <tr align="top">
|
|
jpayne@69
|
204 * <td nowrap valign="top" align="right"><code>op := </code></td>
|
|
jpayne@69
|
205 * <td valign="top"><code>'&' | '-'<br>
|
|
jpayne@69
|
206 * </code></td>
|
|
jpayne@69
|
207 * </tr>
|
|
jpayne@69
|
208 * <tr align="top">
|
|
jpayne@69
|
209 * <td nowrap valign="top" align="right"><code>special := </code></td>
|
|
jpayne@69
|
210 * <td valign="top"><code>'[' | ']' | '-'<br>
|
|
jpayne@69
|
211 * </code></td>
|
|
jpayne@69
|
212 * </tr>
|
|
jpayne@69
|
213 * <tr align="top">
|
|
jpayne@69
|
214 * <td nowrap valign="top" align="right"><code>char := </code></td>
|
|
jpayne@69
|
215 * <td valign="top"><em>any character that is not</em><code> special<br>
|
|
jpayne@69
|
216 * | ('\' </code><em>any character</em><code>)<br>
|
|
jpayne@69
|
217 * | ('\\u' hex hex hex hex)<br>
|
|
jpayne@69
|
218 * </code></td>
|
|
jpayne@69
|
219 * </tr>
|
|
jpayne@69
|
220 * <tr align="top">
|
|
jpayne@69
|
221 * <td nowrap valign="top" align="right"><code>hex := </code></td>
|
|
jpayne@69
|
222 * <td valign="top"><em>any character for which
|
|
jpayne@69
|
223 * </em><code>Character.digit(c, 16)</code><em>
|
|
jpayne@69
|
224 * returns a non-negative result</em></td>
|
|
jpayne@69
|
225 * </tr>
|
|
jpayne@69
|
226 * <tr>
|
|
jpayne@69
|
227 * <td nowrap valign="top" align="right"><code>property := </code></td>
|
|
jpayne@69
|
228 * <td valign="top"><em>a Unicode property set pattern</em></td>
|
|
jpayne@69
|
229 * </tr>
|
|
jpayne@69
|
230 * </table>
|
|
jpayne@69
|
231 * <br>
|
|
jpayne@69
|
232 * <table border="1">
|
|
jpayne@69
|
233 * <tr>
|
|
jpayne@69
|
234 * <td>Legend: <table>
|
|
jpayne@69
|
235 * <tr>
|
|
jpayne@69
|
236 * <td nowrap valign="top"><code>a := b</code></td>
|
|
jpayne@69
|
237 * <td width="20" valign="top"> </td>
|
|
jpayne@69
|
238 * <td valign="top"><code>a</code> may be replaced by <code>b</code> </td>
|
|
jpayne@69
|
239 * </tr>
|
|
jpayne@69
|
240 * <tr>
|
|
jpayne@69
|
241 * <td nowrap valign="top"><code>a?</code></td>
|
|
jpayne@69
|
242 * <td valign="top"></td>
|
|
jpayne@69
|
243 * <td valign="top">zero or one instance of <code>a</code><br>
|
|
jpayne@69
|
244 * </td>
|
|
jpayne@69
|
245 * </tr>
|
|
jpayne@69
|
246 * <tr>
|
|
jpayne@69
|
247 * <td nowrap valign="top"><code>a*</code></td>
|
|
jpayne@69
|
248 * <td valign="top"></td>
|
|
jpayne@69
|
249 * <td valign="top">one or more instances of <code>a</code><br>
|
|
jpayne@69
|
250 * </td>
|
|
jpayne@69
|
251 * </tr>
|
|
jpayne@69
|
252 * <tr>
|
|
jpayne@69
|
253 * <td nowrap valign="top"><code>a | b</code></td>
|
|
jpayne@69
|
254 * <td valign="top"></td>
|
|
jpayne@69
|
255 * <td valign="top">either <code>a</code> or <code>b</code><br>
|
|
jpayne@69
|
256 * </td>
|
|
jpayne@69
|
257 * </tr>
|
|
jpayne@69
|
258 * <tr>
|
|
jpayne@69
|
259 * <td nowrap valign="top"><code>'a'</code></td>
|
|
jpayne@69
|
260 * <td valign="top"></td>
|
|
jpayne@69
|
261 * <td valign="top">the literal string between the quotes </td>
|
|
jpayne@69
|
262 * </tr>
|
|
jpayne@69
|
263 * </table>
|
|
jpayne@69
|
264 * </td>
|
|
jpayne@69
|
265 * </tr>
|
|
jpayne@69
|
266 * </table>
|
|
jpayne@69
|
267 * \htmlonly</blockquote>\endhtmlonly
|
|
jpayne@69
|
268 *
|
|
jpayne@69
|
269 * <p>Note:
|
|
jpayne@69
|
270 * - Most UnicodeSet methods do not take a UErrorCode parameter because
|
|
jpayne@69
|
271 * there are usually very few opportunities for failure other than a shortage
|
|
jpayne@69
|
272 * of memory, error codes in low-level C++ string methods would be inconvenient,
|
|
jpayne@69
|
273 * and the error code as the last parameter (ICU convention) would prevent
|
|
jpayne@69
|
274 * the use of default parameter values.
|
|
jpayne@69
|
275 * Instead, such methods set the UnicodeSet into a "bogus" state
|
|
jpayne@69
|
276 * (see isBogus()) if an error occurs.
|
|
jpayne@69
|
277 *
|
|
jpayne@69
|
278 * @author Alan Liu
|
|
jpayne@69
|
279 * @stable ICU 2.0
|
|
jpayne@69
|
280 */
|
|
jpayne@69
|
281 class U_COMMON_API UnicodeSet U_FINAL : public UnicodeFilter {
|
|
jpayne@69
|
282 private:
|
|
jpayne@69
|
283 /**
|
|
jpayne@69
|
284 * Enough for sets with few ranges.
|
|
jpayne@69
|
285 * For example, White_Space has 10 ranges, list length 21.
|
|
jpayne@69
|
286 */
|
|
jpayne@69
|
287 static constexpr int32_t INITIAL_CAPACITY = 25;
|
|
jpayne@69
|
288 // fFlags constant
|
|
jpayne@69
|
289 static constexpr uint8_t kIsBogus = 1; // This set is bogus (i.e. not valid)
|
|
jpayne@69
|
290
|
|
jpayne@69
|
291 UChar32* list = stackList; // MUST be terminated with HIGH
|
|
jpayne@69
|
292 int32_t capacity = INITIAL_CAPACITY; // capacity of list
|
|
jpayne@69
|
293 int32_t len = 1; // length of list used; 1 <= len <= capacity
|
|
jpayne@69
|
294 uint8_t fFlags = 0; // Bit flag (see constants above)
|
|
jpayne@69
|
295
|
|
jpayne@69
|
296 BMPSet *bmpSet = nullptr; // The set is frozen iff either bmpSet or stringSpan is not NULL.
|
|
jpayne@69
|
297 UChar32* buffer = nullptr; // internal buffer, may be NULL
|
|
jpayne@69
|
298 int32_t bufferCapacity = 0; // capacity of buffer
|
|
jpayne@69
|
299
|
|
jpayne@69
|
300 /**
|
|
jpayne@69
|
301 * The pattern representation of this set. This may not be the
|
|
jpayne@69
|
302 * most economical pattern. It is the pattern supplied to
|
|
jpayne@69
|
303 * applyPattern(), with variables substituted and whitespace
|
|
jpayne@69
|
304 * removed. For sets constructed without applyPattern(), or
|
|
jpayne@69
|
305 * modified using the non-pattern API, this string will be empty,
|
|
jpayne@69
|
306 * indicating that toPattern() must generate a pattern
|
|
jpayne@69
|
307 * representation from the inversion list.
|
|
jpayne@69
|
308 */
|
|
jpayne@69
|
309 char16_t *pat = nullptr;
|
|
jpayne@69
|
310 int32_t patLen = 0;
|
|
jpayne@69
|
311
|
|
jpayne@69
|
312 UVector* strings = nullptr; // maintained in sorted order
|
|
jpayne@69
|
313 UnicodeSetStringSpan *stringSpan = nullptr;
|
|
jpayne@69
|
314
|
|
jpayne@69
|
315 /**
|
|
jpayne@69
|
316 * Initial list array.
|
|
jpayne@69
|
317 * Avoids some heap allocations, and list is never nullptr.
|
|
jpayne@69
|
318 * Increases the object size a bit.
|
|
jpayne@69
|
319 */
|
|
jpayne@69
|
320 UChar32 stackList[INITIAL_CAPACITY];
|
|
jpayne@69
|
321
|
|
jpayne@69
|
322 public:
|
|
jpayne@69
|
323 /**
|
|
jpayne@69
|
324 * Determine if this object contains a valid set.
|
|
jpayne@69
|
325 * A bogus set has no value. It is different from an empty set.
|
|
jpayne@69
|
326 * It can be used to indicate that no set value is available.
|
|
jpayne@69
|
327 *
|
|
jpayne@69
|
328 * @return TRUE if the set is bogus/invalid, FALSE otherwise
|
|
jpayne@69
|
329 * @see setToBogus()
|
|
jpayne@69
|
330 * @stable ICU 4.0
|
|
jpayne@69
|
331 */
|
|
jpayne@69
|
332 inline UBool isBogus(void) const;
|
|
jpayne@69
|
333
|
|
jpayne@69
|
334 /**
|
|
jpayne@69
|
335 * Make this UnicodeSet object invalid.
|
|
jpayne@69
|
336 * The string will test TRUE with isBogus().
|
|
jpayne@69
|
337 *
|
|
jpayne@69
|
338 * A bogus set has no value. It is different from an empty set.
|
|
jpayne@69
|
339 * It can be used to indicate that no set value is available.
|
|
jpayne@69
|
340 *
|
|
jpayne@69
|
341 * This utility function is used throughout the UnicodeSet
|
|
jpayne@69
|
342 * implementation to indicate that a UnicodeSet operation failed,
|
|
jpayne@69
|
343 * and may be used in other functions,
|
|
jpayne@69
|
344 * especially but not exclusively when such functions do not
|
|
jpayne@69
|
345 * take a UErrorCode for simplicity.
|
|
jpayne@69
|
346 *
|
|
jpayne@69
|
347 * @see isBogus()
|
|
jpayne@69
|
348 * @stable ICU 4.0
|
|
jpayne@69
|
349 */
|
|
jpayne@69
|
350 void setToBogus();
|
|
jpayne@69
|
351
|
|
jpayne@69
|
352 public:
|
|
jpayne@69
|
353
|
|
jpayne@69
|
354 enum {
|
|
jpayne@69
|
355 /**
|
|
jpayne@69
|
356 * Minimum value that can be stored in a UnicodeSet.
|
|
jpayne@69
|
357 * @stable ICU 2.4
|
|
jpayne@69
|
358 */
|
|
jpayne@69
|
359 MIN_VALUE = 0,
|
|
jpayne@69
|
360
|
|
jpayne@69
|
361 /**
|
|
jpayne@69
|
362 * Maximum value that can be stored in a UnicodeSet.
|
|
jpayne@69
|
363 * @stable ICU 2.4
|
|
jpayne@69
|
364 */
|
|
jpayne@69
|
365 MAX_VALUE = 0x10ffff
|
|
jpayne@69
|
366 };
|
|
jpayne@69
|
367
|
|
jpayne@69
|
368 //----------------------------------------------------------------
|
|
jpayne@69
|
369 // Constructors &c
|
|
jpayne@69
|
370 //----------------------------------------------------------------
|
|
jpayne@69
|
371
|
|
jpayne@69
|
372 public:
|
|
jpayne@69
|
373
|
|
jpayne@69
|
374 /**
|
|
jpayne@69
|
375 * Constructs an empty set.
|
|
jpayne@69
|
376 * @stable ICU 2.0
|
|
jpayne@69
|
377 */
|
|
jpayne@69
|
378 UnicodeSet();
|
|
jpayne@69
|
379
|
|
jpayne@69
|
380 /**
|
|
jpayne@69
|
381 * Constructs a set containing the given range. If <code>end <
|
|
jpayne@69
|
382 * start</code> then an empty set is created.
|
|
jpayne@69
|
383 *
|
|
jpayne@69
|
384 * @param start first character, inclusive, of range
|
|
jpayne@69
|
385 * @param end last character, inclusive, of range
|
|
jpayne@69
|
386 * @stable ICU 2.4
|
|
jpayne@69
|
387 */
|
|
jpayne@69
|
388 UnicodeSet(UChar32 start, UChar32 end);
|
|
jpayne@69
|
389
|
|
jpayne@69
|
390 #ifndef U_HIDE_INTERNAL_API
|
|
jpayne@69
|
391 /**
|
|
jpayne@69
|
392 * @internal
|
|
jpayne@69
|
393 */
|
|
jpayne@69
|
394 enum ESerialization {
|
|
jpayne@69
|
395 kSerialized /* result of serialize() */
|
|
jpayne@69
|
396 };
|
|
jpayne@69
|
397
|
|
jpayne@69
|
398 /**
|
|
jpayne@69
|
399 * Constructs a set from the output of serialize().
|
|
jpayne@69
|
400 *
|
|
jpayne@69
|
401 * @param buffer the 16 bit array
|
|
jpayne@69
|
402 * @param bufferLen the original length returned from serialize()
|
|
jpayne@69
|
403 * @param serialization the value 'kSerialized'
|
|
jpayne@69
|
404 * @param status error code
|
|
jpayne@69
|
405 *
|
|
jpayne@69
|
406 * @internal
|
|
jpayne@69
|
407 */
|
|
jpayne@69
|
408 UnicodeSet(const uint16_t buffer[], int32_t bufferLen,
|
|
jpayne@69
|
409 ESerialization serialization, UErrorCode &status);
|
|
jpayne@69
|
410 #endif /* U_HIDE_INTERNAL_API */
|
|
jpayne@69
|
411
|
|
jpayne@69
|
412 /**
|
|
jpayne@69
|
413 * Constructs a set from the given pattern. See the class
|
|
jpayne@69
|
414 * description for the syntax of the pattern language.
|
|
jpayne@69
|
415 * @param pattern a string specifying what characters are in the set
|
|
jpayne@69
|
416 * @param status returns <code>U_ILLEGAL_ARGUMENT_ERROR</code> if the pattern
|
|
jpayne@69
|
417 * contains a syntax error.
|
|
jpayne@69
|
418 * @stable ICU 2.0
|
|
jpayne@69
|
419 */
|
|
jpayne@69
|
420 UnicodeSet(const UnicodeString& pattern,
|
|
jpayne@69
|
421 UErrorCode& status);
|
|
jpayne@69
|
422
|
|
jpayne@69
|
423 #ifndef U_HIDE_INTERNAL_API
|
|
jpayne@69
|
424 /**
|
|
jpayne@69
|
425 * Constructs a set from the given pattern. See the class
|
|
jpayne@69
|
426 * description for the syntax of the pattern language.
|
|
jpayne@69
|
427 * @param pattern a string specifying what characters are in the set
|
|
jpayne@69
|
428 * @param options bitmask for options to apply to the pattern.
|
|
jpayne@69
|
429 * Valid options are USET_IGNORE_SPACE and USET_CASE_INSENSITIVE.
|
|
jpayne@69
|
430 * @param symbols a symbol table mapping variable names to values
|
|
jpayne@69
|
431 * and stand-in characters to UnicodeSets; may be NULL
|
|
jpayne@69
|
432 * @param status returns <code>U_ILLEGAL_ARGUMENT_ERROR</code> if the pattern
|
|
jpayne@69
|
433 * contains a syntax error.
|
|
jpayne@69
|
434 * @internal
|
|
jpayne@69
|
435 */
|
|
jpayne@69
|
436 UnicodeSet(const UnicodeString& pattern,
|
|
jpayne@69
|
437 uint32_t options,
|
|
jpayne@69
|
438 const SymbolTable* symbols,
|
|
jpayne@69
|
439 UErrorCode& status);
|
|
jpayne@69
|
440 #endif /* U_HIDE_INTERNAL_API */
|
|
jpayne@69
|
441
|
|
jpayne@69
|
442 /**
|
|
jpayne@69
|
443 * Constructs a set from the given pattern. See the class description
|
|
jpayne@69
|
444 * for the syntax of the pattern language.
|
|
jpayne@69
|
445 * @param pattern a string specifying what characters are in the set
|
|
jpayne@69
|
446 * @param pos on input, the position in pattern at which to start parsing.
|
|
jpayne@69
|
447 * On output, the position after the last character parsed.
|
|
jpayne@69
|
448 * @param options bitmask for options to apply to the pattern.
|
|
jpayne@69
|
449 * Valid options are USET_IGNORE_SPACE and USET_CASE_INSENSITIVE.
|
|
jpayne@69
|
450 * @param symbols a symbol table mapping variable names to values
|
|
jpayne@69
|
451 * and stand-in characters to UnicodeSets; may be NULL
|
|
jpayne@69
|
452 * @param status input-output error code
|
|
jpayne@69
|
453 * @stable ICU 2.8
|
|
jpayne@69
|
454 */
|
|
jpayne@69
|
455 UnicodeSet(const UnicodeString& pattern, ParsePosition& pos,
|
|
jpayne@69
|
456 uint32_t options,
|
|
jpayne@69
|
457 const SymbolTable* symbols,
|
|
jpayne@69
|
458 UErrorCode& status);
|
|
jpayne@69
|
459
|
|
jpayne@69
|
460 /**
|
|
jpayne@69
|
461 * Constructs a set that is identical to the given UnicodeSet.
|
|
jpayne@69
|
462 * @stable ICU 2.0
|
|
jpayne@69
|
463 */
|
|
jpayne@69
|
464 UnicodeSet(const UnicodeSet& o);
|
|
jpayne@69
|
465
|
|
jpayne@69
|
466 /**
|
|
jpayne@69
|
467 * Destructs the set.
|
|
jpayne@69
|
468 * @stable ICU 2.0
|
|
jpayne@69
|
469 */
|
|
jpayne@69
|
470 virtual ~UnicodeSet();
|
|
jpayne@69
|
471
|
|
jpayne@69
|
472 /**
|
|
jpayne@69
|
473 * Assigns this object to be a copy of another.
|
|
jpayne@69
|
474 * A frozen set will not be modified.
|
|
jpayne@69
|
475 * @stable ICU 2.0
|
|
jpayne@69
|
476 */
|
|
jpayne@69
|
477 UnicodeSet& operator=(const UnicodeSet& o);
|
|
jpayne@69
|
478
|
|
jpayne@69
|
479 /**
|
|
jpayne@69
|
480 * Compares the specified object with this set for equality. Returns
|
|
jpayne@69
|
481 * <tt>true</tt> if the two sets
|
|
jpayne@69
|
482 * have the same size, and every member of the specified set is
|
|
jpayne@69
|
483 * contained in this set (or equivalently, every member of this set is
|
|
jpayne@69
|
484 * contained in the specified set).
|
|
jpayne@69
|
485 *
|
|
jpayne@69
|
486 * @param o set to be compared for equality with this set.
|
|
jpayne@69
|
487 * @return <tt>true</tt> if the specified set is equal to this set.
|
|
jpayne@69
|
488 * @stable ICU 2.0
|
|
jpayne@69
|
489 */
|
|
jpayne@69
|
490 virtual UBool operator==(const UnicodeSet& o) const;
|
|
jpayne@69
|
491
|
|
jpayne@69
|
492 /**
|
|
jpayne@69
|
493 * Compares the specified object with this set for equality. Returns
|
|
jpayne@69
|
494 * <tt>true</tt> if the specified set is not equal to this set.
|
|
jpayne@69
|
495 * @stable ICU 2.0
|
|
jpayne@69
|
496 */
|
|
jpayne@69
|
497 inline UBool operator!=(const UnicodeSet& o) const;
|
|
jpayne@69
|
498
|
|
jpayne@69
|
499 /**
|
|
jpayne@69
|
500 * Returns a copy of this object. All UnicodeFunctor objects have
|
|
jpayne@69
|
501 * to support cloning in order to allow classes using
|
|
jpayne@69
|
502 * UnicodeFunctors, such as Transliterator, to implement cloning.
|
|
jpayne@69
|
503 * If this set is frozen, then the clone will be frozen as well.
|
|
jpayne@69
|
504 * Use cloneAsThawed() for a mutable clone of a frozen set.
|
|
jpayne@69
|
505 * @see cloneAsThawed
|
|
jpayne@69
|
506 * @stable ICU 2.0
|
|
jpayne@69
|
507 */
|
|
jpayne@69
|
508 virtual UnicodeSet* clone() const;
|
|
jpayne@69
|
509
|
|
jpayne@69
|
510 /**
|
|
jpayne@69
|
511 * Returns the hash code value for this set.
|
|
jpayne@69
|
512 *
|
|
jpayne@69
|
513 * @return the hash code value for this set.
|
|
jpayne@69
|
514 * @see Object#hashCode()
|
|
jpayne@69
|
515 * @stable ICU 2.0
|
|
jpayne@69
|
516 */
|
|
jpayne@69
|
517 virtual int32_t hashCode(void) const;
|
|
jpayne@69
|
518
|
|
jpayne@69
|
519 /**
|
|
jpayne@69
|
520 * Get a UnicodeSet pointer from a USet
|
|
jpayne@69
|
521 *
|
|
jpayne@69
|
522 * @param uset a USet (the ICU plain C type for UnicodeSet)
|
|
jpayne@69
|
523 * @return the corresponding UnicodeSet pointer.
|
|
jpayne@69
|
524 *
|
|
jpayne@69
|
525 * @stable ICU 4.2
|
|
jpayne@69
|
526 */
|
|
jpayne@69
|
527 inline static UnicodeSet *fromUSet(USet *uset);
|
|
jpayne@69
|
528
|
|
jpayne@69
|
529 /**
|
|
jpayne@69
|
530 * Get a UnicodeSet pointer from a const USet
|
|
jpayne@69
|
531 *
|
|
jpayne@69
|
532 * @param uset a const USet (the ICU plain C type for UnicodeSet)
|
|
jpayne@69
|
533 * @return the corresponding UnicodeSet pointer.
|
|
jpayne@69
|
534 *
|
|
jpayne@69
|
535 * @stable ICU 4.2
|
|
jpayne@69
|
536 */
|
|
jpayne@69
|
537 inline static const UnicodeSet *fromUSet(const USet *uset);
|
|
jpayne@69
|
538
|
|
jpayne@69
|
539 /**
|
|
jpayne@69
|
540 * Produce a USet * pointer for this UnicodeSet.
|
|
jpayne@69
|
541 * USet is the plain C type for UnicodeSet
|
|
jpayne@69
|
542 *
|
|
jpayne@69
|
543 * @return a USet pointer for this UnicodeSet
|
|
jpayne@69
|
544 * @stable ICU 4.2
|
|
jpayne@69
|
545 */
|
|
jpayne@69
|
546 inline USet *toUSet();
|
|
jpayne@69
|
547
|
|
jpayne@69
|
548
|
|
jpayne@69
|
549 /**
|
|
jpayne@69
|
550 * Produce a const USet * pointer for this UnicodeSet.
|
|
jpayne@69
|
551 * USet is the plain C type for UnicodeSet
|
|
jpayne@69
|
552 *
|
|
jpayne@69
|
553 * @return a const USet pointer for this UnicodeSet
|
|
jpayne@69
|
554 * @stable ICU 4.2
|
|
jpayne@69
|
555 */
|
|
jpayne@69
|
556 inline const USet * toUSet() const;
|
|
jpayne@69
|
557
|
|
jpayne@69
|
558
|
|
jpayne@69
|
559 //----------------------------------------------------------------
|
|
jpayne@69
|
560 // Freezable API
|
|
jpayne@69
|
561 //----------------------------------------------------------------
|
|
jpayne@69
|
562
|
|
jpayne@69
|
563 /**
|
|
jpayne@69
|
564 * Determines whether the set has been frozen (made immutable) or not.
|
|
jpayne@69
|
565 * See the ICU4J Freezable interface for details.
|
|
jpayne@69
|
566 * @return TRUE/FALSE for whether the set has been frozen
|
|
jpayne@69
|
567 * @see freeze
|
|
jpayne@69
|
568 * @see cloneAsThawed
|
|
jpayne@69
|
569 * @stable ICU 3.8
|
|
jpayne@69
|
570 */
|
|
jpayne@69
|
571 inline UBool isFrozen() const;
|
|
jpayne@69
|
572
|
|
jpayne@69
|
573 /**
|
|
jpayne@69
|
574 * Freeze the set (make it immutable).
|
|
jpayne@69
|
575 * Once frozen, it cannot be unfrozen and is therefore thread-safe
|
|
jpayne@69
|
576 * until it is deleted.
|
|
jpayne@69
|
577 * See the ICU4J Freezable interface for details.
|
|
jpayne@69
|
578 * Freezing the set may also make some operations faster, for example
|
|
jpayne@69
|
579 * contains() and span().
|
|
jpayne@69
|
580 * A frozen set will not be modified. (It remains frozen.)
|
|
jpayne@69
|
581 * @return this set.
|
|
jpayne@69
|
582 * @see isFrozen
|
|
jpayne@69
|
583 * @see cloneAsThawed
|
|
jpayne@69
|
584 * @stable ICU 3.8
|
|
jpayne@69
|
585 */
|
|
jpayne@69
|
586 UnicodeSet *freeze();
|
|
jpayne@69
|
587
|
|
jpayne@69
|
588 /**
|
|
jpayne@69
|
589 * Clone the set and make the clone mutable.
|
|
jpayne@69
|
590 * See the ICU4J Freezable interface for details.
|
|
jpayne@69
|
591 * @return the mutable clone
|
|
jpayne@69
|
592 * @see freeze
|
|
jpayne@69
|
593 * @see isFrozen
|
|
jpayne@69
|
594 * @stable ICU 3.8
|
|
jpayne@69
|
595 */
|
|
jpayne@69
|
596 UnicodeSet *cloneAsThawed() const;
|
|
jpayne@69
|
597
|
|
jpayne@69
|
598 //----------------------------------------------------------------
|
|
jpayne@69
|
599 // Public API
|
|
jpayne@69
|
600 //----------------------------------------------------------------
|
|
jpayne@69
|
601
|
|
jpayne@69
|
602 /**
|
|
jpayne@69
|
603 * Make this object represent the range `start - end`.
|
|
jpayne@69
|
604 * If `end > start` then this object is set to an empty range.
|
|
jpayne@69
|
605 * A frozen set will not be modified.
|
|
jpayne@69
|
606 *
|
|
jpayne@69
|
607 * @param start first character in the set, inclusive
|
|
jpayne@69
|
608 * @param end last character in the set, inclusive
|
|
jpayne@69
|
609 * @stable ICU 2.4
|
|
jpayne@69
|
610 */
|
|
jpayne@69
|
611 UnicodeSet& set(UChar32 start, UChar32 end);
|
|
jpayne@69
|
612
|
|
jpayne@69
|
613 /**
|
|
jpayne@69
|
614 * Return true if the given position, in the given pattern, appears
|
|
jpayne@69
|
615 * to be the start of a UnicodeSet pattern.
|
|
jpayne@69
|
616 * @stable ICU 2.4
|
|
jpayne@69
|
617 */
|
|
jpayne@69
|
618 static UBool resemblesPattern(const UnicodeString& pattern,
|
|
jpayne@69
|
619 int32_t pos);
|
|
jpayne@69
|
620
|
|
jpayne@69
|
621 /**
|
|
jpayne@69
|
622 * Modifies this set to represent the set specified by the given
|
|
jpayne@69
|
623 * pattern, ignoring Unicode Pattern_White_Space characters.
|
|
jpayne@69
|
624 * See the class description for the syntax of the pattern language.
|
|
jpayne@69
|
625 * A frozen set will not be modified.
|
|
jpayne@69
|
626 * @param pattern a string specifying what characters are in the set
|
|
jpayne@69
|
627 * @param status returns <code>U_ILLEGAL_ARGUMENT_ERROR</code> if the pattern
|
|
jpayne@69
|
628 * contains a syntax error.
|
|
jpayne@69
|
629 * <em> Empties the set passed before applying the pattern.</em>
|
|
jpayne@69
|
630 * @return a reference to this
|
|
jpayne@69
|
631 * @stable ICU 2.0
|
|
jpayne@69
|
632 */
|
|
jpayne@69
|
633 UnicodeSet& applyPattern(const UnicodeString& pattern,
|
|
jpayne@69
|
634 UErrorCode& status);
|
|
jpayne@69
|
635
|
|
jpayne@69
|
636 #ifndef U_HIDE_INTERNAL_API
|
|
jpayne@69
|
637 /**
|
|
jpayne@69
|
638 * Modifies this set to represent the set specified by the given
|
|
jpayne@69
|
639 * pattern, optionally ignoring Unicode Pattern_White_Space characters.
|
|
jpayne@69
|
640 * See the class description for the syntax of the pattern language.
|
|
jpayne@69
|
641 * A frozen set will not be modified.
|
|
jpayne@69
|
642 * @param pattern a string specifying what characters are in the set
|
|
jpayne@69
|
643 * @param options bitmask for options to apply to the pattern.
|
|
jpayne@69
|
644 * Valid options are USET_IGNORE_SPACE and USET_CASE_INSENSITIVE.
|
|
jpayne@69
|
645 * @param symbols a symbol table mapping variable names to
|
|
jpayne@69
|
646 * values and stand-ins to UnicodeSets; may be NULL
|
|
jpayne@69
|
647 * @param status returns <code>U_ILLEGAL_ARGUMENT_ERROR</code> if the pattern
|
|
jpayne@69
|
648 * contains a syntax error.
|
|
jpayne@69
|
649 *<em> Empties the set passed before applying the pattern.</em>
|
|
jpayne@69
|
650 * @return a reference to this
|
|
jpayne@69
|
651 * @internal
|
|
jpayne@69
|
652 */
|
|
jpayne@69
|
653 UnicodeSet& applyPattern(const UnicodeString& pattern,
|
|
jpayne@69
|
654 uint32_t options,
|
|
jpayne@69
|
655 const SymbolTable* symbols,
|
|
jpayne@69
|
656 UErrorCode& status);
|
|
jpayne@69
|
657 #endif /* U_HIDE_INTERNAL_API */
|
|
jpayne@69
|
658
|
|
jpayne@69
|
659 /**
|
|
jpayne@69
|
660 * Parses the given pattern, starting at the given position. The
|
|
jpayne@69
|
661 * character at pattern.charAt(pos.getIndex()) must be '[', or the
|
|
jpayne@69
|
662 * parse fails. Parsing continues until the corresponding closing
|
|
jpayne@69
|
663 * ']'. If a syntax error is encountered between the opening and
|
|
jpayne@69
|
664 * closing brace, the parse fails. Upon return from a successful
|
|
jpayne@69
|
665 * parse, the ParsePosition is updated to point to the character
|
|
jpayne@69
|
666 * following the closing ']', and a StringBuffer containing a
|
|
jpayne@69
|
667 * pairs list for the parsed pattern is returned. This method calls
|
|
jpayne@69
|
668 * itself recursively to parse embedded subpatterns.
|
|
jpayne@69
|
669 *<em> Empties the set passed before applying the pattern.</em>
|
|
jpayne@69
|
670 * A frozen set will not be modified.
|
|
jpayne@69
|
671 *
|
|
jpayne@69
|
672 * @param pattern the string containing the pattern to be parsed.
|
|
jpayne@69
|
673 * The portion of the string from pos.getIndex(), which must be a
|
|
jpayne@69
|
674 * '[', to the corresponding closing ']', is parsed.
|
|
jpayne@69
|
675 * @param pos upon entry, the position at which to being parsing.
|
|
jpayne@69
|
676 * The character at pattern.charAt(pos.getIndex()) must be a '['.
|
|
jpayne@69
|
677 * Upon return from a successful parse, pos.getIndex() is either
|
|
jpayne@69
|
678 * the character after the closing ']' of the parsed pattern, or
|
|
jpayne@69
|
679 * pattern.length() if the closing ']' is the last character of
|
|
jpayne@69
|
680 * the pattern string.
|
|
jpayne@69
|
681 * @param options bitmask for options to apply to the pattern.
|
|
jpayne@69
|
682 * Valid options are USET_IGNORE_SPACE and USET_CASE_INSENSITIVE.
|
|
jpayne@69
|
683 * @param symbols a symbol table mapping variable names to
|
|
jpayne@69
|
684 * values and stand-ins to UnicodeSets; may be NULL
|
|
jpayne@69
|
685 * @param status returns <code>U_ILLEGAL_ARGUMENT_ERROR</code> if the pattern
|
|
jpayne@69
|
686 * contains a syntax error.
|
|
jpayne@69
|
687 * @return a reference to this
|
|
jpayne@69
|
688 * @stable ICU 2.8
|
|
jpayne@69
|
689 */
|
|
jpayne@69
|
690 UnicodeSet& applyPattern(const UnicodeString& pattern,
|
|
jpayne@69
|
691 ParsePosition& pos,
|
|
jpayne@69
|
692 uint32_t options,
|
|
jpayne@69
|
693 const SymbolTable* symbols,
|
|
jpayne@69
|
694 UErrorCode& status);
|
|
jpayne@69
|
695
|
|
jpayne@69
|
696 /**
|
|
jpayne@69
|
697 * Returns a string representation of this set. If the result of
|
|
jpayne@69
|
698 * calling this function is passed to a UnicodeSet constructor, it
|
|
jpayne@69
|
699 * will produce another set that is equal to this one.
|
|
jpayne@69
|
700 * A frozen set will not be modified.
|
|
jpayne@69
|
701 * @param result the string to receive the rules. Previous
|
|
jpayne@69
|
702 * contents will be deleted.
|
|
jpayne@69
|
703 * @param escapeUnprintable if TRUE then convert unprintable
|
|
jpayne@69
|
704 * character to their hex escape representations, \\uxxxx or
|
|
jpayne@69
|
705 * \\Uxxxxxxxx. Unprintable characters are those other than
|
|
jpayne@69
|
706 * U+000A, U+0020..U+007E.
|
|
jpayne@69
|
707 * @stable ICU 2.0
|
|
jpayne@69
|
708 */
|
|
jpayne@69
|
709 virtual UnicodeString& toPattern(UnicodeString& result,
|
|
jpayne@69
|
710 UBool escapeUnprintable = FALSE) const;
|
|
jpayne@69
|
711
|
|
jpayne@69
|
712 /**
|
|
jpayne@69
|
713 * Modifies this set to contain those code points which have the given value
|
|
jpayne@69
|
714 * for the given binary or enumerated property, as returned by
|
|
jpayne@69
|
715 * u_getIntPropertyValue. Prior contents of this set are lost.
|
|
jpayne@69
|
716 * A frozen set will not be modified.
|
|
jpayne@69
|
717 *
|
|
jpayne@69
|
718 * @param prop a property in the range UCHAR_BIN_START..UCHAR_BIN_LIMIT-1
|
|
jpayne@69
|
719 * or UCHAR_INT_START..UCHAR_INT_LIMIT-1
|
|
jpayne@69
|
720 * or UCHAR_MASK_START..UCHAR_MASK_LIMIT-1.
|
|
jpayne@69
|
721 *
|
|
jpayne@69
|
722 * @param value a value in the range u_getIntPropertyMinValue(prop)..
|
|
jpayne@69
|
723 * u_getIntPropertyMaxValue(prop), with one exception. If prop is
|
|
jpayne@69
|
724 * UCHAR_GENERAL_CATEGORY_MASK, then value should not be a UCharCategory, but
|
|
jpayne@69
|
725 * rather a mask value produced by U_GET_GC_MASK(). This allows grouped
|
|
jpayne@69
|
726 * categories such as [:L:] to be represented.
|
|
jpayne@69
|
727 *
|
|
jpayne@69
|
728 * @param ec error code input/output parameter
|
|
jpayne@69
|
729 *
|
|
jpayne@69
|
730 * @return a reference to this set
|
|
jpayne@69
|
731 *
|
|
jpayne@69
|
732 * @stable ICU 2.4
|
|
jpayne@69
|
733 */
|
|
jpayne@69
|
734 UnicodeSet& applyIntPropertyValue(UProperty prop,
|
|
jpayne@69
|
735 int32_t value,
|
|
jpayne@69
|
736 UErrorCode& ec);
|
|
jpayne@69
|
737
|
|
jpayne@69
|
738 /**
|
|
jpayne@69
|
739 * Modifies this set to contain those code points which have the
|
|
jpayne@69
|
740 * given value for the given property. Prior contents of this
|
|
jpayne@69
|
741 * set are lost.
|
|
jpayne@69
|
742 * A frozen set will not be modified.
|
|
jpayne@69
|
743 *
|
|
jpayne@69
|
744 * @param prop a property alias, either short or long. The name is matched
|
|
jpayne@69
|
745 * loosely. See PropertyAliases.txt for names and a description of loose
|
|
jpayne@69
|
746 * matching. If the value string is empty, then this string is interpreted
|
|
jpayne@69
|
747 * as either a General_Category value alias, a Script value alias, a binary
|
|
jpayne@69
|
748 * property alias, or a special ID. Special IDs are matched loosely and
|
|
jpayne@69
|
749 * correspond to the following sets:
|
|
jpayne@69
|
750 *
|
|
jpayne@69
|
751 * "ANY" = [\\u0000-\\U0010FFFF],
|
|
jpayne@69
|
752 * "ASCII" = [\\u0000-\\u007F],
|
|
jpayne@69
|
753 * "Assigned" = [:^Cn:].
|
|
jpayne@69
|
754 *
|
|
jpayne@69
|
755 * @param value a value alias, either short or long. The name is matched
|
|
jpayne@69
|
756 * loosely. See PropertyValueAliases.txt for names and a description of
|
|
jpayne@69
|
757 * loose matching. In addition to aliases listed, numeric values and
|
|
jpayne@69
|
758 * canonical combining classes may be expressed numerically, e.g., ("nv",
|
|
jpayne@69
|
759 * "0.5") or ("ccc", "220"). The value string may also be empty.
|
|
jpayne@69
|
760 *
|
|
jpayne@69
|
761 * @param ec error code input/output parameter
|
|
jpayne@69
|
762 *
|
|
jpayne@69
|
763 * @return a reference to this set
|
|
jpayne@69
|
764 *
|
|
jpayne@69
|
765 * @stable ICU 2.4
|
|
jpayne@69
|
766 */
|
|
jpayne@69
|
767 UnicodeSet& applyPropertyAlias(const UnicodeString& prop,
|
|
jpayne@69
|
768 const UnicodeString& value,
|
|
jpayne@69
|
769 UErrorCode& ec);
|
|
jpayne@69
|
770
|
|
jpayne@69
|
771 /**
|
|
jpayne@69
|
772 * Returns the number of elements in this set (its cardinality).
|
|
jpayne@69
|
773 * Note than the elements of a set may include both individual
|
|
jpayne@69
|
774 * codepoints and strings.
|
|
jpayne@69
|
775 *
|
|
jpayne@69
|
776 * @return the number of elements in this set (its cardinality).
|
|
jpayne@69
|
777 * @stable ICU 2.0
|
|
jpayne@69
|
778 */
|
|
jpayne@69
|
779 virtual int32_t size(void) const;
|
|
jpayne@69
|
780
|
|
jpayne@69
|
781 /**
|
|
jpayne@69
|
782 * Returns <tt>true</tt> if this set contains no elements.
|
|
jpayne@69
|
783 *
|
|
jpayne@69
|
784 * @return <tt>true</tt> if this set contains no elements.
|
|
jpayne@69
|
785 * @stable ICU 2.0
|
|
jpayne@69
|
786 */
|
|
jpayne@69
|
787 virtual UBool isEmpty(void) const;
|
|
jpayne@69
|
788
|
|
jpayne@69
|
789 /**
|
|
jpayne@69
|
790 * Returns true if this set contains the given character.
|
|
jpayne@69
|
791 * This function works faster with a frozen set.
|
|
jpayne@69
|
792 * @param c character to be checked for containment
|
|
jpayne@69
|
793 * @return true if the test condition is met
|
|
jpayne@69
|
794 * @stable ICU 2.0
|
|
jpayne@69
|
795 */
|
|
jpayne@69
|
796 virtual UBool contains(UChar32 c) const;
|
|
jpayne@69
|
797
|
|
jpayne@69
|
798 /**
|
|
jpayne@69
|
799 * Returns true if this set contains every character
|
|
jpayne@69
|
800 * of the given range.
|
|
jpayne@69
|
801 * @param start first character, inclusive, of the range
|
|
jpayne@69
|
802 * @param end last character, inclusive, of the range
|
|
jpayne@69
|
803 * @return true if the test condition is met
|
|
jpayne@69
|
804 * @stable ICU 2.0
|
|
jpayne@69
|
805 */
|
|
jpayne@69
|
806 virtual UBool contains(UChar32 start, UChar32 end) const;
|
|
jpayne@69
|
807
|
|
jpayne@69
|
808 /**
|
|
jpayne@69
|
809 * Returns <tt>true</tt> if this set contains the given
|
|
jpayne@69
|
810 * multicharacter string.
|
|
jpayne@69
|
811 * @param s string to be checked for containment
|
|
jpayne@69
|
812 * @return <tt>true</tt> if this set contains the specified string
|
|
jpayne@69
|
813 * @stable ICU 2.4
|
|
jpayne@69
|
814 */
|
|
jpayne@69
|
815 UBool contains(const UnicodeString& s) const;
|
|
jpayne@69
|
816
|
|
jpayne@69
|
817 /**
|
|
jpayne@69
|
818 * Returns true if this set contains all the characters and strings
|
|
jpayne@69
|
819 * of the given set.
|
|
jpayne@69
|
820 * @param c set to be checked for containment
|
|
jpayne@69
|
821 * @return true if the test condition is met
|
|
jpayne@69
|
822 * @stable ICU 2.4
|
|
jpayne@69
|
823 */
|
|
jpayne@69
|
824 virtual UBool containsAll(const UnicodeSet& c) const;
|
|
jpayne@69
|
825
|
|
jpayne@69
|
826 /**
|
|
jpayne@69
|
827 * Returns true if this set contains all the characters
|
|
jpayne@69
|
828 * of the given string.
|
|
jpayne@69
|
829 * @param s string containing characters to be checked for containment
|
|
jpayne@69
|
830 * @return true if the test condition is met
|
|
jpayne@69
|
831 * @stable ICU 2.4
|
|
jpayne@69
|
832 */
|
|
jpayne@69
|
833 UBool containsAll(const UnicodeString& s) const;
|
|
jpayne@69
|
834
|
|
jpayne@69
|
835 /**
|
|
jpayne@69
|
836 * Returns true if this set contains none of the characters
|
|
jpayne@69
|
837 * of the given range.
|
|
jpayne@69
|
838 * @param start first character, inclusive, of the range
|
|
jpayne@69
|
839 * @param end last character, inclusive, of the range
|
|
jpayne@69
|
840 * @return true if the test condition is met
|
|
jpayne@69
|
841 * @stable ICU 2.4
|
|
jpayne@69
|
842 */
|
|
jpayne@69
|
843 UBool containsNone(UChar32 start, UChar32 end) const;
|
|
jpayne@69
|
844
|
|
jpayne@69
|
845 /**
|
|
jpayne@69
|
846 * Returns true if this set contains none of the characters and strings
|
|
jpayne@69
|
847 * of the given set.
|
|
jpayne@69
|
848 * @param c set to be checked for containment
|
|
jpayne@69
|
849 * @return true if the test condition is met
|
|
jpayne@69
|
850 * @stable ICU 2.4
|
|
jpayne@69
|
851 */
|
|
jpayne@69
|
852 UBool containsNone(const UnicodeSet& c) const;
|
|
jpayne@69
|
853
|
|
jpayne@69
|
854 /**
|
|
jpayne@69
|
855 * Returns true if this set contains none of the characters
|
|
jpayne@69
|
856 * of the given string.
|
|
jpayne@69
|
857 * @param s string containing characters to be checked for containment
|
|
jpayne@69
|
858 * @return true if the test condition is met
|
|
jpayne@69
|
859 * @stable ICU 2.4
|
|
jpayne@69
|
860 */
|
|
jpayne@69
|
861 UBool containsNone(const UnicodeString& s) const;
|
|
jpayne@69
|
862
|
|
jpayne@69
|
863 /**
|
|
jpayne@69
|
864 * Returns true if this set contains one or more of the characters
|
|
jpayne@69
|
865 * in the given range.
|
|
jpayne@69
|
866 * @param start first character, inclusive, of the range
|
|
jpayne@69
|
867 * @param end last character, inclusive, of the range
|
|
jpayne@69
|
868 * @return true if the condition is met
|
|
jpayne@69
|
869 * @stable ICU 2.4
|
|
jpayne@69
|
870 */
|
|
jpayne@69
|
871 inline UBool containsSome(UChar32 start, UChar32 end) const;
|
|
jpayne@69
|
872
|
|
jpayne@69
|
873 /**
|
|
jpayne@69
|
874 * Returns true if this set contains one or more of the characters
|
|
jpayne@69
|
875 * and strings of the given set.
|
|
jpayne@69
|
876 * @param s The set to be checked for containment
|
|
jpayne@69
|
877 * @return true if the condition is met
|
|
jpayne@69
|
878 * @stable ICU 2.4
|
|
jpayne@69
|
879 */
|
|
jpayne@69
|
880 inline UBool containsSome(const UnicodeSet& s) const;
|
|
jpayne@69
|
881
|
|
jpayne@69
|
882 /**
|
|
jpayne@69
|
883 * Returns true if this set contains one or more of the characters
|
|
jpayne@69
|
884 * of the given string.
|
|
jpayne@69
|
885 * @param s string containing characters to be checked for containment
|
|
jpayne@69
|
886 * @return true if the condition is met
|
|
jpayne@69
|
887 * @stable ICU 2.4
|
|
jpayne@69
|
888 */
|
|
jpayne@69
|
889 inline UBool containsSome(const UnicodeString& s) const;
|
|
jpayne@69
|
890
|
|
jpayne@69
|
891 /**
|
|
jpayne@69
|
892 * Returns the length of the initial substring of the input string which
|
|
jpayne@69
|
893 * consists only of characters and strings that are contained in this set
|
|
jpayne@69
|
894 * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE),
|
|
jpayne@69
|
895 * or only of characters and strings that are not contained
|
|
jpayne@69
|
896 * in this set (USET_SPAN_NOT_CONTAINED).
|
|
jpayne@69
|
897 * See USetSpanCondition for details.
|
|
jpayne@69
|
898 * Similar to the strspn() C library function.
|
|
jpayne@69
|
899 * Unpaired surrogates are treated according to contains() of their surrogate code points.
|
|
jpayne@69
|
900 * This function works faster with a frozen set and with a non-negative string length argument.
|
|
jpayne@69
|
901 * @param s start of the string
|
|
jpayne@69
|
902 * @param length of the string; can be -1 for NUL-terminated
|
|
jpayne@69
|
903 * @param spanCondition specifies the containment condition
|
|
jpayne@69
|
904 * @return the length of the initial substring according to the spanCondition;
|
|
jpayne@69
|
905 * 0 if the start of the string does not fit the spanCondition
|
|
jpayne@69
|
906 * @stable ICU 3.8
|
|
jpayne@69
|
907 * @see USetSpanCondition
|
|
jpayne@69
|
908 */
|
|
jpayne@69
|
909 int32_t span(const char16_t *s, int32_t length, USetSpanCondition spanCondition) const;
|
|
jpayne@69
|
910
|
|
jpayne@69
|
911 /**
|
|
jpayne@69
|
912 * Returns the end of the substring of the input string according to the USetSpanCondition.
|
|
jpayne@69
|
913 * Same as <code>start+span(s.getBuffer()+start, s.length()-start, spanCondition)</code>
|
|
jpayne@69
|
914 * after pinning start to 0<=start<=s.length().
|
|
jpayne@69
|
915 * @param s the string
|
|
jpayne@69
|
916 * @param start the start index in the string for the span operation
|
|
jpayne@69
|
917 * @param spanCondition specifies the containment condition
|
|
jpayne@69
|
918 * @return the exclusive end of the substring according to the spanCondition;
|
|
jpayne@69
|
919 * the substring s.tempSubStringBetween(start, end) fulfills the spanCondition
|
|
jpayne@69
|
920 * @stable ICU 4.4
|
|
jpayne@69
|
921 * @see USetSpanCondition
|
|
jpayne@69
|
922 */
|
|
jpayne@69
|
923 inline int32_t span(const UnicodeString &s, int32_t start, USetSpanCondition spanCondition) const;
|
|
jpayne@69
|
924
|
|
jpayne@69
|
925 /**
|
|
jpayne@69
|
926 * Returns the start of the trailing substring of the input string which
|
|
jpayne@69
|
927 * consists only of characters and strings that are contained in this set
|
|
jpayne@69
|
928 * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE),
|
|
jpayne@69
|
929 * or only of characters and strings that are not contained
|
|
jpayne@69
|
930 * in this set (USET_SPAN_NOT_CONTAINED).
|
|
jpayne@69
|
931 * See USetSpanCondition for details.
|
|
jpayne@69
|
932 * Unpaired surrogates are treated according to contains() of their surrogate code points.
|
|
jpayne@69
|
933 * This function works faster with a frozen set and with a non-negative string length argument.
|
|
jpayne@69
|
934 * @param s start of the string
|
|
jpayne@69
|
935 * @param length of the string; can be -1 for NUL-terminated
|
|
jpayne@69
|
936 * @param spanCondition specifies the containment condition
|
|
jpayne@69
|
937 * @return the start of the trailing substring according to the spanCondition;
|
|
jpayne@69
|
938 * the string length if the end of the string does not fit the spanCondition
|
|
jpayne@69
|
939 * @stable ICU 3.8
|
|
jpayne@69
|
940 * @see USetSpanCondition
|
|
jpayne@69
|
941 */
|
|
jpayne@69
|
942 int32_t spanBack(const char16_t *s, int32_t length, USetSpanCondition spanCondition) const;
|
|
jpayne@69
|
943
|
|
jpayne@69
|
944 /**
|
|
jpayne@69
|
945 * Returns the start of the substring of the input string according to the USetSpanCondition.
|
|
jpayne@69
|
946 * Same as <code>spanBack(s.getBuffer(), limit, spanCondition)</code>
|
|
jpayne@69
|
947 * after pinning limit to 0<=end<=s.length().
|
|
jpayne@69
|
948 * @param s the string
|
|
jpayne@69
|
949 * @param limit the exclusive-end index in the string for the span operation
|
|
jpayne@69
|
950 * (use s.length() or INT32_MAX for spanning back from the end of the string)
|
|
jpayne@69
|
951 * @param spanCondition specifies the containment condition
|
|
jpayne@69
|
952 * @return the start of the substring according to the spanCondition;
|
|
jpayne@69
|
953 * the substring s.tempSubStringBetween(start, limit) fulfills the spanCondition
|
|
jpayne@69
|
954 * @stable ICU 4.4
|
|
jpayne@69
|
955 * @see USetSpanCondition
|
|
jpayne@69
|
956 */
|
|
jpayne@69
|
957 inline int32_t spanBack(const UnicodeString &s, int32_t limit, USetSpanCondition spanCondition) const;
|
|
jpayne@69
|
958
|
|
jpayne@69
|
959 /**
|
|
jpayne@69
|
960 * Returns the length of the initial substring of the input string which
|
|
jpayne@69
|
961 * consists only of characters and strings that are contained in this set
|
|
jpayne@69
|
962 * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE),
|
|
jpayne@69
|
963 * or only of characters and strings that are not contained
|
|
jpayne@69
|
964 * in this set (USET_SPAN_NOT_CONTAINED).
|
|
jpayne@69
|
965 * See USetSpanCondition for details.
|
|
jpayne@69
|
966 * Similar to the strspn() C library function.
|
|
jpayne@69
|
967 * Malformed byte sequences are treated according to contains(0xfffd).
|
|
jpayne@69
|
968 * This function works faster with a frozen set and with a non-negative string length argument.
|
|
jpayne@69
|
969 * @param s start of the string (UTF-8)
|
|
jpayne@69
|
970 * @param length of the string; can be -1 for NUL-terminated
|
|
jpayne@69
|
971 * @param spanCondition specifies the containment condition
|
|
jpayne@69
|
972 * @return the length of the initial substring according to the spanCondition;
|
|
jpayne@69
|
973 * 0 if the start of the string does not fit the spanCondition
|
|
jpayne@69
|
974 * @stable ICU 3.8
|
|
jpayne@69
|
975 * @see USetSpanCondition
|
|
jpayne@69
|
976 */
|
|
jpayne@69
|
977 int32_t spanUTF8(const char *s, int32_t length, USetSpanCondition spanCondition) const;
|
|
jpayne@69
|
978
|
|
jpayne@69
|
979 /**
|
|
jpayne@69
|
980 * Returns the start of the trailing substring of the input string which
|
|
jpayne@69
|
981 * consists only of characters and strings that are contained in this set
|
|
jpayne@69
|
982 * (USET_SPAN_CONTAINED, USET_SPAN_SIMPLE),
|
|
jpayne@69
|
983 * or only of characters and strings that are not contained
|
|
jpayne@69
|
984 * in this set (USET_SPAN_NOT_CONTAINED).
|
|
jpayne@69
|
985 * See USetSpanCondition for details.
|
|
jpayne@69
|
986 * Malformed byte sequences are treated according to contains(0xfffd).
|
|
jpayne@69
|
987 * This function works faster with a frozen set and with a non-negative string length argument.
|
|
jpayne@69
|
988 * @param s start of the string (UTF-8)
|
|
jpayne@69
|
989 * @param length of the string; can be -1 for NUL-terminated
|
|
jpayne@69
|
990 * @param spanCondition specifies the containment condition
|
|
jpayne@69
|
991 * @return the start of the trailing substring according to the spanCondition;
|
|
jpayne@69
|
992 * the string length if the end of the string does not fit the spanCondition
|
|
jpayne@69
|
993 * @stable ICU 3.8
|
|
jpayne@69
|
994 * @see USetSpanCondition
|
|
jpayne@69
|
995 */
|
|
jpayne@69
|
996 int32_t spanBackUTF8(const char *s, int32_t length, USetSpanCondition spanCondition) const;
|
|
jpayne@69
|
997
|
|
jpayne@69
|
998 /**
|
|
jpayne@69
|
999 * Implement UnicodeMatcher::matches()
|
|
jpayne@69
|
1000 * @stable ICU 2.4
|
|
jpayne@69
|
1001 */
|
|
jpayne@69
|
1002 virtual UMatchDegree matches(const Replaceable& text,
|
|
jpayne@69
|
1003 int32_t& offset,
|
|
jpayne@69
|
1004 int32_t limit,
|
|
jpayne@69
|
1005 UBool incremental);
|
|
jpayne@69
|
1006
|
|
jpayne@69
|
1007 private:
|
|
jpayne@69
|
1008 /**
|
|
jpayne@69
|
1009 * Returns the longest match for s in text at the given position.
|
|
jpayne@69
|
1010 * If limit > start then match forward from start+1 to limit
|
|
jpayne@69
|
1011 * matching all characters except s.charAt(0). If limit < start,
|
|
jpayne@69
|
1012 * go backward starting from start-1 matching all characters
|
|
jpayne@69
|
1013 * except s.charAt(s.length()-1). This method assumes that the
|
|
jpayne@69
|
1014 * first character, text.charAt(start), matches s, so it does not
|
|
jpayne@69
|
1015 * check it.
|
|
jpayne@69
|
1016 * @param text the text to match
|
|
jpayne@69
|
1017 * @param start the first character to match. In the forward
|
|
jpayne@69
|
1018 * direction, text.charAt(start) is matched against s.charAt(0).
|
|
jpayne@69
|
1019 * In the reverse direction, it is matched against
|
|
jpayne@69
|
1020 * s.charAt(s.length()-1).
|
|
jpayne@69
|
1021 * @param limit the limit offset for matching, either last+1 in
|
|
jpayne@69
|
1022 * the forward direction, or last-1 in the reverse direction,
|
|
jpayne@69
|
1023 * where last is the index of the last character to match.
|
|
jpayne@69
|
1024 * @param s
|
|
jpayne@69
|
1025 * @return If part of s matches up to the limit, return |limit -
|
|
jpayne@69
|
1026 * start|. If all of s matches before reaching the limit, return
|
|
jpayne@69
|
1027 * s.length(). If there is a mismatch between s and text, return
|
|
jpayne@69
|
1028 * 0
|
|
jpayne@69
|
1029 */
|
|
jpayne@69
|
1030 static int32_t matchRest(const Replaceable& text,
|
|
jpayne@69
|
1031 int32_t start, int32_t limit,
|
|
jpayne@69
|
1032 const UnicodeString& s);
|
|
jpayne@69
|
1033
|
|
jpayne@69
|
1034 /**
|
|
jpayne@69
|
1035 * Returns the smallest value i such that c < list[i]. Caller
|
|
jpayne@69
|
1036 * must ensure that c is a legal value or this method will enter
|
|
jpayne@69
|
1037 * an infinite loop. This method performs a binary search.
|
|
jpayne@69
|
1038 * @param c a character in the range MIN_VALUE..MAX_VALUE
|
|
jpayne@69
|
1039 * inclusive
|
|
jpayne@69
|
1040 * @return the smallest integer i in the range 0..len-1,
|
|
jpayne@69
|
1041 * inclusive, such that c < list[i]
|
|
jpayne@69
|
1042 */
|
|
jpayne@69
|
1043 int32_t findCodePoint(UChar32 c) const;
|
|
jpayne@69
|
1044
|
|
jpayne@69
|
1045 public:
|
|
jpayne@69
|
1046
|
|
jpayne@69
|
1047 /**
|
|
jpayne@69
|
1048 * Implementation of UnicodeMatcher API. Union the set of all
|
|
jpayne@69
|
1049 * characters that may be matched by this object into the given
|
|
jpayne@69
|
1050 * set.
|
|
jpayne@69
|
1051 * @param toUnionTo the set into which to union the source characters
|
|
jpayne@69
|
1052 * @stable ICU 2.4
|
|
jpayne@69
|
1053 */
|
|
jpayne@69
|
1054 virtual void addMatchSetTo(UnicodeSet& toUnionTo) const;
|
|
jpayne@69
|
1055
|
|
jpayne@69
|
1056 /**
|
|
jpayne@69
|
1057 * Returns the index of the given character within this set, where
|
|
jpayne@69
|
1058 * the set is ordered by ascending code point. If the character
|
|
jpayne@69
|
1059 * is not in this set, return -1. The inverse of this method is
|
|
jpayne@69
|
1060 * <code>charAt()</code>.
|
|
jpayne@69
|
1061 * @return an index from 0..size()-1, or -1
|
|
jpayne@69
|
1062 * @stable ICU 2.4
|
|
jpayne@69
|
1063 */
|
|
jpayne@69
|
1064 int32_t indexOf(UChar32 c) const;
|
|
jpayne@69
|
1065
|
|
jpayne@69
|
1066 /**
|
|
jpayne@69
|
1067 * Returns the character at the given index within this set, where
|
|
jpayne@69
|
1068 * the set is ordered by ascending code point. If the index is
|
|
jpayne@69
|
1069 * out of range, return (UChar32)-1. The inverse of this method is
|
|
jpayne@69
|
1070 * <code>indexOf()</code>.
|
|
jpayne@69
|
1071 * @param index an index from 0..size()-1
|
|
jpayne@69
|
1072 * @return the character at the given index, or (UChar32)-1.
|
|
jpayne@69
|
1073 * @stable ICU 2.4
|
|
jpayne@69
|
1074 */
|
|
jpayne@69
|
1075 UChar32 charAt(int32_t index) const;
|
|
jpayne@69
|
1076
|
|
jpayne@69
|
1077 /**
|
|
jpayne@69
|
1078 * Adds the specified range to this set if it is not already
|
|
jpayne@69
|
1079 * present. If this set already contains the specified range,
|
|
jpayne@69
|
1080 * the call leaves this set unchanged. If <code>end > start</code>
|
|
jpayne@69
|
1081 * then an empty range is added, leaving the set unchanged.
|
|
jpayne@69
|
1082 * This is equivalent to a boolean logic OR, or a set UNION.
|
|
jpayne@69
|
1083 * A frozen set will not be modified.
|
|
jpayne@69
|
1084 *
|
|
jpayne@69
|
1085 * @param start first character, inclusive, of range to be added
|
|
jpayne@69
|
1086 * to this set.
|
|
jpayne@69
|
1087 * @param end last character, inclusive, of range to be added
|
|
jpayne@69
|
1088 * to this set.
|
|
jpayne@69
|
1089 * @stable ICU 2.0
|
|
jpayne@69
|
1090 */
|
|
jpayne@69
|
1091 virtual UnicodeSet& add(UChar32 start, UChar32 end);
|
|
jpayne@69
|
1092
|
|
jpayne@69
|
1093 /**
|
|
jpayne@69
|
1094 * Adds the specified character to this set if it is not already
|
|
jpayne@69
|
1095 * present. If this set already contains the specified character,
|
|
jpayne@69
|
1096 * the call leaves this set unchanged.
|
|
jpayne@69
|
1097 * A frozen set will not be modified.
|
|
jpayne@69
|
1098 * @stable ICU 2.0
|
|
jpayne@69
|
1099 */
|
|
jpayne@69
|
1100 UnicodeSet& add(UChar32 c);
|
|
jpayne@69
|
1101
|
|
jpayne@69
|
1102 /**
|
|
jpayne@69
|
1103 * Adds the specified multicharacter to this set if it is not already
|
|
jpayne@69
|
1104 * present. If this set already contains the multicharacter,
|
|
jpayne@69
|
1105 * the call leaves this set unchanged.
|
|
jpayne@69
|
1106 * Thus "ch" => {"ch"}
|
|
jpayne@69
|
1107 * <br><b>Warning: you cannot add an empty string ("") to a UnicodeSet.</b>
|
|
jpayne@69
|
1108 * A frozen set will not be modified.
|
|
jpayne@69
|
1109 * @param s the source string
|
|
jpayne@69
|
1110 * @return this object, for chaining
|
|
jpayne@69
|
1111 * @stable ICU 2.4
|
|
jpayne@69
|
1112 */
|
|
jpayne@69
|
1113 UnicodeSet& add(const UnicodeString& s);
|
|
jpayne@69
|
1114
|
|
jpayne@69
|
1115 private:
|
|
jpayne@69
|
1116 /**
|
|
jpayne@69
|
1117 * @return a code point IF the string consists of a single one.
|
|
jpayne@69
|
1118 * otherwise returns -1.
|
|
jpayne@69
|
1119 * @param s string to test
|
|
jpayne@69
|
1120 */
|
|
jpayne@69
|
1121 static int32_t getSingleCP(const UnicodeString& s);
|
|
jpayne@69
|
1122
|
|
jpayne@69
|
1123 void _add(const UnicodeString& s);
|
|
jpayne@69
|
1124
|
|
jpayne@69
|
1125 public:
|
|
jpayne@69
|
1126 /**
|
|
jpayne@69
|
1127 * Adds each of the characters in this string to the set. Thus "ch" => {"c", "h"}
|
|
jpayne@69
|
1128 * If this set already any particular character, it has no effect on that character.
|
|
jpayne@69
|
1129 * A frozen set will not be modified.
|
|
jpayne@69
|
1130 * @param s the source string
|
|
jpayne@69
|
1131 * @return this object, for chaining
|
|
jpayne@69
|
1132 * @stable ICU 2.4
|
|
jpayne@69
|
1133 */
|
|
jpayne@69
|
1134 UnicodeSet& addAll(const UnicodeString& s);
|
|
jpayne@69
|
1135
|
|
jpayne@69
|
1136 /**
|
|
jpayne@69
|
1137 * Retains EACH of the characters in this string. Note: "ch" == {"c", "h"}
|
|
jpayne@69
|
1138 * If this set already any particular character, it has no effect on that character.
|
|
jpayne@69
|
1139 * A frozen set will not be modified.
|
|
jpayne@69
|
1140 * @param s the source string
|
|
jpayne@69
|
1141 * @return this object, for chaining
|
|
jpayne@69
|
1142 * @stable ICU 2.4
|
|
jpayne@69
|
1143 */
|
|
jpayne@69
|
1144 UnicodeSet& retainAll(const UnicodeString& s);
|
|
jpayne@69
|
1145
|
|
jpayne@69
|
1146 /**
|
|
jpayne@69
|
1147 * Complement EACH of the characters in this string. Note: "ch" == {"c", "h"}
|
|
jpayne@69
|
1148 * If this set already any particular character, it has no effect on that character.
|
|
jpayne@69
|
1149 * A frozen set will not be modified.
|
|
jpayne@69
|
1150 * @param s the source string
|
|
jpayne@69
|
1151 * @return this object, for chaining
|
|
jpayne@69
|
1152 * @stable ICU 2.4
|
|
jpayne@69
|
1153 */
|
|
jpayne@69
|
1154 UnicodeSet& complementAll(const UnicodeString& s);
|
|
jpayne@69
|
1155
|
|
jpayne@69
|
1156 /**
|
|
jpayne@69
|
1157 * Remove EACH of the characters in this string. Note: "ch" == {"c", "h"}
|
|
jpayne@69
|
1158 * If this set already any particular character, it has no effect on that character.
|
|
jpayne@69
|
1159 * A frozen set will not be modified.
|
|
jpayne@69
|
1160 * @param s the source string
|
|
jpayne@69
|
1161 * @return this object, for chaining
|
|
jpayne@69
|
1162 * @stable ICU 2.4
|
|
jpayne@69
|
1163 */
|
|
jpayne@69
|
1164 UnicodeSet& removeAll(const UnicodeString& s);
|
|
jpayne@69
|
1165
|
|
jpayne@69
|
1166 /**
|
|
jpayne@69
|
1167 * Makes a set from a multicharacter string. Thus "ch" => {"ch"}
|
|
jpayne@69
|
1168 * <br><b>Warning: you cannot add an empty string ("") to a UnicodeSet.</b>
|
|
jpayne@69
|
1169 * @param s the source string
|
|
jpayne@69
|
1170 * @return a newly created set containing the given string.
|
|
jpayne@69
|
1171 * The caller owns the return object and is responsible for deleting it.
|
|
jpayne@69
|
1172 * @stable ICU 2.4
|
|
jpayne@69
|
1173 */
|
|
jpayne@69
|
1174 static UnicodeSet* U_EXPORT2 createFrom(const UnicodeString& s);
|
|
jpayne@69
|
1175
|
|
jpayne@69
|
1176
|
|
jpayne@69
|
1177 /**
|
|
jpayne@69
|
1178 * Makes a set from each of the characters in the string. Thus "ch" => {"c", "h"}
|
|
jpayne@69
|
1179 * @param s the source string
|
|
jpayne@69
|
1180 * @return a newly created set containing the given characters
|
|
jpayne@69
|
1181 * The caller owns the return object and is responsible for deleting it.
|
|
jpayne@69
|
1182 * @stable ICU 2.4
|
|
jpayne@69
|
1183 */
|
|
jpayne@69
|
1184 static UnicodeSet* U_EXPORT2 createFromAll(const UnicodeString& s);
|
|
jpayne@69
|
1185
|
|
jpayne@69
|
1186 /**
|
|
jpayne@69
|
1187 * Retain only the elements in this set that are contained in the
|
|
jpayne@69
|
1188 * specified range. If <code>end > start</code> then an empty range is
|
|
jpayne@69
|
1189 * retained, leaving the set empty. This is equivalent to
|
|
jpayne@69
|
1190 * a boolean logic AND, or a set INTERSECTION.
|
|
jpayne@69
|
1191 * A frozen set will not be modified.
|
|
jpayne@69
|
1192 *
|
|
jpayne@69
|
1193 * @param start first character, inclusive, of range to be retained
|
|
jpayne@69
|
1194 * to this set.
|
|
jpayne@69
|
1195 * @param end last character, inclusive, of range to be retained
|
|
jpayne@69
|
1196 * to this set.
|
|
jpayne@69
|
1197 * @stable ICU 2.0
|
|
jpayne@69
|
1198 */
|
|
jpayne@69
|
1199 virtual UnicodeSet& retain(UChar32 start, UChar32 end);
|
|
jpayne@69
|
1200
|
|
jpayne@69
|
1201
|
|
jpayne@69
|
1202 /**
|
|
jpayne@69
|
1203 * Retain the specified character from this set if it is present.
|
|
jpayne@69
|
1204 * A frozen set will not be modified.
|
|
jpayne@69
|
1205 * @stable ICU 2.0
|
|
jpayne@69
|
1206 */
|
|
jpayne@69
|
1207 UnicodeSet& retain(UChar32 c);
|
|
jpayne@69
|
1208
|
|
jpayne@69
|
1209 /**
|
|
jpayne@69
|
1210 * Removes the specified range from this set if it is present.
|
|
jpayne@69
|
1211 * The set will not contain the specified range once the call
|
|
jpayne@69
|
1212 * returns. If <code>end > start</code> then an empty range is
|
|
jpayne@69
|
1213 * removed, leaving the set unchanged.
|
|
jpayne@69
|
1214 * A frozen set will not be modified.
|
|
jpayne@69
|
1215 *
|
|
jpayne@69
|
1216 * @param start first character, inclusive, of range to be removed
|
|
jpayne@69
|
1217 * from this set.
|
|
jpayne@69
|
1218 * @param end last character, inclusive, of range to be removed
|
|
jpayne@69
|
1219 * from this set.
|
|
jpayne@69
|
1220 * @stable ICU 2.0
|
|
jpayne@69
|
1221 */
|
|
jpayne@69
|
1222 virtual UnicodeSet& remove(UChar32 start, UChar32 end);
|
|
jpayne@69
|
1223
|
|
jpayne@69
|
1224 /**
|
|
jpayne@69
|
1225 * Removes the specified character from this set if it is present.
|
|
jpayne@69
|
1226 * The set will not contain the specified range once the call
|
|
jpayne@69
|
1227 * returns.
|
|
jpayne@69
|
1228 * A frozen set will not be modified.
|
|
jpayne@69
|
1229 * @stable ICU 2.0
|
|
jpayne@69
|
1230 */
|
|
jpayne@69
|
1231 UnicodeSet& remove(UChar32 c);
|
|
jpayne@69
|
1232
|
|
jpayne@69
|
1233 /**
|
|
jpayne@69
|
1234 * Removes the specified string from this set if it is present.
|
|
jpayne@69
|
1235 * The set will not contain the specified character once the call
|
|
jpayne@69
|
1236 * returns.
|
|
jpayne@69
|
1237 * A frozen set will not be modified.
|
|
jpayne@69
|
1238 * @param s the source string
|
|
jpayne@69
|
1239 * @return this object, for chaining
|
|
jpayne@69
|
1240 * @stable ICU 2.4
|
|
jpayne@69
|
1241 */
|
|
jpayne@69
|
1242 UnicodeSet& remove(const UnicodeString& s);
|
|
jpayne@69
|
1243
|
|
jpayne@69
|
1244 /**
|
|
jpayne@69
|
1245 * Inverts this set. This operation modifies this set so that
|
|
jpayne@69
|
1246 * its value is its complement. This is equivalent to
|
|
jpayne@69
|
1247 * <code>complement(MIN_VALUE, MAX_VALUE)</code>.
|
|
jpayne@69
|
1248 * A frozen set will not be modified.
|
|
jpayne@69
|
1249 * @stable ICU 2.0
|
|
jpayne@69
|
1250 */
|
|
jpayne@69
|
1251 virtual UnicodeSet& complement(void);
|
|
jpayne@69
|
1252
|
|
jpayne@69
|
1253 /**
|
|
jpayne@69
|
1254 * Complements the specified range in this set. Any character in
|
|
jpayne@69
|
1255 * the range will be removed if it is in this set, or will be
|
|
jpayne@69
|
1256 * added if it is not in this set. If <code>end > start</code>
|
|
jpayne@69
|
1257 * then an empty range is complemented, leaving the set unchanged.
|
|
jpayne@69
|
1258 * This is equivalent to a boolean logic XOR.
|
|
jpayne@69
|
1259 * A frozen set will not be modified.
|
|
jpayne@69
|
1260 *
|
|
jpayne@69
|
1261 * @param start first character, inclusive, of range to be removed
|
|
jpayne@69
|
1262 * from this set.
|
|
jpayne@69
|
1263 * @param end last character, inclusive, of range to be removed
|
|
jpayne@69
|
1264 * from this set.
|
|
jpayne@69
|
1265 * @stable ICU 2.0
|
|
jpayne@69
|
1266 */
|
|
jpayne@69
|
1267 virtual UnicodeSet& complement(UChar32 start, UChar32 end);
|
|
jpayne@69
|
1268
|
|
jpayne@69
|
1269 /**
|
|
jpayne@69
|
1270 * Complements the specified character in this set. The character
|
|
jpayne@69
|
1271 * will be removed if it is in this set, or will be added if it is
|
|
jpayne@69
|
1272 * not in this set.
|
|
jpayne@69
|
1273 * A frozen set will not be modified.
|
|
jpayne@69
|
1274 * @stable ICU 2.0
|
|
jpayne@69
|
1275 */
|
|
jpayne@69
|
1276 UnicodeSet& complement(UChar32 c);
|
|
jpayne@69
|
1277
|
|
jpayne@69
|
1278 /**
|
|
jpayne@69
|
1279 * Complement the specified string in this set.
|
|
jpayne@69
|
1280 * The set will not contain the specified string once the call
|
|
jpayne@69
|
1281 * returns.
|
|
jpayne@69
|
1282 * <br><b>Warning: you cannot add an empty string ("") to a UnicodeSet.</b>
|
|
jpayne@69
|
1283 * A frozen set will not be modified.
|
|
jpayne@69
|
1284 * @param s the string to complement
|
|
jpayne@69
|
1285 * @return this object, for chaining
|
|
jpayne@69
|
1286 * @stable ICU 2.4
|
|
jpayne@69
|
1287 */
|
|
jpayne@69
|
1288 UnicodeSet& complement(const UnicodeString& s);
|
|
jpayne@69
|
1289
|
|
jpayne@69
|
1290 /**
|
|
jpayne@69
|
1291 * Adds all of the elements in the specified set to this set if
|
|
jpayne@69
|
1292 * they're not already present. This operation effectively
|
|
jpayne@69
|
1293 * modifies this set so that its value is the <i>union</i> of the two
|
|
jpayne@69
|
1294 * sets. The behavior of this operation is unspecified if the specified
|
|
jpayne@69
|
1295 * collection is modified while the operation is in progress.
|
|
jpayne@69
|
1296 * A frozen set will not be modified.
|
|
jpayne@69
|
1297 *
|
|
jpayne@69
|
1298 * @param c set whose elements are to be added to this set.
|
|
jpayne@69
|
1299 * @see #add(UChar32, UChar32)
|
|
jpayne@69
|
1300 * @stable ICU 2.0
|
|
jpayne@69
|
1301 */
|
|
jpayne@69
|
1302 virtual UnicodeSet& addAll(const UnicodeSet& c);
|
|
jpayne@69
|
1303
|
|
jpayne@69
|
1304 /**
|
|
jpayne@69
|
1305 * Retains only the elements in this set that are contained in the
|
|
jpayne@69
|
1306 * specified set. In other words, removes from this set all of
|
|
jpayne@69
|
1307 * its elements that are not contained in the specified set. This
|
|
jpayne@69
|
1308 * operation effectively modifies this set so that its value is
|
|
jpayne@69
|
1309 * the <i>intersection</i> of the two sets.
|
|
jpayne@69
|
1310 * A frozen set will not be modified.
|
|
jpayne@69
|
1311 *
|
|
jpayne@69
|
1312 * @param c set that defines which elements this set will retain.
|
|
jpayne@69
|
1313 * @stable ICU 2.0
|
|
jpayne@69
|
1314 */
|
|
jpayne@69
|
1315 virtual UnicodeSet& retainAll(const UnicodeSet& c);
|
|
jpayne@69
|
1316
|
|
jpayne@69
|
1317 /**
|
|
jpayne@69
|
1318 * Removes from this set all of its elements that are contained in the
|
|
jpayne@69
|
1319 * specified set. This operation effectively modifies this
|
|
jpayne@69
|
1320 * set so that its value is the <i>asymmetric set difference</i> of
|
|
jpayne@69
|
1321 * the two sets.
|
|
jpayne@69
|
1322 * A frozen set will not be modified.
|
|
jpayne@69
|
1323 *
|
|
jpayne@69
|
1324 * @param c set that defines which elements will be removed from
|
|
jpayne@69
|
1325 * this set.
|
|
jpayne@69
|
1326 * @stable ICU 2.0
|
|
jpayne@69
|
1327 */
|
|
jpayne@69
|
1328 virtual UnicodeSet& removeAll(const UnicodeSet& c);
|
|
jpayne@69
|
1329
|
|
jpayne@69
|
1330 /**
|
|
jpayne@69
|
1331 * Complements in this set all elements contained in the specified
|
|
jpayne@69
|
1332 * set. Any character in the other set will be removed if it is
|
|
jpayne@69
|
1333 * in this set, or will be added if it is not in this set.
|
|
jpayne@69
|
1334 * A frozen set will not be modified.
|
|
jpayne@69
|
1335 *
|
|
jpayne@69
|
1336 * @param c set that defines which elements will be xor'ed from
|
|
jpayne@69
|
1337 * this set.
|
|
jpayne@69
|
1338 * @stable ICU 2.4
|
|
jpayne@69
|
1339 */
|
|
jpayne@69
|
1340 virtual UnicodeSet& complementAll(const UnicodeSet& c);
|
|
jpayne@69
|
1341
|
|
jpayne@69
|
1342 /**
|
|
jpayne@69
|
1343 * Removes all of the elements from this set. This set will be
|
|
jpayne@69
|
1344 * empty after this call returns.
|
|
jpayne@69
|
1345 * A frozen set will not be modified.
|
|
jpayne@69
|
1346 * @stable ICU 2.0
|
|
jpayne@69
|
1347 */
|
|
jpayne@69
|
1348 virtual UnicodeSet& clear(void);
|
|
jpayne@69
|
1349
|
|
jpayne@69
|
1350 /**
|
|
jpayne@69
|
1351 * Close this set over the given attribute. For the attribute
|
|
jpayne@69
|
1352 * USET_CASE, the result is to modify this set so that:
|
|
jpayne@69
|
1353 *
|
|
jpayne@69
|
1354 * 1. For each character or string 'a' in this set, all strings or
|
|
jpayne@69
|
1355 * characters 'b' such that foldCase(a) == foldCase(b) are added
|
|
jpayne@69
|
1356 * to this set.
|
|
jpayne@69
|
1357 *
|
|
jpayne@69
|
1358 * 2. For each string 'e' in the resulting set, if e !=
|
|
jpayne@69
|
1359 * foldCase(e), 'e' will be removed.
|
|
jpayne@69
|
1360 *
|
|
jpayne@69
|
1361 * Example: [aq\\u00DF{Bc}{bC}{Fi}] => [aAqQ\\u00DF\\uFB01{ss}{bc}{fi}]
|
|
jpayne@69
|
1362 *
|
|
jpayne@69
|
1363 * (Here foldCase(x) refers to the operation u_strFoldCase, and a
|
|
jpayne@69
|
1364 * == b denotes that the contents are the same, not pointer
|
|
jpayne@69
|
1365 * comparison.)
|
|
jpayne@69
|
1366 *
|
|
jpayne@69
|
1367 * A frozen set will not be modified.
|
|
jpayne@69
|
1368 *
|
|
jpayne@69
|
1369 * @param attribute bitmask for attributes to close over.
|
|
jpayne@69
|
1370 * Currently only the USET_CASE bit is supported. Any undefined bits
|
|
jpayne@69
|
1371 * are ignored.
|
|
jpayne@69
|
1372 * @return a reference to this set.
|
|
jpayne@69
|
1373 * @stable ICU 4.2
|
|
jpayne@69
|
1374 */
|
|
jpayne@69
|
1375 UnicodeSet& closeOver(int32_t attribute);
|
|
jpayne@69
|
1376
|
|
jpayne@69
|
1377 /**
|
|
jpayne@69
|
1378 * Remove all strings from this set.
|
|
jpayne@69
|
1379 *
|
|
jpayne@69
|
1380 * @return a reference to this set.
|
|
jpayne@69
|
1381 * @stable ICU 4.2
|
|
jpayne@69
|
1382 */
|
|
jpayne@69
|
1383 virtual UnicodeSet &removeAllStrings();
|
|
jpayne@69
|
1384
|
|
jpayne@69
|
1385 /**
|
|
jpayne@69
|
1386 * Iteration method that returns the number of ranges contained in
|
|
jpayne@69
|
1387 * this set.
|
|
jpayne@69
|
1388 * @see #getRangeStart
|
|
jpayne@69
|
1389 * @see #getRangeEnd
|
|
jpayne@69
|
1390 * @stable ICU 2.4
|
|
jpayne@69
|
1391 */
|
|
jpayne@69
|
1392 virtual int32_t getRangeCount(void) const;
|
|
jpayne@69
|
1393
|
|
jpayne@69
|
1394 /**
|
|
jpayne@69
|
1395 * Iteration method that returns the first character in the
|
|
jpayne@69
|
1396 * specified range of this set.
|
|
jpayne@69
|
1397 * @see #getRangeCount
|
|
jpayne@69
|
1398 * @see #getRangeEnd
|
|
jpayne@69
|
1399 * @stable ICU 2.4
|
|
jpayne@69
|
1400 */
|
|
jpayne@69
|
1401 virtual UChar32 getRangeStart(int32_t index) const;
|
|
jpayne@69
|
1402
|
|
jpayne@69
|
1403 /**
|
|
jpayne@69
|
1404 * Iteration method that returns the last character in the
|
|
jpayne@69
|
1405 * specified range of this set.
|
|
jpayne@69
|
1406 * @see #getRangeStart
|
|
jpayne@69
|
1407 * @see #getRangeEnd
|
|
jpayne@69
|
1408 * @stable ICU 2.4
|
|
jpayne@69
|
1409 */
|
|
jpayne@69
|
1410 virtual UChar32 getRangeEnd(int32_t index) const;
|
|
jpayne@69
|
1411
|
|
jpayne@69
|
1412 /**
|
|
jpayne@69
|
1413 * Serializes this set into an array of 16-bit integers. Serialization
|
|
jpayne@69
|
1414 * (currently) only records the characters in the set; multicharacter
|
|
jpayne@69
|
1415 * strings are ignored.
|
|
jpayne@69
|
1416 *
|
|
jpayne@69
|
1417 * The array has following format (each line is one 16-bit
|
|
jpayne@69
|
1418 * integer):
|
|
jpayne@69
|
1419 *
|
|
jpayne@69
|
1420 * length = (n+2*m) | (m!=0?0x8000:0)
|
|
jpayne@69
|
1421 * bmpLength = n; present if m!=0
|
|
jpayne@69
|
1422 * bmp[0]
|
|
jpayne@69
|
1423 * bmp[1]
|
|
jpayne@69
|
1424 * ...
|
|
jpayne@69
|
1425 * bmp[n-1]
|
|
jpayne@69
|
1426 * supp-high[0]
|
|
jpayne@69
|
1427 * supp-low[0]
|
|
jpayne@69
|
1428 * supp-high[1]
|
|
jpayne@69
|
1429 * supp-low[1]
|
|
jpayne@69
|
1430 * ...
|
|
jpayne@69
|
1431 * supp-high[m-1]
|
|
jpayne@69
|
1432 * supp-low[m-1]
|
|
jpayne@69
|
1433 *
|
|
jpayne@69
|
1434 * The array starts with a header. After the header are n bmp
|
|
jpayne@69
|
1435 * code points, then m supplementary code points. Either n or m
|
|
jpayne@69
|
1436 * or both may be zero. n+2*m is always <= 0x7FFF.
|
|
jpayne@69
|
1437 *
|
|
jpayne@69
|
1438 * If there are no supplementary characters (if m==0) then the
|
|
jpayne@69
|
1439 * header is one 16-bit integer, 'length', with value n.
|
|
jpayne@69
|
1440 *
|
|
jpayne@69
|
1441 * If there are supplementary characters (if m!=0) then the header
|
|
jpayne@69
|
1442 * is two 16-bit integers. The first, 'length', has value
|
|
jpayne@69
|
1443 * (n+2*m)|0x8000. The second, 'bmpLength', has value n.
|
|
jpayne@69
|
1444 *
|
|
jpayne@69
|
1445 * After the header the code points are stored in ascending order.
|
|
jpayne@69
|
1446 * Supplementary code points are stored as most significant 16
|
|
jpayne@69
|
1447 * bits followed by least significant 16 bits.
|
|
jpayne@69
|
1448 *
|
|
jpayne@69
|
1449 * @param dest pointer to buffer of destCapacity 16-bit integers.
|
|
jpayne@69
|
1450 * May be NULL only if destCapacity is zero.
|
|
jpayne@69
|
1451 * @param destCapacity size of dest, or zero. Must not be negative.
|
|
jpayne@69
|
1452 * @param ec error code. Will be set to U_INDEX_OUTOFBOUNDS_ERROR
|
|
jpayne@69
|
1453 * if n+2*m > 0x7FFF. Will be set to U_BUFFER_OVERFLOW_ERROR if
|
|
jpayne@69
|
1454 * n+2*m+(m!=0?2:1) > destCapacity.
|
|
jpayne@69
|
1455 * @return the total length of the serialized format, including
|
|
jpayne@69
|
1456 * the header, that is, n+2*m+(m!=0?2:1), or 0 on error other
|
|
jpayne@69
|
1457 * than U_BUFFER_OVERFLOW_ERROR.
|
|
jpayne@69
|
1458 * @stable ICU 2.4
|
|
jpayne@69
|
1459 */
|
|
jpayne@69
|
1460 int32_t serialize(uint16_t *dest, int32_t destCapacity, UErrorCode& ec) const;
|
|
jpayne@69
|
1461
|
|
jpayne@69
|
1462 /**
|
|
jpayne@69
|
1463 * Reallocate this objects internal structures to take up the least
|
|
jpayne@69
|
1464 * possible space, without changing this object's value.
|
|
jpayne@69
|
1465 * A frozen set will not be modified.
|
|
jpayne@69
|
1466 * @stable ICU 2.4
|
|
jpayne@69
|
1467 */
|
|
jpayne@69
|
1468 virtual UnicodeSet& compact();
|
|
jpayne@69
|
1469
|
|
jpayne@69
|
1470 /**
|
|
jpayne@69
|
1471 * Return the class ID for this class. This is useful only for
|
|
jpayne@69
|
1472 * comparing to a return value from getDynamicClassID(). For example:
|
|
jpayne@69
|
1473 * <pre>
|
|
jpayne@69
|
1474 * . Base* polymorphic_pointer = createPolymorphicObject();
|
|
jpayne@69
|
1475 * . if (polymorphic_pointer->getDynamicClassID() ==
|
|
jpayne@69
|
1476 * . Derived::getStaticClassID()) ...
|
|
jpayne@69
|
1477 * </pre>
|
|
jpayne@69
|
1478 * @return The class ID for all objects of this class.
|
|
jpayne@69
|
1479 * @stable ICU 2.0
|
|
jpayne@69
|
1480 */
|
|
jpayne@69
|
1481 static UClassID U_EXPORT2 getStaticClassID(void);
|
|
jpayne@69
|
1482
|
|
jpayne@69
|
1483 /**
|
|
jpayne@69
|
1484 * Implement UnicodeFunctor API.
|
|
jpayne@69
|
1485 *
|
|
jpayne@69
|
1486 * @return The class ID for this object. All objects of a given
|
|
jpayne@69
|
1487 * class have the same class ID. Objects of other classes have
|
|
jpayne@69
|
1488 * different class IDs.
|
|
jpayne@69
|
1489 * @stable ICU 2.4
|
|
jpayne@69
|
1490 */
|
|
jpayne@69
|
1491 virtual UClassID getDynamicClassID(void) const;
|
|
jpayne@69
|
1492
|
|
jpayne@69
|
1493 private:
|
|
jpayne@69
|
1494
|
|
jpayne@69
|
1495 // Private API for the USet API
|
|
jpayne@69
|
1496
|
|
jpayne@69
|
1497 friend class USetAccess;
|
|
jpayne@69
|
1498
|
|
jpayne@69
|
1499 const UnicodeString* getString(int32_t index) const;
|
|
jpayne@69
|
1500
|
|
jpayne@69
|
1501 //----------------------------------------------------------------
|
|
jpayne@69
|
1502 // RuleBasedTransliterator support
|
|
jpayne@69
|
1503 //----------------------------------------------------------------
|
|
jpayne@69
|
1504
|
|
jpayne@69
|
1505 private:
|
|
jpayne@69
|
1506
|
|
jpayne@69
|
1507 /**
|
|
jpayne@69
|
1508 * Returns <tt>true</tt> if this set contains any character whose low byte
|
|
jpayne@69
|
1509 * is the given value. This is used by <tt>RuleBasedTransliterator</tt> for
|
|
jpayne@69
|
1510 * indexing.
|
|
jpayne@69
|
1511 */
|
|
jpayne@69
|
1512 virtual UBool matchesIndexValue(uint8_t v) const;
|
|
jpayne@69
|
1513
|
|
jpayne@69
|
1514 private:
|
|
jpayne@69
|
1515 friend class RBBIRuleScanner;
|
|
jpayne@69
|
1516
|
|
jpayne@69
|
1517 //----------------------------------------------------------------
|
|
jpayne@69
|
1518 // Implementation: Clone as thawed (see ICU4J Freezable)
|
|
jpayne@69
|
1519 //----------------------------------------------------------------
|
|
jpayne@69
|
1520
|
|
jpayne@69
|
1521 UnicodeSet(const UnicodeSet& o, UBool /* asThawed */);
|
|
jpayne@69
|
1522 UnicodeSet& copyFrom(const UnicodeSet& o, UBool asThawed);
|
|
jpayne@69
|
1523
|
|
jpayne@69
|
1524 //----------------------------------------------------------------
|
|
jpayne@69
|
1525 // Implementation: Pattern parsing
|
|
jpayne@69
|
1526 //----------------------------------------------------------------
|
|
jpayne@69
|
1527
|
|
jpayne@69
|
1528 void applyPatternIgnoreSpace(const UnicodeString& pattern,
|
|
jpayne@69
|
1529 ParsePosition& pos,
|
|
jpayne@69
|
1530 const SymbolTable* symbols,
|
|
jpayne@69
|
1531 UErrorCode& status);
|
|
jpayne@69
|
1532
|
|
jpayne@69
|
1533 void applyPattern(RuleCharacterIterator& chars,
|
|
jpayne@69
|
1534 const SymbolTable* symbols,
|
|
jpayne@69
|
1535 UnicodeString& rebuiltPat,
|
|
jpayne@69
|
1536 uint32_t options,
|
|
jpayne@69
|
1537 UnicodeSet& (UnicodeSet::*caseClosure)(int32_t attribute),
|
|
jpayne@69
|
1538 int32_t depth,
|
|
jpayne@69
|
1539 UErrorCode& ec);
|
|
jpayne@69
|
1540
|
|
jpayne@69
|
1541 //----------------------------------------------------------------
|
|
jpayne@69
|
1542 // Implementation: Utility methods
|
|
jpayne@69
|
1543 //----------------------------------------------------------------
|
|
jpayne@69
|
1544
|
|
jpayne@69
|
1545 static int32_t nextCapacity(int32_t minCapacity);
|
|
jpayne@69
|
1546
|
|
jpayne@69
|
1547 bool ensureCapacity(int32_t newLen);
|
|
jpayne@69
|
1548
|
|
jpayne@69
|
1549 bool ensureBufferCapacity(int32_t newLen);
|
|
jpayne@69
|
1550
|
|
jpayne@69
|
1551 void swapBuffers(void);
|
|
jpayne@69
|
1552
|
|
jpayne@69
|
1553 UBool allocateStrings(UErrorCode &status);
|
|
jpayne@69
|
1554 UBool hasStrings() const;
|
|
jpayne@69
|
1555 int32_t stringsSize() const;
|
|
jpayne@69
|
1556 UBool stringsContains(const UnicodeString &s) const;
|
|
jpayne@69
|
1557
|
|
jpayne@69
|
1558 UnicodeString& _toPattern(UnicodeString& result,
|
|
jpayne@69
|
1559 UBool escapeUnprintable) const;
|
|
jpayne@69
|
1560
|
|
jpayne@69
|
1561 UnicodeString& _generatePattern(UnicodeString& result,
|
|
jpayne@69
|
1562 UBool escapeUnprintable) const;
|
|
jpayne@69
|
1563
|
|
jpayne@69
|
1564 static void _appendToPat(UnicodeString& buf, const UnicodeString& s, UBool escapeUnprintable);
|
|
jpayne@69
|
1565
|
|
jpayne@69
|
1566 static void _appendToPat(UnicodeString& buf, UChar32 c, UBool escapeUnprintable);
|
|
jpayne@69
|
1567
|
|
jpayne@69
|
1568 //----------------------------------------------------------------
|
|
jpayne@69
|
1569 // Implementation: Fundamental operators
|
|
jpayne@69
|
1570 //----------------------------------------------------------------
|
|
jpayne@69
|
1571
|
|
jpayne@69
|
1572 void exclusiveOr(const UChar32* other, int32_t otherLen, int8_t polarity);
|
|
jpayne@69
|
1573
|
|
jpayne@69
|
1574 void add(const UChar32* other, int32_t otherLen, int8_t polarity);
|
|
jpayne@69
|
1575
|
|
jpayne@69
|
1576 void retain(const UChar32* other, int32_t otherLen, int8_t polarity);
|
|
jpayne@69
|
1577
|
|
jpayne@69
|
1578 /**
|
|
jpayne@69
|
1579 * Return true if the given position, in the given pattern, appears
|
|
jpayne@69
|
1580 * to be the start of a property set pattern [:foo:], \\p{foo}, or
|
|
jpayne@69
|
1581 * \\P{foo}, or \\N{name}.
|
|
jpayne@69
|
1582 */
|
|
jpayne@69
|
1583 static UBool resemblesPropertyPattern(const UnicodeString& pattern,
|
|
jpayne@69
|
1584 int32_t pos);
|
|
jpayne@69
|
1585
|
|
jpayne@69
|
1586 static UBool resemblesPropertyPattern(RuleCharacterIterator& chars,
|
|
jpayne@69
|
1587 int32_t iterOpts);
|
|
jpayne@69
|
1588
|
|
jpayne@69
|
1589 /**
|
|
jpayne@69
|
1590 * Parse the given property pattern at the given parse position
|
|
jpayne@69
|
1591 * and set this UnicodeSet to the result.
|
|
jpayne@69
|
1592 *
|
|
jpayne@69
|
1593 * The original design document is out of date, but still useful.
|
|
jpayne@69
|
1594 * Ignore the property and value names:
|
|
jpayne@69
|
1595 * http://source.icu-project.org/repos/icu/icuhtml/trunk/design/unicodeset_properties.html
|
|
jpayne@69
|
1596 *
|
|
jpayne@69
|
1597 * Recognized syntax:
|
|
jpayne@69
|
1598 *
|
|
jpayne@69
|
1599 * [:foo:] [:^foo:] - white space not allowed within "[:" or ":]"
|
|
jpayne@69
|
1600 * \\p{foo} \\P{foo} - white space not allowed within "\\p" or "\\P"
|
|
jpayne@69
|
1601 * \\N{name} - white space not allowed within "\\N"
|
|
jpayne@69
|
1602 *
|
|
jpayne@69
|
1603 * Other than the above restrictions, Unicode Pattern_White_Space characters are ignored.
|
|
jpayne@69
|
1604 * Case is ignored except in "\\p" and "\\P" and "\\N". In 'name' leading
|
|
jpayne@69
|
1605 * and trailing space is deleted, and internal runs of whitespace
|
|
jpayne@69
|
1606 * are collapsed to a single space.
|
|
jpayne@69
|
1607 *
|
|
jpayne@69
|
1608 * We support binary properties, enumerated properties, and the
|
|
jpayne@69
|
1609 * following non-enumerated properties:
|
|
jpayne@69
|
1610 *
|
|
jpayne@69
|
1611 * Numeric_Value
|
|
jpayne@69
|
1612 * Name
|
|
jpayne@69
|
1613 * Unicode_1_Name
|
|
jpayne@69
|
1614 *
|
|
jpayne@69
|
1615 * @param pattern the pattern string
|
|
jpayne@69
|
1616 * @param ppos on entry, the position at which to begin parsing.
|
|
jpayne@69
|
1617 * This should be one of the locations marked '^':
|
|
jpayne@69
|
1618 *
|
|
jpayne@69
|
1619 * [:blah:] \\p{blah} \\P{blah} \\N{name}
|
|
jpayne@69
|
1620 * ^ % ^ % ^ % ^ %
|
|
jpayne@69
|
1621 *
|
|
jpayne@69
|
1622 * On return, the position after the last character parsed, that is,
|
|
jpayne@69
|
1623 * the locations marked '%'. If the parse fails, ppos is returned
|
|
jpayne@69
|
1624 * unchanged.
|
|
jpayne@69
|
1625 * @param ec status
|
|
jpayne@69
|
1626 * @return a reference to this.
|
|
jpayne@69
|
1627 */
|
|
jpayne@69
|
1628 UnicodeSet& applyPropertyPattern(const UnicodeString& pattern,
|
|
jpayne@69
|
1629 ParsePosition& ppos,
|
|
jpayne@69
|
1630 UErrorCode &ec);
|
|
jpayne@69
|
1631
|
|
jpayne@69
|
1632 void applyPropertyPattern(RuleCharacterIterator& chars,
|
|
jpayne@69
|
1633 UnicodeString& rebuiltPat,
|
|
jpayne@69
|
1634 UErrorCode& ec);
|
|
jpayne@69
|
1635
|
|
jpayne@69
|
1636 static const UnicodeSet* getInclusions(int32_t src, UErrorCode &status);
|
|
jpayne@69
|
1637
|
|
jpayne@69
|
1638 /**
|
|
jpayne@69
|
1639 * A filter that returns TRUE if the given code point should be
|
|
jpayne@69
|
1640 * included in the UnicodeSet being constructed.
|
|
jpayne@69
|
1641 */
|
|
jpayne@69
|
1642 typedef UBool (*Filter)(UChar32 codePoint, void* context);
|
|
jpayne@69
|
1643
|
|
jpayne@69
|
1644 /**
|
|
jpayne@69
|
1645 * Given a filter, set this UnicodeSet to the code points
|
|
jpayne@69
|
1646 * contained by that filter. The filter MUST be
|
|
jpayne@69
|
1647 * property-conformant. That is, if it returns value v for one
|
|
jpayne@69
|
1648 * code point, then it must return v for all affiliated code
|
|
jpayne@69
|
1649 * points, as defined by the inclusions list. See
|
|
jpayne@69
|
1650 * getInclusions().
|
|
jpayne@69
|
1651 * src is a UPropertySource value.
|
|
jpayne@69
|
1652 */
|
|
jpayne@69
|
1653 void applyFilter(Filter filter,
|
|
jpayne@69
|
1654 void* context,
|
|
jpayne@69
|
1655 const UnicodeSet* inclusions,
|
|
jpayne@69
|
1656 UErrorCode &status);
|
|
jpayne@69
|
1657
|
|
jpayne@69
|
1658 // UCPMap is now stable ICU 63
|
|
jpayne@69
|
1659 void applyIntPropertyValue(const UCPMap *map,
|
|
jpayne@69
|
1660 UCPMapValueFilter *filter, const void *context,
|
|
jpayne@69
|
1661 UErrorCode &errorCode);
|
|
jpayne@69
|
1662
|
|
jpayne@69
|
1663 /**
|
|
jpayne@69
|
1664 * Set the new pattern to cache.
|
|
jpayne@69
|
1665 */
|
|
jpayne@69
|
1666 void setPattern(const UnicodeString& newPat) {
|
|
jpayne@69
|
1667 setPattern(newPat.getBuffer(), newPat.length());
|
|
jpayne@69
|
1668 }
|
|
jpayne@69
|
1669 void setPattern(const char16_t *newPat, int32_t newPatLen);
|
|
jpayne@69
|
1670 /**
|
|
jpayne@69
|
1671 * Release existing cached pattern.
|
|
jpayne@69
|
1672 */
|
|
jpayne@69
|
1673 void releasePattern();
|
|
jpayne@69
|
1674
|
|
jpayne@69
|
1675 friend class UnicodeSetIterator;
|
|
jpayne@69
|
1676 };
|
|
jpayne@69
|
1677
|
|
jpayne@69
|
1678
|
|
jpayne@69
|
1679
|
|
jpayne@69
|
1680 inline UBool UnicodeSet::operator!=(const UnicodeSet& o) const {
|
|
jpayne@69
|
1681 return !operator==(o);
|
|
jpayne@69
|
1682 }
|
|
jpayne@69
|
1683
|
|
jpayne@69
|
1684 inline UBool UnicodeSet::isFrozen() const {
|
|
jpayne@69
|
1685 return (UBool)(bmpSet!=NULL || stringSpan!=NULL);
|
|
jpayne@69
|
1686 }
|
|
jpayne@69
|
1687
|
|
jpayne@69
|
1688 inline UBool UnicodeSet::containsSome(UChar32 start, UChar32 end) const {
|
|
jpayne@69
|
1689 return !containsNone(start, end);
|
|
jpayne@69
|
1690 }
|
|
jpayne@69
|
1691
|
|
jpayne@69
|
1692 inline UBool UnicodeSet::containsSome(const UnicodeSet& s) const {
|
|
jpayne@69
|
1693 return !containsNone(s);
|
|
jpayne@69
|
1694 }
|
|
jpayne@69
|
1695
|
|
jpayne@69
|
1696 inline UBool UnicodeSet::containsSome(const UnicodeString& s) const {
|
|
jpayne@69
|
1697 return !containsNone(s);
|
|
jpayne@69
|
1698 }
|
|
jpayne@69
|
1699
|
|
jpayne@69
|
1700 inline UBool UnicodeSet::isBogus() const {
|
|
jpayne@69
|
1701 return (UBool)(fFlags & kIsBogus);
|
|
jpayne@69
|
1702 }
|
|
jpayne@69
|
1703
|
|
jpayne@69
|
1704 inline UnicodeSet *UnicodeSet::fromUSet(USet *uset) {
|
|
jpayne@69
|
1705 return reinterpret_cast<UnicodeSet *>(uset);
|
|
jpayne@69
|
1706 }
|
|
jpayne@69
|
1707
|
|
jpayne@69
|
1708 inline const UnicodeSet *UnicodeSet::fromUSet(const USet *uset) {
|
|
jpayne@69
|
1709 return reinterpret_cast<const UnicodeSet *>(uset);
|
|
jpayne@69
|
1710 }
|
|
jpayne@69
|
1711
|
|
jpayne@69
|
1712 inline USet *UnicodeSet::toUSet() {
|
|
jpayne@69
|
1713 return reinterpret_cast<USet *>(this);
|
|
jpayne@69
|
1714 }
|
|
jpayne@69
|
1715
|
|
jpayne@69
|
1716 inline const USet *UnicodeSet::toUSet() const {
|
|
jpayne@69
|
1717 return reinterpret_cast<const USet *>(this);
|
|
jpayne@69
|
1718 }
|
|
jpayne@69
|
1719
|
|
jpayne@69
|
1720 inline int32_t UnicodeSet::span(const UnicodeString &s, int32_t start, USetSpanCondition spanCondition) const {
|
|
jpayne@69
|
1721 int32_t sLength=s.length();
|
|
jpayne@69
|
1722 if(start<0) {
|
|
jpayne@69
|
1723 start=0;
|
|
jpayne@69
|
1724 } else if(start>sLength) {
|
|
jpayne@69
|
1725 start=sLength;
|
|
jpayne@69
|
1726 }
|
|
jpayne@69
|
1727 return start+span(s.getBuffer()+start, sLength-start, spanCondition);
|
|
jpayne@69
|
1728 }
|
|
jpayne@69
|
1729
|
|
jpayne@69
|
1730 inline int32_t UnicodeSet::spanBack(const UnicodeString &s, int32_t limit, USetSpanCondition spanCondition) const {
|
|
jpayne@69
|
1731 int32_t sLength=s.length();
|
|
jpayne@69
|
1732 if(limit<0) {
|
|
jpayne@69
|
1733 limit=0;
|
|
jpayne@69
|
1734 } else if(limit>sLength) {
|
|
jpayne@69
|
1735 limit=sLength;
|
|
jpayne@69
|
1736 }
|
|
jpayne@69
|
1737 return spanBack(s.getBuffer(), limit, spanCondition);
|
|
jpayne@69
|
1738 }
|
|
jpayne@69
|
1739
|
|
jpayne@69
|
1740 U_NAMESPACE_END
|
|
jpayne@69
|
1741
|
|
jpayne@69
|
1742 #endif /* U_SHOW_CPLUSPLUS_API */
|
|
jpayne@69
|
1743
|
|
jpayne@69
|
1744 #endif
|
