Mercurial > repos > rliterman > csp2

annotate CSP2/CSP2_env/env-d9b9114564458d9d-741b3de822f2aaca6c6caa4325c4afce/include/unicode/unimatch.h @ 69:33d812a61356

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 2e9511a184a1ca667c7be0c6321a36dc4e3d116d
author jpayne
date Tue, 18 Mar 2025 17:55:14 -0400
parents
children
rev   line source
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 * Copyright (C) 2001-2005, International Business Machines Corporation and others. All Rights Reserved.
jpayne@69 5 **********************************************************************
jpayne@69 6 * Date Name Description
jpayne@69 7 * 07/18/01 aliu Creation.
jpayne@69 8 **********************************************************************
jpayne@69 9 */
jpayne@69 10 #ifndef UNIMATCH_H
jpayne@69 11 #define UNIMATCH_H
jpayne@69 12
jpayne@69 13 #include "unicode/utypes.h"
jpayne@69 14
jpayne@69 15 /**
jpayne@69 16 * \file
jpayne@69 17 * \brief C++ API: Unicode Matcher
jpayne@69 18 */
jpayne@69 19
jpayne@69 20 #if U_SHOW_CPLUSPLUS_API
jpayne@69 21
jpayne@69 22 U_NAMESPACE_BEGIN
jpayne@69 23
jpayne@69 24 class Replaceable;
jpayne@69 25 class UnicodeString;
jpayne@69 26 class UnicodeSet;
jpayne@69 27
jpayne@69 28 /**
jpayne@69 29 * Constants returned by <code>UnicodeMatcher::matches()</code>
jpayne@69 30 * indicating the degree of match.
jpayne@69 31 * @stable ICU 2.4
jpayne@69 32 */
jpayne@69 33 enum UMatchDegree {
jpayne@69 34 /**
jpayne@69 35 * Constant returned by <code>matches()</code> indicating a
jpayne@69 36 * mismatch between the text and this matcher. The text contains
jpayne@69 37 * a character which does not match, or the text does not contain
jpayne@69 38 * all desired characters for a non-incremental match.
jpayne@69 39 * @stable ICU 2.4
jpayne@69 40 */
jpayne@69 41 U_MISMATCH,
jpayne@69 42
jpayne@69 43 /**
jpayne@69 44 * Constant returned by <code>matches()</code> indicating a
jpayne@69 45 * partial match between the text and this matcher. This value is
jpayne@69 46 * only returned for incremental match operations. All characters
jpayne@69 47 * of the text match, but more characters are required for a
jpayne@69 48 * complete match. Alternatively, for variable-length matchers,
jpayne@69 49 * all characters of the text match, and if more characters were
jpayne@69 50 * supplied at limit, they might also match.
jpayne@69 51 * @stable ICU 2.4
jpayne@69 52 */
jpayne@69 53 U_PARTIAL_MATCH,
jpayne@69 54
jpayne@69 55 /**
jpayne@69 56 * Constant returned by <code>matches()</code> indicating a
jpayne@69 57 * complete match between the text and this matcher. For an
jpayne@69 58 * incremental variable-length match, this value is returned if
jpayne@69 59 * the given text matches, and it is known that additional
jpayne@69 60 * characters would not alter the extent of the match.
jpayne@69 61 * @stable ICU 2.4
jpayne@69 62 */
jpayne@69 63 U_MATCH
jpayne@69 64 };
jpayne@69 65
jpayne@69 66 /**
jpayne@69 67 * <code>UnicodeMatcher</code> defines a protocol for objects that can
jpayne@69 68 * match a range of characters in a Replaceable string.
jpayne@69 69 * @stable ICU 2.4
jpayne@69 70 */
jpayne@69 71 class U_COMMON_API UnicodeMatcher /* not : public UObject because this is an interface/mixin class */ {
jpayne@69 72
jpayne@69 73 public:
jpayne@69 74 /**
jpayne@69 75 * Destructor.
jpayne@69 76 * @stable ICU 2.4
jpayne@69 77 */
jpayne@69 78 virtual ~UnicodeMatcher();
jpayne@69 79
jpayne@69 80 /**
jpayne@69 81 * Return a UMatchDegree value indicating the degree of match for
jpayne@69 82 * the given text at the given offset. Zero, one, or more
jpayne@69 83 * characters may be matched.
jpayne@69 84 *
jpayne@69 85 * Matching in the forward direction is indicated by limit >
jpayne@69 86 * offset. Characters from offset forwards to limit-1 will be
jpayne@69 87 * considered for matching.
jpayne@69 88 *
jpayne@69 89 * Matching in the reverse direction is indicated by limit <
jpayne@69 90 * offset. Characters from offset backwards to limit+1 will be
jpayne@69 91 * considered for matching.
jpayne@69 92 *
jpayne@69 93 * If limit == offset then the only match possible is a zero
jpayne@69 94 * character match (which subclasses may implement if desired).
jpayne@69 95 *
jpayne@69 96 * As a side effect, advance the offset parameter to the limit of
jpayne@69 97 * the matched substring. In the forward direction, this will be
jpayne@69 98 * the index of the last matched character plus one. In the
jpayne@69 99 * reverse direction, this will be the index of the last matched
jpayne@69 100 * character minus one.
jpayne@69 101 *
jpayne@69 102 * <p>Note: This method is not const because some classes may
jpayne@69 103 * modify their state as the result of a match.
jpayne@69 104 *
jpayne@69 105 * @param text the text to be matched
jpayne@69 106 * @param offset on input, the index into text at which to begin
jpayne@69 107 * matching. On output, the limit of the matched text. The
jpayne@69 108 * number of matched characters is the output value of offset
jpayne@69 109 * minus the input value. Offset should always point to the
jpayne@69 110 * HIGH SURROGATE (leading code unit) of a pair of surrogates,
jpayne@69 111 * both on entry and upon return.
jpayne@69 112 * @param limit the limit index of text to be matched. Greater
jpayne@69 113 * than offset for a forward direction match, less than offset for
jpayne@69 114 * a backward direction match. The last character to be
jpayne@69 115 * considered for matching will be text.charAt(limit-1) in the
jpayne@69 116 * forward direction or text.charAt(limit+1) in the backward
jpayne@69 117 * direction.
jpayne@69 118 * @param incremental if TRUE, then assume further characters may
jpayne@69 119 * be inserted at limit and check for partial matching. Otherwise
jpayne@69 120 * assume the text as given is complete.
jpayne@69 121 * @return a match degree value indicating a full match, a partial
jpayne@69 122 * match, or a mismatch. If incremental is FALSE then
jpayne@69 123 * U_PARTIAL_MATCH should never be returned.
jpayne@69 124 * @stable ICU 2.4
jpayne@69 125 */
jpayne@69 126 virtual UMatchDegree matches(const Replaceable& text,
jpayne@69 127 int32_t& offset,
jpayne@69 128 int32_t limit,
jpayne@69 129 UBool incremental) = 0;
jpayne@69 130
jpayne@69 131 /**
jpayne@69 132 * Returns a string representation of this matcher. If the result of
jpayne@69 133 * calling this function is passed to the appropriate parser, it
jpayne@69 134 * will produce another matcher that is equal to this one.
jpayne@69 135 * @param result the string to receive the pattern. Previous
jpayne@69 136 * contents will be deleted.
jpayne@69 137 * @param escapeUnprintable if TRUE then convert unprintable
jpayne@69 138 * character to their hex escape representations, \\uxxxx or
jpayne@69 139 * \\Uxxxxxxxx. Unprintable characters are those other than
jpayne@69 140 * U+000A, U+0020..U+007E.
jpayne@69 141 * @stable ICU 2.4
jpayne@69 142 */
jpayne@69 143 virtual UnicodeString& toPattern(UnicodeString& result,
jpayne@69 144 UBool escapeUnprintable = FALSE) const = 0;
jpayne@69 145
jpayne@69 146 /**
jpayne@69 147 * Returns TRUE if this matcher will match a character c, where c
jpayne@69 148 * & 0xFF == v, at offset, in the forward direction (with limit >
jpayne@69 149 * offset). This is used by <tt>RuleBasedTransliterator</tt> for
jpayne@69 150 * indexing.
jpayne@69 151 * @stable ICU 2.4
jpayne@69 152 */
jpayne@69 153 virtual UBool matchesIndexValue(uint8_t v) const = 0;
jpayne@69 154
jpayne@69 155 /**
jpayne@69 156 * Union the set of all characters that may be matched by this object
jpayne@69 157 * into the given set.
jpayne@69 158 * @param toUnionTo the set into which to union the source characters
jpayne@69 159 * @stable ICU 2.4
jpayne@69 160 */
jpayne@69 161 virtual void addMatchSetTo(UnicodeSet& toUnionTo) const = 0;
jpayne@69 162 };
jpayne@69 163
jpayne@69 164 U_NAMESPACE_END
jpayne@69 165
jpayne@69 166 #endif /* U_SHOW_CPLUSPLUS_API */
jpayne@69 167
jpayne@69 168 #endif