blob: df6924cc127ba62023b2e1b38fdab6f0225c641d [file] [log] [blame]
Victor Chang73229502020-09-17 13:39:19 +01001// © 2016 and later: Unicode, Inc. and others.
2// License & terms of use: http://www.unicode.org/copyright.html
3/*
4 **********************************************************************
5 * Copyright (C) 2001-2011, International Business Machines Corporation
6 * and others. All Rights Reserved.
7 **********************************************************************
8 * Date Name Description
9 * 07/26/01 aliu Creation.
10 **********************************************************************
11 */
12#ifndef QUANT_H
13#define QUANT_H
14
15#include "unicode/utypes.h"
16
17#if !UCONFIG_NO_TRANSLITERATION
18
19#include "unicode/unifunct.h"
20#include "unicode/unimatch.h"
21
22U_NAMESPACE_BEGIN
23
24class Quantifier : public UnicodeFunctor, public UnicodeMatcher {
25
26 public:
27
28 enum { MAX = 0x7FFFFFFF };
29
30 Quantifier(UnicodeFunctor *adoptedMatcher,
31 uint32_t minCount, uint32_t maxCount);
32
33 Quantifier(const Quantifier& o);
34
35 virtual ~Quantifier();
36
37 /**
38 * UnicodeFunctor API. Cast 'this' to a UnicodeMatcher* pointer
39 * and return the pointer.
40 * @return the UnicodeMatcher pointer.
41 */
42 virtual UnicodeMatcher* toMatcher() const;
43
44 /**
45 * Implement UnicodeFunctor
46 * @return a copy of the object.
47 */
48 virtual Quantifier* clone() const;
49
50 /**
51 * Implement UnicodeMatcher
52 * @param text the text to be matched
53 * @param offset on input, the index into text at which to begin
54 * matching. On output, the limit of the matched text. The
55 * number of matched characters is the output value of offset
56 * minus the input value. Offset should always point to the
57 * HIGH SURROGATE (leading code unit) of a pair of surrogates,
58 * both on entry and upon return.
59 * @param limit the limit index of text to be matched. Greater
60 * than offset for a forward direction match, less than offset for
61 * a backward direction match. The last character to be
62 * considered for matching will be text.charAt(limit-1) in the
63 * forward direction or text.charAt(limit+1) in the backward
64 * direction.
Victor Changce4bf3c2021-01-19 16:34:24 +000065 * @param incremental if true, then assume further characters may
Victor Chang73229502020-09-17 13:39:19 +010066 * be inserted at limit and check for partial matching. Otherwise
67 * assume the text as given is complete.
68 * @return a match degree value indicating a full match, a partial
Victor Changce4bf3c2021-01-19 16:34:24 +000069 * match, or a mismatch. If incremental is false then
Victor Chang73229502020-09-17 13:39:19 +010070 * U_PARTIAL_MATCH should never be returned.
71 */
72 virtual UMatchDegree matches(const Replaceable& text,
73 int32_t& offset,
74 int32_t limit,
75 UBool incremental);
76
77 /**
78 * Implement UnicodeMatcher
79 * @param result Output param to receive the pattern.
80 * @param escapeUnprintable if True then escape the unprintable characters.
81 * @return A reference to 'result'.
82 */
83 virtual UnicodeString& toPattern(UnicodeString& result,
Victor Changce4bf3c2021-01-19 16:34:24 +000084 UBool escapeUnprintable = false) const;
Victor Chang73229502020-09-17 13:39:19 +010085
86 /**
87 * Implement UnicodeMatcher
88 * @param v the given index value.
89 * @return true if this rule matches the given index value.
90 */
91 virtual UBool matchesIndexValue(uint8_t v) const;
92
93 /**
94 * Implement UnicodeMatcher
95 */
96 virtual void addMatchSetTo(UnicodeSet& toUnionTo) const;
97
98 /**
99 * UnicodeFunctor API
100 */
101 virtual void setData(const TransliterationRuleData*);
102
103 /**
104 * ICU "poor man's RTTI", returns a UClassID for the actual class.
105 */
106 virtual UClassID getDynamicClassID() const;
107
108 /**
109 * ICU "poor man's RTTI", returns a UClassID for this class.
110 */
111 static UClassID U_EXPORT2 getStaticClassID();
112
113 private:
114
115 UnicodeFunctor* matcher; // owned
116
117 uint32_t minCount;
118
119 uint32_t maxCount;
120};
121
122U_NAMESPACE_END
123
124#endif /* #if !UCONFIG_NO_TRANSLITERATION */
125
126#endif