|
1 /* |
|
2 ********************************************************************** |
|
3 * Copyright (C) 2001-2011, International Business Machines Corporation |
|
4 * and others. All Rights Reserved. |
|
5 ********************************************************************** |
|
6 * Date Name Description |
|
7 * 07/26/01 aliu Creation. |
|
8 ********************************************************************** |
|
9 */ |
|
10 #ifndef QUANT_H |
|
11 #define QUANT_H |
|
12 |
|
13 #include "unicode/utypes.h" |
|
14 |
|
15 #if !UCONFIG_NO_TRANSLITERATION |
|
16 |
|
17 #include "unicode/unifunct.h" |
|
18 #include "unicode/unimatch.h" |
|
19 |
|
20 U_NAMESPACE_BEGIN |
|
21 |
|
22 class Quantifier : public UnicodeFunctor, public UnicodeMatcher { |
|
23 |
|
24 public: |
|
25 |
|
26 enum { MAX = 0x7FFFFFFF }; |
|
27 |
|
28 Quantifier(UnicodeFunctor *adoptedMatcher, |
|
29 uint32_t minCount, uint32_t maxCount); |
|
30 |
|
31 Quantifier(const Quantifier& o); |
|
32 |
|
33 virtual ~Quantifier(); |
|
34 |
|
35 /** |
|
36 * UnicodeFunctor API. Cast 'this' to a UnicodeMatcher* pointer |
|
37 * and return the pointer. |
|
38 * @return the UnicodeMatcher pointer. |
|
39 */ |
|
40 virtual UnicodeMatcher* toMatcher() const; |
|
41 |
|
42 /** |
|
43 * Implement UnicodeFunctor |
|
44 * @return a copy of the object. |
|
45 */ |
|
46 virtual UnicodeFunctor* clone() const; |
|
47 |
|
48 /** |
|
49 * Implement UnicodeMatcher |
|
50 * @param text the text to be matched |
|
51 * @param offset on input, the index into text at which to begin |
|
52 * matching. On output, the limit of the matched text. The |
|
53 * number of matched characters is the output value of offset |
|
54 * minus the input value. Offset should always point to the |
|
55 * HIGH SURROGATE (leading code unit) of a pair of surrogates, |
|
56 * both on entry and upon return. |
|
57 * @param limit the limit index of text to be matched. Greater |
|
58 * than offset for a forward direction match, less than offset for |
|
59 * a backward direction match. The last character to be |
|
60 * considered for matching will be text.charAt(limit-1) in the |
|
61 * forward direction or text.charAt(limit+1) in the backward |
|
62 * direction. |
|
63 * @param incremental if TRUE, then assume further characters may |
|
64 * be inserted at limit and check for partial matching. Otherwise |
|
65 * assume the text as given is complete. |
|
66 * @return a match degree value indicating a full match, a partial |
|
67 * match, or a mismatch. If incremental is FALSE then |
|
68 * U_PARTIAL_MATCH should never be returned. |
|
69 */ |
|
70 virtual UMatchDegree matches(const Replaceable& text, |
|
71 int32_t& offset, |
|
72 int32_t limit, |
|
73 UBool incremental); |
|
74 |
|
75 /** |
|
76 * Implement UnicodeMatcher |
|
77 * @param result Output param to receive the pattern. |
|
78 * @param escapeUnprintable if True then escape the unprintable characters. |
|
79 * @return A reference to 'result'. |
|
80 */ |
|
81 virtual UnicodeString& toPattern(UnicodeString& result, |
|
82 UBool escapeUnprintable = FALSE) const; |
|
83 |
|
84 /** |
|
85 * Implement UnicodeMatcher |
|
86 * @param v the given index value. |
|
87 * @return true if this rule matches the given index value. |
|
88 */ |
|
89 virtual UBool matchesIndexValue(uint8_t v) const; |
|
90 |
|
91 /** |
|
92 * Implement UnicodeMatcher |
|
93 */ |
|
94 virtual void addMatchSetTo(UnicodeSet& toUnionTo) const; |
|
95 |
|
96 /** |
|
97 * UnicodeFunctor API |
|
98 */ |
|
99 virtual void setData(const TransliterationRuleData*); |
|
100 |
|
101 /** |
|
102 * ICU "poor man's RTTI", returns a UClassID for the actual class. |
|
103 */ |
|
104 virtual UClassID getDynamicClassID() const; |
|
105 |
|
106 /** |
|
107 * ICU "poor man's RTTI", returns a UClassID for this class. |
|
108 */ |
|
109 static UClassID U_EXPORT2 getStaticClassID(); |
|
110 |
|
111 private: |
|
112 |
|
113 UnicodeFunctor* matcher; // owned |
|
114 |
|
115 uint32_t minCount; |
|
116 |
|
117 uint32_t maxCount; |
|
118 }; |
|
119 |
|
120 U_NAMESPACE_END |
|
121 |
|
122 #endif /* #if !UCONFIG_NO_TRANSLITERATION */ |
|
123 |
|
124 #endif |