xref: /aosp_15_r20/external/icu/libicu/cts_headers/unicode/ucharstriebuilder.h (revision 0e209d3975ff4a8c132096b14b0e9364a753506e)
1*0e209d39SAndroid Build Coastguard Worker // © 2016 and later: Unicode, Inc. and others.
2*0e209d39SAndroid Build Coastguard Worker // License & terms of use: http://www.unicode.org/copyright.html
3*0e209d39SAndroid Build Coastguard Worker /*
4*0e209d39SAndroid Build Coastguard Worker *******************************************************************************
5*0e209d39SAndroid Build Coastguard Worker *   Copyright (C) 2010-2016, International Business Machines
6*0e209d39SAndroid Build Coastguard Worker *   Corporation and others.  All Rights Reserved.
7*0e209d39SAndroid Build Coastguard Worker *******************************************************************************
8*0e209d39SAndroid Build Coastguard Worker *   file name:  ucharstriebuilder.h
9*0e209d39SAndroid Build Coastguard Worker *   encoding:   UTF-8
10*0e209d39SAndroid Build Coastguard Worker *   tab size:   8 (not used)
11*0e209d39SAndroid Build Coastguard Worker *   indentation:4
12*0e209d39SAndroid Build Coastguard Worker *
13*0e209d39SAndroid Build Coastguard Worker *   created on: 2010nov14
14*0e209d39SAndroid Build Coastguard Worker *   created by: Markus W. Scherer
15*0e209d39SAndroid Build Coastguard Worker */
16*0e209d39SAndroid Build Coastguard Worker 
17*0e209d39SAndroid Build Coastguard Worker #ifndef __UCHARSTRIEBUILDER_H__
18*0e209d39SAndroid Build Coastguard Worker #define __UCHARSTRIEBUILDER_H__
19*0e209d39SAndroid Build Coastguard Worker 
20*0e209d39SAndroid Build Coastguard Worker #include "unicode/utypes.h"
21*0e209d39SAndroid Build Coastguard Worker 
22*0e209d39SAndroid Build Coastguard Worker #if U_SHOW_CPLUSPLUS_API
23*0e209d39SAndroid Build Coastguard Worker 
24*0e209d39SAndroid Build Coastguard Worker #include "unicode/stringtriebuilder.h"
25*0e209d39SAndroid Build Coastguard Worker #include "unicode/ucharstrie.h"
26*0e209d39SAndroid Build Coastguard Worker #include "unicode/unistr.h"
27*0e209d39SAndroid Build Coastguard Worker 
28*0e209d39SAndroid Build Coastguard Worker /**
29*0e209d39SAndroid Build Coastguard Worker  * \file
30*0e209d39SAndroid Build Coastguard Worker  * \brief C++ API: Builder for icu::UCharsTrie
31*0e209d39SAndroid Build Coastguard Worker  */
32*0e209d39SAndroid Build Coastguard Worker 
33*0e209d39SAndroid Build Coastguard Worker U_NAMESPACE_BEGIN
34*0e209d39SAndroid Build Coastguard Worker 
35*0e209d39SAndroid Build Coastguard Worker class UCharsTrieElement;
36*0e209d39SAndroid Build Coastguard Worker 
37*0e209d39SAndroid Build Coastguard Worker /**
38*0e209d39SAndroid Build Coastguard Worker  * Builder class for UCharsTrie.
39*0e209d39SAndroid Build Coastguard Worker  *
40*0e209d39SAndroid Build Coastguard Worker  * This class is not intended for public subclassing.
41*0e209d39SAndroid Build Coastguard Worker  * @stable ICU 4.8
42*0e209d39SAndroid Build Coastguard Worker  */
43*0e209d39SAndroid Build Coastguard Worker class U_COMMON_API UCharsTrieBuilder : public StringTrieBuilder {
44*0e209d39SAndroid Build Coastguard Worker public:
45*0e209d39SAndroid Build Coastguard Worker     /**
46*0e209d39SAndroid Build Coastguard Worker      * Constructs an empty builder.
47*0e209d39SAndroid Build Coastguard Worker      * @param errorCode Standard ICU error code.
48*0e209d39SAndroid Build Coastguard Worker      * @stable ICU 4.8
49*0e209d39SAndroid Build Coastguard Worker      */
50*0e209d39SAndroid Build Coastguard Worker     UCharsTrieBuilder(UErrorCode &errorCode);
51*0e209d39SAndroid Build Coastguard Worker 
52*0e209d39SAndroid Build Coastguard Worker     /**
53*0e209d39SAndroid Build Coastguard Worker      * Destructor.
54*0e209d39SAndroid Build Coastguard Worker      * @stable ICU 4.8
55*0e209d39SAndroid Build Coastguard Worker      */
56*0e209d39SAndroid Build Coastguard Worker     virtual ~UCharsTrieBuilder();
57*0e209d39SAndroid Build Coastguard Worker 
58*0e209d39SAndroid Build Coastguard Worker     /**
59*0e209d39SAndroid Build Coastguard Worker      * Adds a (string, value) pair.
60*0e209d39SAndroid Build Coastguard Worker      * The string must be unique.
61*0e209d39SAndroid Build Coastguard Worker      * The string contents will be copied; the builder does not keep
62*0e209d39SAndroid Build Coastguard Worker      * a reference to the input UnicodeString or its buffer.
63*0e209d39SAndroid Build Coastguard Worker      * @param s The input string.
64*0e209d39SAndroid Build Coastguard Worker      * @param value The value associated with this string.
65*0e209d39SAndroid Build Coastguard Worker      * @param errorCode Standard ICU error code. Its input value must
66*0e209d39SAndroid Build Coastguard Worker      *                  pass the U_SUCCESS() test, or else the function returns
67*0e209d39SAndroid Build Coastguard Worker      *                  immediately. Check for U_FAILURE() on output or use with
68*0e209d39SAndroid Build Coastguard Worker      *                  function chaining. (See User Guide for details.)
69*0e209d39SAndroid Build Coastguard Worker      * @return *this
70*0e209d39SAndroid Build Coastguard Worker      * @stable ICU 4.8
71*0e209d39SAndroid Build Coastguard Worker      */
72*0e209d39SAndroid Build Coastguard Worker     UCharsTrieBuilder &add(const UnicodeString &s, int32_t value, UErrorCode &errorCode);
73*0e209d39SAndroid Build Coastguard Worker 
74*0e209d39SAndroid Build Coastguard Worker     /**
75*0e209d39SAndroid Build Coastguard Worker      * Builds a UCharsTrie for the add()ed data.
76*0e209d39SAndroid Build Coastguard Worker      * Once built, no further data can be add()ed until clear() is called.
77*0e209d39SAndroid Build Coastguard Worker      *
78*0e209d39SAndroid Build Coastguard Worker      * A UCharsTrie cannot be empty. At least one (string, value) pair
79*0e209d39SAndroid Build Coastguard Worker      * must have been add()ed.
80*0e209d39SAndroid Build Coastguard Worker      *
81*0e209d39SAndroid Build Coastguard Worker      * This method passes ownership of the builder's internal result array to the new trie object.
82*0e209d39SAndroid Build Coastguard Worker      * Another call to any build() variant will re-serialize the trie.
83*0e209d39SAndroid Build Coastguard Worker      * After clear() has been called, a new array will be used as well.
84*0e209d39SAndroid Build Coastguard Worker      * @param buildOption Build option, see UStringTrieBuildOption.
85*0e209d39SAndroid Build Coastguard Worker      * @param errorCode Standard ICU error code. Its input value must
86*0e209d39SAndroid Build Coastguard Worker      *                  pass the U_SUCCESS() test, or else the function returns
87*0e209d39SAndroid Build Coastguard Worker      *                  immediately. Check for U_FAILURE() on output or use with
88*0e209d39SAndroid Build Coastguard Worker      *                  function chaining. (See User Guide for details.)
89*0e209d39SAndroid Build Coastguard Worker      * @return A new UCharsTrie for the add()ed data.
90*0e209d39SAndroid Build Coastguard Worker      * @stable ICU 4.8
91*0e209d39SAndroid Build Coastguard Worker      */
92*0e209d39SAndroid Build Coastguard Worker     UCharsTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
93*0e209d39SAndroid Build Coastguard Worker 
94*0e209d39SAndroid Build Coastguard Worker     /**
95*0e209d39SAndroid Build Coastguard Worker      * Builds a UCharsTrie for the add()ed data and char16_t-serializes it.
96*0e209d39SAndroid Build Coastguard Worker      * Once built, no further data can be add()ed until clear() is called.
97*0e209d39SAndroid Build Coastguard Worker      *
98*0e209d39SAndroid Build Coastguard Worker      * A UCharsTrie cannot be empty. At least one (string, value) pair
99*0e209d39SAndroid Build Coastguard Worker      * must have been add()ed.
100*0e209d39SAndroid Build Coastguard Worker      *
101*0e209d39SAndroid Build Coastguard Worker      * Multiple calls to buildUnicodeString() set the UnicodeStrings to the
102*0e209d39SAndroid Build Coastguard Worker      * builder's same char16_t array, without rebuilding.
103*0e209d39SAndroid Build Coastguard Worker      * If buildUnicodeString() is called after build(), the trie will be
104*0e209d39SAndroid Build Coastguard Worker      * re-serialized into a new array (because build() passes on ownership).
105*0e209d39SAndroid Build Coastguard Worker      * If build() is called after buildUnicodeString(), the trie object returned
106*0e209d39SAndroid Build Coastguard Worker      * by build() will become the owner of the underlying data for the
107*0e209d39SAndroid Build Coastguard Worker      * previously returned UnicodeString.
108*0e209d39SAndroid Build Coastguard Worker      * After clear() has been called, a new array will be used as well.
109*0e209d39SAndroid Build Coastguard Worker      * @param buildOption Build option, see UStringTrieBuildOption.
110*0e209d39SAndroid Build Coastguard Worker      * @param result A UnicodeString which will be set to the char16_t-serialized
111*0e209d39SAndroid Build Coastguard Worker      *               UCharsTrie for the add()ed data.
112*0e209d39SAndroid Build Coastguard Worker      * @param errorCode Standard ICU error code. Its input value must
113*0e209d39SAndroid Build Coastguard Worker      *                  pass the U_SUCCESS() test, or else the function returns
114*0e209d39SAndroid Build Coastguard Worker      *                  immediately. Check for U_FAILURE() on output or use with
115*0e209d39SAndroid Build Coastguard Worker      *                  function chaining. (See User Guide for details.)
116*0e209d39SAndroid Build Coastguard Worker      * @return result
117*0e209d39SAndroid Build Coastguard Worker      * @stable ICU 4.8
118*0e209d39SAndroid Build Coastguard Worker      */
119*0e209d39SAndroid Build Coastguard Worker     UnicodeString &buildUnicodeString(UStringTrieBuildOption buildOption, UnicodeString &result,
120*0e209d39SAndroid Build Coastguard Worker                                       UErrorCode &errorCode);
121*0e209d39SAndroid Build Coastguard Worker 
122*0e209d39SAndroid Build Coastguard Worker     /**
123*0e209d39SAndroid Build Coastguard Worker      * Removes all (string, value) pairs.
124*0e209d39SAndroid Build Coastguard Worker      * New data can then be add()ed and a new trie can be built.
125*0e209d39SAndroid Build Coastguard Worker      * @return *this
126*0e209d39SAndroid Build Coastguard Worker      * @stable ICU 4.8
127*0e209d39SAndroid Build Coastguard Worker      */
clear()128*0e209d39SAndroid Build Coastguard Worker     UCharsTrieBuilder &clear() {
129*0e209d39SAndroid Build Coastguard Worker         strings.remove();
130*0e209d39SAndroid Build Coastguard Worker         elementsLength=0;
131*0e209d39SAndroid Build Coastguard Worker         ucharsLength=0;
132*0e209d39SAndroid Build Coastguard Worker         return *this;
133*0e209d39SAndroid Build Coastguard Worker     }
134*0e209d39SAndroid Build Coastguard Worker 
135*0e209d39SAndroid Build Coastguard Worker private:
136*0e209d39SAndroid Build Coastguard Worker     UCharsTrieBuilder(const UCharsTrieBuilder &other) = delete;  // no copy constructor
137*0e209d39SAndroid Build Coastguard Worker     UCharsTrieBuilder &operator=(const UCharsTrieBuilder &other) = delete;  // no assignment operator
138*0e209d39SAndroid Build Coastguard Worker 
139*0e209d39SAndroid Build Coastguard Worker     void buildUChars(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
140*0e209d39SAndroid Build Coastguard Worker 
141*0e209d39SAndroid Build Coastguard Worker     virtual int32_t getElementStringLength(int32_t i) const override;
142*0e209d39SAndroid Build Coastguard Worker     virtual char16_t getElementUnit(int32_t i, int32_t unitIndex) const override;
143*0e209d39SAndroid Build Coastguard Worker     virtual int32_t getElementValue(int32_t i) const override;
144*0e209d39SAndroid Build Coastguard Worker 
145*0e209d39SAndroid Build Coastguard Worker     virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t unitIndex) const override;
146*0e209d39SAndroid Build Coastguard Worker 
147*0e209d39SAndroid Build Coastguard Worker     virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t unitIndex) const override;
148*0e209d39SAndroid Build Coastguard Worker     virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t unitIndex, int32_t count) const override;
149*0e209d39SAndroid Build Coastguard Worker     virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t unitIndex, char16_t unit) const override;
150*0e209d39SAndroid Build Coastguard Worker 
matchNodesCanHaveValues()151*0e209d39SAndroid Build Coastguard Worker     virtual UBool matchNodesCanHaveValues() const override { return true; }
152*0e209d39SAndroid Build Coastguard Worker 
getMaxBranchLinearSubNodeLength()153*0e209d39SAndroid Build Coastguard Worker     virtual int32_t getMaxBranchLinearSubNodeLength() const override { return UCharsTrie::kMaxBranchLinearSubNodeLength; }
getMinLinearMatch()154*0e209d39SAndroid Build Coastguard Worker     virtual int32_t getMinLinearMatch() const override { return UCharsTrie::kMinLinearMatch; }
getMaxLinearMatchLength()155*0e209d39SAndroid Build Coastguard Worker     virtual int32_t getMaxLinearMatchLength() const override { return UCharsTrie::kMaxLinearMatchLength; }
156*0e209d39SAndroid Build Coastguard Worker 
157*0e209d39SAndroid Build Coastguard Worker     class UCTLinearMatchNode : public LinearMatchNode {
158*0e209d39SAndroid Build Coastguard Worker     public:
159*0e209d39SAndroid Build Coastguard Worker         UCTLinearMatchNode(const char16_t *units, int32_t len, Node *nextNode);
160*0e209d39SAndroid Build Coastguard Worker         virtual bool operator==(const Node &other) const override;
161*0e209d39SAndroid Build Coastguard Worker         virtual void write(StringTrieBuilder &builder) override;
162*0e209d39SAndroid Build Coastguard Worker     private:
163*0e209d39SAndroid Build Coastguard Worker         const char16_t *s;
164*0e209d39SAndroid Build Coastguard Worker     };
165*0e209d39SAndroid Build Coastguard Worker 
166*0e209d39SAndroid Build Coastguard Worker     virtual Node *createLinearMatchNode(int32_t i, int32_t unitIndex, int32_t length,
167*0e209d39SAndroid Build Coastguard Worker                                         Node *nextNode) const override;
168*0e209d39SAndroid Build Coastguard Worker 
169*0e209d39SAndroid Build Coastguard Worker     UBool ensureCapacity(int32_t length);
170*0e209d39SAndroid Build Coastguard Worker     virtual int32_t write(int32_t unit) override;
171*0e209d39SAndroid Build Coastguard Worker     int32_t write(const char16_t *s, int32_t length);
172*0e209d39SAndroid Build Coastguard Worker     virtual int32_t writeElementUnits(int32_t i, int32_t unitIndex, int32_t length) override;
173*0e209d39SAndroid Build Coastguard Worker     virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal) override;
174*0e209d39SAndroid Build Coastguard Worker     virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node) override;
175*0e209d39SAndroid Build Coastguard Worker     virtual int32_t writeDeltaTo(int32_t jumpTarget) override;
176*0e209d39SAndroid Build Coastguard Worker 
177*0e209d39SAndroid Build Coastguard Worker     UnicodeString strings;
178*0e209d39SAndroid Build Coastguard Worker     UCharsTrieElement *elements;
179*0e209d39SAndroid Build Coastguard Worker     int32_t elementsCapacity;
180*0e209d39SAndroid Build Coastguard Worker     int32_t elementsLength;
181*0e209d39SAndroid Build Coastguard Worker 
182*0e209d39SAndroid Build Coastguard Worker     // char16_t serialization of the trie.
183*0e209d39SAndroid Build Coastguard Worker     // Grows from the back: ucharsLength measures from the end of the buffer!
184*0e209d39SAndroid Build Coastguard Worker     char16_t *uchars;
185*0e209d39SAndroid Build Coastguard Worker     int32_t ucharsCapacity;
186*0e209d39SAndroid Build Coastguard Worker     int32_t ucharsLength;
187*0e209d39SAndroid Build Coastguard Worker };
188*0e209d39SAndroid Build Coastguard Worker 
189*0e209d39SAndroid Build Coastguard Worker U_NAMESPACE_END
190*0e209d39SAndroid Build Coastguard Worker 
191*0e209d39SAndroid Build Coastguard Worker #endif /* U_SHOW_CPLUSPLUS_API */
192*0e209d39SAndroid Build Coastguard Worker 
193*0e209d39SAndroid Build Coastguard Worker #endif  // __UCHARSTRIEBUILDER_H__
194