1 /*
2 *******************************************************************************
3 *   Copyright (C) 2010-2014, International Business Machines
4 *   Corporation and others.  All Rights Reserved.
5 *******************************************************************************
6 *   file name:  ucharstriebuilder.h
7 *   encoding:   US-ASCII
8 *   tab size:   8 (not used)
9 *   indentation:4
10 *
11 *   created on: 2010nov14
12 *   created by: Markus W. Scherer
13 */
14 
15 #ifndef __UCHARSTRIEBUILDER_H__
16 #define __UCHARSTRIEBUILDER_H__
17 
18 #include "unicode/utypes.h"
19 #include "unicode/stringtriebuilder.h"
20 #include "unicode/ucharstrie.h"
21 #include "unicode/unistr.h"
22 
23 /**
24  * \file
25  * \brief C++ API: Builder for icu::UCharsTrie
26  */
27 
28 U_NAMESPACE_BEGIN
29 
30 class UCharsTrieElement;
31 
32 /**
33  * Builder class for UCharsTrie.
34  *
35  * This class is not intended for public subclassing.
36  * @stable ICU 4.8
37  */
38 class U_COMMON_API UCharsTrieBuilder : public StringTrieBuilder {
39 public:
40     /**
41      * Constructs an empty builder.
42      * @param errorCode Standard ICU error code.
43      * @stable ICU 4.8
44      */
45     UCharsTrieBuilder(UErrorCode &errorCode);
46 
47     /**
48      * Destructor.
49      * @stable ICU 4.8
50      */
51     virtual ~UCharsTrieBuilder();
52 
53     /**
54      * Adds a (string, value) pair.
55      * The string must be unique.
56      * The string contents will be copied; the builder does not keep
57      * a reference to the input UnicodeString or its buffer.
58      * @param s The input string.
59      * @param value The value associated with this string.
60      * @param errorCode Standard ICU error code. Its input value must
61      *                  pass the U_SUCCESS() test, or else the function returns
62      *                  immediately. Check for U_FAILURE() on output or use with
63      *                  function chaining. (See User Guide for details.)
64      * @return *this
65      * @stable ICU 4.8
66      */
67     UCharsTrieBuilder &add(const UnicodeString &s, int32_t value, UErrorCode &errorCode);
68 
69     /**
70      * Builds a UCharsTrie for the add()ed data.
71      * Once built, no further data can be add()ed until clear() is called.
72      *
73      * A UCharsTrie cannot be empty. At least one (string, value) pair
74      * must have been add()ed.
75      *
76      * This method passes ownership of the builder's internal result array to the new trie object.
77      * Another call to any build() variant will re-serialize the trie.
78      * After clear() has been called, a new array will be used as well.
79      * @param buildOption Build option, see UStringTrieBuildOption.
80      * @param errorCode Standard ICU error code. Its input value must
81      *                  pass the U_SUCCESS() test, or else the function returns
82      *                  immediately. Check for U_FAILURE() on output or use with
83      *                  function chaining. (See User Guide for details.)
84      * @return A new UCharsTrie for the add()ed data.
85      * @stable ICU 4.8
86      */
87     UCharsTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
88 
89     /**
90      * Builds a UCharsTrie for the add()ed data and UChar-serializes it.
91      * Once built, no further data can be add()ed until clear() is called.
92      *
93      * A UCharsTrie cannot be empty. At least one (string, value) pair
94      * must have been add()ed.
95      *
96      * Multiple calls to buildUnicodeString() set the UnicodeStrings to the
97      * builder's same UChar array, without rebuilding.
98      * If buildUnicodeString() is called after build(), the trie will be
99      * re-serialized into a new array.
100      * If build() is called after buildUnicodeString(), the trie object will become
101      * the owner of the previously returned array.
102      * After clear() has been called, a new array will be used as well.
103      * @param buildOption Build option, see UStringTrieBuildOption.
104      * @param result A UnicodeString which will be set to the UChar-serialized
105      *               UCharsTrie for the add()ed data.
106      * @param errorCode Standard ICU error code. Its input value must
107      *                  pass the U_SUCCESS() test, or else the function returns
108      *                  immediately. Check for U_FAILURE() on output or use with
109      *                  function chaining. (See User Guide for details.)
110      * @return result
111      * @stable ICU 4.8
112      */
113     UnicodeString &buildUnicodeString(UStringTrieBuildOption buildOption, UnicodeString &result,
114                                       UErrorCode &errorCode);
115 
116     /**
117      * Removes all (string, value) pairs.
118      * New data can then be add()ed and a new trie can be built.
119      * @return *this
120      * @stable ICU 4.8
121      */
clear()122     UCharsTrieBuilder &clear() {
123         strings.remove();
124         elementsLength=0;
125         ucharsLength=0;
126         return *this;
127     }
128 
129 private:
130     UCharsTrieBuilder(const UCharsTrieBuilder &other);  // no copy constructor
131     UCharsTrieBuilder &operator=(const UCharsTrieBuilder &other);  // no assignment operator
132 
133     void buildUChars(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
134 
135     virtual int32_t getElementStringLength(int32_t i) const;
136     virtual UChar getElementUnit(int32_t i, int32_t unitIndex) const;
137     virtual int32_t getElementValue(int32_t i) const;
138 
139     virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t unitIndex) const;
140 
141     virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t unitIndex) const;
142     virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t unitIndex, int32_t count) const;
143     virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t unitIndex, UChar unit) const;
144 
matchNodesCanHaveValues()145     virtual UBool matchNodesCanHaveValues() const { return TRUE; }
146 
getMaxBranchLinearSubNodeLength()147     virtual int32_t getMaxBranchLinearSubNodeLength() const { return UCharsTrie::kMaxBranchLinearSubNodeLength; }
getMinLinearMatch()148     virtual int32_t getMinLinearMatch() const { return UCharsTrie::kMinLinearMatch; }
getMaxLinearMatchLength()149     virtual int32_t getMaxLinearMatchLength() const { return UCharsTrie::kMaxLinearMatchLength; }
150 
151     class UCTLinearMatchNode : public LinearMatchNode {
152     public:
153         UCTLinearMatchNode(const UChar *units, int32_t len, Node *nextNode);
154         virtual UBool operator==(const Node &other) const;
155         virtual void write(StringTrieBuilder &builder);
156     private:
157         const UChar *s;
158     };
159 
160     virtual Node *createLinearMatchNode(int32_t i, int32_t unitIndex, int32_t length,
161                                         Node *nextNode) const;
162 
163     UBool ensureCapacity(int32_t length);
164     virtual int32_t write(int32_t unit);
165     int32_t write(const UChar *s, int32_t length);
166     virtual int32_t writeElementUnits(int32_t i, int32_t unitIndex, int32_t length);
167     virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal);
168     virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node);
169     virtual int32_t writeDeltaTo(int32_t jumpTarget);
170 
171     UnicodeString strings;
172     UCharsTrieElement *elements;
173     int32_t elementsCapacity;
174     int32_t elementsLength;
175 
176     // UChar serialization of the trie.
177     // Grows from the back: ucharsLength measures from the end of the buffer!
178     UChar *uchars;
179     int32_t ucharsCapacity;
180     int32_t ucharsLength;
181 };
182 
183 U_NAMESPACE_END
184 
185 #endif  // __UCHARSTRIEBUILDER_H__
186