icuSources/common/unicode/ucharstriebuilder.h

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