icuSources/common/unicode/bytestriebuilder.h

   1 /*
   2 *******************************************************************************
   3 *   Copyright (C) 2010-2012, International Business Machines
   4 *   Corporation and others.  All Rights Reserved.
   5 *******************************************************************************
   6 *   file name:  bytestriebuilder.h
   7 *   encoding:   US-ASCII
   8 *   tab size:   8 (not used)
   9 *   indentation:4
  10 *
  11 *   created on: 2010sep25
  12 *   created by: Markus W. Scherer
  13 */
  14
  15 /**
  16  * \file
  17  * \brief C++ API: Builder for icu::BytesTrie
  18  */
  19
  20 #ifndef __BYTESTRIEBUILDER_H__
  21 #define __BYTESTRIEBUILDER_H__
  22
  23 #include "unicode/utypes.h"
  24 #include "unicode/bytestrie.h"
  25 #include "unicode/stringpiece.h"
  26 #include "unicode/stringtriebuilder.h"
  27
  28 U_NAMESPACE_BEGIN
  29
  30 class BytesTrieElement;
  31 class CharString;
  32
  33 /**
  34  * Builder class for BytesTrie.
  35  *
  36  * This class is not intended for public subclassing.
  37  * @stable ICU 4.8
  38  */
  39 class U_COMMON_API BytesTrieBuilder : public StringTrieBuilder {
  40 public:
  41     /**
  42      * Constructs an empty builder.
  43      * @param errorCode Standard ICU error code.
  44      * @stable ICU 4.8
  45      */
  46     BytesTrieBuilder(UErrorCode &errorCode);
  47
  48     /**
  49      * Destructor.
  50      * @stable ICU 4.8
  51      */
  52     virtual ~BytesTrieBuilder();
  53
  54     /**
  55      * Adds a (byte sequence, value) pair.
  56      * The byte sequence must be unique.
  57      * The bytes will be copied; the builder does not keep
  58      * a reference to the input StringPiece or its data().
  59      * @param s The input byte sequence.
  60      * @param value The value associated with this byte sequence.
  61      * @param errorCode Standard ICU error code. Its input value must
  62      *                  pass the U_SUCCESS() test, or else the function returns
  63      *                  immediately. Check for U_FAILURE() on output or use with
  64      *                  function chaining. (See User Guide for details.)
  65      * @return *this
  66      * @stable ICU 4.8
  67      */
  68     BytesTrieBuilder &add(const StringPiece &s, int32_t value, UErrorCode &errorCode);
  69
  70     /**
  71      * Builds a BytesTrie for the add()ed data.
  72      * Once built, no further data can be add()ed until clear() is called.
  73      *
  74      * This method passes ownership of the builder's internal result array to the new trie object.
  75      * Another call to any build() variant will re-serialize the trie.
  76      * After clear() has been called, a new array will be used as well.
  77      * @param buildOption Build option, see UStringTrieBuildOption.
  78      * @param errorCode Standard ICU error code. Its input value must
  79      *                  pass the U_SUCCESS() test, or else the function returns
  80      *                  immediately. Check for U_FAILURE() on output or use with
  81      *                  function chaining. (See User Guide for details.)
  82      * @return A new BytesTrie for the add()ed data.
  83      * @stable ICU 4.8
  84      */
  85     BytesTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
  86
  87     /**
  88      * Builds a BytesTrie for the add()ed data and byte-serializes it.
  89      * Once built, no further data can be add()ed until clear() is called.
  90      *
  91      * Multiple calls to buildStringPiece() return StringPieces referring to the
  92      * builder's same byte array, without rebuilding.
  93      * If buildStringPiece() is called after build(), the trie will be
  94      * re-serialized into a new array.
  95      * If build() is called after buildStringPiece(), the trie object will become
  96      * the owner of the previously returned array.
  97      * After clear() has been called, a new array will be used as well.
  98      * @param buildOption Build option, see UStringTrieBuildOption.
  99      * @param errorCode Standard ICU error code. Its input value must
 100      *                  pass the U_SUCCESS() test, or else the function returns
 101      *                  immediately. Check for U_FAILURE() on output or use with
 102      *                  function chaining. (See User Guide for details.)
 103      * @return A StringPiece which refers to the byte-serialized BytesTrie for the add()ed data.
 104      * @stable ICU 4.8
 105      */
 106     StringPiece buildStringPiece(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
 107
 108     /**
 109      * Removes all (byte sequence, value) pairs.
 110      * New data can then be add()ed and a new trie can be built.
 111      * @return *this
 112      * @stable ICU 4.8
 113      */
 114     BytesTrieBuilder &clear();
 115
 116 private:
 117     BytesTrieBuilder(const BytesTrieBuilder &other);  // no copy constructor
 118     BytesTrieBuilder &operator=(const BytesTrieBuilder &other);  // no assignment operator
 119
 120     void buildBytes(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
 121
 122     virtual int32_t getElementStringLength(int32_t i) const;
 123     virtual UChar getElementUnit(int32_t i, int32_t byteIndex) const;
 124     virtual int32_t getElementValue(int32_t i) const;
 125
 126     virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t byteIndex) const;
 127
 128     virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t byteIndex) const;
 129     virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t byteIndex, int32_t count) const;
 130     virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t byteIndex, UChar byte) const;
 131
 132     virtual UBool matchNodesCanHaveValues() const { return FALSE; }
 133
 134     virtual int32_t getMaxBranchLinearSubNodeLength() const { return BytesTrie::kMaxBranchLinearSubNodeLength; }
 135     virtual int32_t getMinLinearMatch() const { return BytesTrie::kMinLinearMatch; }
 136     virtual int32_t getMaxLinearMatchLength() const { return BytesTrie::kMaxLinearMatchLength; }
 137
 138 #ifndef U_HIDE_INTERNAL_API
 139     /**
 140      * @internal
 141      */
 142     class BTLinearMatchNode : public LinearMatchNode {
 143     public:
 144         BTLinearMatchNode(const char *units, int32_t len, Node *nextNode);
 145         virtual UBool operator==(const Node &other) const;
 146         virtual void write(StringTrieBuilder &builder);
 147     private:
 148         const char *s;
 149     };
 150 #endif  /* U_HIDE_INTERNAL_API */
 151
 152     virtual Node *createLinearMatchNode(int32_t i, int32_t byteIndex, int32_t length,
 153                                         Node *nextNode) const;
 154
 155     UBool ensureCapacity(int32_t length);
 156     virtual int32_t write(int32_t byte);
 157     int32_t write(const char *b, int32_t length);
 158     virtual int32_t writeElementUnits(int32_t i, int32_t byteIndex, int32_t length);
 159     virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal);
 160     virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node);
 161     virtual int32_t writeDeltaTo(int32_t jumpTarget);
 162
 163     CharString *strings;  // Pointer not object so we need not #include internal charstr.h.
 164     BytesTrieElement *elements;
 165     int32_t elementsCapacity;
 166     int32_t elementsLength;
 167
 168     // Byte serialization of the trie.
 169     // Grows from the back: bytesLength measures from the end of the buffer!
 170     char *bytes;
 171     int32_t bytesCapacity;
 172     int32_t bytesLength;
 173 };
 174
 175 U_NAMESPACE_END
 176
 177 #endif  // __BYTESTRIEBUILDER_H__