icuSources/common/unicode/bytestriebuilder.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:  bytestriebuilder.h
   9 *   encoding:   UTF-8
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 2010sep25
  14 *   created by: Markus W. Scherer
  15 */
  16
  17 /**
  18  * \file
  19  * \brief C++ API: Builder for icu::BytesTrie
  20  */
  21
  22 #ifndef __BYTESTRIEBUILDER_H__
  23 #define __BYTESTRIEBUILDER_H__
  24
  25 #include "unicode/utypes.h"
  26
  27 #if U_SHOW_CPLUSPLUS_API
  28
  29 #include "unicode/bytestrie.h"
  30 #include "unicode/stringpiece.h"
  31 #include "unicode/stringtriebuilder.h"
  32
  33 U_NAMESPACE_BEGIN
  34
  35 class BytesTrieElement;
  36 class CharString;
  37 /**
  38  * Builder class for BytesTrie.
  39  *
  40  * This class is not intended for public subclassing.
  41  * @stable ICU 4.8
  42  */
  43 class U_COMMON_API BytesTrieBuilder : public StringTrieBuilder {
  44 public:
  45     /**
  46      * Constructs an empty builder.
  47      * @param errorCode Standard ICU error code.
  48      * @stable ICU 4.8
  49      */
  50     BytesTrieBuilder(UErrorCode &errorCode);
  51
  52     /**
  53      * Destructor.
  54      * @stable ICU 4.8
  55      */
  56     virtual ~BytesTrieBuilder();
  57
  58     /**
  59      * Adds a (byte sequence, value) pair.
  60      * The byte sequence must be unique.
  61      * The bytes will be copied; the builder does not keep
  62      * a reference to the input StringPiece or its data().
  63      * @param s The input byte sequence.
  64      * @param value The value associated with this byte sequence.
  65      * @param errorCode Standard ICU error code. Its input value must
  66      *                  pass the U_SUCCESS() test, or else the function returns
  67      *                  immediately. Check for U_FAILURE() on output or use with
  68      *                  function chaining. (See User Guide for details.)
  69      * @return *this
  70      * @stable ICU 4.8
  71      */
  72     BytesTrieBuilder &add(StringPiece s, int32_t value, UErrorCode &errorCode);
  73
  74     /**
  75      * Builds a BytesTrie for the add()ed data.
  76      * Once built, no further data can be add()ed until clear() is called.
  77      *
  78      * A BytesTrie cannot be empty. At least one (byte sequence, value) pair
  79      * must have been add()ed.
  80      *
  81      * This method passes ownership of the builder's internal result array to the new trie object.
  82      * Another call to any build() variant will re-serialize the trie.
  83      * After clear() has been called, a new array will be used as well.
  84      * @param buildOption Build option, see UStringTrieBuildOption.
  85      * @param errorCode Standard ICU error code. Its input value must
  86      *                  pass the U_SUCCESS() test, or else the function returns
  87      *                  immediately. Check for U_FAILURE() on output or use with
  88      *                  function chaining. (See User Guide for details.)
  89      * @return A new BytesTrie for the add()ed data.
  90      * @stable ICU 4.8
  91      */
  92     BytesTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
  93
  94     /**
  95      * Builds a BytesTrie for the add()ed data and byte-serializes it.
  96      * Once built, no further data can be add()ed until clear() is called.
  97      *
  98      * A BytesTrie cannot be empty. At least one (byte sequence, value) pair
  99      * must have been add()ed.
 100      *
 101      * Multiple calls to buildStringPiece() return StringPieces referring to the
 102      * builder's same byte array, without rebuilding.
 103      * If buildStringPiece() is called after build(), the trie will be
 104      * re-serialized into a new array.
 105      * If build() is called after buildStringPiece(), the trie object will become
 106      * the owner of the previously returned array.
 107      * After clear() has been called, a new array will be used as well.
 108      * @param buildOption Build option, see UStringTrieBuildOption.
 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 A StringPiece which refers to the byte-serialized BytesTrie for the add()ed data.
 114      * @stable ICU 4.8
 115      */
 116     StringPiece buildStringPiece(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
 117
 118     /**
 119      * Removes all (byte sequence, value) pairs.
 120      * New data can then be add()ed and a new trie can be built.
 121      * @return *this
 122      * @stable ICU 4.8
 123      */
 124     BytesTrieBuilder &clear();
 125
 126 private:
 127     BytesTrieBuilder(const BytesTrieBuilder &other);  // no copy constructor
 128     BytesTrieBuilder &operator=(const BytesTrieBuilder &other);  // no assignment operator
 129
 130     void buildBytes(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
 131
 132     virtual int32_t getElementStringLength(int32_t i) const;
 133     virtual char16_t getElementUnit(int32_t i, int32_t byteIndex) const;
 134     virtual int32_t getElementValue(int32_t i) const;
 135
 136     virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t byteIndex) const;
 137
 138     virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t byteIndex) const;
 139     virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t byteIndex, int32_t count) const;
 140     virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t byteIndex, char16_t byte) const;
 141
 142     virtual UBool matchNodesCanHaveValues() const { return FALSE; }
 143
 144     virtual int32_t getMaxBranchLinearSubNodeLength() const { return BytesTrie::kMaxBranchLinearSubNodeLength; }
 145     virtual int32_t getMinLinearMatch() const { return BytesTrie::kMinLinearMatch; }
 146     virtual int32_t getMaxLinearMatchLength() const { return BytesTrie::kMaxLinearMatchLength; }
 147
 148     /**
 149      * @internal (private)
 150      */
 151     class BTLinearMatchNode : public LinearMatchNode {
 152     public:
 153         BTLinearMatchNode(const char *units, int32_t len, Node *nextNode);
 154         virtual UBool operator==(const Node &other) const;
 155         virtual void write(StringTrieBuilder &builder);
 156     private:
 157         const char *s;
 158     };
 159
 160     virtual Node *createLinearMatchNode(int32_t i, int32_t byteIndex, int32_t length,
 161                                         Node *nextNode) const;
 162
 163     UBool ensureCapacity(int32_t length);
 164     virtual int32_t write(int32_t byte);
 165     int32_t write(const char *b, int32_t length);
 166     virtual int32_t writeElementUnits(int32_t i, int32_t byteIndex, 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     CharString *strings;  // Pointer not object so we need not #include internal charstr.h.
 172     BytesTrieElement *elements;
 173     int32_t elementsCapacity;
 174     int32_t elementsLength;
 175
 176     // Byte serialization of the trie.
 177     // Grows from the back: bytesLength measures from the end of the buffer!
 178     char *bytes;
 179     int32_t bytesCapacity;
 180     int32_t bytesLength;
 181 };
 182
 183 U_NAMESPACE_END
 184
 185 #endif /* U_SHOW_CPLUSPLUS_API */
 186
 187 #endif  // __BYTESTRIEBUILDER_H__