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 #include "unicode/bytestrie.h"
  27 #include "unicode/stringpiece.h"
  28 #include "unicode/stringtriebuilder.h"
  29
  30 #if U_SHOW_CPLUSPLUS_API
  31 U_NAMESPACE_BEGIN
  32
  33 class BytesTrieElement;
  34 class CharString;
  35 /**
  36  * Builder class for BytesTrie.
  37  *
  38  * This class is not intended for public subclassing.
  39  * @stable ICU 4.8
  40  */
  41 class U_COMMON_API BytesTrieBuilder : public StringTrieBuilder {
  42 public:
  43     /**
  44      * Constructs an empty builder.
  45      * @param errorCode Standard ICU error code.
  46      * @stable ICU 4.8
  47      */
  48     BytesTrieBuilder(UErrorCode &errorCode);
  49
  50     /**
  51      * Destructor.
  52      * @stable ICU 4.8
  53      */
  54     virtual ~BytesTrieBuilder();
  55
  56     /**
  57      * Adds a (byte sequence, value) pair.
  58      * The byte sequence must be unique.
  59      * The bytes will be copied; the builder does not keep
  60      * a reference to the input StringPiece or its data().
  61      * @param s The input byte sequence.
  62      * @param value The value associated with this byte sequence.
  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     BytesTrieBuilder &add(StringPiece s, int32_t value, UErrorCode &errorCode);
  71
  72     /**
  73      * Builds a BytesTrie for the add()ed data.
  74      * Once built, no further data can be add()ed until clear() is called.
  75      *
  76      * A BytesTrie cannot be empty. At least one (byte sequence, 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 BytesTrie for the add()ed data.
  88      * @stable ICU 4.8
  89      */
  90     BytesTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
  91
  92     /**
  93      * Builds a BytesTrie for the add()ed data and byte-serializes it.
  94      * Once built, no further data can be add()ed until clear() is called.
  95      *
  96      * A BytesTrie cannot be empty. At least one (byte sequence, value) pair
  97      * must have been add()ed.
  98      *
  99      * Multiple calls to buildStringPiece() return StringPieces referring to the
 100      * builder's same byte array, without rebuilding.
 101      * If buildStringPiece() is called after build(), the trie will be
 102      * re-serialized into a new array.
 103      * If build() is called after buildStringPiece(), 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 errorCode Standard ICU error code. Its input value must
 108      *                  pass the U_SUCCESS() test, or else the function returns
 109      *                  immediately. Check for U_FAILURE() on output or use with
 110      *                  function chaining. (See User Guide for details.)
 111      * @return A StringPiece which refers to the byte-serialized BytesTrie for the add()ed data.
 112      * @stable ICU 4.8
 113      */
 114     StringPiece buildStringPiece(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
 115
 116     /**
 117      * Removes all (byte sequence, 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      */
 122     BytesTrieBuilder &clear();
 123
 124 private:
 125     BytesTrieBuilder(const BytesTrieBuilder &other);  // no copy constructor
 126     BytesTrieBuilder &operator=(const BytesTrieBuilder &other);  // no assignment operator
 127
 128     void buildBytes(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
 129
 130     virtual int32_t getElementStringLength(int32_t i) const;
 131     virtual char16_t getElementUnit(int32_t i, int32_t byteIndex) const;
 132     virtual int32_t getElementValue(int32_t i) const;
 133
 134     virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t byteIndex) const;
 135
 136     virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t byteIndex) const;
 137     virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t byteIndex, int32_t count) const;
 138     virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t byteIndex, char16_t byte) const;
 139
 140     virtual UBool matchNodesCanHaveValues() const { return FALSE; }
 141
 142     virtual int32_t getMaxBranchLinearSubNodeLength() const { return BytesTrie::kMaxBranchLinearSubNodeLength; }
 143     virtual int32_t getMinLinearMatch() const { return BytesTrie::kMinLinearMatch; }
 144     virtual int32_t getMaxLinearMatchLength() const { return BytesTrie::kMaxLinearMatchLength; }
 145
 146     /**
 147      * @internal (private)
 148      */
 149     class BTLinearMatchNode : public LinearMatchNode {
 150     public:
 151         BTLinearMatchNode(const char *units, int32_t len, Node *nextNode);
 152         virtual UBool operator==(const Node &other) const;
 153         virtual void write(StringTrieBuilder &builder);
 154     private:
 155         const char *s;
 156     };
 157
 158     virtual Node *createLinearMatchNode(int32_t i, int32_t byteIndex, int32_t length,
 159                                         Node *nextNode) const;
 160
 161     UBool ensureCapacity(int32_t length);
 162     virtual int32_t write(int32_t byte);
 163     int32_t write(const char *b, int32_t length);
 164     virtual int32_t writeElementUnits(int32_t i, int32_t byteIndex, int32_t length);
 165     virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal);
 166     virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node);
 167     virtual int32_t writeDeltaTo(int32_t jumpTarget);
 168
 169     CharString *strings;  // Pointer not object so we need not #include internal charstr.h.
 170     BytesTrieElement *elements;
 171     int32_t elementsCapacity;
 172     int32_t elementsLength;
 173
 174     // Byte serialization of the trie.
 175     // Grows from the back: bytesLength measures from the end of the buffer!
 176     char *bytes;
 177     int32_t bytesCapacity;
 178     int32_t bytesLength;
 179 };
 180
 181 U_NAMESPACE_END
 182 #endif // U_SHOW_CPLUSPLUS_API
 183
 184 #endif  // __BYTESTRIEBUILDER_H__