icuSources/common/unicode/ucharstriebuilder.h

   1 /*
   2 *******************************************************************************
   3 *   Copyright (C) 2010-2012, 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 U_NAMESPACE_BEGIN
  24
  25 class UCharsTrieElement;
  26
  27 /**
  28  * Builder class for UCharsTrie.
  29  *
  30  * This class is not intended for public subclassing.
  31  * @stable ICU 4.8
  32  */
  33 class U_COMMON_API UCharsTrieBuilder : public StringTrieBuilder {
  34 public:
  35     /**
  36      * Constructs an empty builder.
  37      * @param errorCode Standard ICU error code.
  38      * @stable ICU 4.8
  39      */
  40     UCharsTrieBuilder(UErrorCode &errorCode);
  41
  42     /**
  43      * Destructor.
  44      * @stable ICU 4.8
  45      */
  46     virtual ~UCharsTrieBuilder();
  47
  48     /**
  49      * Adds a (string, value) pair.
  50      * The string must be unique.
  51      * The string contents will be copied; the builder does not keep
  52      * a reference to the input UnicodeString or its buffer.
  53      * @param s The input string.
  54      * @param value The value associated with this string.
  55      * @param errorCode Standard ICU error code. Its input value must
  56      *                  pass the U_SUCCESS() test, or else the function returns
  57      *                  immediately. Check for U_FAILURE() on output or use with
  58      *                  function chaining. (See User Guide for details.)
  59      * @return *this
  60      * @stable ICU 4.8
  61      */
  62     UCharsTrieBuilder &add(const UnicodeString &s, int32_t value, UErrorCode &errorCode);
  63
  64     /**
  65      * Builds a UCharsTrie for the add()ed data.
  66      * Once built, no further data can be add()ed until clear() is called.
  67      *
  68      * This method passes ownership of the builder's internal result array to the new trie object.
  69      * Another call to any build() variant will re-serialize the trie.
  70      * After clear() has been called, a new array will be used as well.
  71      * @param buildOption Build option, see UStringTrieBuildOption.
  72      * @param errorCode Standard ICU error code. Its input value must
  73      *                  pass the U_SUCCESS() test, or else the function returns
  74      *                  immediately. Check for U_FAILURE() on output or use with
  75      *                  function chaining. (See User Guide for details.)
  76      * @return A new UCharsTrie for the add()ed data.
  77      * @stable ICU 4.8
  78      */
  79     UCharsTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
  80
  81     /**
  82      * Builds a UCharsTrie for the add()ed data and UChar-serializes it.
  83      * Once built, no further data can be add()ed until clear() is called.
  84      *
  85      * Multiple calls to buildUnicodeString() set the UnicodeStrings to the
  86      * builder's same UChar array, without rebuilding.
  87      * If buildUnicodeString() is called after build(), the trie will be
  88      * re-serialized into a new array.
  89      * If build() is called after buildUnicodeString(), the trie object will become
  90      * the owner of the previously returned array.
  91      * After clear() has been called, a new array will be used as well.
  92      * @param buildOption Build option, see UStringTrieBuildOption.
  93      * @param result A UnicodeString which will be set to the UChar-serialized
  94      *               UCharsTrie for the add()ed data.
  95      * @param errorCode Standard ICU error code. Its input value must
  96      *                  pass the U_SUCCESS() test, or else the function returns
  97      *                  immediately. Check for U_FAILURE() on output or use with
  98      *                  function chaining. (See User Guide for details.)
  99      * @return result
 100      * @stable ICU 4.8
 101      */
 102     UnicodeString &buildUnicodeString(UStringTrieBuildOption buildOption, UnicodeString &result,
 103                                       UErrorCode &errorCode);
 104
 105     /**
 106      * Removes all (string, value) pairs.
 107      * New data can then be add()ed and a new trie can be built.
 108      * @return *this
 109      * @stable ICU 4.8
 110      */
 111     UCharsTrieBuilder &clear() {
 112         strings.remove();
 113         elementsLength=0;
 114         ucharsLength=0;
 115         return *this;
 116     }
 117
 118 private:
 119     UCharsTrieBuilder(const UCharsTrieBuilder &other);  // no copy constructor
 120     UCharsTrieBuilder &operator=(const UCharsTrieBuilder &other);  // no assignment operator
 121
 122     void buildUChars(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
 123
 124     virtual int32_t getElementStringLength(int32_t i) const;
 125     virtual UChar getElementUnit(int32_t i, int32_t unitIndex) const;
 126     virtual int32_t getElementValue(int32_t i) const;
 127
 128     virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t unitIndex) const;
 129
 130     virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t unitIndex) const;
 131     virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t unitIndex, int32_t count) const;
 132     virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t unitIndex, UChar unit) const;
 133
 134     virtual UBool matchNodesCanHaveValues() const { return TRUE; }
 135
 136     virtual int32_t getMaxBranchLinearSubNodeLength() const { return UCharsTrie::kMaxBranchLinearSubNodeLength; }
 137     virtual int32_t getMinLinearMatch() const { return UCharsTrie::kMinLinearMatch; }
 138     virtual int32_t getMaxLinearMatchLength() const { return UCharsTrie::kMaxLinearMatchLength; }
 139
 140 #ifndef U_HIDE_INTERNAL_API
 141     class UCTLinearMatchNode : public LinearMatchNode {
 142     public:
 143         UCTLinearMatchNode(const UChar *units, int32_t len, Node *nextNode);
 144         virtual UBool operator==(const Node &other) const;
 145         virtual void write(StringTrieBuilder &builder);
 146     private:
 147         const UChar *s;
 148     };
 149 #endif
 150
 151     virtual Node *createLinearMatchNode(int32_t i, int32_t unitIndex, int32_t length,
 152                                         Node *nextNode) const;
 153
 154     UBool ensureCapacity(int32_t length);
 155     virtual int32_t write(int32_t unit);
 156     int32_t write(const UChar *s, int32_t length);
 157     virtual int32_t writeElementUnits(int32_t i, int32_t unitIndex, int32_t length);
 158     virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal);
 159     virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node);
 160     virtual int32_t writeDeltaTo(int32_t jumpTarget);
 161
 162     UnicodeString strings;
 163     UCharsTrieElement *elements;
 164     int32_t elementsCapacity;
 165     int32_t elementsLength;
 166
 167     // UChar serialization of the trie.
 168     // Grows from the back: ucharsLength measures from the end of the buffer!
 169     UChar *uchars;
 170     int32_t ucharsCapacity;
 171     int32_t ucharsLength;
 172 };
 173
 174 U_NAMESPACE_END
 175
 176 #endif  // __UCHARSTRIEBUILDER_H__