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