1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/unicode/ucharstriebuilder.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,187 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +* Copyright (C) 2010-2013, International Business Machines
1.7 +* Corporation and others. All Rights Reserved.
1.8 +*******************************************************************************
1.9 +* file name: ucharstriebuilder.h
1.10 +* encoding: US-ASCII
1.11 +* tab size: 8 (not used)
1.12 +* indentation:4
1.13 +*
1.14 +* created on: 2010nov14
1.15 +* created by: Markus W. Scherer
1.16 +*/
1.17 +
1.18 +#ifndef __UCHARSTRIEBUILDER_H__
1.19 +#define __UCHARSTRIEBUILDER_H__
1.20 +
1.21 +#include "unicode/utypes.h"
1.22 +#include "unicode/stringtriebuilder.h"
1.23 +#include "unicode/ucharstrie.h"
1.24 +#include "unicode/unistr.h"
1.25 +
1.26 +/**
1.27 + * \file
1.28 + * \brief C++ API: Builder for icu::UCharsTrie
1.29 + */
1.30 +
1.31 +U_NAMESPACE_BEGIN
1.32 +
1.33 +class UCharsTrieElement;
1.34 +
1.35 +/**
1.36 + * Builder class for UCharsTrie.
1.37 + *
1.38 + * This class is not intended for public subclassing.
1.39 + * @stable ICU 4.8
1.40 + */
1.41 +class U_COMMON_API UCharsTrieBuilder : public StringTrieBuilder {
1.42 +public:
1.43 + /**
1.44 + * Constructs an empty builder.
1.45 + * @param errorCode Standard ICU error code.
1.46 + * @stable ICU 4.8
1.47 + */
1.48 + UCharsTrieBuilder(UErrorCode &errorCode);
1.49 +
1.50 + /**
1.51 + * Destructor.
1.52 + * @stable ICU 4.8
1.53 + */
1.54 + virtual ~UCharsTrieBuilder();
1.55 +
1.56 + /**
1.57 + * Adds a (string, value) pair.
1.58 + * The string must be unique.
1.59 + * The string contents will be copied; the builder does not keep
1.60 + * a reference to the input UnicodeString or its buffer.
1.61 + * @param s The input string.
1.62 + * @param value The value associated with this string.
1.63 + * @param errorCode Standard ICU error code. Its input value must
1.64 + * pass the U_SUCCESS() test, or else the function returns
1.65 + * immediately. Check for U_FAILURE() on output or use with
1.66 + * function chaining. (See User Guide for details.)
1.67 + * @return *this
1.68 + * @stable ICU 4.8
1.69 + */
1.70 + UCharsTrieBuilder &add(const UnicodeString &s, int32_t value, UErrorCode &errorCode);
1.71 +
1.72 + /**
1.73 + * Builds a UCharsTrie for the add()ed data.
1.74 + * Once built, no further data can be add()ed until clear() is called.
1.75 + *
1.76 + * A UCharsTrie cannot be empty. At least one (string, value) pair
1.77 + * must have been add()ed.
1.78 + *
1.79 + * This method passes ownership of the builder's internal result array to the new trie object.
1.80 + * Another call to any build() variant will re-serialize the trie.
1.81 + * After clear() has been called, a new array will be used as well.
1.82 + * @param buildOption Build option, see UStringTrieBuildOption.
1.83 + * @param errorCode Standard ICU error code. Its input value must
1.84 + * pass the U_SUCCESS() test, or else the function returns
1.85 + * immediately. Check for U_FAILURE() on output or use with
1.86 + * function chaining. (See User Guide for details.)
1.87 + * @return A new UCharsTrie for the add()ed data.
1.88 + * @stable ICU 4.8
1.89 + */
1.90 + UCharsTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
1.91 +
1.92 + /**
1.93 + * Builds a UCharsTrie for the add()ed data and UChar-serializes it.
1.94 + * Once built, no further data can be add()ed until clear() is called.
1.95 + *
1.96 + * A UCharsTrie cannot be empty. At least one (string, value) pair
1.97 + * must have been add()ed.
1.98 + *
1.99 + * Multiple calls to buildUnicodeString() set the UnicodeStrings to the
1.100 + * builder's same UChar array, without rebuilding.
1.101 + * If buildUnicodeString() is called after build(), the trie will be
1.102 + * re-serialized into a new array.
1.103 + * If build() is called after buildUnicodeString(), the trie object will become
1.104 + * the owner of the previously returned array.
1.105 + * After clear() has been called, a new array will be used as well.
1.106 + * @param buildOption Build option, see UStringTrieBuildOption.
1.107 + * @param result A UnicodeString which will be set to the UChar-serialized
1.108 + * UCharsTrie for the add()ed data.
1.109 + * @param errorCode Standard ICU error code. Its input value must
1.110 + * pass the U_SUCCESS() test, or else the function returns
1.111 + * immediately. Check for U_FAILURE() on output or use with
1.112 + * function chaining. (See User Guide for details.)
1.113 + * @return result
1.114 + * @stable ICU 4.8
1.115 + */
1.116 + UnicodeString &buildUnicodeString(UStringTrieBuildOption buildOption, UnicodeString &result,
1.117 + UErrorCode &errorCode);
1.118 +
1.119 + /**
1.120 + * Removes all (string, value) pairs.
1.121 + * New data can then be add()ed and a new trie can be built.
1.122 + * @return *this
1.123 + * @stable ICU 4.8
1.124 + */
1.125 + UCharsTrieBuilder &clear() {
1.126 + strings.remove();
1.127 + elementsLength=0;
1.128 + ucharsLength=0;
1.129 + return *this;
1.130 + }
1.131 +
1.132 +private:
1.133 + UCharsTrieBuilder(const UCharsTrieBuilder &other); // no copy constructor
1.134 + UCharsTrieBuilder &operator=(const UCharsTrieBuilder &other); // no assignment operator
1.135 +
1.136 + void buildUChars(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
1.137 +
1.138 + virtual int32_t getElementStringLength(int32_t i) const;
1.139 + virtual UChar getElementUnit(int32_t i, int32_t unitIndex) const;
1.140 + virtual int32_t getElementValue(int32_t i) const;
1.141 +
1.142 + virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t unitIndex) const;
1.143 +
1.144 + virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t unitIndex) const;
1.145 + virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t unitIndex, int32_t count) const;
1.146 + virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t unitIndex, UChar unit) const;
1.147 +
1.148 + virtual UBool matchNodesCanHaveValues() const { return TRUE; }
1.149 +
1.150 + virtual int32_t getMaxBranchLinearSubNodeLength() const { return UCharsTrie::kMaxBranchLinearSubNodeLength; }
1.151 + virtual int32_t getMinLinearMatch() const { return UCharsTrie::kMinLinearMatch; }
1.152 + virtual int32_t getMaxLinearMatchLength() const { return UCharsTrie::kMaxLinearMatchLength; }
1.153 +
1.154 +#ifndef U_HIDE_INTERNAL_API
1.155 + class UCTLinearMatchNode : public LinearMatchNode {
1.156 + public:
1.157 + UCTLinearMatchNode(const UChar *units, int32_t len, Node *nextNode);
1.158 + virtual UBool operator==(const Node &other) const;
1.159 + virtual void write(StringTrieBuilder &builder);
1.160 + private:
1.161 + const UChar *s;
1.162 + };
1.163 +#endif
1.164 +
1.165 + virtual Node *createLinearMatchNode(int32_t i, int32_t unitIndex, int32_t length,
1.166 + Node *nextNode) const;
1.167 +
1.168 + UBool ensureCapacity(int32_t length);
1.169 + virtual int32_t write(int32_t unit);
1.170 + int32_t write(const UChar *s, int32_t length);
1.171 + virtual int32_t writeElementUnits(int32_t i, int32_t unitIndex, int32_t length);
1.172 + virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal);
1.173 + virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node);
1.174 + virtual int32_t writeDeltaTo(int32_t jumpTarget);
1.175 +
1.176 + UnicodeString strings;
1.177 + UCharsTrieElement *elements;
1.178 + int32_t elementsCapacity;
1.179 + int32_t elementsLength;
1.180 +
1.181 + // UChar serialization of the trie.
1.182 + // Grows from the back: ucharsLength measures from the end of the buffer!
1.183 + UChar *uchars;
1.184 + int32_t ucharsCapacity;
1.185 + int32_t ucharsLength;
1.186 +};
1.187 +
1.188 +U_NAMESPACE_END
1.189 +
1.190 +#endif // __UCHARSTRIEBUILDER_H__
