1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/unicode/bytestriebuilder.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,183 @@
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: bytestriebuilder.h
1.10 +* encoding: US-ASCII
1.11 +* tab size: 8 (not used)
1.12 +* indentation:4
1.13 +*
1.14 +* created on: 2010sep25
1.15 +* created by: Markus W. Scherer
1.16 +*/
1.17 +
1.18 +/**
1.19 + * \file
1.20 + * \brief C++ API: Builder for icu::BytesTrie
1.21 + */
1.22 +
1.23 +#ifndef __BYTESTRIEBUILDER_H__
1.24 +#define __BYTESTRIEBUILDER_H__
1.25 +
1.26 +#include "unicode/utypes.h"
1.27 +#include "unicode/bytestrie.h"
1.28 +#include "unicode/stringpiece.h"
1.29 +#include "unicode/stringtriebuilder.h"
1.30 +
1.31 +U_NAMESPACE_BEGIN
1.32 +
1.33 +class BytesTrieElement;
1.34 +class CharString;
1.35 +
1.36 +/**
1.37 + * Builder class for BytesTrie.
1.38 + *
1.39 + * This class is not intended for public subclassing.
1.40 + * @stable ICU 4.8
1.41 + */
1.42 +class U_COMMON_API BytesTrieBuilder : public StringTrieBuilder {
1.43 +public:
1.44 + /**
1.45 + * Constructs an empty builder.
1.46 + * @param errorCode Standard ICU error code.
1.47 + * @stable ICU 4.8
1.48 + */
1.49 + BytesTrieBuilder(UErrorCode &errorCode);
1.50 +
1.51 + /**
1.52 + * Destructor.
1.53 + * @stable ICU 4.8
1.54 + */
1.55 + virtual ~BytesTrieBuilder();
1.56 +
1.57 + /**
1.58 + * Adds a (byte sequence, value) pair.
1.59 + * The byte sequence must be unique.
1.60 + * The bytes will be copied; the builder does not keep
1.61 + * a reference to the input StringPiece or its data().
1.62 + * @param s The input byte sequence.
1.63 + * @param value The value associated with this byte sequence.
1.64 + * @param errorCode Standard ICU error code. Its input value must
1.65 + * pass the U_SUCCESS() test, or else the function returns
1.66 + * immediately. Check for U_FAILURE() on output or use with
1.67 + * function chaining. (See User Guide for details.)
1.68 + * @return *this
1.69 + * @stable ICU 4.8
1.70 + */
1.71 + BytesTrieBuilder &add(const StringPiece &s, int32_t value, UErrorCode &errorCode);
1.72 +
1.73 + /**
1.74 + * Builds a BytesTrie for the add()ed data.
1.75 + * Once built, no further data can be add()ed until clear() is called.
1.76 + *
1.77 + * A BytesTrie cannot be empty. At least one (byte sequence, value) pair
1.78 + * must have been add()ed.
1.79 + *
1.80 + * This method passes ownership of the builder's internal result array to the new trie object.
1.81 + * Another call to any build() variant will re-serialize the trie.
1.82 + * After clear() has been called, a new array will be used as well.
1.83 + * @param buildOption Build option, see UStringTrieBuildOption.
1.84 + * @param errorCode Standard ICU error code. Its input value must
1.85 + * pass the U_SUCCESS() test, or else the function returns
1.86 + * immediately. Check for U_FAILURE() on output or use with
1.87 + * function chaining. (See User Guide for details.)
1.88 + * @return A new BytesTrie for the add()ed data.
1.89 + * @stable ICU 4.8
1.90 + */
1.91 + BytesTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
1.92 +
1.93 + /**
1.94 + * Builds a BytesTrie for the add()ed data and byte-serializes it.
1.95 + * Once built, no further data can be add()ed until clear() is called.
1.96 + *
1.97 + * A BytesTrie cannot be empty. At least one (byte sequence, value) pair
1.98 + * must have been add()ed.
1.99 + *
1.100 + * Multiple calls to buildStringPiece() return StringPieces referring to the
1.101 + * builder's same byte array, without rebuilding.
1.102 + * If buildStringPiece() is called after build(), the trie will be
1.103 + * re-serialized into a new array.
1.104 + * If build() is called after buildStringPiece(), the trie object will become
1.105 + * the owner of the previously returned array.
1.106 + * After clear() has been called, a new array will be used as well.
1.107 + * @param buildOption Build option, see UStringTrieBuildOption.
1.108 + * @param errorCode Standard ICU error code. Its input value must
1.109 + * pass the U_SUCCESS() test, or else the function returns
1.110 + * immediately. Check for U_FAILURE() on output or use with
1.111 + * function chaining. (See User Guide for details.)
1.112 + * @return A StringPiece which refers to the byte-serialized BytesTrie for the add()ed data.
1.113 + * @stable ICU 4.8
1.114 + */
1.115 + StringPiece buildStringPiece(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
1.116 +
1.117 + /**
1.118 + * Removes all (byte sequence, value) pairs.
1.119 + * New data can then be add()ed and a new trie can be built.
1.120 + * @return *this
1.121 + * @stable ICU 4.8
1.122 + */
1.123 + BytesTrieBuilder &clear();
1.124 +
1.125 +private:
1.126 + BytesTrieBuilder(const BytesTrieBuilder &other); // no copy constructor
1.127 + BytesTrieBuilder &operator=(const BytesTrieBuilder &other); // no assignment operator
1.128 +
1.129 + void buildBytes(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
1.130 +
1.131 + virtual int32_t getElementStringLength(int32_t i) const;
1.132 + virtual UChar getElementUnit(int32_t i, int32_t byteIndex) const;
1.133 + virtual int32_t getElementValue(int32_t i) const;
1.134 +
1.135 + virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t byteIndex) const;
1.136 +
1.137 + virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t byteIndex) const;
1.138 + virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t byteIndex, int32_t count) const;
1.139 + virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t byteIndex, UChar byte) const;
1.140 +
1.141 + virtual UBool matchNodesCanHaveValues() const { return FALSE; }
1.142 +
1.143 + virtual int32_t getMaxBranchLinearSubNodeLength() const { return BytesTrie::kMaxBranchLinearSubNodeLength; }
1.144 + virtual int32_t getMinLinearMatch() const { return BytesTrie::kMinLinearMatch; }
1.145 + virtual int32_t getMaxLinearMatchLength() const { return BytesTrie::kMaxLinearMatchLength; }
1.146 +
1.147 +#ifndef U_HIDE_INTERNAL_API
1.148 + /**
1.149 + * @internal
1.150 + */
1.151 + class BTLinearMatchNode : public LinearMatchNode {
1.152 + public:
1.153 + BTLinearMatchNode(const char *units, int32_t len, Node *nextNode);
1.154 + virtual UBool operator==(const Node &other) const;
1.155 + virtual void write(StringTrieBuilder &builder);
1.156 + private:
1.157 + const char *s;
1.158 + };
1.159 +#endif /* U_HIDE_INTERNAL_API */
1.160 +
1.161 + virtual Node *createLinearMatchNode(int32_t i, int32_t byteIndex, int32_t length,
1.162 + Node *nextNode) const;
1.163 +
1.164 + UBool ensureCapacity(int32_t length);
1.165 + virtual int32_t write(int32_t byte);
1.166 + int32_t write(const char *b, int32_t length);
1.167 + virtual int32_t writeElementUnits(int32_t i, int32_t byteIndex, int32_t length);
1.168 + virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal);
1.169 + virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node);
1.170 + virtual int32_t writeDeltaTo(int32_t jumpTarget);
1.171 +
1.172 + CharString *strings; // Pointer not object so we need not #include internal charstr.h.
1.173 + BytesTrieElement *elements;
1.174 + int32_t elementsCapacity;
1.175 + int32_t elementsLength;
1.176 +
1.177 + // Byte serialization of the trie.
1.178 + // Grows from the back: bytesLength measures from the end of the buffer!
1.179 + char *bytes;
1.180 + int32_t bytesCapacity;
1.181 + int32_t bytesLength;
1.182 +};
1.183 +
1.184 +U_NAMESPACE_END
1.185 +
1.186 +#endif // __BYTESTRIEBUILDER_H__
