michael@0: /*
michael@0: *******************************************************************************
michael@0: *   Copyright (C) 2010-2013, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *******************************************************************************
michael@0: *   file name:  bytestriebuilder.h
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 2010sep25
michael@0: *   created by: Markus W. Scherer
michael@0: */
michael@0: 
michael@0: /**
michael@0:  * \file
michael@0:  * \brief C++ API: Builder for icu::BytesTrie
michael@0:  */
michael@0: 
michael@0: #ifndef __BYTESTRIEBUILDER_H__
michael@0: #define __BYTESTRIEBUILDER_H__
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/bytestrie.h"
michael@0: #include "unicode/stringpiece.h"
michael@0: #include "unicode/stringtriebuilder.h"
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: class BytesTrieElement;
michael@0: class CharString;
michael@0: 
michael@0: /**
michael@0:  * Builder class for BytesTrie.
michael@0:  *
michael@0:  * This class is not intended for public subclassing.
michael@0:  * @stable ICU 4.8
michael@0:  */
michael@0: class U_COMMON_API BytesTrieBuilder : public StringTrieBuilder {
michael@0: public:
michael@0:     /**
michael@0:      * Constructs an empty builder.
michael@0:      * @param errorCode Standard ICU error code.
michael@0:      * @stable ICU 4.8
michael@0:      */
michael@0:     BytesTrieBuilder(UErrorCode &errorCode);
michael@0: 
michael@0:     /**
michael@0:      * Destructor.
michael@0:      * @stable ICU 4.8
michael@0:      */
michael@0:     virtual ~BytesTrieBuilder();
michael@0: 
michael@0:     /**
michael@0:      * Adds a (byte sequence, value) pair.
michael@0:      * The byte sequence must be unique.
michael@0:      * The bytes will be copied; the builder does not keep
michael@0:      * a reference to the input StringPiece or its data().
michael@0:      * @param s The input byte sequence.
michael@0:      * @param value The value associated with this byte sequence.
michael@0:      * @param errorCode Standard ICU error code. Its input value must
michael@0:      *                  pass the U_SUCCESS() test, or else the function returns
michael@0:      *                  immediately. Check for U_FAILURE() on output or use with
michael@0:      *                  function chaining. (See User Guide for details.)
michael@0:      * @return *this
michael@0:      * @stable ICU 4.8
michael@0:      */
michael@0:     BytesTrieBuilder &add(const StringPiece &s, int32_t value, UErrorCode &errorCode);
michael@0: 
michael@0:     /**
michael@0:      * Builds a BytesTrie for the add()ed data.
michael@0:      * Once built, no further data can be add()ed until clear() is called.
michael@0:      *
michael@0:      * A BytesTrie cannot be empty. At least one (byte sequence, value) pair
michael@0:      * must have been add()ed.
michael@0:      *
michael@0:      * This method passes ownership of the builder's internal result array to the new trie object.
michael@0:      * Another call to any build() variant will re-serialize the trie.
michael@0:      * After clear() has been called, a new array will be used as well.
michael@0:      * @param buildOption Build option, see UStringTrieBuildOption.
michael@0:      * @param errorCode Standard ICU error code. Its input value must
michael@0:      *                  pass the U_SUCCESS() test, or else the function returns
michael@0:      *                  immediately. Check for U_FAILURE() on output or use with
michael@0:      *                  function chaining. (See User Guide for details.)
michael@0:      * @return A new BytesTrie for the add()ed data.
michael@0:      * @stable ICU 4.8
michael@0:      */
michael@0:     BytesTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
michael@0: 
michael@0:     /**
michael@0:      * Builds a BytesTrie for the add()ed data and byte-serializes it.
michael@0:      * Once built, no further data can be add()ed until clear() is called.
michael@0:      *
michael@0:      * A BytesTrie cannot be empty. At least one (byte sequence, value) pair
michael@0:      * must have been add()ed.
michael@0:      *
michael@0:      * Multiple calls to buildStringPiece() return StringPieces referring to the
michael@0:      * builder's same byte array, without rebuilding.
michael@0:      * If buildStringPiece() is called after build(), the trie will be
michael@0:      * re-serialized into a new array.
michael@0:      * If build() is called after buildStringPiece(), the trie object will become
michael@0:      * the owner of the previously returned array.
michael@0:      * After clear() has been called, a new array will be used as well.
michael@0:      * @param buildOption Build option, see UStringTrieBuildOption.
michael@0:      * @param errorCode Standard ICU error code. Its input value must
michael@0:      *                  pass the U_SUCCESS() test, or else the function returns
michael@0:      *                  immediately. Check for U_FAILURE() on output or use with
michael@0:      *                  function chaining. (See User Guide for details.)
michael@0:      * @return A StringPiece which refers to the byte-serialized BytesTrie for the add()ed data.
michael@0:      * @stable ICU 4.8
michael@0:      */
michael@0:     StringPiece buildStringPiece(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
michael@0: 
michael@0:     /**
michael@0:      * Removes all (byte sequence, value) pairs.
michael@0:      * New data can then be add()ed and a new trie can be built.
michael@0:      * @return *this
michael@0:      * @stable ICU 4.8
michael@0:      */
michael@0:     BytesTrieBuilder &clear();
michael@0: 
michael@0: private:
michael@0:     BytesTrieBuilder(const BytesTrieBuilder &other);  // no copy constructor
michael@0:     BytesTrieBuilder &operator=(const BytesTrieBuilder &other);  // no assignment operator
michael@0: 
michael@0:     void buildBytes(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
michael@0: 
michael@0:     virtual int32_t getElementStringLength(int32_t i) const;
michael@0:     virtual UChar getElementUnit(int32_t i, int32_t byteIndex) const;
michael@0:     virtual int32_t getElementValue(int32_t i) const;
michael@0: 
michael@0:     virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t byteIndex) const;
michael@0: 
michael@0:     virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t byteIndex) const;
michael@0:     virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t byteIndex, int32_t count) const;
michael@0:     virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t byteIndex, UChar byte) const;
michael@0: 
michael@0:     virtual UBool matchNodesCanHaveValues() const { return FALSE; }
michael@0: 
michael@0:     virtual int32_t getMaxBranchLinearSubNodeLength() const { return BytesTrie::kMaxBranchLinearSubNodeLength; }
michael@0:     virtual int32_t getMinLinearMatch() const { return BytesTrie::kMinLinearMatch; }
michael@0:     virtual int32_t getMaxLinearMatchLength() const { return BytesTrie::kMaxLinearMatchLength; }
michael@0: 
michael@0: #ifndef U_HIDE_INTERNAL_API
michael@0:     /**
michael@0:      * @internal
michael@0:      */
michael@0:     class BTLinearMatchNode : public LinearMatchNode {
michael@0:     public:
michael@0:         BTLinearMatchNode(const char *units, int32_t len, Node *nextNode);
michael@0:         virtual UBool operator==(const Node &other) const;
michael@0:         virtual void write(StringTrieBuilder &builder);
michael@0:     private:
michael@0:         const char *s;
michael@0:     };
michael@0: #endif  /* U_HIDE_INTERNAL_API */
michael@0: 
michael@0:     virtual Node *createLinearMatchNode(int32_t i, int32_t byteIndex, int32_t length,
michael@0:                                         Node *nextNode) const;
michael@0: 
michael@0:     UBool ensureCapacity(int32_t length);
michael@0:     virtual int32_t write(int32_t byte);
michael@0:     int32_t write(const char *b, int32_t length);
michael@0:     virtual int32_t writeElementUnits(int32_t i, int32_t byteIndex, int32_t length);
michael@0:     virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal);
michael@0:     virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node);
michael@0:     virtual int32_t writeDeltaTo(int32_t jumpTarget);
michael@0: 
michael@0:     CharString *strings;  // Pointer not object so we need not #include internal charstr.h.
michael@0:     BytesTrieElement *elements;
michael@0:     int32_t elementsCapacity;
michael@0:     int32_t elementsLength;
michael@0: 
michael@0:     // Byte serialization of the trie.
michael@0:     // Grows from the back: bytesLength measures from the end of the buffer!
michael@0:     char *bytes;
michael@0:     int32_t bytesCapacity;
michael@0:     int32_t bytesLength;
michael@0: };
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: 
michael@0: #endif  // __BYTESTRIEBUILDER_H__