1b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho/*
2b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*******************************************************************************
3b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*   Copyright (C) 2010-2011, International Business Machines
4b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*   Corporation and others.  All Rights Reserved.
5b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*******************************************************************************
6b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*   file name:  bytestriebuilder.h
7b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*   encoding:   US-ASCII
8b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*   tab size:   8 (not used)
9b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*   indentation:4
10b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*
11b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*   created on: 2010sep25
12b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*   created by: Markus W. Scherer
13b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*/
14b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
15b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho#ifndef __BYTESTRIEBUILDER_H__
16b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho#define __BYTESTRIEBUILDER_H__
17b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
18b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho#include "unicode/utypes.h"
19b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho#include "unicode/bytestrie.h"
20b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho#include "unicode/stringpiece.h"
21b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho#include "unicode/stringtriebuilder.h"
22b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
23b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehoU_NAMESPACE_BEGIN
24b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
25b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehoclass BytesTrieElement;
26b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehoclass CharString;
27b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
28b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho/**
29b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Builder class for BytesTrie.
30b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho *
31b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * This class is not intended for public subclassing.
32b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @draft ICU 4.8
33b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho */
34b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehoclass U_COMMON_API BytesTrieBuilder : public StringTrieBuilder {
35b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehopublic:
36b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    /**
37b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * Constructs an empty builder.
38b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @param errorCode Standard ICU error code.
39b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @draft ICU 4.8
40b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     */
41b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    BytesTrieBuilder(UErrorCode &errorCode);
42b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
43b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    /**
44b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * Destructor.
45b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @draft ICU 4.8
46b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     */
47b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual ~BytesTrieBuilder();
48b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
49b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    /**
50b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * Adds a (byte sequence, value) pair.
51b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * The byte sequence must be unique.
52b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * The bytes will be copied; the builder does not keep
53b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * a reference to the input StringPiece or its data().
54b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @param s The input byte sequence.
55b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @param value The value associated with this byte sequence.
56b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @param errorCode Standard ICU error code. Its input value must
57b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     *                  pass the U_SUCCESS() test, or else the function returns
58b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     *                  immediately. Check for U_FAILURE() on output or use with
59b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     *                  function chaining. (See User Guide for details.)
60b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @return *this
61b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @draft ICU 4.8
62b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     */
63b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    BytesTrieBuilder &add(const StringPiece &s, int32_t value, UErrorCode &errorCode);
64b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
65b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    /**
66b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * Builds a BytesTrie for the add()ed data.
67b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * Once built, no further data can be add()ed until clear() is called.
68b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     *
69b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * This method passes ownership of the builder's internal result array to the new trie object.
70b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * Another call to any build() variant will re-serialize the trie.
71b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * After clear() has been called, a new array will be used as well.
72b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @param buildOption Build option, see UStringTrieBuildOption.
73b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @param errorCode Standard ICU error code. Its input value must
74b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     *                  pass the U_SUCCESS() test, or else the function returns
75b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     *                  immediately. Check for U_FAILURE() on output or use with
76b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     *                  function chaining. (See User Guide for details.)
77b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @return A new BytesTrie for the add()ed data.
78b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @draft ICU 4.8
79b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     */
80b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    BytesTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
81b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
82b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    /**
83b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * Builds a BytesTrie for the add()ed data and byte-serializes it.
84b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * Once built, no further data can be add()ed until clear() is called.
85b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     *
86b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * Multiple calls to buildStringPiece() return StringPieces referring to the
87b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * builder's same byte array, without rebuilding.
88b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * If buildStringPiece() is called after build(), the trie will be
89b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * re-serialized into a new array.
90b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * If build() is called after buildStringPiece(), the trie object will become
91b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * the owner of the previously returned array.
92b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * After clear() has been called, a new array will be used as well.
93b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @param buildOption Build option, see UStringTrieBuildOption.
94b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @param errorCode Standard ICU error code. Its input value must
95b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     *                  pass the U_SUCCESS() test, or else the function returns
96b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     *                  immediately. Check for U_FAILURE() on output or use with
97b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     *                  function chaining. (See User Guide for details.)
98b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @return A StringPiece which refers to the byte-serialized BytesTrie for the add()ed data.
99b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @draft ICU 4.8
100b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     */
101b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    StringPiece buildStringPiece(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
102b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
103b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    /**
104b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * Removes all (byte sequence, value) pairs.
105b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * New data can then be add()ed and a new trie can be built.
106b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @return *this
107b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @draft ICU 4.8
108b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     */
109b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    BytesTrieBuilder &clear();
110b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
111b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehoprivate:
112b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    BytesTrieBuilder(const BytesTrieBuilder &other);  // no copy constructor
113b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    BytesTrieBuilder &operator=(const BytesTrieBuilder &other);  // no assignment operator
114b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
115b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    void buildBytes(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
116b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
117b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t getElementStringLength(int32_t i) const;
118b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual UChar getElementUnit(int32_t i, int32_t byteIndex) const;
119b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t getElementValue(int32_t i) const;
120b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
121b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t byteIndex) const;
122b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
123b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t byteIndex) const;
124b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t byteIndex, int32_t count) const;
125b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t byteIndex, UChar byte) const;
126b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
127b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual UBool matchNodesCanHaveValues() const { return FALSE; }
128b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
129b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t getMaxBranchLinearSubNodeLength() const { return BytesTrie::kMaxBranchLinearSubNodeLength; }
130b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t getMinLinearMatch() const { return BytesTrie::kMinLinearMatch; }
131b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t getMaxLinearMatchLength() const { return BytesTrie::kMaxLinearMatchLength; }
132b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
133b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    /**
134b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     * @internal
135b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho     */
136b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    class BTLinearMatchNode : public LinearMatchNode {
137b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    public:
138b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho        BTLinearMatchNode(const char *units, int32_t len, Node *nextNode);
139b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho        virtual UBool operator==(const Node &other) const;
140b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho        virtual void write(StringTrieBuilder &builder);
141b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    private:
142b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho        const char *s;
143b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    };
144b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
145b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual Node *createLinearMatchNode(int32_t i, int32_t byteIndex, int32_t length,
146b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho                                        Node *nextNode) const;
147b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
148b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    UBool ensureCapacity(int32_t length);
149b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t write(int32_t byte);
150b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    int32_t write(const char *b, int32_t length);
151b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t writeElementUnits(int32_t i, int32_t byteIndex, int32_t length);
152b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal);
153b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node);
154b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    virtual int32_t writeDeltaTo(int32_t jumpTarget);
155b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
156b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    CharString *strings;  // Pointer not object so we need not #include internal charstr.h.
157b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    BytesTrieElement *elements;
158b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    int32_t elementsCapacity;
159b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    int32_t elementsLength;
160b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
161b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    // Byte serialization of the trie.
162b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    // Grows from the back: bytesLength measures from the end of the buffer!
163b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    char *bytes;
164b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    int32_t bytesCapacity;
165b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho    int32_t bytesLength;
166b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho};
167b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
168b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehoU_NAMESPACE_END
169b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho
170b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho#endif  // __BYTESTRIEBUILDER_H__
171