1/*
2*******************************************************************************
3* Copyright (C) 2013-2014, International Business Machines
4* Corporation and others.  All Rights Reserved.
5*******************************************************************************
6* collationfastlatinbuilder.h
7*
8* created on: 2013aug09
9* created by: Markus W. Scherer
10*/
11
12#ifndef __COLLATIONFASTLATINBUILDER_H__
13#define __COLLATIONFASTLATINBUILDER_H__
14
15#include "unicode/utypes.h"
16
17#if !UCONFIG_NO_COLLATION
18
19#include "unicode/unistr.h"
20#include "unicode/uobject.h"
21#include "collation.h"
22#include "collationfastlatin.h"
23#include "uvectr64.h"
24
25U_NAMESPACE_BEGIN
26
27struct CollationData;
28
29class U_I18N_API CollationFastLatinBuilder : public UObject {
30public:
31    CollationFastLatinBuilder(UErrorCode &errorCode);
32    ~CollationFastLatinBuilder();
33
34    UBool forData(const CollationData &data, UErrorCode &errorCode);
35
36    const uint16_t *getTable() const {
37        return reinterpret_cast<const uint16_t *>(result.getBuffer());
38    }
39    int32_t lengthOfTable() const { return result.length(); }
40
41private:
42    UBool loadGroups(const CollationData &data, UErrorCode &errorCode);
43    UBool inSameGroup(uint32_t p, uint32_t q) const;
44
45    void resetCEs();
46    void getCEs(const CollationData &data, UErrorCode &errorCode);
47    UBool getCEsFromCE32(const CollationData &data, UChar32 c, uint32_t ce32,
48                         UErrorCode &errorCode);
49    UBool getCEsFromContractionCE32(const CollationData &data, uint32_t ce32,
50                                    UErrorCode &errorCode);
51    void addContractionEntry(int32_t x, int64_t cce0, int64_t cce1, UErrorCode &errorCode);
52    void addUniqueCE(int64_t ce, UErrorCode &errorCode);
53    uint32_t getMiniCE(int64_t ce) const;
54    UBool encodeUniqueCEs(UErrorCode &errorCode);
55    UBool encodeCharCEs(UErrorCode &errorCode);
56    UBool encodeContractions(UErrorCode &errorCode);
57    uint32_t encodeTwoCEs(int64_t first, int64_t second) const;
58
59    static UBool isContractionCharCE(int64_t ce) {
60        return (uint32_t)(ce >> 32) == Collation::NO_CE_PRIMARY && ce != Collation::NO_CE;
61    }
62
63    static const uint32_t CONTRACTION_FLAG = 0x80000000;
64
65    // temporary "buffer"
66    int64_t ce0, ce1;
67
68    int64_t charCEs[CollationFastLatin::NUM_FAST_CHARS][2];
69
70    UVector64 contractionCEs;
71    UVector64 uniqueCEs;
72
73    /** One 16-bit mini CE per unique CE. */
74    uint16_t *miniCEs;
75
76    // These are constant for a given list of CollationData.scripts.
77    uint32_t firstDigitPrimary;
78    uint32_t firstLatinPrimary;
79    uint32_t lastLatinPrimary;
80    // This determines the first normal primary weight which is mapped to
81    // a short mini primary. It must be >=firstDigitPrimary.
82    uint32_t firstShortPrimary;
83
84    UBool shortPrimaryOverflow;
85
86    UnicodeString result;
87    int32_t headerLength;
88};
89
90U_NAMESPACE_END
91
92#endif  // !UCONFIG_NO_COLLATION
93#endif  // __COLLATIONFASTLATINBUILDER_H__
94