icuSources/i18n/collationfastlatinbuilder.h

   1 /*
   2 *******************************************************************************
   3 * Copyright (C) 2013-2016, 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/ucol.h"
  20 #include "unicode/unistr.h"
  21 #include "unicode/uobject.h"
  22 #include "collation.h"
  23 #include "collationfastlatin.h"
  24 #include "uvectr64.h"
  25
  26 U_NAMESPACE_BEGIN
  27
  28 struct CollationData;
  29
  30 class U_I18N_API CollationFastLatinBuilder : public UObject {
  31 public:
  32     CollationFastLatinBuilder(UErrorCode &errorCode);
  33     ~CollationFastLatinBuilder();
  34
  35     UBool forData(const CollationData &data, UErrorCode &errorCode);
  36
  37     const uint16_t *getTable() const {
  38         return reinterpret_cast<const uint16_t *>(result.getBuffer());
  39     }
  40     int32_t lengthOfTable() const { return result.length(); }
  41
  42 private:
  43     // space, punct, symbol, currency (not digit)
  44     enum { NUM_SPECIAL_GROUPS = UCOL_REORDER_CODE_CURRENCY - UCOL_REORDER_CODE_FIRST + 1 };
  45
  46     UBool loadGroups(const CollationData &data, UErrorCode &errorCode);
  47     UBool inSameGroup(uint32_t p, uint32_t q) const;
  48
  49     void resetCEs();
  50     void getCEs(const CollationData &data, UErrorCode &errorCode);
  51     UBool getCEsFromCE32(const CollationData &data, UChar32 c, uint32_t ce32,
  52                          UErrorCode &errorCode);
  53     UBool getCEsFromContractionCE32(const CollationData &data, uint32_t ce32,
  54                                     UErrorCode &errorCode);
  55     void addContractionEntry(int32_t x, int64_t cce0, int64_t cce1, UErrorCode &errorCode);
  56     void addUniqueCE(int64_t ce, UErrorCode &errorCode);
  57     uint32_t getMiniCE(int64_t ce) const;
  58     UBool encodeUniqueCEs(UErrorCode &errorCode);
  59     UBool encodeCharCEs(UErrorCode &errorCode);
  60     UBool encodeContractions(UErrorCode &errorCode);
  61     uint32_t encodeTwoCEs(int64_t first, int64_t second) const;
  62
  63     static UBool isContractionCharCE(int64_t ce) {
  64         return (uint32_t)(ce >> 32) == Collation::NO_CE_PRIMARY && ce != Collation::NO_CE;
  65     }
  66
  67     static const uint32_t CONTRACTION_FLAG = 0x80000000;
  68
  69     // temporary "buffer"
  70     int64_t ce0, ce1;
  71
  72     int64_t charCEs[CollationFastLatin::NUM_FAST_CHARS][2];
  73
  74     UVector64 contractionCEs;
  75     UVector64 uniqueCEs;
  76
  77     /** One 16-bit mini CE per unique CE. */
  78     uint16_t *miniCEs;
  79
  80     // These are constant for a given root collator.
  81     uint32_t lastSpecialPrimaries[NUM_SPECIAL_GROUPS];
  82     uint32_t firstDigitPrimary;
  83     uint32_t firstLatinPrimary;
  84     uint32_t lastLatinPrimary;
  85     // This determines the first normal primary weight which is mapped to
  86     // a short mini primary. It must be >=firstDigitPrimary.
  87     uint32_t firstShortPrimary;
  88
  89     UBool shortPrimaryOverflow;
  90
  91     UnicodeString result;
  92     int32_t headerLength;
  93 };
  94
  95 U_NAMESPACE_END
  96
  97 #endif  // !UCONFIG_NO_COLLATION
  98 #endif  // __COLLATIONFASTLATINBUILDER_H__