icuSources/tools/gennorm2/n2builder.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 *
   6 *   Copyright (C) 2009-2014, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 *******************************************************************************
  10 *   file name:  n2builder.h
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 2009nov25
  16 *   created by: Markus W. Scherer
  17 */
  18
  19 #ifndef __N2BUILDER_H__
  20 #define __N2BUILDER_H__
  21
  22 #include "unicode/utypes.h"
  23
  24 #if !UCONFIG_NO_NORMALIZATION
  25
  26 #include "unicode/errorcode.h"
  27 #include "unicode/unistr.h"
  28 #include "normalizer2impl.h"  // for IX_COUNT
  29 #include "toolutil.h"
  30 #include "utrie2.h"
  31
  32 U_NAMESPACE_BEGIN
  33
  34 extern UBool beVerbose, haveCopyright;
  35
  36 struct Norm;
  37
  38 class BuilderReorderingBuffer;
  39 class ExtraDataWriter;
  40
  41 class Normalizer2DataBuilder {
  42 public:
  43     Normalizer2DataBuilder(UErrorCode &errorCode);
  44     ~Normalizer2DataBuilder();
  45
  46     enum OverrideHandling {
  47         OVERRIDE_NONE,
  48         OVERRIDE_ANY,
  49         OVERRIDE_PREVIOUS
  50     };
  51
  52     void setOverrideHandling(OverrideHandling oh);
  53
  54     enum Optimization {
  55         OPTIMIZE_NORMAL,
  56         OPTIMIZE_FAST
  57     };
  58
  59     void setOptimization(Optimization opt) { optimization=opt; }
  60
  61     void setCC(UChar32 c, uint8_t cc);
  62     void setOneWayMapping(UChar32 c, const UnicodeString &m);
  63     void setRoundTripMapping(UChar32 c, const UnicodeString &m);
  64     void removeMapping(UChar32 c);
  65
  66     void setUnicodeVersion(const char *v);
  67
  68     void writeBinaryFile(const char *filename);
  69     void writeCSourceFile(const char *filename);
  70
  71 private:
  72     friend class CompositionBuilder;
  73     friend class Decomposer;
  74     friend class ExtraDataWriter;
  75     friend class Norm16Writer;
  76
  77     // No copy constructor nor assignment operator.
  78     Normalizer2DataBuilder(const Normalizer2DataBuilder &other);
  79     Normalizer2DataBuilder &operator=(const Normalizer2DataBuilder &other);
  80
  81     Norm *allocNorm();
  82     Norm *getNorm(UChar32 c);
  83     Norm *createNorm(UChar32 c);
  84     Norm *checkNormForMapping(Norm *p, UChar32 c);  // check for permitted overrides
  85
  86     const Norm &getNormRef(UChar32 c) const;
  87     uint8_t getCC(UChar32 c) const;
  88     UBool combinesWithCCBetween(const Norm &norm, uint8_t lowCC, uint8_t highCC) const;
  89     UChar32 combine(const Norm &norm, UChar32 trail) const;
  90
  91     void addComposition(UChar32 start, UChar32 end, uint32_t value);
  92     UBool decompose(UChar32 start, UChar32 end, uint32_t value);
  93     void reorder(Norm *p, BuilderReorderingBuffer &buffer);
  94     UBool hasNoCompBoundaryAfter(BuilderReorderingBuffer &buffer);
  95     void setHangulData();
  96     int32_t writeMapping(UChar32 c, const Norm *p, UnicodeString &dataString);
  97     void writeCompositions(UChar32 c, const Norm *p, UnicodeString &dataString);
  98     void writeExtraData(UChar32 c, uint32_t value, ExtraDataWriter &writer);
  99     int32_t getCenterNoNoDelta() {
 100         return indexes[Normalizer2Impl::IX_MIN_MAYBE_YES]-Normalizer2Impl::MAX_DELTA-1;
 101     }
 102     void writeNorm16(UChar32 start, UChar32 end, uint32_t value);
 103     void processData();
 104
 105     UTrie2 *normTrie;
 106     UToolMemory *normMem;
 107     Norm *norms;
 108
 109     int32_t phase;
 110     OverrideHandling overrideHandling;
 111
 112     Optimization optimization;
 113
 114     int32_t indexes[Normalizer2Impl::IX_COUNT];
 115     UTrie2 *norm16Trie;
 116     int32_t norm16TrieLength;
 117     UnicodeString extraData;
 118     uint8_t smallFCD[0x100];
 119
 120     UVersionInfo unicodeVersion;
 121 };
 122
 123 U_NAMESPACE_END
 124
 125 #endif // #if !UCONFIG_NO_NORMALIZATION
 126
 127 #endif  // __N2BUILDER_H__