icuSources/i18n/uitercollationiterator.h

   1 /*
   2 *******************************************************************************
   3 * Copyright (C) 2012-2016, International Business Machines
   4 * Corporation and others.  All Rights Reserved.
   5 *******************************************************************************
   6 * uitercollationiterator.h
   7 *
   8 * created on: 2012sep23 (from utf16collationiterator.h)
   9 * created by: Markus W. Scherer
  10 */
  11
  12 #ifndef __UITERCOLLATIONITERATOR_H__
  13 #define __UITERCOLLATIONITERATOR_H__
  14
  15 #include "unicode/utypes.h"
  16
  17 #if !UCONFIG_NO_COLLATION
  18
  19 #include "unicode/uiter.h"
  20 #include "cmemory.h"
  21 #include "collation.h"
  22 #include "collationdata.h"
  23 #include "collationiterator.h"
  24 #include "normalizer2impl.h"
  25
  26 U_NAMESPACE_BEGIN
  27
  28 /**
  29  * UCharIterator-based collation element and character iterator.
  30  * Handles normalized text inline, with length or NUL-terminated.
  31  * Unnormalized text is handled by a subclass.
  32  */
  33 class U_I18N_API UIterCollationIterator : public CollationIterator {
  34 public:
  35     UIterCollationIterator(const CollationData *d, UBool numeric, UCharIterator &ui)
  36             : CollationIterator(d, numeric), iter(ui) {}
  37
  38     virtual ~UIterCollationIterator();
  39
  40     virtual void resetToOffset(int32_t newOffset);
  41
  42     virtual int32_t getOffset() const;
  43
  44     virtual UChar32 nextCodePoint(UErrorCode &errorCode);
  45
  46     virtual UChar32 previousCodePoint(UErrorCode &errorCode);
  47
  48 protected:
  49     virtual uint32_t handleNextCE32(UChar32 &c, UErrorCode &errorCode);
  50
  51     virtual UChar handleGetTrailSurrogate();
  52
  53     virtual void forwardNumCodePoints(int32_t num, UErrorCode &errorCode);
  54
  55     virtual void backwardNumCodePoints(int32_t num, UErrorCode &errorCode);
  56
  57     UCharIterator &iter;
  58 };
  59
  60 /**
  61  * Incrementally checks the input text for FCD and normalizes where necessary.
  62  */
  63 class U_I18N_API FCDUIterCollationIterator : public UIterCollationIterator {
  64 public:
  65     FCDUIterCollationIterator(const CollationData *data, UBool numeric, UCharIterator &ui, int32_t startIndex)
  66             : UIterCollationIterator(data, numeric, ui),
  67               state(ITER_CHECK_FWD), start(startIndex),
  68               nfcImpl(data->nfcImpl) {}
  69
  70     virtual ~FCDUIterCollationIterator();
  71
  72     virtual void resetToOffset(int32_t newOffset);
  73
  74     virtual int32_t getOffset() const;
  75
  76     virtual UChar32 nextCodePoint(UErrorCode &errorCode);
  77
  78     virtual UChar32 previousCodePoint(UErrorCode &errorCode);
  79
  80 protected:
  81     virtual uint32_t handleNextCE32(UChar32 &c, UErrorCode &errorCode);
  82
  83     virtual UChar handleGetTrailSurrogate();
  84
  85     virtual void forwardNumCodePoints(int32_t num, UErrorCode &errorCode);
  86
  87     virtual void backwardNumCodePoints(int32_t num, UErrorCode &errorCode);
  88
  89 private:
  90     /**
  91      * Switches to forward checking if possible.
  92      */
  93     void switchToForward();
  94
  95     /**
  96      * Extends the FCD text segment forward or normalizes around pos.
  97      * @return TRUE if success
  98      */
  99     UBool nextSegment(UErrorCode &errorCode);
 100
 101     /**
 102      * Switches to backward checking.
 103      */
 104     void switchToBackward();
 105
 106     /**
 107      * Extends the FCD text segment backward or normalizes around pos.
 108      * @return TRUE if success
 109      */
 110     UBool previousSegment(UErrorCode &errorCode);
 111
 112     UBool normalize(const UnicodeString &s, UErrorCode &errorCode);
 113
 114     enum State {
 115         /**
 116          * The input text [start..(iter index)[ passes the FCD check.
 117          * Moving forward checks incrementally.
 118          * pos & limit are undefined.
 119          */
 120         ITER_CHECK_FWD,
 121         /**
 122          * The input text [(iter index)..limit[ passes the FCD check.
 123          * Moving backward checks incrementally.
 124          * start & pos are undefined.
 125          */
 126         ITER_CHECK_BWD,
 127         /**
 128          * The input text [start..limit[ passes the FCD check.
 129          * pos tracks the current text index.
 130          */
 131         ITER_IN_FCD_SEGMENT,
 132         /**
 133          * The input text [start..limit[ failed the FCD check and was normalized.
 134          * pos tracks the current index in the normalized string.
 135          * The text iterator is at the limit index.
 136          */
 137         IN_NORM_ITER_AT_LIMIT,
 138         /**
 139          * The input text [start..limit[ failed the FCD check and was normalized.
 140          * pos tracks the current index in the normalized string.
 141          * The text iterator is at the start index.
 142          */
 143         IN_NORM_ITER_AT_START
 144     };
 145
 146     State state;
 147
 148     int32_t start;
 149     int32_t pos;
 150     int32_t limit;
 151
 152     const Normalizer2Impl &nfcImpl;
 153     UnicodeString normalized;
 154 };
 155
 156 U_NAMESPACE_END
 157
 158 #endif  // !UCONFIG_NO_COLLATION
 159 #endif  // __UITERCOLLATIONITERATOR_H__