icuSources/common/bmpset.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) 2007, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 ******************************************************************************
  10 *   file name:  bmpset.h
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 2007jan29
  16 *   created by: Markus W. Scherer
  17 */
  18
  19 #ifndef __BMPSET_H__
  20 #define __BMPSET_H__
  21
  22 #include "unicode/utypes.h"
  23 #include "unicode/uniset.h"
  24
  25 U_NAMESPACE_BEGIN
  26
  27 /*
  28  * Helper class for frozen UnicodeSets, implements contains() and span()
  29  * optimized for BMP code points. Structured to be UTF-8-friendly.
  30  *
  31  * ASCII: Look up bytes.
  32  * 2-byte characters: Bits organized vertically.
  33  * 3-byte characters: Use zero/one/mixed data per 64-block in U+0000..U+FFFF,
  34  *                    with mixed for illegal ranges.
  35  * Supplementary characters: Call contains() on the parent set.
  36  */
  37 class BMPSet : public UMemory {
  38 public:
  39     BMPSet(const int32_t *parentList, int32_t parentListLength);
  40     BMPSet(const BMPSet &otherBMPSet, const int32_t *newParentList, int32_t newParentListLength);
  41     virtual ~BMPSet();
  42
  43     virtual UBool contains(UChar32 c) const;
  44
  45     /*
  46      * Span the initial substring for which each character c has spanCondition==contains(c).
  47      * It must be s<limit and spanCondition==0 or 1.
  48      * @return The string pointer which limits the span.
  49      */
  50     const UChar *span(const UChar *s, const UChar *limit, USetSpanCondition spanCondition) const;
  51
  52     /*
  53      * Span the trailing substring for which each character c has spanCondition==contains(c).
  54      * It must be s<limit and spanCondition==0 or 1.
  55      * @return The string pointer which starts the span.
  56      */
  57     const UChar *spanBack(const UChar *s, const UChar *limit, USetSpanCondition spanCondition) const;
  58
  59     /*
  60      * Span the initial substring for which each character c has spanCondition==contains(c).
  61      * It must be length>0 and spanCondition==0 or 1.
  62      * @return The string pointer which limits the span.
  63      */
  64     const uint8_t *spanUTF8(const uint8_t *s, int32_t length, USetSpanCondition spanCondition) const;
  65
  66     /*
  67      * Span the trailing substring for which each character c has spanCondition==contains(c).
  68      * It must be length>0 and spanCondition==0 or 1.
  69      * @return The start of the span.
  70      */
  71     int32_t spanBackUTF8(const uint8_t *s, int32_t length, USetSpanCondition spanCondition) const;
  72
  73 private:
  74     void initBits();
  75     void overrideIllegal();
  76
  77     /**
  78      * Same as UnicodeSet::findCodePoint(UChar32 c) const except that the
  79      * binary search is restricted for finding code points in a certain range.
  80      *
  81      * For restricting the search for finding in the range start..end,
  82      * pass in
  83      *   lo=findCodePoint(start) and
  84      *   hi=findCodePoint(end)
  85      * with 0<=lo<=hi<len.
  86      * findCodePoint(c) defaults to lo=0 and hi=len-1.
  87      *
  88      * @param c a character in a subrange of MIN_VALUE..MAX_VALUE
  89      * @param lo The lowest index to be returned.
  90      * @param hi The highest index to be returned.
  91      * @return the smallest integer i in the range lo..hi,
  92      *         inclusive, such that c < list[i]
  93      */
  94     int32_t findCodePoint(UChar32 c, int32_t lo, int32_t hi) const;
  95
  96     inline UBool containsSlow(UChar32 c, int32_t lo, int32_t hi) const;
  97
  98     /*
  99      * One byte per ASCII character, or trail byte in lead position.
 100      * 0 or 1 for ASCII characters.
 101      * The value for trail bytes is the result of contains(FFFD)
 102      * for faster validity checking at runtime.
 103      */
 104     UBool asciiBytes[0xc0];
 105
 106     /*
 107      * One bit per code point from U+0000..U+07FF.
 108      * The bits are organized vertically; consecutive code points
 109      * correspond to the same bit positions in consecutive table words.
 110      * With code point parts
 111      *   lead=c{10..6}
 112      *   trail=c{5..0}
 113      * it is set.contains(c)==(table7FF[trail] bit lead)
 114      *
 115      * Bits for 0..7F (non-shortest forms) are set to the result of contains(FFFD)
 116      * for faster validity checking at runtime.
 117      */
 118     uint32_t table7FF[64];
 119
 120     /*
 121      * One bit per 64 BMP code points.
 122      * The bits are organized vertically; consecutive 64-code point blocks
 123      * correspond to the same bit position in consecutive table words.
 124      * With code point parts
 125      *   lead=c{15..12}
 126      *   t1=c{11..6}
 127      * test bits (lead+16) and lead in bmpBlockBits[t1].
 128      * If the upper bit is 0, then the lower bit indicates if contains(c)
 129      * for all code points in the 64-block.
 130      * If the upper bit is 1, then the block is mixed and set.contains(c)
 131      * must be called.
 132      *
 133      * Bits for 0..7FF (non-shortest forms) and D800..DFFF are set to
 134      * the result of contains(FFFD) for faster validity checking at runtime.
 135      */
 136     uint32_t bmpBlockBits[64];
 137
 138     /*
 139      * Inversion list indexes for restricted binary searches in
 140      * findCodePoint(), from
 141      * findCodePoint(U+0800, U+1000, U+2000, .., U+F000, U+10000).
 142      * U+0800 is the first 3-byte-UTF-8 code point. Code points below U+0800 are
 143      * always looked up in the bit tables.
 144      * The last pair of indexes is for finding supplementary code points.
 145      */
 146     int32_t list4kStarts[18];
 147
 148     /*
 149      * The inversion list of the parent set, for the slower contains() implementation
 150      * for mixed BMP blocks and for supplementary code points.
 151      * The list is terminated with list[listLength-1]=0x110000.
 152      */
 153     const int32_t *list;
 154     int32_t listLength;
 155 };
 156
 157 inline UBool BMPSet::containsSlow(UChar32 c, int32_t lo, int32_t hi) const {
 158     return (UBool)(findCodePoint(c, lo, hi) & 1);
 159 }
 160
 161 U_NAMESPACE_END
 162
 163 #endif