1 // Copyright (C) 2016 and later: Unicode, Inc. and others. 2 // License & terms of use: http://www.unicode.org/copyright.html 3 /* 4 ******************************************************************************* 5 * Copyright (C) 2001-2014, International Business Machines 6 * Corporation and others. All Rights Reserved. 7 ******************************************************************************* 8 * file name: bocsu.h 9 * encoding: US-ASCII 10 * tab size: 8 (not used) 11 * indentation:4 12 * 13 * Author: Markus W. Scherer 14 * 15 * Modification history: 16 * 05/18/2001 weiv Made into separate module 17 */ 18 19 #ifndef BOCSU_H 20 #define BOCSU_H 21 22 #include "unicode/utypes.h" 23 24 #if !UCONFIG_NO_COLLATION 25 26 U_NAMESPACE_BEGIN 27 28 class ByteSink; 29 30 U_NAMESPACE_END 31 32 /* 33 * "BOCSU" 34 * Binary Ordered Compression Scheme for Unicode 35 * 36 * Specific application: 37 * 38 * Encode a Unicode string for the identical level of a sort key. 39 * Restrictions: 40 * - byte stream (unsigned 8-bit bytes) 41 * - lexical order of the identical-level run must be 42 * the same as code point order for the string 43 * - avoid byte values 0, 1, 2 44 * 45 * Method: Slope Detection 46 * Remember the previous code point (initial 0). 47 * For each cp in the string, encode the difference to the previous one. 48 * 49 * With a compact encoding of differences, this yields good results for 50 * small scripts and UTF-like results otherwise. 51 * 52 * Encoding of differences: 53 * - Similar to a UTF, encoding the length of the byte sequence in the lead bytes. 54 * - Does not need to be friendly for decoding or random access 55 * (trail byte values may overlap with lead/single byte values). 56 * - The signedness must be encoded as the most significant part. 57 * 58 * We encode differences with few bytes if their absolute values are small. 59 * For correct ordering, we must treat the entire value range -10ffff..+10ffff 60 * in ascending order, which forbids encoding the sign and the absolute value separately. 61 * Instead, we split the lead byte range in the middle and encode non-negative values 62 * going up and negative values going down. 63 * 64 * For very small absolute values, the difference is added to a middle byte value 65 * for single-byte encoded differences. 66 * For somewhat larger absolute values, the difference is divided by the number 67 * of byte values available, the modulo is used for one trail byte, and the remainder 68 * is added to a lead byte avoiding the single-byte range. 69 * For large absolute values, the difference is similarly encoded in three bytes. 70 * 71 * This encoding does not use byte values 0, 1, 2, but uses all other byte values 72 * for lead/single bytes so that the middle range of single bytes is as large 73 * as possible. 74 * Note that the lead byte ranges overlap some, but that the sequences as a whole 75 * are well ordered. I.e., even if the lead byte is the same for sequences of different 76 * lengths, the trail bytes establish correct order. 77 * It would be possible to encode slightly larger ranges for each length (>1) by 78 * subtracting the lower bound of the range. However, that would also slow down the 79 * calculation. 80 * 81 * For the actual string encoding, an optimization moves the previous code point value 82 * to the middle of its Unicode script block to minimize the differences in 83 * same-script text runs. 84 */ 85 86 /* Do not use byte values 0, 1, 2 because they are separators in sort keys. */ 87 #define SLOPE_MIN 3 88 #define SLOPE_MAX 0xff 89 #define SLOPE_MIDDLE 0x81 90 91 #define SLOPE_TAIL_COUNT (SLOPE_MAX-SLOPE_MIN+1) 92 93 #define SLOPE_MAX_BYTES 4 94 95 /* 96 * Number of lead bytes: 97 * 1 middle byte for 0 98 * 2*80=160 single bytes for !=0 99 * 2*42=84 for double-byte values 100 * 2*3=6 for 3-byte values 101 * 2*1=2 for 4-byte values 102 * 103 * The sum must be <=SLOPE_TAIL_COUNT. 104 * 105 * Why these numbers? 106 * - There should be >=128 single-byte values to cover 128-blocks 107 * with small scripts. 108 * - There should be >=20902 single/double-byte values to cover Unihan. 109 * - It helps CJK Extension B some if there are 3-byte values that cover 110 * the distance between them and Unihan. 111 * This also helps to jump among distant places in the BMP. 112 * - Four-byte values are necessary to cover the rest of Unicode. 113 * 114 * Symmetrical lead byte counts are for convenience. 115 * With an equal distribution of even and odd differences there is also 116 * no advantage to asymmetrical lead byte counts. 117 */ 118 #define SLOPE_SINGLE 80 119 #define SLOPE_LEAD_2 42 120 #define SLOPE_LEAD_3 3 121 #define SLOPE_LEAD_4 1 122 123 /* The difference value range for single-byters. */ 124 #define SLOPE_REACH_POS_1 SLOPE_SINGLE 125 #define SLOPE_REACH_NEG_1 (-SLOPE_SINGLE) 126 127 /* The difference value range for double-byters. */ 128 #define SLOPE_REACH_POS_2 (SLOPE_LEAD_2*SLOPE_TAIL_COUNT+(SLOPE_LEAD_2-1)) 129 #define SLOPE_REACH_NEG_2 (-SLOPE_REACH_POS_2-1) 130 131 /* The difference value range for 3-byters. */ 132 #define SLOPE_REACH_POS_3 (SLOPE_LEAD_3*SLOPE_TAIL_COUNT*SLOPE_TAIL_COUNT+(SLOPE_LEAD_3-1)*SLOPE_TAIL_COUNT+(SLOPE_TAIL_COUNT-1)) 133 #define SLOPE_REACH_NEG_3 (-SLOPE_REACH_POS_3-1) 134 135 /* The lead byte start values. */ 136 #define SLOPE_START_POS_2 (SLOPE_MIDDLE+SLOPE_SINGLE+1) 137 #define SLOPE_START_POS_3 (SLOPE_START_POS_2+SLOPE_LEAD_2) 138 139 #define SLOPE_START_NEG_2 (SLOPE_MIDDLE+SLOPE_REACH_NEG_1) 140 #define SLOPE_START_NEG_3 (SLOPE_START_NEG_2-SLOPE_LEAD_2) 141 142 /* 143 * Integer division and modulo with negative numerators 144 * yields negative modulo results and quotients that are one more than 145 * what we need here. 146 */ 147 #define NEGDIVMOD(n, d, m) { \ 148 (m)=(n)%(d); \ 149 (n)/=(d); \ 150 if((m)<0) { \ 151 --(n); \ 152 (m)+=(d); \ 153 } \ 154 } 155 156 U_CFUNC UChar32 157 u_writeIdenticalLevelRun(UChar32 prev, const UChar *s, int32_t length, icu::ByteSink &sink); 158 159 #endif /* #if !UCONFIG_NO_COLLATION */ 160 161 #endif 162