1*0e209d39SAndroid Build Coastguard Worker // © 2016 and later: Unicode, Inc. and others. 2*0e209d39SAndroid Build Coastguard Worker // License & terms of use: http://www.unicode.org/copyright.html 3*0e209d39SAndroid Build Coastguard Worker /* 4*0e209d39SAndroid Build Coastguard Worker ******************************************************************************* 5*0e209d39SAndroid Build Coastguard Worker * Copyright (C) 2010-2015, International Business Machines 6*0e209d39SAndroid Build Coastguard Worker * Corporation and others. All Rights Reserved. 7*0e209d39SAndroid Build Coastguard Worker ******************************************************************************* 8*0e209d39SAndroid Build Coastguard Worker * collation.h 9*0e209d39SAndroid Build Coastguard Worker * 10*0e209d39SAndroid Build Coastguard Worker * created on: 2010oct27 11*0e209d39SAndroid Build Coastguard Worker * created by: Markus W. Scherer 12*0e209d39SAndroid Build Coastguard Worker */ 13*0e209d39SAndroid Build Coastguard Worker 14*0e209d39SAndroid Build Coastguard Worker #ifndef __COLLATION_H__ 15*0e209d39SAndroid Build Coastguard Worker #define __COLLATION_H__ 16*0e209d39SAndroid Build Coastguard Worker 17*0e209d39SAndroid Build Coastguard Worker #include "unicode/utypes.h" 18*0e209d39SAndroid Build Coastguard Worker 19*0e209d39SAndroid Build Coastguard Worker #if !UCONFIG_NO_COLLATION 20*0e209d39SAndroid Build Coastguard Worker 21*0e209d39SAndroid Build Coastguard Worker U_NAMESPACE_BEGIN 22*0e209d39SAndroid Build Coastguard Worker 23*0e209d39SAndroid Build Coastguard Worker /** 24*0e209d39SAndroid Build Coastguard Worker * Collation v2 basic definitions and static helper functions. 25*0e209d39SAndroid Build Coastguard Worker * 26*0e209d39SAndroid Build Coastguard Worker * Data structures except for expansion tables store 32-bit CEs which are 27*0e209d39SAndroid Build Coastguard Worker * either specials (see tags below) or are compact forms of 64-bit CEs. 28*0e209d39SAndroid Build Coastguard Worker */ 29*0e209d39SAndroid Build Coastguard Worker class U_I18N_API Collation { 30*0e209d39SAndroid Build Coastguard Worker public: 31*0e209d39SAndroid Build Coastguard Worker // Special sort key bytes for all levels. 32*0e209d39SAndroid Build Coastguard Worker static const uint8_t TERMINATOR_BYTE = 0; 33*0e209d39SAndroid Build Coastguard Worker static const uint8_t LEVEL_SEPARATOR_BYTE = 1; 34*0e209d39SAndroid Build Coastguard Worker 35*0e209d39SAndroid Build Coastguard Worker /** The secondary/tertiary lower limit for tailoring before any root elements. */ 36*0e209d39SAndroid Build Coastguard Worker static const uint32_t BEFORE_WEIGHT16 = 0x0100; 37*0e209d39SAndroid Build Coastguard Worker 38*0e209d39SAndroid Build Coastguard Worker /** 39*0e209d39SAndroid Build Coastguard Worker * Merge-sort-key separator. 40*0e209d39SAndroid Build Coastguard Worker * Same as the unique primary and identical-level weights of U+FFFE. 41*0e209d39SAndroid Build Coastguard Worker * Must not be used as primary compression low terminator. 42*0e209d39SAndroid Build Coastguard Worker * Otherwise usable. 43*0e209d39SAndroid Build Coastguard Worker */ 44*0e209d39SAndroid Build Coastguard Worker static const uint8_t MERGE_SEPARATOR_BYTE = 2; 45*0e209d39SAndroid Build Coastguard Worker static const uint32_t MERGE_SEPARATOR_PRIMARY = 0x02000000; // U+FFFE 46*0e209d39SAndroid Build Coastguard Worker static const uint32_t MERGE_SEPARATOR_CE32 = 0x02000505; // U+FFFE 47*0e209d39SAndroid Build Coastguard Worker 48*0e209d39SAndroid Build Coastguard Worker /** 49*0e209d39SAndroid Build Coastguard Worker * Primary compression low terminator, must be greater than MERGE_SEPARATOR_BYTE. 50*0e209d39SAndroid Build Coastguard Worker * Reserved value in primary second byte if the lead byte is compressible. 51*0e209d39SAndroid Build Coastguard Worker * Otherwise usable in all CE weight bytes. 52*0e209d39SAndroid Build Coastguard Worker */ 53*0e209d39SAndroid Build Coastguard Worker static const uint8_t PRIMARY_COMPRESSION_LOW_BYTE = 3; 54*0e209d39SAndroid Build Coastguard Worker /** 55*0e209d39SAndroid Build Coastguard Worker * Primary compression high terminator. 56*0e209d39SAndroid Build Coastguard Worker * Reserved value in primary second byte if the lead byte is compressible. 57*0e209d39SAndroid Build Coastguard Worker * Otherwise usable in all CE weight bytes. 58*0e209d39SAndroid Build Coastguard Worker */ 59*0e209d39SAndroid Build Coastguard Worker static const uint8_t PRIMARY_COMPRESSION_HIGH_BYTE = 0xff; 60*0e209d39SAndroid Build Coastguard Worker 61*0e209d39SAndroid Build Coastguard Worker /** Default secondary/tertiary weight lead byte. */ 62*0e209d39SAndroid Build Coastguard Worker static const uint8_t COMMON_BYTE = 5; 63*0e209d39SAndroid Build Coastguard Worker static const uint32_t COMMON_WEIGHT16 = 0x0500; 64*0e209d39SAndroid Build Coastguard Worker /** Middle 16 bits of a CE with a common secondary weight. */ 65*0e209d39SAndroid Build Coastguard Worker static const uint32_t COMMON_SECONDARY_CE = 0x05000000; 66*0e209d39SAndroid Build Coastguard Worker /** Lower 16 bits of a CE with a common tertiary weight. */ 67*0e209d39SAndroid Build Coastguard Worker static const uint32_t COMMON_TERTIARY_CE = 0x0500; 68*0e209d39SAndroid Build Coastguard Worker /** Lower 32 bits of a CE with common secondary and tertiary weights. */ 69*0e209d39SAndroid Build Coastguard Worker static const uint32_t COMMON_SEC_AND_TER_CE = 0x05000500; 70*0e209d39SAndroid Build Coastguard Worker 71*0e209d39SAndroid Build Coastguard Worker static const uint32_t SECONDARY_MASK = 0xffff0000; 72*0e209d39SAndroid Build Coastguard Worker static const uint32_t CASE_MASK = 0xc000; 73*0e209d39SAndroid Build Coastguard Worker static const uint32_t SECONDARY_AND_CASE_MASK = SECONDARY_MASK | CASE_MASK; 74*0e209d39SAndroid Build Coastguard Worker /** Only the 2*6 bits for the pure tertiary weight. */ 75*0e209d39SAndroid Build Coastguard Worker static const uint32_t ONLY_TERTIARY_MASK = 0x3f3f; 76*0e209d39SAndroid Build Coastguard Worker /** Only the secondary & tertiary bits; no case, no quaternary. */ 77*0e209d39SAndroid Build Coastguard Worker static const uint32_t ONLY_SEC_TER_MASK = SECONDARY_MASK | ONLY_TERTIARY_MASK; 78*0e209d39SAndroid Build Coastguard Worker /** Case bits and tertiary bits. */ 79*0e209d39SAndroid Build Coastguard Worker static const uint32_t CASE_AND_TERTIARY_MASK = CASE_MASK | ONLY_TERTIARY_MASK; 80*0e209d39SAndroid Build Coastguard Worker static const uint32_t QUATERNARY_MASK = 0xc0; 81*0e209d39SAndroid Build Coastguard Worker /** Case bits and quaternary bits. */ 82*0e209d39SAndroid Build Coastguard Worker static const uint32_t CASE_AND_QUATERNARY_MASK = CASE_MASK | QUATERNARY_MASK; 83*0e209d39SAndroid Build Coastguard Worker 84*0e209d39SAndroid Build Coastguard Worker static const uint8_t UNASSIGNED_IMPLICIT_BYTE = 0xfe; // compressible 85*0e209d39SAndroid Build Coastguard Worker /** 86*0e209d39SAndroid Build Coastguard Worker * First unassigned: AlphabeticIndex overflow boundary. 87*0e209d39SAndroid Build Coastguard Worker * We want a 3-byte primary so that it fits into the root elements table. 88*0e209d39SAndroid Build Coastguard Worker * 89*0e209d39SAndroid Build Coastguard Worker * This 3-byte primary will not collide with 90*0e209d39SAndroid Build Coastguard Worker * any unassigned-implicit 4-byte primaries because 91*0e209d39SAndroid Build Coastguard Worker * the first few hundred Unicode code points all have real mappings. 92*0e209d39SAndroid Build Coastguard Worker */ 93*0e209d39SAndroid Build Coastguard Worker static const uint32_t FIRST_UNASSIGNED_PRIMARY = 0xfe040200; 94*0e209d39SAndroid Build Coastguard Worker 95*0e209d39SAndroid Build Coastguard Worker static const uint8_t TRAIL_WEIGHT_BYTE = 0xff; // not compressible 96*0e209d39SAndroid Build Coastguard Worker static const uint32_t FIRST_TRAILING_PRIMARY = 0xff020200; // [first trailing] 97*0e209d39SAndroid Build Coastguard Worker static const uint32_t MAX_PRIMARY = 0xffff0000; // U+FFFF 98*0e209d39SAndroid Build Coastguard Worker static const uint32_t MAX_REGULAR_CE32 = 0xffff0505; // U+FFFF 99*0e209d39SAndroid Build Coastguard Worker 100*0e209d39SAndroid Build Coastguard Worker // CE32 value for U+FFFD as well as illegal UTF-8 byte sequences (which behave like U+FFFD). 101*0e209d39SAndroid Build Coastguard Worker // We use the third-highest primary weight for U+FFFD (as in UCA 6.3+). 102*0e209d39SAndroid Build Coastguard Worker static const uint32_t FFFD_PRIMARY = MAX_PRIMARY - 0x20000; 103*0e209d39SAndroid Build Coastguard Worker static const uint32_t FFFD_CE32 = MAX_REGULAR_CE32 - 0x20000; 104*0e209d39SAndroid Build Coastguard Worker 105*0e209d39SAndroid Build Coastguard Worker /** 106*0e209d39SAndroid Build Coastguard Worker * A CE32 is special if its low byte is this or greater. 107*0e209d39SAndroid Build Coastguard Worker * Impossible case bits 11 mark special CE32s. 108*0e209d39SAndroid Build Coastguard Worker * This value itself is used to indicate a fallback to the base collator. 109*0e209d39SAndroid Build Coastguard Worker */ 110*0e209d39SAndroid Build Coastguard Worker static const uint8_t SPECIAL_CE32_LOW_BYTE = 0xc0; 111*0e209d39SAndroid Build Coastguard Worker static const uint32_t FALLBACK_CE32 = SPECIAL_CE32_LOW_BYTE; 112*0e209d39SAndroid Build Coastguard Worker /** 113*0e209d39SAndroid Build Coastguard Worker * Low byte of a long-primary special CE32. 114*0e209d39SAndroid Build Coastguard Worker */ 115*0e209d39SAndroid Build Coastguard Worker static const uint8_t LONG_PRIMARY_CE32_LOW_BYTE = 0xc1; // SPECIAL_CE32_LOW_BYTE | LONG_PRIMARY_TAG 116*0e209d39SAndroid Build Coastguard Worker 117*0e209d39SAndroid Build Coastguard Worker static const uint32_t UNASSIGNED_CE32 = 0xffffffff; // Compute an unassigned-implicit CE. 118*0e209d39SAndroid Build Coastguard Worker 119*0e209d39SAndroid Build Coastguard Worker static const uint32_t NO_CE32 = 1; 120*0e209d39SAndroid Build Coastguard Worker 121*0e209d39SAndroid Build Coastguard Worker /** No CE: End of input. Only used in runtime code, not stored in data. */ 122*0e209d39SAndroid Build Coastguard Worker static const uint32_t NO_CE_PRIMARY = 1; // not a left-adjusted weight 123*0e209d39SAndroid Build Coastguard Worker static const uint32_t NO_CE_WEIGHT16 = 0x0100; // weight of LEVEL_SEPARATOR_BYTE 124*0e209d39SAndroid Build Coastguard Worker static const int64_t NO_CE = INT64_C(0x101000100); // NO_CE_PRIMARY, NO_CE_WEIGHT16, NO_CE_WEIGHT16 125*0e209d39SAndroid Build Coastguard Worker 126*0e209d39SAndroid Build Coastguard Worker /** Sort key levels. */ 127*0e209d39SAndroid Build Coastguard Worker enum Level { 128*0e209d39SAndroid Build Coastguard Worker /** Unspecified level. */ 129*0e209d39SAndroid Build Coastguard Worker NO_LEVEL, 130*0e209d39SAndroid Build Coastguard Worker PRIMARY_LEVEL, 131*0e209d39SAndroid Build Coastguard Worker SECONDARY_LEVEL, 132*0e209d39SAndroid Build Coastguard Worker CASE_LEVEL, 133*0e209d39SAndroid Build Coastguard Worker TERTIARY_LEVEL, 134*0e209d39SAndroid Build Coastguard Worker QUATERNARY_LEVEL, 135*0e209d39SAndroid Build Coastguard Worker IDENTICAL_LEVEL, 136*0e209d39SAndroid Build Coastguard Worker /** Beyond sort key bytes. */ 137*0e209d39SAndroid Build Coastguard Worker ZERO_LEVEL 138*0e209d39SAndroid Build Coastguard Worker }; 139*0e209d39SAndroid Build Coastguard Worker 140*0e209d39SAndroid Build Coastguard Worker /** 141*0e209d39SAndroid Build Coastguard Worker * Sort key level flags: xx_FLAG = 1 << xx_LEVEL. 142*0e209d39SAndroid Build Coastguard Worker * In Java, use enum Level with flag() getters, or use EnumSet rather than hand-made bit sets. 143*0e209d39SAndroid Build Coastguard Worker */ 144*0e209d39SAndroid Build Coastguard Worker static const uint32_t NO_LEVEL_FLAG = 1; 145*0e209d39SAndroid Build Coastguard Worker static const uint32_t PRIMARY_LEVEL_FLAG = 2; 146*0e209d39SAndroid Build Coastguard Worker static const uint32_t SECONDARY_LEVEL_FLAG = 4; 147*0e209d39SAndroid Build Coastguard Worker static const uint32_t CASE_LEVEL_FLAG = 8; 148*0e209d39SAndroid Build Coastguard Worker static const uint32_t TERTIARY_LEVEL_FLAG = 0x10; 149*0e209d39SAndroid Build Coastguard Worker static const uint32_t QUATERNARY_LEVEL_FLAG = 0x20; 150*0e209d39SAndroid Build Coastguard Worker static const uint32_t IDENTICAL_LEVEL_FLAG = 0x40; 151*0e209d39SAndroid Build Coastguard Worker static const uint32_t ZERO_LEVEL_FLAG = 0x80; 152*0e209d39SAndroid Build Coastguard Worker 153*0e209d39SAndroid Build Coastguard Worker /** 154*0e209d39SAndroid Build Coastguard Worker * Special-CE32 tags, from bits 3..0 of a special 32-bit CE. 155*0e209d39SAndroid Build Coastguard Worker * Bits 31..8 are available for tag-specific data. 156*0e209d39SAndroid Build Coastguard Worker * Bits 5..4: Reserved. May be used in the future to indicate lccc!=0 and tccc!=0. 157*0e209d39SAndroid Build Coastguard Worker */ 158*0e209d39SAndroid Build Coastguard Worker enum { 159*0e209d39SAndroid Build Coastguard Worker /** 160*0e209d39SAndroid Build Coastguard Worker * Fall back to the base collator. 161*0e209d39SAndroid Build Coastguard Worker * This is the tag value in SPECIAL_CE32_LOW_BYTE and FALLBACK_CE32. 162*0e209d39SAndroid Build Coastguard Worker * Bits 31..8: Unused, 0. 163*0e209d39SAndroid Build Coastguard Worker */ 164*0e209d39SAndroid Build Coastguard Worker FALLBACK_TAG = 0, 165*0e209d39SAndroid Build Coastguard Worker /** 166*0e209d39SAndroid Build Coastguard Worker * Long-primary CE with COMMON_SEC_AND_TER_CE. 167*0e209d39SAndroid Build Coastguard Worker * Bits 31..8: Three-byte primary. 168*0e209d39SAndroid Build Coastguard Worker */ 169*0e209d39SAndroid Build Coastguard Worker LONG_PRIMARY_TAG = 1, 170*0e209d39SAndroid Build Coastguard Worker /** 171*0e209d39SAndroid Build Coastguard Worker * Long-secondary CE with zero primary. 172*0e209d39SAndroid Build Coastguard Worker * Bits 31..16: Secondary weight. 173*0e209d39SAndroid Build Coastguard Worker * Bits 15.. 8: Tertiary weight. 174*0e209d39SAndroid Build Coastguard Worker */ 175*0e209d39SAndroid Build Coastguard Worker LONG_SECONDARY_TAG = 2, 176*0e209d39SAndroid Build Coastguard Worker /** 177*0e209d39SAndroid Build Coastguard Worker * Unused. 178*0e209d39SAndroid Build Coastguard Worker * May be used in the future for single-byte secondary CEs (SHORT_SECONDARY_TAG), 179*0e209d39SAndroid Build Coastguard Worker * storing the secondary in bits 31..24, the ccc in bits 23..16, 180*0e209d39SAndroid Build Coastguard Worker * and the tertiary in bits 15..8. 181*0e209d39SAndroid Build Coastguard Worker */ 182*0e209d39SAndroid Build Coastguard Worker RESERVED_TAG_3 = 3, 183*0e209d39SAndroid Build Coastguard Worker /** 184*0e209d39SAndroid Build Coastguard Worker * Latin mini expansions of two simple CEs [pp, 05, tt] [00, ss, 05]. 185*0e209d39SAndroid Build Coastguard Worker * Bits 31..24: Single-byte primary weight pp of the first CE. 186*0e209d39SAndroid Build Coastguard Worker * Bits 23..16: Tertiary weight tt of the first CE. 187*0e209d39SAndroid Build Coastguard Worker * Bits 15.. 8: Secondary weight ss of the second CE. 188*0e209d39SAndroid Build Coastguard Worker */ 189*0e209d39SAndroid Build Coastguard Worker LATIN_EXPANSION_TAG = 4, 190*0e209d39SAndroid Build Coastguard Worker /** 191*0e209d39SAndroid Build Coastguard Worker * Points to one or more simple/long-primary/long-secondary 32-bit CE32s. 192*0e209d39SAndroid Build Coastguard Worker * Bits 31..13: Index into uint32_t table. 193*0e209d39SAndroid Build Coastguard Worker * Bits 12.. 8: Length=1..31. 194*0e209d39SAndroid Build Coastguard Worker */ 195*0e209d39SAndroid Build Coastguard Worker EXPANSION32_TAG = 5, 196*0e209d39SAndroid Build Coastguard Worker /** 197*0e209d39SAndroid Build Coastguard Worker * Points to one or more 64-bit CEs. 198*0e209d39SAndroid Build Coastguard Worker * Bits 31..13: Index into CE table. 199*0e209d39SAndroid Build Coastguard Worker * Bits 12.. 8: Length=1..31. 200*0e209d39SAndroid Build Coastguard Worker */ 201*0e209d39SAndroid Build Coastguard Worker EXPANSION_TAG = 6, 202*0e209d39SAndroid Build Coastguard Worker /** 203*0e209d39SAndroid Build Coastguard Worker * Builder data, used only in the CollationDataBuilder, not in runtime data. 204*0e209d39SAndroid Build Coastguard Worker * 205*0e209d39SAndroid Build Coastguard Worker * If bit 8 is 0: Builder context, points to a list of context-sensitive mappings. 206*0e209d39SAndroid Build Coastguard Worker * Bits 31..13: Index to the builder's list of ConditionalCE32 for this character. 207*0e209d39SAndroid Build Coastguard Worker * Bits 12.. 9: Unused, 0. 208*0e209d39SAndroid Build Coastguard Worker * 209*0e209d39SAndroid Build Coastguard Worker * If bit 8 is 1 (IS_BUILDER_JAMO_CE32): Builder-only jamoCE32 value. 210*0e209d39SAndroid Build Coastguard Worker * The builder fetches the Jamo CE32 from the trie. 211*0e209d39SAndroid Build Coastguard Worker * Bits 31..13: Jamo code point. 212*0e209d39SAndroid Build Coastguard Worker * Bits 12.. 9: Unused, 0. 213*0e209d39SAndroid Build Coastguard Worker */ 214*0e209d39SAndroid Build Coastguard Worker BUILDER_DATA_TAG = 7, 215*0e209d39SAndroid Build Coastguard Worker /** 216*0e209d39SAndroid Build Coastguard Worker * Points to prefix trie. 217*0e209d39SAndroid Build Coastguard Worker * Bits 31..13: Index into prefix/contraction data. 218*0e209d39SAndroid Build Coastguard Worker * Bits 12.. 8: Unused, 0. 219*0e209d39SAndroid Build Coastguard Worker */ 220*0e209d39SAndroid Build Coastguard Worker PREFIX_TAG = 8, 221*0e209d39SAndroid Build Coastguard Worker /** 222*0e209d39SAndroid Build Coastguard Worker * Points to contraction data. 223*0e209d39SAndroid Build Coastguard Worker * Bits 31..13: Index into prefix/contraction data. 224*0e209d39SAndroid Build Coastguard Worker * Bit 12: Unused, 0. 225*0e209d39SAndroid Build Coastguard Worker * Bit 11: CONTRACT_HAS_STARTER flag. (Used by ICU4X only.) 226*0e209d39SAndroid Build Coastguard Worker * Bit 10: CONTRACT_TRAILING_CCC flag. 227*0e209d39SAndroid Build Coastguard Worker * Bit 9: CONTRACT_NEXT_CCC flag. 228*0e209d39SAndroid Build Coastguard Worker * Bit 8: CONTRACT_SINGLE_CP_NO_MATCH flag. 229*0e209d39SAndroid Build Coastguard Worker */ 230*0e209d39SAndroid Build Coastguard Worker CONTRACTION_TAG = 9, 231*0e209d39SAndroid Build Coastguard Worker /** 232*0e209d39SAndroid Build Coastguard Worker * Decimal digit. 233*0e209d39SAndroid Build Coastguard Worker * Bits 31..13: Index into uint32_t table for non-numeric-collation CE32. 234*0e209d39SAndroid Build Coastguard Worker * Bit 12: Unused, 0. 235*0e209d39SAndroid Build Coastguard Worker * Bits 11.. 8: Digit value 0..9. 236*0e209d39SAndroid Build Coastguard Worker */ 237*0e209d39SAndroid Build Coastguard Worker DIGIT_TAG = 10, 238*0e209d39SAndroid Build Coastguard Worker /** 239*0e209d39SAndroid Build Coastguard Worker * Tag for U+0000, for moving the NUL-termination handling 240*0e209d39SAndroid Build Coastguard Worker * from the regular fastpath into specials-handling code. 241*0e209d39SAndroid Build Coastguard Worker * Bits 31..8: Unused, 0. 242*0e209d39SAndroid Build Coastguard Worker */ 243*0e209d39SAndroid Build Coastguard Worker U0000_TAG = 11, 244*0e209d39SAndroid Build Coastguard Worker /** 245*0e209d39SAndroid Build Coastguard Worker * Tag for a Hangul syllable. 246*0e209d39SAndroid Build Coastguard Worker * Bits 31..9: Unused, 0. 247*0e209d39SAndroid Build Coastguard Worker * Bit 8: HANGUL_NO_SPECIAL_JAMO flag. 248*0e209d39SAndroid Build Coastguard Worker */ 249*0e209d39SAndroid Build Coastguard Worker HANGUL_TAG = 12, 250*0e209d39SAndroid Build Coastguard Worker /** 251*0e209d39SAndroid Build Coastguard Worker * Tag for a lead surrogate code unit. 252*0e209d39SAndroid Build Coastguard Worker * Optional optimization for UTF-16 string processing. 253*0e209d39SAndroid Build Coastguard Worker * Bits 31..10: Unused, 0. 254*0e209d39SAndroid Build Coastguard Worker * 9.. 8: =0: All associated supplementary code points are unassigned-implicit. 255*0e209d39SAndroid Build Coastguard Worker * =1: All associated supplementary code points fall back to the base data. 256*0e209d39SAndroid Build Coastguard Worker * else: (Normally 2) Look up the data for the supplementary code point. 257*0e209d39SAndroid Build Coastguard Worker */ 258*0e209d39SAndroid Build Coastguard Worker LEAD_SURROGATE_TAG = 13, 259*0e209d39SAndroid Build Coastguard Worker /** 260*0e209d39SAndroid Build Coastguard Worker * Tag for CEs with primary weights in code point order. 261*0e209d39SAndroid Build Coastguard Worker * Bits 31..13: Index into CE table, for one data "CE". 262*0e209d39SAndroid Build Coastguard Worker * Bits 12.. 8: Unused, 0. 263*0e209d39SAndroid Build Coastguard Worker * 264*0e209d39SAndroid Build Coastguard Worker * This data "CE" has the following bit fields: 265*0e209d39SAndroid Build Coastguard Worker * Bits 63..32: Three-byte primary pppppp00. 266*0e209d39SAndroid Build Coastguard Worker * 31.. 8: Start/base code point of the in-order range. 267*0e209d39SAndroid Build Coastguard Worker * 7: Flag isCompressible primary. 268*0e209d39SAndroid Build Coastguard Worker * 6.. 0: Per-code point primary-weight increment. 269*0e209d39SAndroid Build Coastguard Worker */ 270*0e209d39SAndroid Build Coastguard Worker OFFSET_TAG = 14, 271*0e209d39SAndroid Build Coastguard Worker /** 272*0e209d39SAndroid Build Coastguard Worker * Implicit CE tag. Compute an unassigned-implicit CE. 273*0e209d39SAndroid Build Coastguard Worker * All bits are set (UNASSIGNED_CE32=0xffffffff). 274*0e209d39SAndroid Build Coastguard Worker */ 275*0e209d39SAndroid Build Coastguard Worker IMPLICIT_TAG = 15 276*0e209d39SAndroid Build Coastguard Worker }; 277*0e209d39SAndroid Build Coastguard Worker isAssignedCE32(uint32_t ce32)278*0e209d39SAndroid Build Coastguard Worker static UBool isAssignedCE32(uint32_t ce32) { 279*0e209d39SAndroid Build Coastguard Worker return ce32 != FALLBACK_CE32 && ce32 != UNASSIGNED_CE32; 280*0e209d39SAndroid Build Coastguard Worker } 281*0e209d39SAndroid Build Coastguard Worker 282*0e209d39SAndroid Build Coastguard Worker /** 283*0e209d39SAndroid Build Coastguard Worker * We limit the number of CEs in an expansion 284*0e209d39SAndroid Build Coastguard Worker * so that we can use a small number of length bits in the data structure, 285*0e209d39SAndroid Build Coastguard Worker * and so that an implementation can copy CEs at runtime without growing a destination buffer. 286*0e209d39SAndroid Build Coastguard Worker */ 287*0e209d39SAndroid Build Coastguard Worker static const int32_t MAX_EXPANSION_LENGTH = 31; 288*0e209d39SAndroid Build Coastguard Worker static const int32_t MAX_INDEX = 0x7ffff; 289*0e209d39SAndroid Build Coastguard Worker 290*0e209d39SAndroid Build Coastguard Worker /** 291*0e209d39SAndroid Build Coastguard Worker * Set if there is no match for the single (no-suffix) character itself. 292*0e209d39SAndroid Build Coastguard Worker * This is only possible if there is a prefix. 293*0e209d39SAndroid Build Coastguard Worker * In this case, discontiguous contraction matching cannot add combining marks 294*0e209d39SAndroid Build Coastguard Worker * starting from an empty suffix. 295*0e209d39SAndroid Build Coastguard Worker * The default CE32 is used anyway if there is no suffix match. 296*0e209d39SAndroid Build Coastguard Worker */ 297*0e209d39SAndroid Build Coastguard Worker static const uint32_t CONTRACT_SINGLE_CP_NO_MATCH = 0x100; 298*0e209d39SAndroid Build Coastguard Worker /** Set if the first character of every contraction suffix has lccc!=0. */ 299*0e209d39SAndroid Build Coastguard Worker static const uint32_t CONTRACT_NEXT_CCC = 0x200; 300*0e209d39SAndroid Build Coastguard Worker /** Set if any contraction suffix ends with lccc!=0. */ 301*0e209d39SAndroid Build Coastguard Worker static const uint32_t CONTRACT_TRAILING_CCC = 0x400; 302*0e209d39SAndroid Build Coastguard Worker /** Set if any contraction suffix contains a starter. (Used by ICU4X only.) */ 303*0e209d39SAndroid Build Coastguard Worker static const uint32_t CONTRACT_HAS_STARTER = 0x800; 304*0e209d39SAndroid Build Coastguard Worker 305*0e209d39SAndroid Build Coastguard Worker /** For HANGUL_TAG: None of its Jamo CE32s isSpecialCE32(). */ 306*0e209d39SAndroid Build Coastguard Worker static const uint32_t HANGUL_NO_SPECIAL_JAMO = 0x100; 307*0e209d39SAndroid Build Coastguard Worker 308*0e209d39SAndroid Build Coastguard Worker static const uint32_t LEAD_ALL_UNASSIGNED = 0; 309*0e209d39SAndroid Build Coastguard Worker static const uint32_t LEAD_ALL_FALLBACK = 0x100; 310*0e209d39SAndroid Build Coastguard Worker static const uint32_t LEAD_MIXED = 0x200; 311*0e209d39SAndroid Build Coastguard Worker static const uint32_t LEAD_TYPE_MASK = 0x300; 312*0e209d39SAndroid Build Coastguard Worker makeLongPrimaryCE32(uint32_t p)313*0e209d39SAndroid Build Coastguard Worker static uint32_t makeLongPrimaryCE32(uint32_t p) { return p | LONG_PRIMARY_CE32_LOW_BYTE; } 314*0e209d39SAndroid Build Coastguard Worker 315*0e209d39SAndroid Build Coastguard Worker /** Turns the long-primary CE32 into a primary weight pppppp00. */ primaryFromLongPrimaryCE32(uint32_t ce32)316*0e209d39SAndroid Build Coastguard Worker static inline uint32_t primaryFromLongPrimaryCE32(uint32_t ce32) { 317*0e209d39SAndroid Build Coastguard Worker return ce32 & 0xffffff00; 318*0e209d39SAndroid Build Coastguard Worker } ceFromLongPrimaryCE32(uint32_t ce32)319*0e209d39SAndroid Build Coastguard Worker static inline int64_t ceFromLongPrimaryCE32(uint32_t ce32) { 320*0e209d39SAndroid Build Coastguard Worker return ((int64_t)(ce32 & 0xffffff00) << 32) | COMMON_SEC_AND_TER_CE; 321*0e209d39SAndroid Build Coastguard Worker } 322*0e209d39SAndroid Build Coastguard Worker makeLongSecondaryCE32(uint32_t lower32)323*0e209d39SAndroid Build Coastguard Worker static uint32_t makeLongSecondaryCE32(uint32_t lower32) { 324*0e209d39SAndroid Build Coastguard Worker return lower32 | SPECIAL_CE32_LOW_BYTE | LONG_SECONDARY_TAG; 325*0e209d39SAndroid Build Coastguard Worker } ceFromLongSecondaryCE32(uint32_t ce32)326*0e209d39SAndroid Build Coastguard Worker static inline int64_t ceFromLongSecondaryCE32(uint32_t ce32) { 327*0e209d39SAndroid Build Coastguard Worker return ce32 & 0xffffff00; 328*0e209d39SAndroid Build Coastguard Worker } 329*0e209d39SAndroid Build Coastguard Worker 330*0e209d39SAndroid Build Coastguard Worker /** Makes a special CE32 with tag, index and length. */ makeCE32FromTagIndexAndLength(int32_t tag,int32_t index,int32_t length)331*0e209d39SAndroid Build Coastguard Worker static uint32_t makeCE32FromTagIndexAndLength(int32_t tag, int32_t index, int32_t length) { 332*0e209d39SAndroid Build Coastguard Worker return (index << 13) | (length << 8) | SPECIAL_CE32_LOW_BYTE | tag; 333*0e209d39SAndroid Build Coastguard Worker } 334*0e209d39SAndroid Build Coastguard Worker /** Makes a special CE32 with only tag and index. */ makeCE32FromTagAndIndex(int32_t tag,int32_t index)335*0e209d39SAndroid Build Coastguard Worker static uint32_t makeCE32FromTagAndIndex(int32_t tag, int32_t index) { 336*0e209d39SAndroid Build Coastguard Worker return (index << 13) | SPECIAL_CE32_LOW_BYTE | tag; 337*0e209d39SAndroid Build Coastguard Worker } 338*0e209d39SAndroid Build Coastguard Worker isSpecialCE32(uint32_t ce32)339*0e209d39SAndroid Build Coastguard Worker static inline UBool isSpecialCE32(uint32_t ce32) { 340*0e209d39SAndroid Build Coastguard Worker return (ce32 & 0xff) >= SPECIAL_CE32_LOW_BYTE; 341*0e209d39SAndroid Build Coastguard Worker } 342*0e209d39SAndroid Build Coastguard Worker tagFromCE32(uint32_t ce32)343*0e209d39SAndroid Build Coastguard Worker static inline int32_t tagFromCE32(uint32_t ce32) { 344*0e209d39SAndroid Build Coastguard Worker return (int32_t)(ce32 & 0xf); 345*0e209d39SAndroid Build Coastguard Worker } 346*0e209d39SAndroid Build Coastguard Worker hasCE32Tag(uint32_t ce32,int32_t tag)347*0e209d39SAndroid Build Coastguard Worker static inline UBool hasCE32Tag(uint32_t ce32, int32_t tag) { 348*0e209d39SAndroid Build Coastguard Worker return isSpecialCE32(ce32) && tagFromCE32(ce32) == tag; 349*0e209d39SAndroid Build Coastguard Worker } 350*0e209d39SAndroid Build Coastguard Worker isLongPrimaryCE32(uint32_t ce32)351*0e209d39SAndroid Build Coastguard Worker static inline UBool isLongPrimaryCE32(uint32_t ce32) { 352*0e209d39SAndroid Build Coastguard Worker return hasCE32Tag(ce32, LONG_PRIMARY_TAG); 353*0e209d39SAndroid Build Coastguard Worker } 354*0e209d39SAndroid Build Coastguard Worker isSimpleOrLongCE32(uint32_t ce32)355*0e209d39SAndroid Build Coastguard Worker static UBool isSimpleOrLongCE32(uint32_t ce32) { 356*0e209d39SAndroid Build Coastguard Worker return !isSpecialCE32(ce32) || 357*0e209d39SAndroid Build Coastguard Worker tagFromCE32(ce32) == LONG_PRIMARY_TAG || 358*0e209d39SAndroid Build Coastguard Worker tagFromCE32(ce32) == LONG_SECONDARY_TAG; 359*0e209d39SAndroid Build Coastguard Worker } 360*0e209d39SAndroid Build Coastguard Worker 361*0e209d39SAndroid Build Coastguard Worker /** 362*0e209d39SAndroid Build Coastguard Worker * @return true if the ce32 yields one or more CEs without further data lookups 363*0e209d39SAndroid Build Coastguard Worker */ isSelfContainedCE32(uint32_t ce32)364*0e209d39SAndroid Build Coastguard Worker static UBool isSelfContainedCE32(uint32_t ce32) { 365*0e209d39SAndroid Build Coastguard Worker return !isSpecialCE32(ce32) || 366*0e209d39SAndroid Build Coastguard Worker tagFromCE32(ce32) == LONG_PRIMARY_TAG || 367*0e209d39SAndroid Build Coastguard Worker tagFromCE32(ce32) == LONG_SECONDARY_TAG || 368*0e209d39SAndroid Build Coastguard Worker tagFromCE32(ce32) == LATIN_EXPANSION_TAG; 369*0e209d39SAndroid Build Coastguard Worker } 370*0e209d39SAndroid Build Coastguard Worker isPrefixCE32(uint32_t ce32)371*0e209d39SAndroid Build Coastguard Worker static inline UBool isPrefixCE32(uint32_t ce32) { 372*0e209d39SAndroid Build Coastguard Worker return hasCE32Tag(ce32, PREFIX_TAG); 373*0e209d39SAndroid Build Coastguard Worker } 374*0e209d39SAndroid Build Coastguard Worker isContractionCE32(uint32_t ce32)375*0e209d39SAndroid Build Coastguard Worker static inline UBool isContractionCE32(uint32_t ce32) { 376*0e209d39SAndroid Build Coastguard Worker return hasCE32Tag(ce32, CONTRACTION_TAG); 377*0e209d39SAndroid Build Coastguard Worker } 378*0e209d39SAndroid Build Coastguard Worker ce32HasContext(uint32_t ce32)379*0e209d39SAndroid Build Coastguard Worker static inline UBool ce32HasContext(uint32_t ce32) { 380*0e209d39SAndroid Build Coastguard Worker return isSpecialCE32(ce32) && 381*0e209d39SAndroid Build Coastguard Worker (tagFromCE32(ce32) == PREFIX_TAG || 382*0e209d39SAndroid Build Coastguard Worker tagFromCE32(ce32) == CONTRACTION_TAG); 383*0e209d39SAndroid Build Coastguard Worker } 384*0e209d39SAndroid Build Coastguard Worker 385*0e209d39SAndroid Build Coastguard Worker /** 386*0e209d39SAndroid Build Coastguard Worker * Get the first of the two Latin-expansion CEs encoded in ce32. 387*0e209d39SAndroid Build Coastguard Worker * @see LATIN_EXPANSION_TAG 388*0e209d39SAndroid Build Coastguard Worker */ latinCE0FromCE32(uint32_t ce32)389*0e209d39SAndroid Build Coastguard Worker static inline int64_t latinCE0FromCE32(uint32_t ce32) { 390*0e209d39SAndroid Build Coastguard Worker return ((int64_t)(ce32 & 0xff000000) << 32) | COMMON_SECONDARY_CE | ((ce32 & 0xff0000) >> 8); 391*0e209d39SAndroid Build Coastguard Worker } 392*0e209d39SAndroid Build Coastguard Worker 393*0e209d39SAndroid Build Coastguard Worker /** 394*0e209d39SAndroid Build Coastguard Worker * Get the second of the two Latin-expansion CEs encoded in ce32. 395*0e209d39SAndroid Build Coastguard Worker * @see LATIN_EXPANSION_TAG 396*0e209d39SAndroid Build Coastguard Worker */ latinCE1FromCE32(uint32_t ce32)397*0e209d39SAndroid Build Coastguard Worker static inline int64_t latinCE1FromCE32(uint32_t ce32) { 398*0e209d39SAndroid Build Coastguard Worker return ((ce32 & 0xff00) << 16) | COMMON_TERTIARY_CE; 399*0e209d39SAndroid Build Coastguard Worker } 400*0e209d39SAndroid Build Coastguard Worker 401*0e209d39SAndroid Build Coastguard Worker /** 402*0e209d39SAndroid Build Coastguard Worker * Returns the data index from a special CE32. 403*0e209d39SAndroid Build Coastguard Worker */ indexFromCE32(uint32_t ce32)404*0e209d39SAndroid Build Coastguard Worker static inline int32_t indexFromCE32(uint32_t ce32) { 405*0e209d39SAndroid Build Coastguard Worker return (int32_t)(ce32 >> 13); 406*0e209d39SAndroid Build Coastguard Worker } 407*0e209d39SAndroid Build Coastguard Worker 408*0e209d39SAndroid Build Coastguard Worker /** 409*0e209d39SAndroid Build Coastguard Worker * Returns the data length from a ce32. 410*0e209d39SAndroid Build Coastguard Worker */ lengthFromCE32(uint32_t ce32)411*0e209d39SAndroid Build Coastguard Worker static inline int32_t lengthFromCE32(uint32_t ce32) { 412*0e209d39SAndroid Build Coastguard Worker return (ce32 >> 8) & 31; 413*0e209d39SAndroid Build Coastguard Worker } 414*0e209d39SAndroid Build Coastguard Worker 415*0e209d39SAndroid Build Coastguard Worker /** 416*0e209d39SAndroid Build Coastguard Worker * Returns the digit value from a DIGIT_TAG ce32. 417*0e209d39SAndroid Build Coastguard Worker */ digitFromCE32(uint32_t ce32)418*0e209d39SAndroid Build Coastguard Worker static inline char digitFromCE32(uint32_t ce32) { 419*0e209d39SAndroid Build Coastguard Worker return (char)((ce32 >> 8) & 0xf); 420*0e209d39SAndroid Build Coastguard Worker } 421*0e209d39SAndroid Build Coastguard Worker 422*0e209d39SAndroid Build Coastguard Worker /** Returns a 64-bit CE from a simple CE32 (not special). */ ceFromSimpleCE32(uint32_t ce32)423*0e209d39SAndroid Build Coastguard Worker static inline int64_t ceFromSimpleCE32(uint32_t ce32) { 424*0e209d39SAndroid Build Coastguard Worker // normal form ppppsstt -> pppp0000ss00tt00 425*0e209d39SAndroid Build Coastguard Worker // assert (ce32 & 0xff) < SPECIAL_CE32_LOW_BYTE 426*0e209d39SAndroid Build Coastguard Worker return ((int64_t)(ce32 & 0xffff0000) << 32) | ((ce32 & 0xff00) << 16) | ((ce32 & 0xff) << 8); 427*0e209d39SAndroid Build Coastguard Worker } 428*0e209d39SAndroid Build Coastguard Worker 429*0e209d39SAndroid Build Coastguard Worker /** Returns a 64-bit CE from a simple/long-primary/long-secondary CE32. */ ceFromCE32(uint32_t ce32)430*0e209d39SAndroid Build Coastguard Worker static inline int64_t ceFromCE32(uint32_t ce32) { 431*0e209d39SAndroid Build Coastguard Worker uint32_t tertiary = ce32 & 0xff; 432*0e209d39SAndroid Build Coastguard Worker if(tertiary < SPECIAL_CE32_LOW_BYTE) { 433*0e209d39SAndroid Build Coastguard Worker // normal form ppppsstt -> pppp0000ss00tt00 434*0e209d39SAndroid Build Coastguard Worker return ((int64_t)(ce32 & 0xffff0000) << 32) | ((ce32 & 0xff00) << 16) | (tertiary << 8); 435*0e209d39SAndroid Build Coastguard Worker } else { 436*0e209d39SAndroid Build Coastguard Worker ce32 -= tertiary; 437*0e209d39SAndroid Build Coastguard Worker if((tertiary & 0xf) == LONG_PRIMARY_TAG) { 438*0e209d39SAndroid Build Coastguard Worker // long-primary form ppppppC1 -> pppppp00050000500 439*0e209d39SAndroid Build Coastguard Worker return ((int64_t)ce32 << 32) | COMMON_SEC_AND_TER_CE; 440*0e209d39SAndroid Build Coastguard Worker } else { 441*0e209d39SAndroid Build Coastguard Worker // long-secondary form ssssttC2 -> 00000000sssstt00 442*0e209d39SAndroid Build Coastguard Worker // assert (tertiary & 0xf) == LONG_SECONDARY_TAG 443*0e209d39SAndroid Build Coastguard Worker return ce32; 444*0e209d39SAndroid Build Coastguard Worker } 445*0e209d39SAndroid Build Coastguard Worker } 446*0e209d39SAndroid Build Coastguard Worker } 447*0e209d39SAndroid Build Coastguard Worker 448*0e209d39SAndroid Build Coastguard Worker /** Creates a CE from a primary weight. */ makeCE(uint32_t p)449*0e209d39SAndroid Build Coastguard Worker static inline int64_t makeCE(uint32_t p) { 450*0e209d39SAndroid Build Coastguard Worker return ((int64_t)p << 32) | COMMON_SEC_AND_TER_CE; 451*0e209d39SAndroid Build Coastguard Worker } 452*0e209d39SAndroid Build Coastguard Worker /** 453*0e209d39SAndroid Build Coastguard Worker * Creates a CE from a primary weight, 454*0e209d39SAndroid Build Coastguard Worker * 16-bit secondary/tertiary weights, and a 2-bit quaternary. 455*0e209d39SAndroid Build Coastguard Worker */ makeCE(uint32_t p,uint32_t s,uint32_t t,uint32_t q)456*0e209d39SAndroid Build Coastguard Worker static inline int64_t makeCE(uint32_t p, uint32_t s, uint32_t t, uint32_t q) { 457*0e209d39SAndroid Build Coastguard Worker return ((int64_t)p << 32) | (s << 16) | t | (q << 6); 458*0e209d39SAndroid Build Coastguard Worker } 459*0e209d39SAndroid Build Coastguard Worker 460*0e209d39SAndroid Build Coastguard Worker /** 461*0e209d39SAndroid Build Coastguard Worker * Increments a 2-byte primary by a code point offset. 462*0e209d39SAndroid Build Coastguard Worker */ 463*0e209d39SAndroid Build Coastguard Worker static uint32_t incTwoBytePrimaryByOffset(uint32_t basePrimary, UBool isCompressible, 464*0e209d39SAndroid Build Coastguard Worker int32_t offset); 465*0e209d39SAndroid Build Coastguard Worker 466*0e209d39SAndroid Build Coastguard Worker /** 467*0e209d39SAndroid Build Coastguard Worker * Increments a 3-byte primary by a code point offset. 468*0e209d39SAndroid Build Coastguard Worker */ 469*0e209d39SAndroid Build Coastguard Worker static uint32_t incThreeBytePrimaryByOffset(uint32_t basePrimary, UBool isCompressible, 470*0e209d39SAndroid Build Coastguard Worker int32_t offset); 471*0e209d39SAndroid Build Coastguard Worker 472*0e209d39SAndroid Build Coastguard Worker /** 473*0e209d39SAndroid Build Coastguard Worker * Decrements a 2-byte primary by one range step (1..0x7f). 474*0e209d39SAndroid Build Coastguard Worker */ 475*0e209d39SAndroid Build Coastguard Worker static uint32_t decTwoBytePrimaryByOneStep(uint32_t basePrimary, UBool isCompressible, int32_t step); 476*0e209d39SAndroid Build Coastguard Worker 477*0e209d39SAndroid Build Coastguard Worker /** 478*0e209d39SAndroid Build Coastguard Worker * Decrements a 3-byte primary by one range step (1..0x7f). 479*0e209d39SAndroid Build Coastguard Worker */ 480*0e209d39SAndroid Build Coastguard Worker static uint32_t decThreeBytePrimaryByOneStep(uint32_t basePrimary, UBool isCompressible, int32_t step); 481*0e209d39SAndroid Build Coastguard Worker 482*0e209d39SAndroid Build Coastguard Worker /** 483*0e209d39SAndroid Build Coastguard Worker * Computes a 3-byte primary for c's OFFSET_TAG data "CE". 484*0e209d39SAndroid Build Coastguard Worker */ 485*0e209d39SAndroid Build Coastguard Worker static uint32_t getThreeBytePrimaryForOffsetData(UChar32 c, int64_t dataCE); 486*0e209d39SAndroid Build Coastguard Worker 487*0e209d39SAndroid Build Coastguard Worker /** 488*0e209d39SAndroid Build Coastguard Worker * Returns the unassigned-character implicit primary weight for any valid code point c. 489*0e209d39SAndroid Build Coastguard Worker */ 490*0e209d39SAndroid Build Coastguard Worker static uint32_t unassignedPrimaryFromCodePoint(UChar32 c); 491*0e209d39SAndroid Build Coastguard Worker unassignedCEFromCodePoint(UChar32 c)492*0e209d39SAndroid Build Coastguard Worker static inline int64_t unassignedCEFromCodePoint(UChar32 c) { 493*0e209d39SAndroid Build Coastguard Worker return makeCE(unassignedPrimaryFromCodePoint(c)); 494*0e209d39SAndroid Build Coastguard Worker } 495*0e209d39SAndroid Build Coastguard Worker 496*0e209d39SAndroid Build Coastguard Worker private: 497*0e209d39SAndroid Build Coastguard Worker Collation() = delete; // No instantiation. 498*0e209d39SAndroid Build Coastguard Worker }; 499*0e209d39SAndroid Build Coastguard Worker 500*0e209d39SAndroid Build Coastguard Worker U_NAMESPACE_END 501*0e209d39SAndroid Build Coastguard Worker 502*0e209d39SAndroid Build Coastguard Worker #endif // !UCONFIG_NO_COLLATION 503*0e209d39SAndroid Build Coastguard Worker #endif // __COLLATION_H__ 504