diff options
Diffstat (limited to 'src/unicode_wb_data.c')
| -rw-r--r-- | src/unicode_wb_data.c | 49 |
1 files changed, 35 insertions, 14 deletions
diff --git a/src/unicode_wb_data.c b/src/unicode_wb_data.c index 216c982..117feea 100644 --- a/src/unicode_wb_data.c +++ b/src/unicode_wb_data.c @@ -1,6 +1,6 @@ /* unicode_wb_data.c: Generated by make_unicode_wb_data.py. */ /*- - * Copyright (c) 2019-2021 K.Kosako + * Copyright (c) 2019-2023 K.Kosako * All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -25,7 +25,7 @@ * SUCH DAMAGE. */ -#define WORD_BREAK_PROPERTY_VERSION 140000 +#define WORD_BREAK_PROPERTY_VERSION 150100 /* ALetter @@ -48,7 +48,7 @@ WSegSpace ZWJ */ -static int WB_RANGE_NUM = 1031; +static int WB_RANGE_NUM = 1052; static WB_RANGE_TYPE WB_RANGES[] = { {0x00000a, 0x00000a, WB_LF }, {0x00000b, 0x00000c, WB_Newline }, @@ -105,7 +105,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x0005ef, 0x0005f2, WB_Hebrew_Letter }, {0x0005f3, 0x0005f3, WB_ALetter }, {0x0005f4, 0x0005f4, WB_MidLetter }, - {0x000600, 0x000605, WB_Format }, + {0x000600, 0x000605, WB_Numeric }, {0x00060c, 0x00060d, WB_MidNum }, {0x000610, 0x00061a, WB_Extend }, {0x00061c, 0x00061c, WB_Format }, @@ -119,7 +119,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x000671, 0x0006d3, WB_ALetter }, {0x0006d5, 0x0006d5, WB_ALetter }, {0x0006d6, 0x0006dc, WB_Extend }, - {0x0006dd, 0x0006dd, WB_Format }, + {0x0006dd, 0x0006dd, WB_Numeric }, {0x0006df, 0x0006e4, WB_Extend }, {0x0006e5, 0x0006e6, WB_ALetter }, {0x0006e7, 0x0006e8, WB_Extend }, @@ -128,8 +128,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x0006f0, 0x0006f9, WB_Numeric }, {0x0006fa, 0x0006fc, WB_ALetter }, {0x0006ff, 0x0006ff, WB_ALetter }, - {0x00070f, 0x00070f, WB_Format }, - {0x000710, 0x000710, WB_ALetter }, + {0x00070f, 0x000710, WB_ALetter }, {0x000711, 0x000711, WB_Extend }, {0x000712, 0x00072f, WB_ALetter }, {0x000730, 0x00074a, WB_Extend }, @@ -156,11 +155,11 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x000860, 0x00086a, WB_ALetter }, {0x000870, 0x000887, WB_ALetter }, {0x000889, 0x00088e, WB_ALetter }, - {0x000890, 0x000891, WB_Format }, + {0x000890, 0x000891, WB_Numeric }, {0x000898, 0x00089f, WB_Extend }, {0x0008a0, 0x0008c9, WB_ALetter }, {0x0008ca, 0x0008e1, WB_Extend }, - {0x0008e2, 0x0008e2, WB_Format }, + {0x0008e2, 0x0008e2, WB_Numeric }, {0x0008e3, 0x000903, WB_Extend }, {0x000904, 0x000939, WB_ALetter }, {0x00093a, 0x00093c, WB_Extend }, @@ -299,6 +298,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x000ce2, 0x000ce3, WB_Extend }, {0x000ce6, 0x000cef, WB_Numeric }, {0x000cf1, 0x000cf2, WB_ALetter }, + {0x000cf3, 0x000cf3, WB_Extend }, {0x000d00, 0x000d03, WB_Extend }, {0x000d04, 0x000d0c, WB_ALetter }, {0x000d0e, 0x000d10, WB_ALetter }, @@ -333,7 +333,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x000e50, 0x000e59, WB_Numeric }, {0x000eb1, 0x000eb1, WB_Extend }, {0x000eb4, 0x000ebc, WB_Extend }, - {0x000ec8, 0x000ecd, WB_Extend }, + {0x000ec8, 0x000ece, WB_Extend }, {0x000ed0, 0x000ed9, WB_Numeric }, {0x000f00, 0x000f00, WB_ALetter }, {0x000f18, 0x000f19, WB_Extend }, @@ -758,6 +758,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x010e80, 0x010ea9, WB_ALetter }, {0x010eab, 0x010eac, WB_Extend }, {0x010eb0, 0x010eb1, WB_ALetter }, + {0x010efd, 0x010eff, WB_Extend }, {0x010f00, 0x010f1c, WB_ALetter }, {0x010f27, 0x010f27, WB_ALetter }, {0x010f30, 0x010f45, WB_ALetter }, @@ -777,9 +778,9 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x01107f, 0x011082, WB_Extend }, {0x011083, 0x0110af, WB_ALetter }, {0x0110b0, 0x0110ba, WB_Extend }, - {0x0110bd, 0x0110bd, WB_Format }, + {0x0110bd, 0x0110bd, WB_Numeric }, {0x0110c2, 0x0110c2, WB_Extend }, - {0x0110cd, 0x0110cd, WB_Format }, + {0x0110cd, 0x0110cd, WB_Numeric }, {0x0110d0, 0x0110e8, WB_ALetter }, {0x0110f0, 0x0110f9, WB_Numeric }, {0x011100, 0x011102, WB_Extend }, @@ -805,6 +806,8 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x011213, 0x01122b, WB_ALetter }, {0x01122c, 0x011237, WB_Extend }, {0x01123e, 0x01123e, WB_Extend }, + {0x01123f, 0x011240, WB_ALetter }, + {0x011241, 0x011241, WB_Extend }, {0x011280, 0x011286, WB_ALetter }, {0x011288, 0x011288, WB_ALetter }, {0x01128a, 0x01128d, WB_ALetter }, @@ -923,13 +926,24 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x011da0, 0x011da9, WB_Numeric }, {0x011ee0, 0x011ef2, WB_ALetter }, {0x011ef3, 0x011ef6, WB_Extend }, + {0x011f00, 0x011f01, WB_Extend }, + {0x011f02, 0x011f02, WB_ALetter }, + {0x011f03, 0x011f03, WB_Extend }, + {0x011f04, 0x011f10, WB_ALetter }, + {0x011f12, 0x011f33, WB_ALetter }, + {0x011f34, 0x011f3a, WB_Extend }, + {0x011f3e, 0x011f42, WB_Extend }, + {0x011f50, 0x011f59, WB_Numeric }, {0x011fb0, 0x011fb0, WB_ALetter }, {0x012000, 0x012399, WB_ALetter }, {0x012400, 0x01246e, WB_ALetter }, {0x012480, 0x012543, WB_ALetter }, {0x012f90, 0x012ff0, WB_ALetter }, - {0x013000, 0x01342e, WB_ALetter }, - {0x013430, 0x013438, WB_Format }, + {0x013000, 0x01342f, WB_ALetter }, + {0x013430, 0x01343f, WB_Format }, + {0x013440, 0x013440, WB_Extend }, + {0x013441, 0x013446, WB_ALetter }, + {0x013447, 0x013455, WB_Extend }, {0x014400, 0x014646, WB_ALetter }, {0x016800, 0x016a38, WB_ALetter }, {0x016a40, 0x016a5e, WB_ALetter }, @@ -960,6 +974,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x01affd, 0x01affe, WB_Katakana }, {0x01b000, 0x01b000, WB_Katakana }, {0x01b120, 0x01b122, WB_Katakana }, + {0x01b155, 0x01b155, WB_Katakana }, {0x01b164, 0x01b167, WB_Katakana }, {0x01bc00, 0x01bc6a, WB_ALetter }, {0x01bc70, 0x01bc7c, WB_ALetter }, @@ -1014,11 +1029,14 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x01da9b, 0x01da9f, WB_Extend }, {0x01daa1, 0x01daaf, WB_Extend }, {0x01df00, 0x01df1e, WB_ALetter }, + {0x01df25, 0x01df2a, WB_ALetter }, {0x01e000, 0x01e006, WB_Extend }, {0x01e008, 0x01e018, WB_Extend }, {0x01e01b, 0x01e021, WB_Extend }, {0x01e023, 0x01e024, WB_Extend }, {0x01e026, 0x01e02a, WB_Extend }, + {0x01e030, 0x01e06d, WB_ALetter }, + {0x01e08f, 0x01e08f, WB_Extend }, {0x01e100, 0x01e12c, WB_ALetter }, {0x01e130, 0x01e136, WB_Extend }, {0x01e137, 0x01e13d, WB_ALetter }, @@ -1029,6 +1047,9 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x01e2c0, 0x01e2eb, WB_ALetter }, {0x01e2ec, 0x01e2ef, WB_Extend }, {0x01e2f0, 0x01e2f9, WB_Numeric }, + {0x01e4d0, 0x01e4eb, WB_ALetter }, + {0x01e4ec, 0x01e4ef, WB_Extend }, + {0x01e4f0, 0x01e4f9, WB_Numeric }, {0x01e7e0, 0x01e7e6, WB_ALetter }, {0x01e7e8, 0x01e7eb, WB_ALetter }, {0x01e7ed, 0x01e7ee, WB_ALetter }, |