diff options
Diffstat (limited to 'src/unicode_wb_data.c')
| -rw-r--r-- | src/unicode_wb_data.c | 82 |
1 files changed, 60 insertions, 22 deletions
diff --git a/src/unicode_wb_data.c b/src/unicode_wb_data.c index ce082a2..216c982 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) 2019-2020 K.Kosako + * Copyright (c) 2019-2021 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 130000 +#define WORD_BREAK_PROPERTY_VERSION 140000 /* ALetter @@ -48,7 +48,7 @@ WSegSpace ZWJ */ -static int WB_RANGE_NUM = 993; +static int WB_RANGE_NUM = 1031; static WB_RANGE_TYPE WB_RANGES[] = { {0x00000a, 0x00000a, WB_LF }, {0x00000b, 0x00000c, WB_Newline }, @@ -154,9 +154,12 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x000840, 0x000858, WB_ALetter }, {0x000859, 0x00085b, WB_Extend }, {0x000860, 0x00086a, WB_ALetter }, - {0x0008a0, 0x0008b4, WB_ALetter }, - {0x0008b6, 0x0008c7, WB_ALetter }, - {0x0008d3, 0x0008e1, WB_Extend }, + {0x000870, 0x000887, WB_ALetter }, + {0x000889, 0x00088e, WB_ALetter }, + {0x000890, 0x000891, WB_Format }, + {0x000898, 0x00089f, WB_Extend }, + {0x0008a0, 0x0008c9, WB_ALetter }, + {0x0008ca, 0x0008e1, WB_Extend }, {0x0008e2, 0x0008e2, WB_Format }, {0x0008e3, 0x000903, WB_Extend }, {0x000904, 0x000939, WB_ALetter }, @@ -267,12 +270,14 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x000c0e, 0x000c10, WB_ALetter }, {0x000c12, 0x000c28, WB_ALetter }, {0x000c2a, 0x000c39, WB_ALetter }, + {0x000c3c, 0x000c3c, WB_Extend }, {0x000c3d, 0x000c3d, WB_ALetter }, {0x000c3e, 0x000c44, WB_Extend }, {0x000c46, 0x000c48, WB_Extend }, {0x000c4a, 0x000c4d, WB_Extend }, {0x000c55, 0x000c56, WB_Extend }, {0x000c58, 0x000c5a, WB_ALetter }, + {0x000c5d, 0x000c5d, WB_ALetter }, {0x000c60, 0x000c61, WB_ALetter }, {0x000c62, 0x000c63, WB_Extend }, {0x000c66, 0x000c6f, WB_Numeric }, @@ -289,7 +294,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x000cc6, 0x000cc8, WB_Extend }, {0x000cca, 0x000ccd, WB_Extend }, {0x000cd5, 0x000cd6, WB_Extend }, - {0x000cde, 0x000cde, WB_ALetter }, + {0x000cdd, 0x000cde, WB_ALetter }, {0x000ce0, 0x000ce1, WB_ALetter }, {0x000ce2, 0x000ce3, WB_Extend }, {0x000ce6, 0x000cef, WB_Numeric }, @@ -386,10 +391,9 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x001681, 0x00169a, WB_ALetter }, {0x0016a0, 0x0016ea, WB_ALetter }, {0x0016ee, 0x0016f8, WB_ALetter }, - {0x001700, 0x00170c, WB_ALetter }, - {0x00170e, 0x001711, WB_ALetter }, - {0x001712, 0x001714, WB_Extend }, - {0x001720, 0x001731, WB_ALetter }, + {0x001700, 0x001711, WB_ALetter }, + {0x001712, 0x001715, WB_Extend }, + {0x00171f, 0x001731, WB_ALetter }, {0x001732, 0x001734, WB_Extend }, {0x001740, 0x001751, WB_ALetter }, {0x001752, 0x001753, WB_Extend }, @@ -401,6 +405,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x0017e0, 0x0017e9, WB_Numeric }, {0x00180b, 0x00180d, WB_Extend }, {0x00180e, 0x00180e, WB_Format }, + {0x00180f, 0x00180f, WB_Extend }, {0x001810, 0x001819, WB_Numeric }, {0x001820, 0x001878, WB_ALetter }, {0x001880, 0x001884, WB_ALetter }, @@ -421,11 +426,11 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x001a7f, 0x001a7f, WB_Extend }, {0x001a80, 0x001a89, WB_Numeric }, {0x001a90, 0x001a99, WB_Numeric }, - {0x001ab0, 0x001ac0, WB_Extend }, + {0x001ab0, 0x001ace, WB_Extend }, {0x001b00, 0x001b04, WB_Extend }, {0x001b05, 0x001b33, WB_ALetter }, {0x001b34, 0x001b44, WB_Extend }, - {0x001b45, 0x001b4b, WB_ALetter }, + {0x001b45, 0x001b4c, WB_ALetter }, {0x001b50, 0x001b59, WB_Numeric }, {0x001b6b, 0x001b73, WB_Extend }, {0x001b80, 0x001b82, WB_Extend }, @@ -454,8 +459,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x001cf7, 0x001cf9, WB_Extend }, {0x001cfa, 0x001cfa, WB_ALetter }, {0x001d00, 0x001dbf, WB_ALetter }, - {0x001dc0, 0x001df9, WB_Extend }, - {0x001dfb, 0x001dff, WB_Extend }, + {0x001dc0, 0x001dff, WB_Extend }, {0x001e00, 0x001f15, WB_ALetter }, {0x001f18, 0x001f1d, WB_ALetter }, {0x001f20, 0x001f45, WB_ALetter }, @@ -511,9 +515,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x00214e, 0x00214e, WB_ALetter }, {0x002160, 0x002188, WB_ALetter }, {0x0024b6, 0x0024e9, WB_ALetter }, - {0x002c00, 0x002c2e, WB_ALetter }, - {0x002c30, 0x002c5e, WB_ALetter }, - {0x002c60, 0x002ce4, WB_ALetter }, + {0x002c00, 0x002ce4, WB_ALetter }, {0x002ceb, 0x002cee, WB_ALetter }, {0x002cef, 0x002cf1, WB_Extend }, {0x002cf2, 0x002cf3, WB_ALetter }, @@ -562,9 +564,11 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x00a69e, 0x00a69f, WB_Extend }, {0x00a6a0, 0x00a6ef, WB_ALetter }, {0x00a6f0, 0x00a6f1, WB_Extend }, - {0x00a708, 0x00a7bf, WB_ALetter }, - {0x00a7c2, 0x00a7ca, WB_ALetter }, - {0x00a7f5, 0x00a801, WB_ALetter }, + {0x00a708, 0x00a7ca, WB_ALetter }, + {0x00a7d0, 0x00a7d1, WB_ALetter }, + {0x00a7d3, 0x00a7d3, WB_ALetter }, + {0x00a7d5, 0x00a7d9, WB_ALetter }, + {0x00a7f2, 0x00a801, WB_ALetter }, {0x00a802, 0x00a802, WB_Extend }, {0x00a803, 0x00a805, WB_ALetter }, {0x00a806, 0x00a806, WB_Extend }, @@ -699,9 +703,20 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x0104d8, 0x0104fb, WB_ALetter }, {0x010500, 0x010527, WB_ALetter }, {0x010530, 0x010563, WB_ALetter }, + {0x010570, 0x01057a, WB_ALetter }, + {0x01057c, 0x01058a, WB_ALetter }, + {0x01058c, 0x010592, WB_ALetter }, + {0x010594, 0x010595, WB_ALetter }, + {0x010597, 0x0105a1, WB_ALetter }, + {0x0105a3, 0x0105b1, WB_ALetter }, + {0x0105b3, 0x0105b9, WB_ALetter }, + {0x0105bb, 0x0105bc, WB_ALetter }, {0x010600, 0x010736, WB_ALetter }, {0x010740, 0x010755, WB_ALetter }, {0x010760, 0x010767, WB_ALetter }, + {0x010780, 0x010785, WB_ALetter }, + {0x010787, 0x0107b0, WB_ALetter }, + {0x0107b2, 0x0107ba, WB_ALetter }, {0x010800, 0x010805, WB_ALetter }, {0x010808, 0x010808, WB_ALetter }, {0x01080a, 0x010835, WB_ALetter }, @@ -747,16 +762,23 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x010f27, 0x010f27, WB_ALetter }, {0x010f30, 0x010f45, WB_ALetter }, {0x010f46, 0x010f50, WB_Extend }, + {0x010f70, 0x010f81, WB_ALetter }, + {0x010f82, 0x010f85, WB_Extend }, {0x010fb0, 0x010fc4, WB_ALetter }, {0x010fe0, 0x010ff6, WB_ALetter }, {0x011000, 0x011002, WB_Extend }, {0x011003, 0x011037, WB_ALetter }, {0x011038, 0x011046, WB_Extend }, {0x011066, 0x01106f, WB_Numeric }, + {0x011070, 0x011070, WB_Extend }, + {0x011071, 0x011072, WB_ALetter }, + {0x011073, 0x011074, WB_Extend }, + {0x011075, 0x011075, WB_ALetter }, {0x01107f, 0x011082, WB_Extend }, {0x011083, 0x0110af, WB_ALetter }, {0x0110b0, 0x0110ba, WB_Extend }, {0x0110bd, 0x0110bd, WB_Format }, + {0x0110c2, 0x0110c2, WB_Extend }, {0x0110cd, 0x0110cd, WB_Format }, {0x0110d0, 0x0110e8, WB_ALetter }, {0x0110f0, 0x0110f9, WB_Numeric }, @@ -871,7 +893,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x011a5c, 0x011a89, WB_ALetter }, {0x011a8a, 0x011a99, WB_Extend }, {0x011a9d, 0x011a9d, WB_ALetter }, - {0x011ac0, 0x011af8, WB_ALetter }, + {0x011ab0, 0x011af8, WB_ALetter }, {0x011c00, 0x011c08, WB_ALetter }, {0x011c0a, 0x011c2e, WB_ALetter }, {0x011c2f, 0x011c36, WB_Extend }, @@ -905,12 +927,15 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x012000, 0x012399, WB_ALetter }, {0x012400, 0x01246e, WB_ALetter }, {0x012480, 0x012543, WB_ALetter }, + {0x012f90, 0x012ff0, WB_ALetter }, {0x013000, 0x01342e, WB_ALetter }, {0x013430, 0x013438, WB_Format }, {0x014400, 0x014646, WB_ALetter }, {0x016800, 0x016a38, WB_ALetter }, {0x016a40, 0x016a5e, WB_ALetter }, {0x016a60, 0x016a69, WB_Numeric }, + {0x016a70, 0x016abe, WB_ALetter }, + {0x016ac0, 0x016ac9, WB_Numeric }, {0x016ad0, 0x016aed, WB_ALetter }, {0x016af0, 0x016af4, WB_Extend }, {0x016b00, 0x016b2f, WB_ALetter }, @@ -930,7 +955,11 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x016fe3, 0x016fe3, WB_ALetter }, {0x016fe4, 0x016fe4, WB_Extend }, {0x016ff0, 0x016ff1, WB_Extend }, + {0x01aff0, 0x01aff3, WB_Katakana }, + {0x01aff5, 0x01affb, WB_Katakana }, + {0x01affd, 0x01affe, WB_Katakana }, {0x01b000, 0x01b000, WB_Katakana }, + {0x01b120, 0x01b122, WB_Katakana }, {0x01b164, 0x01b167, WB_Katakana }, {0x01bc00, 0x01bc6a, WB_ALetter }, {0x01bc70, 0x01bc7c, WB_ALetter }, @@ -938,6 +967,8 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x01bc90, 0x01bc99, WB_ALetter }, {0x01bc9d, 0x01bc9e, WB_Extend }, {0x01bca0, 0x01bca3, WB_Format }, + {0x01cf00, 0x01cf2d, WB_Extend }, + {0x01cf30, 0x01cf46, WB_Extend }, {0x01d165, 0x01d169, WB_Extend }, {0x01d16d, 0x01d172, WB_Extend }, {0x01d173, 0x01d17a, WB_Format }, @@ -982,6 +1013,7 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x01da84, 0x01da84, WB_Extend }, {0x01da9b, 0x01da9f, WB_Extend }, {0x01daa1, 0x01daaf, WB_Extend }, + {0x01df00, 0x01df1e, WB_ALetter }, {0x01e000, 0x01e006, WB_Extend }, {0x01e008, 0x01e018, WB_Extend }, {0x01e01b, 0x01e021, WB_Extend }, @@ -992,9 +1024,15 @@ static WB_RANGE_TYPE WB_RANGES[] = { {0x01e137, 0x01e13d, WB_ALetter }, {0x01e140, 0x01e149, WB_Numeric }, {0x01e14e, 0x01e14e, WB_ALetter }, + {0x01e290, 0x01e2ad, WB_ALetter }, + {0x01e2ae, 0x01e2ae, WB_Extend }, {0x01e2c0, 0x01e2eb, WB_ALetter }, {0x01e2ec, 0x01e2ef, WB_Extend }, {0x01e2f0, 0x01e2f9, WB_Numeric }, + {0x01e7e0, 0x01e7e6, WB_ALetter }, + {0x01e7e8, 0x01e7eb, WB_ALetter }, + {0x01e7ed, 0x01e7ee, WB_ALetter }, + {0x01e7f0, 0x01e7fe, WB_ALetter }, {0x01e800, 0x01e8c4, WB_ALetter }, {0x01e8d0, 0x01e8d6, WB_Extend }, {0x01e900, 0x01e943, WB_ALetter }, |