summaryrefslogtreecommitdiff
path: root/src/static_libs/libunibreak/wordbreakdata.c
diff options
context:
space:
mode:
authorTom Hacohen <tom@stosb.com>2015-05-07 10:53:11 +0100
committerTom Hacohen <tom@stosb.com>2015-05-07 10:54:26 +0100
commit7a49d23f90f41c478db9b7beb9763aa0cd74ae46 (patch)
tree8ff009d70d91523d2c5f718ed395c8e23df43c4f /src/static_libs/libunibreak/wordbreakdata.c
parentba77a837a37af0d154d7ceafbb5ab7d4f75090f6 (diff)
Static deps unibreak: update to what will soon be version 3.
Version 3 is not yet released, but this is on track to become it. This is based on commit: a815e11f7ebf35b59278f783227a829ee4692760. @feature.
Diffstat (limited to 'src/static_libs/libunibreak/wordbreakdata.c')
-rw-r--r--src/static_libs/libunibreak/wordbreakdata.c201
1 files changed, 162 insertions, 39 deletions
diff --git a/src/static_libs/libunibreak/wordbreakdata.c b/src/static_libs/libunibreak/wordbreakdata.c
index fe5afe3f33..c6d5694d06 100644
--- a/src/static_libs/libunibreak/wordbreakdata.c
+++ b/src/static_libs/libunibreak/wordbreakdata.c
@@ -1,16 +1,16 @@
1/* The content of this file is generated from: 1/* The content of this file is generated from:
2# WordBreakProperty-6.2.0.txt 2# WordBreakProperty-7.0.0.txt
3# Date: 2012-08-13, 19:12:09 GMT [MD] 3# Date: 2014-02-19, 15:51:39 GMT [MD]
4*/ 4*/
5 5
6#include "linebreak.h"
7#include "wordbreakdef.h" 6#include "wordbreakdef.h"
8 7
9static struct WordBreakProperties wb_prop_default[] = { 8static struct WordBreakProperties wb_prop_default[] = {
10 {0x000A, 0x000A, WBP_LF}, 9 {0x000A, 0x000A, WBP_LF},
11 {0x000B, 0x000C, WBP_Newline}, 10 {0x000B, 0x000C, WBP_Newline},
12 {0x000D, 0x000D, WBP_CR}, 11 {0x000D, 0x000D, WBP_CR},
13 {0x0027, 0x0027, WBP_MidNumLet}, 12 {0x0022, 0x0022, WBP_Double_Quote},
13 {0x0027, 0x0027, WBP_Single_Quote},
14 {0x002C, 0x002C, WBP_MidNum}, 14 {0x002C, 0x002C, WBP_MidNum},
15 {0x002E, 0x002E, WBP_MidNumLet}, 15 {0x002E, 0x002E, WBP_MidNumLet},
16 {0x0030, 0x0039, WBP_Numeric}, 16 {0x0030, 0x0039, WBP_Numeric},
@@ -36,6 +36,7 @@ static struct WordBreakProperties wb_prop_default[] = {
36 {0x0295, 0x02AF, WBP_ALetter}, 36 {0x0295, 0x02AF, WBP_ALetter},
37 {0x02B0, 0x02C1, WBP_ALetter}, 37 {0x02B0, 0x02C1, WBP_ALetter},
38 {0x02C6, 0x02D1, WBP_ALetter}, 38 {0x02C6, 0x02D1, WBP_ALetter},
39 {0x02D7, 0x02D7, WBP_MidLetter},
39 {0x02E0, 0x02E4, WBP_ALetter}, 40 {0x02E0, 0x02E4, WBP_ALetter},
40 {0x02EC, 0x02EC, WBP_ALetter}, 41 {0x02EC, 0x02EC, WBP_ALetter},
41 {0x02EE, 0x02EE, WBP_ALetter}, 42 {0x02EE, 0x02EE, WBP_ALetter},
@@ -46,6 +47,7 @@ static struct WordBreakProperties wb_prop_default[] = {
46 {0x037A, 0x037A, WBP_ALetter}, 47 {0x037A, 0x037A, WBP_ALetter},
47 {0x037B, 0x037D, WBP_ALetter}, 48 {0x037B, 0x037D, WBP_ALetter},
48 {0x037E, 0x037E, WBP_MidNum}, 49 {0x037E, 0x037E, WBP_MidNum},
50 {0x037F, 0x037F, WBP_ALetter},
49 {0x0386, 0x0386, WBP_ALetter}, 51 {0x0386, 0x0386, WBP_ALetter},
50 {0x0387, 0x0387, WBP_MidLetter}, 52 {0x0387, 0x0387, WBP_MidLetter},
51 {0x0388, 0x038A, WBP_ALetter}, 53 {0x0388, 0x038A, WBP_ALetter},
@@ -55,7 +57,7 @@ static struct WordBreakProperties wb_prop_default[] = {
55 {0x03F7, 0x0481, WBP_ALetter}, 57 {0x03F7, 0x0481, WBP_ALetter},
56 {0x0483, 0x0487, WBP_Extend}, 58 {0x0483, 0x0487, WBP_Extend},
57 {0x0488, 0x0489, WBP_Extend}, 59 {0x0488, 0x0489, WBP_Extend},
58 {0x048A, 0x0527, WBP_ALetter}, 60 {0x048A, 0x052F, WBP_ALetter},
59 {0x0531, 0x0556, WBP_ALetter}, 61 {0x0531, 0x0556, WBP_ALetter},
60 {0x0559, 0x0559, WBP_ALetter}, 62 {0x0559, 0x0559, WBP_ALetter},
61 {0x0561, 0x0587, WBP_ALetter}, 63 {0x0561, 0x0587, WBP_ALetter},
@@ -65,13 +67,14 @@ static struct WordBreakProperties wb_prop_default[] = {
65 {0x05C1, 0x05C2, WBP_Extend}, 67 {0x05C1, 0x05C2, WBP_Extend},
66 {0x05C4, 0x05C5, WBP_Extend}, 68 {0x05C4, 0x05C5, WBP_Extend},
67 {0x05C7, 0x05C7, WBP_Extend}, 69 {0x05C7, 0x05C7, WBP_Extend},
68 {0x05D0, 0x05EA, WBP_ALetter}, 70 {0x05D0, 0x05EA, WBP_Hebrew_Letter},
69 {0x05F0, 0x05F2, WBP_ALetter}, 71 {0x05F0, 0x05F2, WBP_Hebrew_Letter},
70 {0x05F3, 0x05F3, WBP_ALetter}, 72 {0x05F3, 0x05F3, WBP_ALetter},
71 {0x05F4, 0x05F4, WBP_MidLetter}, 73 {0x05F4, 0x05F4, WBP_MidLetter},
72 {0x0600, 0x0604, WBP_Format}, 74 {0x0600, 0x0605, WBP_Format},
73 {0x060C, 0x060D, WBP_MidNum}, 75 {0x060C, 0x060D, WBP_MidNum},
74 {0x0610, 0x061A, WBP_Extend}, 76 {0x0610, 0x061A, WBP_Extend},
77 {0x061C, 0x061C, WBP_Format},
75 {0x0620, 0x063F, WBP_ALetter}, 78 {0x0620, 0x063F, WBP_ALetter},
76 {0x0640, 0x0640, WBP_ALetter}, 79 {0x0640, 0x0640, WBP_ALetter},
77 {0x0641, 0x064A, WBP_ALetter}, 80 {0x0641, 0x064A, WBP_ALetter},
@@ -117,10 +120,8 @@ static struct WordBreakProperties wb_prop_default[] = {
117 {0x0829, 0x082D, WBP_Extend}, 120 {0x0829, 0x082D, WBP_Extend},
118 {0x0840, 0x0858, WBP_ALetter}, 121 {0x0840, 0x0858, WBP_ALetter},
119 {0x0859, 0x085B, WBP_Extend}, 122 {0x0859, 0x085B, WBP_Extend},
120 {0x08A0, 0x08A0, WBP_ALetter}, 123 {0x08A0, 0x08B2, WBP_ALetter},
121 {0x08A2, 0x08AC, WBP_ALetter}, 124 {0x08E4, 0x0902, WBP_Extend},
122 {0x08E4, 0x08FE, WBP_Extend},
123 {0x0900, 0x0902, WBP_Extend},
124 {0x0903, 0x0903, WBP_Extend}, 125 {0x0903, 0x0903, WBP_Extend},
125 {0x0904, 0x0939, WBP_ALetter}, 126 {0x0904, 0x0939, WBP_ALetter},
126 {0x093A, 0x093A, WBP_Extend}, 127 {0x093A, 0x093A, WBP_Extend},
@@ -138,8 +139,7 @@ static struct WordBreakProperties wb_prop_default[] = {
138 {0x0962, 0x0963, WBP_Extend}, 139 {0x0962, 0x0963, WBP_Extend},
139 {0x0966, 0x096F, WBP_Numeric}, 140 {0x0966, 0x096F, WBP_Numeric},
140 {0x0971, 0x0971, WBP_ALetter}, 141 {0x0971, 0x0971, WBP_ALetter},
141 {0x0972, 0x0977, WBP_ALetter}, 142 {0x0972, 0x0980, WBP_ALetter},
142 {0x0979, 0x097F, WBP_ALetter},
143 {0x0981, 0x0981, WBP_Extend}, 143 {0x0981, 0x0981, WBP_Extend},
144 {0x0982, 0x0983, WBP_Extend}, 144 {0x0982, 0x0983, WBP_Extend},
145 {0x0985, 0x098C, WBP_ALetter}, 145 {0x0985, 0x098C, WBP_ALetter},
@@ -247,12 +247,12 @@ static struct WordBreakProperties wb_prop_default[] = {
247 {0x0BD0, 0x0BD0, WBP_ALetter}, 247 {0x0BD0, 0x0BD0, WBP_ALetter},
248 {0x0BD7, 0x0BD7, WBP_Extend}, 248 {0x0BD7, 0x0BD7, WBP_Extend},
249 {0x0BE6, 0x0BEF, WBP_Numeric}, 249 {0x0BE6, 0x0BEF, WBP_Numeric},
250 {0x0C00, 0x0C00, WBP_Extend},
250 {0x0C01, 0x0C03, WBP_Extend}, 251 {0x0C01, 0x0C03, WBP_Extend},
251 {0x0C05, 0x0C0C, WBP_ALetter}, 252 {0x0C05, 0x0C0C, WBP_ALetter},
252 {0x0C0E, 0x0C10, WBP_ALetter}, 253 {0x0C0E, 0x0C10, WBP_ALetter},
253 {0x0C12, 0x0C28, WBP_ALetter}, 254 {0x0C12, 0x0C28, WBP_ALetter},
254 {0x0C2A, 0x0C33, WBP_ALetter}, 255 {0x0C2A, 0x0C39, WBP_ALetter},
255 {0x0C35, 0x0C39, WBP_ALetter},
256 {0x0C3D, 0x0C3D, WBP_ALetter}, 256 {0x0C3D, 0x0C3D, WBP_ALetter},
257 {0x0C3E, 0x0C40, WBP_Extend}, 257 {0x0C3E, 0x0C40, WBP_Extend},
258 {0x0C41, 0x0C44, WBP_Extend}, 258 {0x0C41, 0x0C44, WBP_Extend},
@@ -263,6 +263,7 @@ static struct WordBreakProperties wb_prop_default[] = {
263 {0x0C60, 0x0C61, WBP_ALetter}, 263 {0x0C60, 0x0C61, WBP_ALetter},
264 {0x0C62, 0x0C63, WBP_Extend}, 264 {0x0C62, 0x0C63, WBP_Extend},
265 {0x0C66, 0x0C6F, WBP_Numeric}, 265 {0x0C66, 0x0C6F, WBP_Numeric},
266 {0x0C81, 0x0C81, WBP_Extend},
266 {0x0C82, 0x0C83, WBP_Extend}, 267 {0x0C82, 0x0C83, WBP_Extend},
267 {0x0C85, 0x0C8C, WBP_ALetter}, 268 {0x0C85, 0x0C8C, WBP_ALetter},
268 {0x0C8E, 0x0C90, WBP_ALetter}, 269 {0x0C8E, 0x0C90, WBP_ALetter},
@@ -284,6 +285,7 @@ static struct WordBreakProperties wb_prop_default[] = {
284 {0x0CE2, 0x0CE3, WBP_Extend}, 285 {0x0CE2, 0x0CE3, WBP_Extend},
285 {0x0CE6, 0x0CEF, WBP_Numeric}, 286 {0x0CE6, 0x0CEF, WBP_Numeric},
286 {0x0CF1, 0x0CF2, WBP_ALetter}, 287 {0x0CF1, 0x0CF2, WBP_ALetter},
288 {0x0D01, 0x0D01, WBP_Extend},
287 {0x0D02, 0x0D03, WBP_Extend}, 289 {0x0D02, 0x0D03, WBP_Extend},
288 {0x0D05, 0x0D0C, WBP_ALetter}, 290 {0x0D05, 0x0D0C, WBP_ALetter},
289 {0x0D0E, 0x0D10, WBP_ALetter}, 291 {0x0D0E, 0x0D10, WBP_ALetter},
@@ -311,6 +313,7 @@ static struct WordBreakProperties wb_prop_default[] = {
311 {0x0DD2, 0x0DD4, WBP_Extend}, 313 {0x0DD2, 0x0DD4, WBP_Extend},
312 {0x0DD6, 0x0DD6, WBP_Extend}, 314 {0x0DD6, 0x0DD6, WBP_Extend},
313 {0x0DD8, 0x0DDF, WBP_Extend}, 315 {0x0DD8, 0x0DDF, WBP_Extend},
316 {0x0DE6, 0x0DEF, WBP_Numeric},
314 {0x0DF2, 0x0DF3, WBP_Extend}, 317 {0x0DF2, 0x0DF3, WBP_Extend},
315 {0x0E31, 0x0E31, WBP_Extend}, 318 {0x0E31, 0x0E31, WBP_Extend},
316 {0x0E34, 0x0E3A, WBP_Extend}, 319 {0x0E34, 0x0E3A, WBP_Extend},
@@ -391,6 +394,7 @@ static struct WordBreakProperties wb_prop_default[] = {
391 {0x1681, 0x169A, WBP_ALetter}, 394 {0x1681, 0x169A, WBP_ALetter},
392 {0x16A0, 0x16EA, WBP_ALetter}, 395 {0x16A0, 0x16EA, WBP_ALetter},
393 {0x16EE, 0x16F0, WBP_ALetter}, 396 {0x16EE, 0x16F0, WBP_ALetter},
397 {0x16F1, 0x16F8, WBP_ALetter},
394 {0x1700, 0x170C, WBP_ALetter}, 398 {0x1700, 0x170C, WBP_ALetter},
395 {0x170E, 0x1711, WBP_ALetter}, 399 {0x170E, 0x1711, WBP_ALetter},
396 {0x1712, 0x1714, WBP_Extend}, 400 {0x1712, 0x1714, WBP_Extend},
@@ -411,6 +415,7 @@ static struct WordBreakProperties wb_prop_default[] = {
411 {0x17DD, 0x17DD, WBP_Extend}, 415 {0x17DD, 0x17DD, WBP_Extend},
412 {0x17E0, 0x17E9, WBP_Numeric}, 416 {0x17E0, 0x17E9, WBP_Numeric},
413 {0x180B, 0x180D, WBP_Extend}, 417 {0x180B, 0x180D, WBP_Extend},
418 {0x180E, 0x180E, WBP_Format},
414 {0x1810, 0x1819, WBP_Numeric}, 419 {0x1810, 0x1819, WBP_Numeric},
415 {0x1820, 0x1842, WBP_ALetter}, 420 {0x1820, 0x1842, WBP_ALetter},
416 {0x1843, 0x1843, WBP_ALetter}, 421 {0x1843, 0x1843, WBP_ALetter},
@@ -419,7 +424,7 @@ static struct WordBreakProperties wb_prop_default[] = {
419 {0x18A9, 0x18A9, WBP_Extend}, 424 {0x18A9, 0x18A9, WBP_Extend},
420 {0x18AA, 0x18AA, WBP_ALetter}, 425 {0x18AA, 0x18AA, WBP_ALetter},
421 {0x18B0, 0x18F5, WBP_ALetter}, 426 {0x18B0, 0x18F5, WBP_ALetter},
422 {0x1900, 0x191C, WBP_ALetter}, 427 {0x1900, 0x191E, WBP_ALetter},
423 {0x1920, 0x1922, WBP_Extend}, 428 {0x1920, 0x1922, WBP_Extend},
424 {0x1923, 0x1926, WBP_Extend}, 429 {0x1923, 0x1926, WBP_Extend},
425 {0x1927, 0x1928, WBP_Extend}, 430 {0x1927, 0x1928, WBP_Extend},
@@ -434,7 +439,8 @@ static struct WordBreakProperties wb_prop_default[] = {
434 {0x19D0, 0x19D9, WBP_Numeric}, 439 {0x19D0, 0x19D9, WBP_Numeric},
435 {0x1A00, 0x1A16, WBP_ALetter}, 440 {0x1A00, 0x1A16, WBP_ALetter},
436 {0x1A17, 0x1A18, WBP_Extend}, 441 {0x1A17, 0x1A18, WBP_Extend},
437 {0x1A19, 0x1A1B, WBP_Extend}, 442 {0x1A19, 0x1A1A, WBP_Extend},
443 {0x1A1B, 0x1A1B, WBP_Extend},
438 {0x1A55, 0x1A55, WBP_Extend}, 444 {0x1A55, 0x1A55, WBP_Extend},
439 {0x1A56, 0x1A56, WBP_Extend}, 445 {0x1A56, 0x1A56, WBP_Extend},
440 {0x1A57, 0x1A57, WBP_Extend}, 446 {0x1A57, 0x1A57, WBP_Extend},
@@ -449,6 +455,8 @@ static struct WordBreakProperties wb_prop_default[] = {
449 {0x1A7F, 0x1A7F, WBP_Extend}, 455 {0x1A7F, 0x1A7F, WBP_Extend},
450 {0x1A80, 0x1A89, WBP_Numeric}, 456 {0x1A80, 0x1A89, WBP_Numeric},
451 {0x1A90, 0x1A99, WBP_Numeric}, 457 {0x1A90, 0x1A99, WBP_Numeric},
458 {0x1AB0, 0x1ABD, WBP_Extend},
459 {0x1ABE, 0x1ABE, WBP_Extend},
452 {0x1B00, 0x1B03, WBP_Extend}, 460 {0x1B00, 0x1B03, WBP_Extend},
453 {0x1B04, 0x1B04, WBP_Extend}, 461 {0x1B04, 0x1B04, WBP_Extend},
454 {0x1B05, 0x1B33, WBP_ALetter}, 462 {0x1B05, 0x1B33, WBP_ALetter},
@@ -471,8 +479,7 @@ static struct WordBreakProperties wb_prop_default[] = {
471 {0x1BA6, 0x1BA7, WBP_Extend}, 479 {0x1BA6, 0x1BA7, WBP_Extend},
472 {0x1BA8, 0x1BA9, WBP_Extend}, 480 {0x1BA8, 0x1BA9, WBP_Extend},
473 {0x1BAA, 0x1BAA, WBP_Extend}, 481 {0x1BAA, 0x1BAA, WBP_Extend},
474 {0x1BAB, 0x1BAB, WBP_Extend}, 482 {0x1BAB, 0x1BAD, WBP_Extend},
475 {0x1BAC, 0x1BAD, WBP_Extend},
476 {0x1BAE, 0x1BAF, WBP_ALetter}, 483 {0x1BAE, 0x1BAF, WBP_ALetter},
477 {0x1BB0, 0x1BB9, WBP_Numeric}, 484 {0x1BB0, 0x1BB9, WBP_Numeric},
478 {0x1BBA, 0x1BE5, WBP_ALetter}, 485 {0x1BBA, 0x1BE5, WBP_ALetter},
@@ -504,13 +511,14 @@ static struct WordBreakProperties wb_prop_default[] = {
504 {0x1CF2, 0x1CF3, WBP_Extend}, 511 {0x1CF2, 0x1CF3, WBP_Extend},
505 {0x1CF4, 0x1CF4, WBP_Extend}, 512 {0x1CF4, 0x1CF4, WBP_Extend},
506 {0x1CF5, 0x1CF6, WBP_ALetter}, 513 {0x1CF5, 0x1CF6, WBP_ALetter},
514 {0x1CF8, 0x1CF9, WBP_Extend},
507 {0x1D00, 0x1D2B, WBP_ALetter}, 515 {0x1D00, 0x1D2B, WBP_ALetter},
508 {0x1D2C, 0x1D6A, WBP_ALetter}, 516 {0x1D2C, 0x1D6A, WBP_ALetter},
509 {0x1D6B, 0x1D77, WBP_ALetter}, 517 {0x1D6B, 0x1D77, WBP_ALetter},
510 {0x1D78, 0x1D78, WBP_ALetter}, 518 {0x1D78, 0x1D78, WBP_ALetter},
511 {0x1D79, 0x1D9A, WBP_ALetter}, 519 {0x1D79, 0x1D9A, WBP_ALetter},
512 {0x1D9B, 0x1DBF, WBP_ALetter}, 520 {0x1D9B, 0x1DBF, WBP_ALetter},
513 {0x1DC0, 0x1DE6, WBP_Extend}, 521 {0x1DC0, 0x1DF5, WBP_Extend},
514 {0x1DFC, 0x1DFF, WBP_Extend}, 522 {0x1DFC, 0x1DFF, WBP_Extend},
515 {0x1E00, 0x1F15, WBP_ALetter}, 523 {0x1E00, 0x1F15, WBP_ALetter},
516 {0x1F18, 0x1F1D, WBP_ALetter}, 524 {0x1F18, 0x1F1D, WBP_ALetter},
@@ -544,7 +552,7 @@ static struct WordBreakProperties wb_prop_default[] = {
544 {0x2044, 0x2044, WBP_MidNum}, 552 {0x2044, 0x2044, WBP_MidNum},
545 {0x2054, 0x2054, WBP_ExtendNumLet}, 553 {0x2054, 0x2054, WBP_ExtendNumLet},
546 {0x2060, 0x2064, WBP_Format}, 554 {0x2060, 0x2064, WBP_Format},
547 {0x206A, 0x206F, WBP_Format}, 555 {0x2066, 0x206F, WBP_Format},
548 {0x2071, 0x2071, WBP_ALetter}, 556 {0x2071, 0x2071, WBP_ALetter},
549 {0x207F, 0x207F, WBP_ALetter}, 557 {0x207F, 0x207F, WBP_ALetter},
550 {0x2090, 0x209C, WBP_ALetter}, 558 {0x2090, 0x209C, WBP_ALetter},
@@ -631,7 +639,8 @@ static struct WordBreakProperties wb_prop_default[] = {
631 {0xA670, 0xA672, WBP_Extend}, 639 {0xA670, 0xA672, WBP_Extend},
632 {0xA674, 0xA67D, WBP_Extend}, 640 {0xA674, 0xA67D, WBP_Extend},
633 {0xA67F, 0xA67F, WBP_ALetter}, 641 {0xA67F, 0xA67F, WBP_ALetter},
634 {0xA680, 0xA697, WBP_ALetter}, 642 {0xA680, 0xA69B, WBP_ALetter},
643 {0xA69C, 0xA69D, WBP_ALetter},
635 {0xA69F, 0xA69F, WBP_Extend}, 644 {0xA69F, 0xA69F, WBP_Extend},
636 {0xA6A0, 0xA6E5, WBP_ALetter}, 645 {0xA6A0, 0xA6E5, WBP_ALetter},
637 {0xA6E6, 0xA6EF, WBP_ALetter}, 646 {0xA6E6, 0xA6EF, WBP_ALetter},
@@ -642,8 +651,9 @@ static struct WordBreakProperties wb_prop_default[] = {
642 {0xA771, 0xA787, WBP_ALetter}, 651 {0xA771, 0xA787, WBP_ALetter},
643 {0xA788, 0xA788, WBP_ALetter}, 652 {0xA788, 0xA788, WBP_ALetter},
644 {0xA78B, 0xA78E, WBP_ALetter}, 653 {0xA78B, 0xA78E, WBP_ALetter},
645 {0xA790, 0xA793, WBP_ALetter}, 654 {0xA790, 0xA7AD, WBP_ALetter},
646 {0xA7A0, 0xA7AA, WBP_ALetter}, 655 {0xA7B0, 0xA7B1, WBP_ALetter},
656 {0xA7F7, 0xA7F7, WBP_ALetter},
647 {0xA7F8, 0xA7F9, WBP_ALetter}, 657 {0xA7F8, 0xA7F9, WBP_ALetter},
648 {0xA7FA, 0xA7FA, WBP_ALetter}, 658 {0xA7FA, 0xA7FA, WBP_ALetter},
649 {0xA7FB, 0xA801, WBP_ALetter}, 659 {0xA7FB, 0xA801, WBP_ALetter},
@@ -683,6 +693,8 @@ static struct WordBreakProperties wb_prop_default[] = {
683 {0xA9BD, 0xA9C0, WBP_Extend}, 693 {0xA9BD, 0xA9C0, WBP_Extend},
684 {0xA9CF, 0xA9CF, WBP_ALetter}, 694 {0xA9CF, 0xA9CF, WBP_ALetter},
685 {0xA9D0, 0xA9D9, WBP_Numeric}, 695 {0xA9D0, 0xA9D9, WBP_Numeric},
696 {0xA9E5, 0xA9E5, WBP_Extend},
697 {0xA9F0, 0xA9F9, WBP_Numeric},
686 {0xAA00, 0xAA28, WBP_ALetter}, 698 {0xAA00, 0xAA28, WBP_ALetter},
687 {0xAA29, 0xAA2E, WBP_Extend}, 699 {0xAA29, 0xAA2E, WBP_Extend},
688 {0xAA2F, 0xAA30, WBP_Extend}, 700 {0xAA2F, 0xAA30, WBP_Extend},
@@ -696,6 +708,8 @@ static struct WordBreakProperties wb_prop_default[] = {
696 {0xAA4D, 0xAA4D, WBP_Extend}, 708 {0xAA4D, 0xAA4D, WBP_Extend},
697 {0xAA50, 0xAA59, WBP_Numeric}, 709 {0xAA50, 0xAA59, WBP_Numeric},
698 {0xAA7B, 0xAA7B, WBP_Extend}, 710 {0xAA7B, 0xAA7B, WBP_Extend},
711 {0xAA7C, 0xAA7C, WBP_Extend},
712 {0xAA7D, 0xAA7D, WBP_Extend},
699 {0xAAB0, 0xAAB0, WBP_Extend}, 713 {0xAAB0, 0xAAB0, WBP_Extend},
700 {0xAAB2, 0xAAB4, WBP_Extend}, 714 {0xAAB2, 0xAAB4, WBP_Extend},
701 {0xAAB7, 0xAAB8, WBP_Extend}, 715 {0xAAB7, 0xAAB8, WBP_Extend},
@@ -714,6 +728,9 @@ static struct WordBreakProperties wb_prop_default[] = {
714 {0xAB11, 0xAB16, WBP_ALetter}, 728 {0xAB11, 0xAB16, WBP_ALetter},
715 {0xAB20, 0xAB26, WBP_ALetter}, 729 {0xAB20, 0xAB26, WBP_ALetter},
716 {0xAB28, 0xAB2E, WBP_ALetter}, 730 {0xAB28, 0xAB2E, WBP_ALetter},
731 {0xAB30, 0xAB5A, WBP_ALetter},
732 {0xAB5C, 0xAB5F, WBP_ALetter},
733 {0xAB64, 0xAB65, WBP_ALetter},
717 {0xABC0, 0xABE2, WBP_ALetter}, 734 {0xABC0, 0xABE2, WBP_ALetter},
718 {0xABE3, 0xABE4, WBP_Extend}, 735 {0xABE3, 0xABE4, WBP_Extend},
719 {0xABE5, 0xABE5, WBP_Extend}, 736 {0xABE5, 0xABE5, WBP_Extend},
@@ -728,15 +745,16 @@ static struct WordBreakProperties wb_prop_default[] = {
728 {0xD7CB, 0xD7FB, WBP_ALetter}, 745 {0xD7CB, 0xD7FB, WBP_ALetter},
729 {0xFB00, 0xFB06, WBP_ALetter}, 746 {0xFB00, 0xFB06, WBP_ALetter},
730 {0xFB13, 0xFB17, WBP_ALetter}, 747 {0xFB13, 0xFB17, WBP_ALetter},
731 {0xFB1D, 0xFB1D, WBP_ALetter}, 748 {0xFB1D, 0xFB1D, WBP_Hebrew_Letter},
732 {0xFB1E, 0xFB1E, WBP_Extend}, 749 {0xFB1E, 0xFB1E, WBP_Extend},
733 {0xFB1F, 0xFB28, WBP_ALetter}, 750 {0xFB1F, 0xFB28, WBP_Hebrew_Letter},
734 {0xFB2A, 0xFB36, WBP_ALetter}, 751 {0xFB2A, 0xFB36, WBP_Hebrew_Letter},
735 {0xFB38, 0xFB3C, WBP_ALetter}, 752 {0xFB38, 0xFB3C, WBP_Hebrew_Letter},
736 {0xFB3E, 0xFB3E, WBP_ALetter}, 753 {0xFB3E, 0xFB3E, WBP_Hebrew_Letter},
737 {0xFB40, 0xFB41, WBP_ALetter}, 754 {0xFB40, 0xFB41, WBP_Hebrew_Letter},
738 {0xFB43, 0xFB44, WBP_ALetter}, 755 {0xFB43, 0xFB44, WBP_Hebrew_Letter},
739 {0xFB46, 0xFBB1, WBP_ALetter}, 756 {0xFB46, 0xFB4F, WBP_Hebrew_Letter},
757 {0xFB50, 0xFBB1, WBP_ALetter},
740 {0xFBD3, 0xFD3D, WBP_ALetter}, 758 {0xFBD3, 0xFD3D, WBP_ALetter},
741 {0xFD50, 0xFD8F, WBP_ALetter}, 759 {0xFD50, 0xFD8F, WBP_ALetter},
742 {0xFD92, 0xFDC7, WBP_ALetter}, 760 {0xFD92, 0xFDC7, WBP_ALetter},
@@ -745,7 +763,7 @@ static struct WordBreakProperties wb_prop_default[] = {
745 {0xFE10, 0xFE10, WBP_MidNum}, 763 {0xFE10, 0xFE10, WBP_MidNum},
746 {0xFE13, 0xFE13, WBP_MidLetter}, 764 {0xFE13, 0xFE13, WBP_MidLetter},
747 {0xFE14, 0xFE14, WBP_MidNum}, 765 {0xFE14, 0xFE14, WBP_MidNum},
748 {0xFE20, 0xFE26, WBP_Extend}, 766 {0xFE20, 0xFE2D, WBP_Extend},
749 {0xFE33, 0xFE34, WBP_ExtendNumLet}, 767 {0xFE33, 0xFE34, WBP_ExtendNumLet},
750 {0xFE4D, 0xFE4F, WBP_ExtendNumLet}, 768 {0xFE4D, 0xFE4F, WBP_ExtendNumLet},
751 {0xFE50, 0xFE50, WBP_MidNum}, 769 {0xFE50, 0xFE50, WBP_MidNum},
@@ -784,11 +802,14 @@ static struct WordBreakProperties wb_prop_default[] = {
784 {0x101FD, 0x101FD, WBP_Extend}, 802 {0x101FD, 0x101FD, WBP_Extend},
785 {0x10280, 0x1029C, WBP_ALetter}, 803 {0x10280, 0x1029C, WBP_ALetter},
786 {0x102A0, 0x102D0, WBP_ALetter}, 804 {0x102A0, 0x102D0, WBP_ALetter},
787 {0x10300, 0x1031E, WBP_ALetter}, 805 {0x102E0, 0x102E0, WBP_Extend},
806 {0x10300, 0x1031F, WBP_ALetter},
788 {0x10330, 0x10340, WBP_ALetter}, 807 {0x10330, 0x10340, WBP_ALetter},
789 {0x10341, 0x10341, WBP_ALetter}, 808 {0x10341, 0x10341, WBP_ALetter},
790 {0x10342, 0x10349, WBP_ALetter}, 809 {0x10342, 0x10349, WBP_ALetter},
791 {0x1034A, 0x1034A, WBP_ALetter}, 810 {0x1034A, 0x1034A, WBP_ALetter},
811 {0x10350, 0x10375, WBP_ALetter},
812 {0x10376, 0x1037A, WBP_Extend},
792 {0x10380, 0x1039D, WBP_ALetter}, 813 {0x10380, 0x1039D, WBP_ALetter},
793 {0x103A0, 0x103C3, WBP_ALetter}, 814 {0x103A0, 0x103C3, WBP_ALetter},
794 {0x103C8, 0x103CF, WBP_ALetter}, 815 {0x103C8, 0x103CF, WBP_ALetter},
@@ -796,12 +817,19 @@ static struct WordBreakProperties wb_prop_default[] = {
796 {0x10400, 0x1044F, WBP_ALetter}, 817 {0x10400, 0x1044F, WBP_ALetter},
797 {0x10450, 0x1049D, WBP_ALetter}, 818 {0x10450, 0x1049D, WBP_ALetter},
798 {0x104A0, 0x104A9, WBP_Numeric}, 819 {0x104A0, 0x104A9, WBP_Numeric},
820 {0x10500, 0x10527, WBP_ALetter},
821 {0x10530, 0x10563, WBP_ALetter},
822 {0x10600, 0x10736, WBP_ALetter},
823 {0x10740, 0x10755, WBP_ALetter},
824 {0x10760, 0x10767, WBP_ALetter},
799 {0x10800, 0x10805, WBP_ALetter}, 825 {0x10800, 0x10805, WBP_ALetter},
800 {0x10808, 0x10808, WBP_ALetter}, 826 {0x10808, 0x10808, WBP_ALetter},
801 {0x1080A, 0x10835, WBP_ALetter}, 827 {0x1080A, 0x10835, WBP_ALetter},
802 {0x10837, 0x10838, WBP_ALetter}, 828 {0x10837, 0x10838, WBP_ALetter},
803 {0x1083C, 0x1083C, WBP_ALetter}, 829 {0x1083C, 0x1083C, WBP_ALetter},
804 {0x1083F, 0x10855, WBP_ALetter}, 830 {0x1083F, 0x10855, WBP_ALetter},
831 {0x10860, 0x10876, WBP_ALetter},
832 {0x10880, 0x1089E, WBP_ALetter},
805 {0x10900, 0x10915, WBP_ALetter}, 833 {0x10900, 0x10915, WBP_ALetter},
806 {0x10920, 0x10939, WBP_ALetter}, 834 {0x10920, 0x10939, WBP_ALetter},
807 {0x10980, 0x109B7, WBP_ALetter}, 835 {0x10980, 0x109B7, WBP_ALetter},
@@ -816,9 +844,14 @@ static struct WordBreakProperties wb_prop_default[] = {
816 {0x10A38, 0x10A3A, WBP_Extend}, 844 {0x10A38, 0x10A3A, WBP_Extend},
817 {0x10A3F, 0x10A3F, WBP_Extend}, 845 {0x10A3F, 0x10A3F, WBP_Extend},
818 {0x10A60, 0x10A7C, WBP_ALetter}, 846 {0x10A60, 0x10A7C, WBP_ALetter},
847 {0x10A80, 0x10A9C, WBP_ALetter},
848 {0x10AC0, 0x10AC7, WBP_ALetter},
849 {0x10AC9, 0x10AE4, WBP_ALetter},
850 {0x10AE5, 0x10AE6, WBP_Extend},
819 {0x10B00, 0x10B35, WBP_ALetter}, 851 {0x10B00, 0x10B35, WBP_ALetter},
820 {0x10B40, 0x10B55, WBP_ALetter}, 852 {0x10B40, 0x10B55, WBP_ALetter},
821 {0x10B60, 0x10B72, WBP_ALetter}, 853 {0x10B60, 0x10B72, WBP_ALetter},
854 {0x10B80, 0x10B91, WBP_ALetter},
822 {0x10C00, 0x10C48, WBP_ALetter}, 855 {0x10C00, 0x10C48, WBP_ALetter},
823 {0x11000, 0x11000, WBP_Extend}, 856 {0x11000, 0x11000, WBP_Extend},
824 {0x11001, 0x11001, WBP_Extend}, 857 {0x11001, 0x11001, WBP_Extend},
@@ -826,7 +859,7 @@ static struct WordBreakProperties wb_prop_default[] = {
826 {0x11003, 0x11037, WBP_ALetter}, 859 {0x11003, 0x11037, WBP_ALetter},
827 {0x11038, 0x11046, WBP_Extend}, 860 {0x11038, 0x11046, WBP_Extend},
828 {0x11066, 0x1106F, WBP_Numeric}, 861 {0x11066, 0x1106F, WBP_Numeric},
829 {0x11080, 0x11081, WBP_Extend}, 862 {0x1107F, 0x11081, WBP_Extend},
830 {0x11082, 0x11082, WBP_Extend}, 863 {0x11082, 0x11082, WBP_Extend},
831 {0x11083, 0x110AF, WBP_ALetter}, 864 {0x11083, 0x110AF, WBP_ALetter},
832 {0x110B0, 0x110B2, WBP_Extend}, 865 {0x110B0, 0x110B2, WBP_Extend},
@@ -842,6 +875,9 @@ static struct WordBreakProperties wb_prop_default[] = {
842 {0x1112C, 0x1112C, WBP_Extend}, 875 {0x1112C, 0x1112C, WBP_Extend},
843 {0x1112D, 0x11134, WBP_Extend}, 876 {0x1112D, 0x11134, WBP_Extend},
844 {0x11136, 0x1113F, WBP_Numeric}, 877 {0x11136, 0x1113F, WBP_Numeric},
878 {0x11150, 0x11172, WBP_ALetter},
879 {0x11173, 0x11173, WBP_Extend},
880 {0x11176, 0x11176, WBP_ALetter},
845 {0x11180, 0x11181, WBP_Extend}, 881 {0x11180, 0x11181, WBP_Extend},
846 {0x11182, 0x11182, WBP_Extend}, 882 {0x11182, 0x11182, WBP_Extend},
847 {0x11183, 0x111B2, WBP_ALetter}, 883 {0x11183, 0x111B2, WBP_ALetter},
@@ -850,6 +886,68 @@ static struct WordBreakProperties wb_prop_default[] = {
850 {0x111BF, 0x111C0, WBP_Extend}, 886 {0x111BF, 0x111C0, WBP_Extend},
851 {0x111C1, 0x111C4, WBP_ALetter}, 887 {0x111C1, 0x111C4, WBP_ALetter},
852 {0x111D0, 0x111D9, WBP_Numeric}, 888 {0x111D0, 0x111D9, WBP_Numeric},
889 {0x111DA, 0x111DA, WBP_ALetter},
890 {0x11200, 0x11211, WBP_ALetter},
891 {0x11213, 0x1122B, WBP_ALetter},
892 {0x1122C, 0x1122E, WBP_Extend},
893 {0x1122F, 0x11231, WBP_Extend},
894 {0x11232, 0x11233, WBP_Extend},
895 {0x11234, 0x11234, WBP_Extend},
896 {0x11235, 0x11235, WBP_Extend},
897 {0x11236, 0x11237, WBP_Extend},
898 {0x112B0, 0x112DE, WBP_ALetter},
899 {0x112DF, 0x112DF, WBP_Extend},
900 {0x112E0, 0x112E2, WBP_Extend},
901 {0x112E3, 0x112EA, WBP_Extend},
902 {0x112F0, 0x112F9, WBP_Numeric},
903 {0x11301, 0x11301, WBP_Extend},
904 {0x11302, 0x11303, WBP_Extend},
905 {0x11305, 0x1130C, WBP_ALetter},
906 {0x1130F, 0x11310, WBP_ALetter},
907 {0x11313, 0x11328, WBP_ALetter},
908 {0x1132A, 0x11330, WBP_ALetter},
909 {0x11332, 0x11333, WBP_ALetter},
910 {0x11335, 0x11339, WBP_ALetter},
911 {0x1133C, 0x1133C, WBP_Extend},
912 {0x1133D, 0x1133D, WBP_ALetter},
913 {0x1133E, 0x1133F, WBP_Extend},
914 {0x11340, 0x11340, WBP_Extend},
915 {0x11341, 0x11344, WBP_Extend},
916 {0x11347, 0x11348, WBP_Extend},
917 {0x1134B, 0x1134D, WBP_Extend},
918 {0x11357, 0x11357, WBP_Extend},
919 {0x1135D, 0x11361, WBP_ALetter},
920 {0x11362, 0x11363, WBP_Extend},
921 {0x11366, 0x1136C, WBP_Extend},
922 {0x11370, 0x11374, WBP_Extend},
923 {0x11480, 0x114AF, WBP_ALetter},
924 {0x114B0, 0x114B2, WBP_Extend},
925 {0x114B3, 0x114B8, WBP_Extend},
926 {0x114B9, 0x114B9, WBP_Extend},
927 {0x114BA, 0x114BA, WBP_Extend},
928 {0x114BB, 0x114BE, WBP_Extend},
929 {0x114BF, 0x114C0, WBP_Extend},
930 {0x114C1, 0x114C1, WBP_Extend},
931 {0x114C2, 0x114C3, WBP_Extend},
932 {0x114C4, 0x114C5, WBP_ALetter},
933 {0x114C7, 0x114C7, WBP_ALetter},
934 {0x114D0, 0x114D9, WBP_Numeric},
935 {0x11580, 0x115AE, WBP_ALetter},
936 {0x115AF, 0x115B1, WBP_Extend},
937 {0x115B2, 0x115B5, WBP_Extend},
938 {0x115B8, 0x115BB, WBP_Extend},
939 {0x115BC, 0x115BD, WBP_Extend},
940 {0x115BE, 0x115BE, WBP_Extend},
941 {0x115BF, 0x115C0, WBP_Extend},
942 {0x11600, 0x1162F, WBP_ALetter},
943 {0x11630, 0x11632, WBP_Extend},
944 {0x11633, 0x1163A, WBP_Extend},
945 {0x1163B, 0x1163C, WBP_Extend},
946 {0x1163D, 0x1163D, WBP_Extend},
947 {0x1163E, 0x1163E, WBP_Extend},
948 {0x1163F, 0x11640, WBP_Extend},
949 {0x11644, 0x11644, WBP_ALetter},
950 {0x11650, 0x11659, WBP_Numeric},
853 {0x11680, 0x116AA, WBP_ALetter}, 951 {0x11680, 0x116AA, WBP_ALetter},
854 {0x116AB, 0x116AB, WBP_Extend}, 952 {0x116AB, 0x116AB, WBP_Extend},
855 {0x116AC, 0x116AC, WBP_Extend}, 953 {0x116AC, 0x116AC, WBP_Extend},
@@ -859,16 +957,36 @@ static struct WordBreakProperties wb_prop_default[] = {
859 {0x116B6, 0x116B6, WBP_Extend}, 957 {0x116B6, 0x116B6, WBP_Extend},
860 {0x116B7, 0x116B7, WBP_Extend}, 958 {0x116B7, 0x116B7, WBP_Extend},
861 {0x116C0, 0x116C9, WBP_Numeric}, 959 {0x116C0, 0x116C9, WBP_Numeric},
862 {0x12000, 0x1236E, WBP_ALetter}, 960 {0x118A0, 0x118DF, WBP_ALetter},
863 {0x12400, 0x12462, WBP_ALetter}, 961 {0x118E0, 0x118E9, WBP_Numeric},
962 {0x118FF, 0x118FF, WBP_ALetter},
963 {0x11AC0, 0x11AF8, WBP_ALetter},
964 {0x12000, 0x12398, WBP_ALetter},
965 {0x12400, 0x1246E, WBP_ALetter},
864 {0x13000, 0x1342E, WBP_ALetter}, 966 {0x13000, 0x1342E, WBP_ALetter},
865 {0x16800, 0x16A38, WBP_ALetter}, 967 {0x16800, 0x16A38, WBP_ALetter},
968 {0x16A40, 0x16A5E, WBP_ALetter},
969 {0x16A60, 0x16A69, WBP_Numeric},
970 {0x16AD0, 0x16AED, WBP_ALetter},
971 {0x16AF0, 0x16AF4, WBP_Extend},
972 {0x16B00, 0x16B2F, WBP_ALetter},
973 {0x16B30, 0x16B36, WBP_Extend},
974 {0x16B40, 0x16B43, WBP_ALetter},
975 {0x16B50, 0x16B59, WBP_Numeric},
976 {0x16B63, 0x16B77, WBP_ALetter},
977 {0x16B7D, 0x16B8F, WBP_ALetter},
866 {0x16F00, 0x16F44, WBP_ALetter}, 978 {0x16F00, 0x16F44, WBP_ALetter},
867 {0x16F50, 0x16F50, WBP_ALetter}, 979 {0x16F50, 0x16F50, WBP_ALetter},
868 {0x16F51, 0x16F7E, WBP_Extend}, 980 {0x16F51, 0x16F7E, WBP_Extend},
869 {0x16F8F, 0x16F92, WBP_Extend}, 981 {0x16F8F, 0x16F92, WBP_Extend},
870 {0x16F93, 0x16F9F, WBP_ALetter}, 982 {0x16F93, 0x16F9F, WBP_ALetter},
871 {0x1B000, 0x1B000, WBP_Katakana}, 983 {0x1B000, 0x1B000, WBP_Katakana},
984 {0x1BC00, 0x1BC6A, WBP_ALetter},
985 {0x1BC70, 0x1BC7C, WBP_ALetter},
986 {0x1BC80, 0x1BC88, WBP_ALetter},
987 {0x1BC90, 0x1BC99, WBP_ALetter},
988 {0x1BC9D, 0x1BC9E, WBP_Extend},
989 {0x1BCA0, 0x1BCA3, WBP_Format},
872 {0x1D165, 0x1D166, WBP_Extend}, 990 {0x1D165, 0x1D166, WBP_Extend},
873 {0x1D167, 0x1D169, WBP_Extend}, 991 {0x1D167, 0x1D169, WBP_Extend},
874 {0x1D16D, 0x1D172, WBP_Extend}, 992 {0x1D16D, 0x1D172, WBP_Extend},
@@ -908,6 +1026,8 @@ static struct WordBreakProperties wb_prop_default[] = {
908 {0x1D7AA, 0x1D7C2, WBP_ALetter}, 1026 {0x1D7AA, 0x1D7C2, WBP_ALetter},
909 {0x1D7C4, 0x1D7CB, WBP_ALetter}, 1027 {0x1D7C4, 0x1D7CB, WBP_ALetter},
910 {0x1D7CE, 0x1D7FF, WBP_Numeric}, 1028 {0x1D7CE, 0x1D7FF, WBP_Numeric},
1029 {0x1E800, 0x1E8C4, WBP_ALetter},
1030 {0x1E8D0, 0x1E8D6, WBP_Extend},
911 {0x1EE00, 0x1EE03, WBP_ALetter}, 1031 {0x1EE00, 0x1EE03, WBP_ALetter},
912 {0x1EE05, 0x1EE1F, WBP_ALetter}, 1032 {0x1EE05, 0x1EE1F, WBP_ALetter},
913 {0x1EE21, 0x1EE22, WBP_ALetter}, 1033 {0x1EE21, 0x1EE22, WBP_ALetter},
@@ -941,7 +1061,10 @@ static struct WordBreakProperties wb_prop_default[] = {
941 {0x1EEA1, 0x1EEA3, WBP_ALetter}, 1061 {0x1EEA1, 0x1EEA3, WBP_ALetter},
942 {0x1EEA5, 0x1EEA9, WBP_ALetter}, 1062 {0x1EEA5, 0x1EEA9, WBP_ALetter},
943 {0x1EEAB, 0x1EEBB, WBP_ALetter}, 1063 {0x1EEAB, 0x1EEBB, WBP_ALetter},
944 {0x1F1E6, 0x1F1FF, WBP_Regional}, 1064 {0x1F130, 0x1F149, WBP_ALetter},
1065 {0x1F150, 0x1F169, WBP_ALetter},
1066 {0x1F170, 0x1F189, WBP_ALetter},
1067 {0x1F1E6, 0x1F1FF, WBP_Regional_Indicator},
945 {0xE0001, 0xE0001, WBP_Format}, 1068 {0xE0001, 0xE0001, WBP_Format},
946 {0xE0020, 0xE007F, WBP_Format}, 1069 {0xE0020, 0xE007F, WBP_Format},
947 {0xE0100, 0xE01EF, WBP_Extend}, 1070 {0xE0100, 0xE01EF, WBP_Extend},