summaryrefslogtreecommitdiff
path: root/src/static_libs
diff options
context:
space:
mode:
authorTom Hacohen <tom@stosb.com>2013-08-07 11:56:46 +0100
committerTom Hacohen <tom@stosb.com>2013-08-07 11:57:09 +0100
commit4185694ecbe5293fad72f7eb9dd971900decb778 (patch)
tree919dca7f7dbe42d349e25b1f0017b591c7cd2b66 /src/static_libs
parent3cc3d0878a6dc29d0dcac9c2f5495bc8bdd84b7f (diff)
Synced libunibreak local copy with upstream.
Diffstat (limited to 'src/static_libs')
-rw-r--r--src/static_libs/libunibreak/wordbreak.c39
-rw-r--r--src/static_libs/libunibreak/wordbreakdata.c121
-rw-r--r--src/static_libs/libunibreak/wordbreakdef.h31
3 files changed, 146 insertions, 45 deletions
diff --git a/src/static_libs/libunibreak/wordbreak.c b/src/static_libs/libunibreak/wordbreak.c
index 60db99e..f2996c0 100644
--- a/src/static_libs/libunibreak/wordbreak.c
+++ b/src/static_libs/libunibreak/wordbreak.c
@@ -40,7 +40,7 @@
40 * Implementation of the word breaking algorithm as described in Unicode 40 * Implementation of the word breaking algorithm as described in Unicode
41 * Standard Annex 29. 41 * Standard Annex 29.
42 * 42 *
43 * @version 2.2, 2012/02/04 43 * @version 2.3, 2013/05/14
44 * @author Tom Hacohen 44 * @author Tom Hacohen
45 */ 45 */
46 46
@@ -188,7 +188,7 @@ static void set_wordbreaks(
188 188
189 switch (wbcCur) 189 switch (wbcCur)
190 { 190 {
191 case WBP_CR: 191 case WBP_CR:
192 /* WB3b */ 192 /* WB3b */
193 set_brks_to(s, brks, posLast, posCur, len, 193 set_brks_to(s, brks, posLast, posCur, len,
194 WORDBREAK_BREAK, get_next_char); 194 WORDBREAK_BREAK, get_next_char);
@@ -196,7 +196,7 @@ static void set_wordbreaks(
196 posLast = posCur; 196 posLast = posCur;
197 break; 197 break;
198 198
199 case WBP_LF: 199 case WBP_LF:
200 if (wbcSeqStart == WBP_CR) /* WB3 */ 200 if (wbcSeqStart == WBP_CR) /* WB3 */
201 { 201 {
202 set_brks_to(s, brks, posLast, posCur, len, 202 set_brks_to(s, brks, posLast, posCur, len,
@@ -207,7 +207,7 @@ static void set_wordbreaks(
207 } 207 }
208 /* Fall off */ 208 /* Fall off */
209 209
210 case WBP_Newline: 210 case WBP_Newline:
211 /* WB3a,3b */ 211 /* WB3a,3b */
212 set_brks_to(s, brks, posLast, posCur, len, 212 set_brks_to(s, brks, posLast, posCur, len,
213 WORDBREAK_BREAK, get_next_char); 213 WORDBREAK_BREAK, get_next_char);
@@ -215,8 +215,8 @@ static void set_wordbreaks(
215 posLast = posCur; 215 posLast = posCur;
216 break; 216 break;
217 217
218 case WBP_Extend: 218 case WBP_Extend:
219 case WBP_Format: 219 case WBP_Format:
220 /* WB4 - If not the first char/after a newline (WB3a,3b), skip 220 /* WB4 - If not the first char/after a newline (WB3a,3b), skip
221 * this class, set it to be the same as the prev, and mark 221 * this class, set it to be the same as the prev, and mark
222 * brks not to break before them. */ 222 * brks not to break before them. */
@@ -235,7 +235,7 @@ static void set_wordbreaks(
235 } 235 }
236 break; 236 break;
237 237
238 case WBP_Katakana: 238 case WBP_Katakana:
239 if ((wbcSeqStart == WBP_Katakana) || /* WB13 */ 239 if ((wbcSeqStart == WBP_Katakana) || /* WB13 */
240 (wbcSeqStart == WBP_ExtendNumLet)) /* WB13b */ 240 (wbcSeqStart == WBP_ExtendNumLet)) /* WB13b */
241 { 241 {
@@ -252,7 +252,7 @@ static void set_wordbreaks(
252 posLast = posCur; 252 posLast = posCur;
253 break; 253 break;
254 254
255 case WBP_ALetter: 255 case WBP_ALetter:
256 if ((wbcSeqStart == WBP_ALetter) || /* WB5,6,7 */ 256 if ((wbcSeqStart == WBP_ALetter) || /* WB5,6,7 */
257 (wbcLast == WBP_Numeric) || /* WB10 */ 257 (wbcLast == WBP_Numeric) || /* WB10 */
258 (wbcSeqStart == WBP_ExtendNumLet)) /* WB13b */ 258 (wbcSeqStart == WBP_ExtendNumLet)) /* WB13b */
@@ -270,7 +270,7 @@ static void set_wordbreaks(
270 posLast = posCur; 270 posLast = posCur;
271 break; 271 break;
272 272
273 case WBP_MidNumLet: 273 case WBP_MidNumLet:
274 if ((wbcLast == WBP_ALetter) || /* WB6,7 */ 274 if ((wbcLast == WBP_ALetter) || /* WB6,7 */
275 (wbcLast == WBP_Numeric)) /* WB11,12 */ 275 (wbcLast == WBP_Numeric)) /* WB11,12 */
276 { 276 {
@@ -285,7 +285,7 @@ static void set_wordbreaks(
285 } 285 }
286 break; 286 break;
287 287
288 case WBP_MidLetter: 288 case WBP_MidLetter:
289 if (wbcLast == WBP_ALetter) /* WB6,7 */ 289 if (wbcLast == WBP_ALetter) /* WB6,7 */
290 { 290 {
291 /* Go on */ 291 /* Go on */
@@ -299,7 +299,7 @@ static void set_wordbreaks(
299 } 299 }
300 break; 300 break;
301 301
302 case WBP_MidNum: 302 case WBP_MidNum:
303 if (wbcLast == WBP_Numeric) /* WB11,12 */ 303 if (wbcLast == WBP_Numeric) /* WB11,12 */
304 { 304 {
305 /* Go on */ 305 /* Go on */
@@ -313,7 +313,7 @@ static void set_wordbreaks(
313 } 313 }
314 break; 314 break;
315 315
316 case WBP_Numeric: 316 case WBP_Numeric:
317 if ((wbcSeqStart == WBP_Numeric) || /* WB8,11,12 */ 317 if ((wbcSeqStart == WBP_Numeric) || /* WB8,11,12 */
318 (wbcLast == WBP_ALetter) || /* WB9 */ 318 (wbcLast == WBP_ALetter) || /* WB9 */
319 (wbcSeqStart == WBP_ExtendNumLet)) /* WB13b */ 319 (wbcSeqStart == WBP_ExtendNumLet)) /* WB13b */
@@ -331,7 +331,7 @@ static void set_wordbreaks(
331 posLast = posCur; 331 posLast = posCur;
332 break; 332 break;
333 333
334 case WBP_ExtendNumLet: 334 case WBP_ExtendNumLet:
335 /* WB13a,13b */ 335 /* WB13a,13b */
336 if ((wbcSeqStart == wbcLast) && 336 if ((wbcSeqStart == wbcLast) &&
337 ((wbcLast == WBP_ALetter) || 337 ((wbcLast == WBP_ALetter) ||
@@ -352,7 +352,18 @@ static void set_wordbreaks(
352 posLast = posCur; 352 posLast = posCur;
353 break; 353 break;
354 354
355 case WBP_Any: 355 case WBP_Regional:
356 /* WB13c */
357 if (wbcSeqStart == WBP_Regional)
358 {
359 set_brks_to(s, brks, posLast, posCur, len,
360 WORDBREAK_NOBREAK, get_next_char);
361 }
362 wbcSeqStart = wbcCur;
363 posLast = posCur;
364 break;
365
366 case WBP_Any:
356 /* Allow breaks and reset */ 367 /* Allow breaks and reset */
357 set_brks_to(s, brks, posLast, posCur, len, 368 set_brks_to(s, brks, posLast, posCur, len,
358 WORDBREAK_BREAK, get_next_char); 369 WORDBREAK_BREAK, get_next_char);
diff --git a/src/static_libs/libunibreak/wordbreakdata.c b/src/static_libs/libunibreak/wordbreakdata.c
index c42c51b..fe5afe3 100644
--- a/src/static_libs/libunibreak/wordbreakdata.c
+++ b/src/static_libs/libunibreak/wordbreakdata.c
@@ -1,6 +1,6 @@
1/* The content of this file is generated from: 1/* The content of this file is generated from:
2# WordBreakProperty-6.0.0.txt 2# WordBreakProperty-6.2.0.txt
3# Date: 2010-08-19, 00:48:48 GMT [MD] 3# Date: 2012-08-13, 19:12:09 GMT [MD]
4*/ 4*/
5 5
6#include "linebreak.h" 6#include "linebreak.h"
@@ -69,7 +69,7 @@ static struct WordBreakProperties wb_prop_default[] = {
69 {0x05F0, 0x05F2, WBP_ALetter}, 69 {0x05F0, 0x05F2, WBP_ALetter},
70 {0x05F3, 0x05F3, WBP_ALetter}, 70 {0x05F3, 0x05F3, WBP_ALetter},
71 {0x05F4, 0x05F4, WBP_MidLetter}, 71 {0x05F4, 0x05F4, WBP_MidLetter},
72 {0x0600, 0x0603, WBP_Format}, 72 {0x0600, 0x0604, WBP_Format},
73 {0x060C, 0x060D, WBP_MidNum}, 73 {0x060C, 0x060D, WBP_MidNum},
74 {0x0610, 0x061A, WBP_Extend}, 74 {0x0610, 0x061A, WBP_Extend},
75 {0x0620, 0x063F, WBP_ALetter}, 75 {0x0620, 0x063F, WBP_ALetter},
@@ -117,6 +117,9 @@ static struct WordBreakProperties wb_prop_default[] = {
117 {0x0829, 0x082D, WBP_Extend}, 117 {0x0829, 0x082D, WBP_Extend},
118 {0x0840, 0x0858, WBP_ALetter}, 118 {0x0840, 0x0858, WBP_ALetter},
119 {0x0859, 0x085B, WBP_Extend}, 119 {0x0859, 0x085B, WBP_Extend},
120 {0x08A0, 0x08A0, WBP_ALetter},
121 {0x08A2, 0x08AC, WBP_ALetter},
122 {0x08E4, 0x08FE, WBP_Extend},
120 {0x0900, 0x0902, WBP_Extend}, 123 {0x0900, 0x0902, WBP_Extend},
121 {0x0903, 0x0903, WBP_Extend}, 124 {0x0903, 0x0903, WBP_Extend},
122 {0x0904, 0x0939, WBP_ALetter}, 125 {0x0904, 0x0939, WBP_ALetter},
@@ -360,9 +363,11 @@ static struct WordBreakProperties wb_prop_default[] = {
360 {0x109A, 0x109C, WBP_Extend}, 363 {0x109A, 0x109C, WBP_Extend},
361 {0x109D, 0x109D, WBP_Extend}, 364 {0x109D, 0x109D, WBP_Extend},
362 {0x10A0, 0x10C5, WBP_ALetter}, 365 {0x10A0, 0x10C5, WBP_ALetter},
366 {0x10C7, 0x10C7, WBP_ALetter},
367 {0x10CD, 0x10CD, WBP_ALetter},
363 {0x10D0, 0x10FA, WBP_ALetter}, 368 {0x10D0, 0x10FA, WBP_ALetter},
364 {0x10FC, 0x10FC, WBP_ALetter}, 369 {0x10FC, 0x10FC, WBP_ALetter},
365 {0x1100, 0x1248, WBP_ALetter}, 370 {0x10FD, 0x1248, WBP_ALetter},
366 {0x124A, 0x124D, WBP_ALetter}, 371 {0x124A, 0x124D, WBP_ALetter},
367 {0x1250, 0x1256, WBP_ALetter}, 372 {0x1250, 0x1256, WBP_ALetter},
368 {0x1258, 0x1258, WBP_ALetter}, 373 {0x1258, 0x1258, WBP_ALetter},
@@ -396,7 +401,7 @@ static struct WordBreakProperties wb_prop_default[] = {
396 {0x1760, 0x176C, WBP_ALetter}, 401 {0x1760, 0x176C, WBP_ALetter},
397 {0x176E, 0x1770, WBP_ALetter}, 402 {0x176E, 0x1770, WBP_ALetter},
398 {0x1772, 0x1773, WBP_Extend}, 403 {0x1772, 0x1773, WBP_Extend},
399 {0x17B4, 0x17B5, WBP_Format}, 404 {0x17B4, 0x17B5, WBP_Extend},
400 {0x17B6, 0x17B6, WBP_Extend}, 405 {0x17B6, 0x17B6, WBP_Extend},
401 {0x17B7, 0x17BD, WBP_Extend}, 406 {0x17B7, 0x17BD, WBP_Extend},
402 {0x17BE, 0x17C5, WBP_Extend}, 407 {0x17BE, 0x17C5, WBP_Extend},
@@ -466,9 +471,11 @@ static struct WordBreakProperties wb_prop_default[] = {
466 {0x1BA6, 0x1BA7, WBP_Extend}, 471 {0x1BA6, 0x1BA7, WBP_Extend},
467 {0x1BA8, 0x1BA9, WBP_Extend}, 472 {0x1BA8, 0x1BA9, WBP_Extend},
468 {0x1BAA, 0x1BAA, WBP_Extend}, 473 {0x1BAA, 0x1BAA, WBP_Extend},
474 {0x1BAB, 0x1BAB, WBP_Extend},
475 {0x1BAC, 0x1BAD, WBP_Extend},
469 {0x1BAE, 0x1BAF, WBP_ALetter}, 476 {0x1BAE, 0x1BAF, WBP_ALetter},
470 {0x1BB0, 0x1BB9, WBP_Numeric}, 477 {0x1BB0, 0x1BB9, WBP_Numeric},
471 {0x1BC0, 0x1BE5, WBP_ALetter}, 478 {0x1BBA, 0x1BE5, WBP_ALetter},
472 {0x1BE6, 0x1BE6, WBP_Extend}, 479 {0x1BE6, 0x1BE6, WBP_Extend},
473 {0x1BE7, 0x1BE7, WBP_Extend}, 480 {0x1BE7, 0x1BE7, WBP_Extend},
474 {0x1BE8, 0x1BE9, WBP_Extend}, 481 {0x1BE8, 0x1BE9, WBP_Extend},
@@ -494,10 +501,12 @@ static struct WordBreakProperties wb_prop_default[] = {
494 {0x1CE9, 0x1CEC, WBP_ALetter}, 501 {0x1CE9, 0x1CEC, WBP_ALetter},
495 {0x1CED, 0x1CED, WBP_Extend}, 502 {0x1CED, 0x1CED, WBP_Extend},
496 {0x1CEE, 0x1CF1, WBP_ALetter}, 503 {0x1CEE, 0x1CF1, WBP_ALetter},
497 {0x1CF2, 0x1CF2, WBP_Extend}, 504 {0x1CF2, 0x1CF3, WBP_Extend},
505 {0x1CF4, 0x1CF4, WBP_Extend},
506 {0x1CF5, 0x1CF6, WBP_ALetter},
498 {0x1D00, 0x1D2B, WBP_ALetter}, 507 {0x1D00, 0x1D2B, WBP_ALetter},
499 {0x1D2C, 0x1D61, WBP_ALetter}, 508 {0x1D2C, 0x1D6A, WBP_ALetter},
500 {0x1D62, 0x1D77, WBP_ALetter}, 509 {0x1D6B, 0x1D77, WBP_ALetter},
501 {0x1D78, 0x1D78, WBP_ALetter}, 510 {0x1D78, 0x1D78, WBP_ALetter},
502 {0x1D79, 0x1D9A, WBP_ALetter}, 511 {0x1D79, 0x1D9A, WBP_ALetter},
503 {0x1D9B, 0x1DBF, WBP_ALetter}, 512 {0x1D9B, 0x1DBF, WBP_ALetter},
@@ -565,13 +574,16 @@ static struct WordBreakProperties wb_prop_default[] = {
565 {0x24B6, 0x24E9, WBP_ALetter}, 574 {0x24B6, 0x24E9, WBP_ALetter},
566 {0x2C00, 0x2C2E, WBP_ALetter}, 575 {0x2C00, 0x2C2E, WBP_ALetter},
567 {0x2C30, 0x2C5E, WBP_ALetter}, 576 {0x2C30, 0x2C5E, WBP_ALetter},
568 {0x2C60, 0x2C7C, WBP_ALetter}, 577 {0x2C60, 0x2C7B, WBP_ALetter},
569 {0x2C7D, 0x2C7D, WBP_ALetter}, 578 {0x2C7C, 0x2C7D, WBP_ALetter},
570 {0x2C7E, 0x2CE4, WBP_ALetter}, 579 {0x2C7E, 0x2CE4, WBP_ALetter},
571 {0x2CEB, 0x2CEE, WBP_ALetter}, 580 {0x2CEB, 0x2CEE, WBP_ALetter},
572 {0x2CEF, 0x2CF1, WBP_Extend}, 581 {0x2CEF, 0x2CF1, WBP_Extend},
582 {0x2CF2, 0x2CF3, WBP_ALetter},
573 {0x2D00, 0x2D25, WBP_ALetter}, 583 {0x2D00, 0x2D25, WBP_ALetter},
574 {0x2D30, 0x2D65, WBP_ALetter}, 584 {0x2D27, 0x2D27, WBP_ALetter},
585 {0x2D2D, 0x2D2D, WBP_ALetter},
586 {0x2D30, 0x2D67, WBP_ALetter},
575 {0x2D6F, 0x2D6F, WBP_ALetter}, 587 {0x2D6F, 0x2D6F, WBP_ALetter},
576 {0x2D7F, 0x2D7F, WBP_Extend}, 588 {0x2D7F, 0x2D7F, WBP_Extend},
577 {0x2D80, 0x2D96, WBP_ALetter}, 589 {0x2D80, 0x2D96, WBP_ALetter},
@@ -586,7 +598,8 @@ static struct WordBreakProperties wb_prop_default[] = {
586 {0x2DE0, 0x2DFF, WBP_Extend}, 598 {0x2DE0, 0x2DFF, WBP_Extend},
587 {0x2E2F, 0x2E2F, WBP_ALetter}, 599 {0x2E2F, 0x2E2F, WBP_ALetter},
588 {0x3005, 0x3005, WBP_ALetter}, 600 {0x3005, 0x3005, WBP_ALetter},
589 {0x302A, 0x302F, WBP_Extend}, 601 {0x302A, 0x302D, WBP_Extend},
602 {0x302E, 0x302F, WBP_Extend},
590 {0x3031, 0x3035, WBP_Katakana}, 603 {0x3031, 0x3035, WBP_Katakana},
591 {0x303B, 0x303B, WBP_ALetter}, 604 {0x303B, 0x303B, WBP_ALetter},
592 {0x303C, 0x303C, WBP_ALetter}, 605 {0x303C, 0x303C, WBP_ALetter},
@@ -616,9 +629,10 @@ static struct WordBreakProperties wb_prop_default[] = {
616 {0xA66E, 0xA66E, WBP_ALetter}, 629 {0xA66E, 0xA66E, WBP_ALetter},
617 {0xA66F, 0xA66F, WBP_Extend}, 630 {0xA66F, 0xA66F, WBP_Extend},
618 {0xA670, 0xA672, WBP_Extend}, 631 {0xA670, 0xA672, WBP_Extend},
619 {0xA67C, 0xA67D, WBP_Extend}, 632 {0xA674, 0xA67D, WBP_Extend},
620 {0xA67F, 0xA67F, WBP_ALetter}, 633 {0xA67F, 0xA67F, WBP_ALetter},
621 {0xA680, 0xA697, WBP_ALetter}, 634 {0xA680, 0xA697, WBP_ALetter},
635 {0xA69F, 0xA69F, WBP_Extend},
622 {0xA6A0, 0xA6E5, WBP_ALetter}, 636 {0xA6A0, 0xA6E5, WBP_ALetter},
623 {0xA6E6, 0xA6EF, WBP_ALetter}, 637 {0xA6E6, 0xA6EF, WBP_ALetter},
624 {0xA6F0, 0xA6F1, WBP_Extend}, 638 {0xA6F0, 0xA6F1, WBP_Extend},
@@ -628,8 +642,9 @@ static struct WordBreakProperties wb_prop_default[] = {
628 {0xA771, 0xA787, WBP_ALetter}, 642 {0xA771, 0xA787, WBP_ALetter},
629 {0xA788, 0xA788, WBP_ALetter}, 643 {0xA788, 0xA788, WBP_ALetter},
630 {0xA78B, 0xA78E, WBP_ALetter}, 644 {0xA78B, 0xA78E, WBP_ALetter},
631 {0xA790, 0xA791, WBP_ALetter}, 645 {0xA790, 0xA793, WBP_ALetter},
632 {0xA7A0, 0xA7A9, WBP_ALetter}, 646 {0xA7A0, 0xA7AA, WBP_ALetter},
647 {0xA7F8, 0xA7F9, WBP_ALetter},
633 {0xA7FA, 0xA7FA, WBP_ALetter}, 648 {0xA7FA, 0xA7FA, WBP_ALetter},
634 {0xA7FB, 0xA801, WBP_ALetter}, 649 {0xA7FB, 0xA801, WBP_ALetter},
635 {0xA802, 0xA802, WBP_Extend}, 650 {0xA802, 0xA802, WBP_Extend},
@@ -686,6 +701,14 @@ static struct WordBreakProperties wb_prop_default[] = {
686 {0xAAB7, 0xAAB8, WBP_Extend}, 701 {0xAAB7, 0xAAB8, WBP_Extend},
687 {0xAABE, 0xAABF, WBP_Extend}, 702 {0xAABE, 0xAABF, WBP_Extend},
688 {0xAAC1, 0xAAC1, WBP_Extend}, 703 {0xAAC1, 0xAAC1, WBP_Extend},
704 {0xAAE0, 0xAAEA, WBP_ALetter},
705 {0xAAEB, 0xAAEB, WBP_Extend},
706 {0xAAEC, 0xAAED, WBP_Extend},
707 {0xAAEE, 0xAAEF, WBP_Extend},
708 {0xAAF2, 0xAAF2, WBP_ALetter},
709 {0xAAF3, 0xAAF4, WBP_ALetter},
710 {0xAAF5, 0xAAF5, WBP_Extend},
711 {0xAAF6, 0xAAF6, WBP_Extend},
689 {0xAB01, 0xAB06, WBP_ALetter}, 712 {0xAB01, 0xAB06, WBP_ALetter},
690 {0xAB09, 0xAB0E, WBP_ALetter}, 713 {0xAB09, 0xAB0E, WBP_ALetter},
691 {0xAB11, 0xAB16, WBP_ALetter}, 714 {0xAB11, 0xAB16, WBP_ALetter},
@@ -781,6 +804,8 @@ static struct WordBreakProperties wb_prop_default[] = {
781 {0x1083F, 0x10855, WBP_ALetter}, 804 {0x1083F, 0x10855, WBP_ALetter},
782 {0x10900, 0x10915, WBP_ALetter}, 805 {0x10900, 0x10915, WBP_ALetter},
783 {0x10920, 0x10939, WBP_ALetter}, 806 {0x10920, 0x10939, WBP_ALetter},
807 {0x10980, 0x109B7, WBP_ALetter},
808 {0x109BE, 0x109BF, WBP_ALetter},
784 {0x10A00, 0x10A00, WBP_ALetter}, 809 {0x10A00, 0x10A00, WBP_ALetter},
785 {0x10A01, 0x10A03, WBP_Extend}, 810 {0x10A01, 0x10A03, WBP_Extend},
786 {0x10A05, 0x10A06, WBP_Extend}, 811 {0x10A05, 0x10A06, WBP_Extend},
@@ -809,10 +834,40 @@ static struct WordBreakProperties wb_prop_default[] = {
809 {0x110B7, 0x110B8, WBP_Extend}, 834 {0x110B7, 0x110B8, WBP_Extend},
810 {0x110B9, 0x110BA, WBP_Extend}, 835 {0x110B9, 0x110BA, WBP_Extend},
811 {0x110BD, 0x110BD, WBP_Format}, 836 {0x110BD, 0x110BD, WBP_Format},
837 {0x110D0, 0x110E8, WBP_ALetter},
838 {0x110F0, 0x110F9, WBP_Numeric},
839 {0x11100, 0x11102, WBP_Extend},
840 {0x11103, 0x11126, WBP_ALetter},
841 {0x11127, 0x1112B, WBP_Extend},
842 {0x1112C, 0x1112C, WBP_Extend},
843 {0x1112D, 0x11134, WBP_Extend},
844 {0x11136, 0x1113F, WBP_Numeric},
845 {0x11180, 0x11181, WBP_Extend},
846 {0x11182, 0x11182, WBP_Extend},
847 {0x11183, 0x111B2, WBP_ALetter},
848 {0x111B3, 0x111B5, WBP_Extend},
849 {0x111B6, 0x111BE, WBP_Extend},
850 {0x111BF, 0x111C0, WBP_Extend},
851 {0x111C1, 0x111C4, WBP_ALetter},
852 {0x111D0, 0x111D9, WBP_Numeric},
853 {0x11680, 0x116AA, WBP_ALetter},
854 {0x116AB, 0x116AB, WBP_Extend},
855 {0x116AC, 0x116AC, WBP_Extend},
856 {0x116AD, 0x116AD, WBP_Extend},
857 {0x116AE, 0x116AF, WBP_Extend},
858 {0x116B0, 0x116B5, WBP_Extend},
859 {0x116B6, 0x116B6, WBP_Extend},
860 {0x116B7, 0x116B7, WBP_Extend},
861 {0x116C0, 0x116C9, WBP_Numeric},
812 {0x12000, 0x1236E, WBP_ALetter}, 862 {0x12000, 0x1236E, WBP_ALetter},
813 {0x12400, 0x12462, WBP_ALetter}, 863 {0x12400, 0x12462, WBP_ALetter},
814 {0x13000, 0x1342E, WBP_ALetter}, 864 {0x13000, 0x1342E, WBP_ALetter},
815 {0x16800, 0x16A38, WBP_ALetter}, 865 {0x16800, 0x16A38, WBP_ALetter},
866 {0x16F00, 0x16F44, WBP_ALetter},
867 {0x16F50, 0x16F50, WBP_ALetter},
868 {0x16F51, 0x16F7E, WBP_Extend},
869 {0x16F8F, 0x16F92, WBP_Extend},
870 {0x16F93, 0x16F9F, WBP_ALetter},
816 {0x1B000, 0x1B000, WBP_Katakana}, 871 {0x1B000, 0x1B000, WBP_Katakana},
817 {0x1D165, 0x1D166, WBP_Extend}, 872 {0x1D165, 0x1D166, WBP_Extend},
818 {0x1D167, 0x1D169, WBP_Extend}, 873 {0x1D167, 0x1D169, WBP_Extend},
@@ -853,6 +908,40 @@ static struct WordBreakProperties wb_prop_default[] = {
853 {0x1D7AA, 0x1D7C2, WBP_ALetter}, 908 {0x1D7AA, 0x1D7C2, WBP_ALetter},
854 {0x1D7C4, 0x1D7CB, WBP_ALetter}, 909 {0x1D7C4, 0x1D7CB, WBP_ALetter},
855 {0x1D7CE, 0x1D7FF, WBP_Numeric}, 910 {0x1D7CE, 0x1D7FF, WBP_Numeric},
911 {0x1EE00, 0x1EE03, WBP_ALetter},
912 {0x1EE05, 0x1EE1F, WBP_ALetter},
913 {0x1EE21, 0x1EE22, WBP_ALetter},
914 {0x1EE24, 0x1EE24, WBP_ALetter},
915 {0x1EE27, 0x1EE27, WBP_ALetter},
916 {0x1EE29, 0x1EE32, WBP_ALetter},
917 {0x1EE34, 0x1EE37, WBP_ALetter},
918 {0x1EE39, 0x1EE39, WBP_ALetter},
919 {0x1EE3B, 0x1EE3B, WBP_ALetter},
920 {0x1EE42, 0x1EE42, WBP_ALetter},
921 {0x1EE47, 0x1EE47, WBP_ALetter},
922 {0x1EE49, 0x1EE49, WBP_ALetter},
923 {0x1EE4B, 0x1EE4B, WBP_ALetter},
924 {0x1EE4D, 0x1EE4F, WBP_ALetter},
925 {0x1EE51, 0x1EE52, WBP_ALetter},
926 {0x1EE54, 0x1EE54, WBP_ALetter},
927 {0x1EE57, 0x1EE57, WBP_ALetter},
928 {0x1EE59, 0x1EE59, WBP_ALetter},
929 {0x1EE5B, 0x1EE5B, WBP_ALetter},
930 {0x1EE5D, 0x1EE5D, WBP_ALetter},
931 {0x1EE5F, 0x1EE5F, WBP_ALetter},
932 {0x1EE61, 0x1EE62, WBP_ALetter},
933 {0x1EE64, 0x1EE64, WBP_ALetter},
934 {0x1EE67, 0x1EE6A, WBP_ALetter},
935 {0x1EE6C, 0x1EE72, WBP_ALetter},
936 {0x1EE74, 0x1EE77, WBP_ALetter},
937 {0x1EE79, 0x1EE7C, WBP_ALetter},
938 {0x1EE7E, 0x1EE7E, WBP_ALetter},
939 {0x1EE80, 0x1EE89, WBP_ALetter},
940 {0x1EE8B, 0x1EE9B, WBP_ALetter},
941 {0x1EEA1, 0x1EEA3, WBP_ALetter},
942 {0x1EEA5, 0x1EEA9, WBP_ALetter},
943 {0x1EEAB, 0x1EEBB, WBP_ALetter},
944 {0x1F1E6, 0x1F1FF, WBP_Regional},
856 {0xE0001, 0xE0001, WBP_Format}, 945 {0xE0001, 0xE0001, WBP_Format},
857 {0xE0020, 0xE007F, WBP_Format}, 946 {0xE0020, 0xE007F, WBP_Format},
858 {0xE0100, 0xE01EF, WBP_Extend}, 947 {0xE0100, 0xE01EF, WBP_Extend},
diff --git a/src/static_libs/libunibreak/wordbreakdef.h b/src/static_libs/libunibreak/wordbreakdef.h
index 0b4b165..ca8b1b6 100644
--- a/src/static_libs/libunibreak/wordbreakdef.h
+++ b/src/static_libs/libunibreak/wordbreakdef.h
@@ -40,7 +40,7 @@
40 * Definitions of internal data structures, declarations of global 40 * Definitions of internal data structures, declarations of global
41 * variables, and function prototypes for the word breaking algorithm. 41 * variables, and function prototypes for the word breaking algorithm.
42 * 42 *
43 * @version 2.1, 2012/01/18 43 * @version 2.2, 2013/05/14
44 * @author Tom Hacohen 44 * @author Tom Hacohen
45 */ 45 */
46 46
@@ -50,20 +50,21 @@
50 */ 50 */
51enum WordBreakClass 51enum WordBreakClass
52{ 52{
53 WBP_Undefined, 53 WBP_Undefined,
54 WBP_CR, 54 WBP_CR,
55 WBP_LF, 55 WBP_LF,
56 WBP_Newline, 56 WBP_Newline,
57 WBP_Extend, 57 WBP_Extend,
58 WBP_Format, 58 WBP_Format,
59 WBP_Katakana, 59 WBP_Katakana,
60 WBP_ALetter, 60 WBP_ALetter,
61 WBP_MidNumLet, 61 WBP_MidNumLet,
62 WBP_MidLetter, 62 WBP_MidLetter,
63 WBP_MidNum, 63 WBP_MidNum,
64 WBP_Numeric, 64 WBP_Numeric,
65 WBP_ExtendNumLet, 65 WBP_ExtendNumLet,
66 WBP_Any 66 WBP_Regional,
67 WBP_Any
67}; 68};
68 69
69/** 70/**