OLD | NEW |
1 # Copyright (c) 2002-2013 International Business Machines Corporation and | 1 # Copyright (c) 2002-2015 International Business Machines Corporation and |
2 # others. All Rights Reserved. | 2 # others. All Rights Reserved. |
3 # | 3 # |
4 # file: line_ja.txt | 4 # file: line_normal.txt |
5 # | 5 # |
6 # Line Breaking Rules | 6 # Line Breaking Rules |
7 # Implement default line breaking as defined by | 7 # Implement default line breaking as defined by |
8 # Unicode Standard Annex #14 Revision 29 for Unicode 6.2 | 8 # Unicode Standard Annex #14 Revision 34 for Unicode 8.0 |
9 # http://www.unicode.org/reports/tr14/ | 9 # http://www.unicode.org/reports/tr14/ |
| 10 # tailored as noted in 2nd paragraph below.. |
10 # | 11 # |
11 # TODO: Rule LB 8 remains as it was in Unicode 5.2 | 12 # TODO: Rule LB 8 remains as it was in Unicode 5.2 |
12 # This is only because of a limitation of ICU break engine implementatio
n, | 13 # This is only because of a limitation of ICU break engine implementatio
n, |
13 # not because the older behavior is desirable. | 14 # not because the older behavior is desirable. |
| 15 # |
| 16 # This tailors the line break behavior to correspond to CSS |
| 17 # line-break=normal (BCP47 -u-lb-normal) as defined for languages other
than |
| 18 # Chinese & Japanese. |
| 19 # It sets characters of class CJ to behave like ID. |
14 | 20 |
15 # | 21 # |
16 # Character Classes defined by TR 14. | 22 # Character Classes defined by TR 14. |
17 # | 23 # |
18 | 24 |
19 !!chain; | 25 !!chain; |
20 !!LBCMNoChain; | 26 !!LBCMNoChain; |
21 | 27 |
22 | 28 |
23 !!lookAheadHardBreak; | 29 !!lookAheadHardBreak; |
(...skipping 318 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
342 # | 348 # |
343 $HLcm ($HYcm | $BAcm) [^$CB]?; | 349 $HLcm ($HYcm | $BAcm) [^$CB]?; |
344 | 350 |
345 # LB 21b (forward) Don't break between SY and HL | 351 # LB 21b (forward) Don't break between SY and HL |
346 # (break between HL and SY already disallowed by LB 13 above) | 352 # (break between HL and SY already disallowed by LB 13 above) |
347 $SYcm $HLcm; | 353 $SYcm $HLcm; |
348 | 354 |
349 # LB 22 | 355 # LB 22 |
350 ($ALcm | $HLcm) $INcm; | 356 ($ALcm | $HLcm) $INcm; |
351 $CM+ $INcm; # by rule 10, any otherwise unattached CM behaves as AL | 357 $CM+ $INcm; # by rule 10, any otherwise unattached CM behaves as AL |
| 358 $EXcm $INcm; |
352 $IDcm $INcm; | 359 $IDcm $INcm; |
353 $INcm $INcm; | 360 $INcm $INcm; |
354 $NUcm $INcm; | 361 $NUcm $INcm; |
355 | 362 |
356 | 363 |
357 # $LB 23 | 364 # $LB 23 |
358 $IDcm $POcm; | 365 $IDcm $POcm; |
359 $ALcm $NUcm; # includes $LB19 | 366 $ALcm $NUcm; # includes $LB19 |
360 $HLcm $NUcm; | 367 $HLcm $NUcm; |
361 $CM+ $NUcm; # Rule 10, any otherwise unattached CM behaves as AL | 368 $CM+ $NUcm; # Rule 10, any otherwise unattached CM behaves as AL |
(...skipping 215 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
577 [^$CB] $CM* $BB; # | 584 [^$CB] $CM* $BB; # |
578 | 585 |
579 # LB21a | 586 # LB21a |
580 [^$CB] $CM* ($HY | $BA) $CM* $HL; | 587 [^$CB] $CM* ($HY | $BA) $CM* $HL; |
581 | 588 |
582 # LB21b (reverse) | 589 # LB21b (reverse) |
583 $CM* $HL $CM* $SY; | 590 $CM* $HL $CM* $SY; |
584 | 591 |
585 # LB 22 | 592 # LB 22 |
586 $CM* $IN $CM* ($ALPlus | $HL); | 593 $CM* $IN $CM* ($ALPlus | $HL); |
| 594 $CM* $IN $CM* $EX; |
587 $CM* $IN $CM* $ID; | 595 $CM* $IN $CM* $ID; |
588 $CM* $IN $CM* $IN; | 596 $CM* $IN $CM* $IN; |
589 $CM* $IN $CM* $NU; | 597 $CM* $IN $CM* $NU; |
590 | 598 |
591 # LB 23 | 599 # LB 23 |
592 $CM* $PO $CM* $ID; | 600 $CM* $PO $CM* $ID; |
593 $CM* $NU $CM* ($ALPlus | $HL); | 601 $CM* $NU $CM* ($ALPlus | $HL); |
594 $CM* ($ALPlus | $HL) $CM* $NU; | 602 $CM* ($ALPlus | $HL) $CM* $NU; |
595 | 603 |
596 # LB 24 | 604 # LB 24 |
(...skipping 68 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
665 # of context. | 673 # of context. |
666 # | 674 # |
667 # It might be slightly more efficient to have specific rules | 675 # It might be slightly more efficient to have specific rules |
668 # instead of one generic one, but only if we could | 676 # instead of one generic one, but only if we could |
669 # turn off rule chaining. We don't want to move more | 677 # turn off rule chaining. We don't want to move more |
670 # than necessary. | 678 # than necessary. |
671 # | 679 # |
672 [$CM $OP $QU $CL $CP $B2 $PR $HY $BA $SP $dictionary]+ [^$CM $OP $QU $CL $CP $B2
$PR $HY $BA $dictionary]; | 680 [$CM $OP $QU $CL $CP $B2 $PR $HY $BA $SP $dictionary]+ [^$CM $OP $QU $CL $CP $B2
$PR $HY $BA $dictionary]; |
673 $dictionary $dictionary; | 681 $dictionary $dictionary; |
674 | 682 |
OLD | NEW |