1 # Copyright (c) 2002-2011 International Business Machines Corporation and 2 # others. All Rights Reserved. 3 # 4 # file: line_ja.txt 5 # 6 # Line Breaking Rules 7 # Implement default line breaking as defined by 8 # Unicode Standard Annex #14 Revision 24 for Unicode 6.0 9 # http://www.unicode.org/reports/tr14/ 10 # 11 # TODO: Rule LB 8 remains as it was in Unicode 5.2 12 # This is only because of a limitation of ICU break engine implementation, 13 # not because the older behavior is desirable. 14 15 # 16 # Character Classes defined by TR 14. 17 # 18 19 !!chain; 20 !!LBCMNoChain; 21 22 23 !!lookAheadHardBreak; 24 # 25 # !!lookAheadHardBreak Described here because it is (as yet) undocumented elsewhere 26 # and only used for the line break rules. 27 # 28 # It is used in the implementation of rule LB 10 29 # which says to treat any combining mark that is not attached to a base 30 # character as if it were of class AL (alphabetic). 31 # 32 # The problem occurs in the reverse rules. 33 # 34 # Consider a sequence like, with correct breaks as shown 35 # LF ID CM AL AL 36 # ^ ^ ^ 37 # Then consider the sequence without the initial ID (ideographic) 38 # LF CM AL AL 39 # ^ ^ 40 # Our CM, which in the first example was attached to the ideograph, 41 # is now unattached, becomes an alpha, and joins in with the other 42 # alphas. 43 # 44 # When iterating forwards, these sequences do not present any problems 45 # When iterating backwards, we need to look ahead when encountering 46 # a CM to see whether it attaches to something further on or not. 47 # (Look-ahead in a reverse rule is looking towards the start) 48 # 49 # If the CM is unattached, we need to force a break. 50 # 51 # !!lookAheadHardBreak forces the run time state machine to 52 # stop immediately when a look ahead rule ( '/' operator) matches, 53 # and set the match position to that of the look-ahead operator, 54 # no matter what other rules may be in play at the time. 55 # 56 # See rule LB 19 for an example. 57 # 58 59 $SmallHira = [\u3041 \u3043 \u3045 \u3047 \u3049 \u3063 \u3083 \u3085 \u3087 \u308E \u3095 \u3096]; 60 $SmallKata = [\u30A1 \u30A3 \u30A5 \u30A7 \u30A9 \u30C3 \u30E3 \u30E5 \u30E7 \u30EE \u30F5 \u30F6]; 61 $SmallKataExt = [\u31F0 \u31F1 \u31F2 \u31F3 \u31F4 \u31F5 \u31F6 \u31F7 \u31F8 \u31F9 \u31FA \u31FB \u31FC \u31FD \u31FE \u31FF]; 62 $SmallKanaAndProlongedMark = [[$SmallHira] [$SmallKata] [$SmallKataExt] [\u30FC]]; 63 64 $AI = [:LineBreak = Ambiguous:]; 65 $AL = [:LineBreak = Alphabetic:]; 66 $BA = [:LineBreak = Break_After:]; 67 $BB = [:LineBreak = Break_Before:]; 68 $BK = [:LineBreak = Mandatory_Break:]; 69 $B2 = [:LineBreak = Break_Both:]; 70 $CB = [:LineBreak = Contingent_Break:]; 71 $CL = [:LineBreak = Close_Punctuation:]; 72 $CM = [:LineBreak = Combining_Mark:]; 73 $CP = [:LineBreak = Close_Parenthesis:]; 74 $CR = [:LineBreak = Carriage_Return:]; 75 $EX = [:LineBreak = Exclamation:]; 76 $GL = [:LineBreak = Glue:]; 77 $HY = [:LineBreak = Hyphen:]; 78 $H2 = [:LineBreak = H2:]; 79 $H3 = [:LineBreak = H3:]; 80 $ID = [[:LineBreak = Ideographic:] [$SmallKanaAndProlongedMark]]; 81 $IN = [:LineBreak = Inseperable:]; 82 $IS = [:LineBreak = Infix_Numeric:]; 83 $JL = [:LineBreak = JL:]; 84 $JV = [:LineBreak = JV:]; 85 $JT = [:LineBreak = JT:]; 86 $LF = [:LineBreak = Line_Feed:]; 87 $NL = [:LineBreak = Next_Line:]; 88 $NS = [[:LineBreak = Nonstarter:] - [$SmallKanaAndProlongedMark]]; 89 $NU = [:LineBreak = Numeric:]; 90 $OP = [:LineBreak = Open_Punctuation:]; 91 $PO = [:LineBreak = Postfix_Numeric:]; 92 $PR = [:LineBreak = Prefix_Numeric:]; 93 $QU = [:LineBreak = Quotation:]; 94 $SA = [:LineBreak = Complex_Context:]; 95 $SG = [:LineBreak = Surrogate:]; 96 $SP = [:LineBreak = Space:]; 97 $SY = [:LineBreak = Break_Symbols:]; 98 $WJ = [:LineBreak = Word_Joiner:]; 99 $XX = [:LineBreak = Unknown:]; 100 $ZW = [:LineBreak = ZWSpace:]; 101 102 # Dictionary character set, for triggering language-based break engines. Currently 103 # limited to LineBreak=Complex_Context. Note that this set only works in Unicode 104 # 5.0 or later as the definition of Complex_Context was corrected to include all 105 # characters requiring dictionary break. 106 107 $dictionary = [:LineBreak = Complex_Context:]; 108 109 # 110 # Rule LB1. By default, treat AI (characters with ambiguous east Asian width), 111 # SA (South East Asian: Thai, Lao, Khmer) 112 # SG (Unpaired Surrogates) 113 # XX (Unknown, unassigned) 114 # as $AL (Alphabetic) 115 # 116 $ALPlus = [$AL $AI $SA $SG $XX]; 117 118 # 119 # Combining Marks. X $CM* behaves as if it were X. Rule LB6. 120 # 121 $ALcm = $ALPlus $CM*; 122 $BAcm = $BA $CM*; 123 $BBcm = $BB $CM*; 124 $B2cm = $B2 $CM*; 125 $CLcm = $CL $CM*; 126 $CPcm = $CP $CM*; 127 $EXcm = $EX $CM*; 128 $GLcm = $GL $CM*; 129 $HYcm = $HY $CM*; 130 $H2cm = $H2 $CM*; 131 $H3cm = $H3 $CM*; 132 $IDcm = $ID $CM*; 133 $INcm = $IN $CM*; 134 $IScm = $IS $CM*; 135 $JLcm = $JL $CM*; 136 $JVcm = $JV $CM*; 137 $JTcm = $JT $CM*; 138 $NScm = $NS $CM*; 139 $NUcm = $NU $CM*; 140 $OPcm = $OP $CM*; 141 $POcm = $PO $CM*; 142 $PRcm = $PR $CM*; 143 $QUcm = $QU $CM*; 144 $SYcm = $SY $CM*; 145 $WJcm = $WJ $CM*; 146 147 ## ------------------------------------------------- 148 149 !!forward; 150 151 # 152 # Each class of character can stand by itself as an unbroken token, with trailing combining stuff 153 # 154 $ALPlus $CM+; 155 $BA $CM+; 156 $BB $CM+; 157 $B2 $CM+; 158 $CL $CM+; 159 $CP $CM+; 160 $EX $CM+; 161 $GL $CM+; 162 $HY $CM+; 163 $H2 $CM+; 164 $H3 $CM+; 165 $ID $CM+; 166 $IN $CM+; 167 $IS $CM+; 168 $JL $CM+; 169 $JV $CM+; 170 $JT $CM+; 171 $NS $CM+; 172 $NU $CM+; 173 $OP $CM+; 174 $PO $CM+; 175 $PR $CM+; 176 $QU $CM+; 177 $SY $CM+; 178 $WJ $CM+; 179 180 # 181 # CAN_CM is the set of characters that may combine with CM combining chars. 182 # Note that Linebreak UAX 14's concept of a combining char and the rules 183 # for what they can combine with are _very_ different from the rest of Unicode. 184 # 185 # Note that $CM itself is left out of this set. If CM is needed as a base 186 # it must be listed separately in the rule. 187 # 188 $CAN_CM = [^$SP $BK $CR $LF $NL $ZW $CM]; # Bases that can take CMs 189 $CANT_CM = [ $SP $BK $CR $LF $NL $ZW $CM]; # Bases that can't take CMs 190 191 # 192 # AL_FOLLOW set of chars that can unconditionally follow an AL 193 # Needed in rules where stand-alone $CM s are treated as AL. 194 # Chaining is disabled with CM because it causes other failures, 195 # so for this one case we need to manually list out longer sequences. 196 # 197 $AL_FOLLOW_NOCM = [$BK $CR $LF $NL $ZW $SP]; 198 $AL_FOLLOW_CM = [$CL $CP $EX $IS $SY $WJ $GL $OP $QU $BA $HY $NS $IN $NU $ALPlus]; 199 $AL_FOLLOW = [$AL_FOLLOW_NOCM $AL_FOLLOW_CM]; 200 201 202 # 203 # Rule LB 4, 5 Mandatory (Hard) breaks. 204 # 205 $LB4Breaks = [$BK $CR $LF $NL]; 206 $LB4NonBreaks = [^$BK $CR $LF $NL]; 207 $CR $LF {100}; 208 209 # 210 # LB 6 Do not break before hard line breaks. 211 # 212 $LB4NonBreaks? $LB4Breaks {100}; # LB 5 do not break before hard breaks. 213 $CAN_CM $CM* $LB4Breaks {100}; 214 $CM+ $LB4Breaks {100}; 215 216 # LB 7 x SP 217 # x ZW 218 $LB4NonBreaks [$SP $ZW]; 219 $CAN_CM $CM* [$SP $ZW]; 220 $CM+ [$SP $ZW]; 221 222 # 223 # LB 8 Break after zero width space 224 # TODO: ZW SP* <break> 225 # An engine change is required to write the reverse rule for this. 226 # For now, leave the Unicode 5.2 rule, ZW <break> 227 # 228 $LB8Breaks = [$LB4Breaks $ZW]; 229 $LB8NonBreaks = [[$LB4NonBreaks] - [$ZW]]; 230 231 232 # LB 9 Combining marks. X $CM needs to behave like X, where X is not $SP, $BK $CR $LF $NL 233 # $CM not covered by the above needs to behave like $AL 234 # See definition of $CAN_CM. 235 236 $CAN_CM $CM+; # Stick together any combining sequences that don't match other rules. 237 $CM+; 238 239 # 240 # LB 11 Do not break before or after WORD JOINER & related characters. 241 # 242 $CAN_CM $CM* $WJcm; 243 $LB8NonBreaks $WJcm; 244 $CM+ $WJcm; 245 246 $WJcm $CANT_CM; 247 $WJcm $CAN_CM $CM*; 248 249 # 250 # LB 12 Do not break after NBSP and related characters. 251 # GL x 252 # 253 $GLcm $CAN_CM $CM*; 254 $GLcm $CANT_CM; 255 256 # 257 # LB 12a Do not break before NBSP and related characters ... 258 # [^SP BA HY] x GL 259 # 260 [[$LB8NonBreaks] - [$SP $BA $HY]] $CM* $GLcm; 261 $CM+ GLcm; 262 263 264 265 # 266 # LB 13 Don't break before ']' or '!' or ';' or '/', even after spaces. 267 # 268 $LB8NonBreaks $CL; 269 $CAN_CM $CM* $CL; 270 $CM+ $CL; # by rule 10, stand-alone CM behaves as AL 271 272 $LB8NonBreaks $CP; 273 $CAN_CM $CM* $CP; 274 $CM+ $CP; # by rule 10, stand-alone CM behaves as AL 275 276 $LB8NonBreaks $EX; 277 $CAN_CM $CM* $EX; 278 $CM+ $EX; # by rule 10, stand-alone CM behaves as AL 279 280 $LB8NonBreaks $IS; 281 $CAN_CM $CM* $IS; 282 $CM+ $IS; # by rule 10, stand-alone CM behaves as AL 283 284 $LB8NonBreaks $SY; 285 $CAN_CM $CM* $SY; 286 $CM+ $SY; # by rule 10, stand-alone CM behaves as AL 287 288 289 # 290 # LB 14 Do not break after OP, even after spaces 291 # 292 $OPcm $SP* $CAN_CM $CM*; 293 $OPcm $SP* $CANT_CM; 294 295 $OPcm $SP+ $CM+ $AL_FOLLOW?; # by rule 10, stand-alone CM behaves as AL 296 297 # LB 15 298 $QUcm $SP* $OPcm; 299 300 # LB 16 301 ($CLcm | $CPcm) $SP* $NScm; 302 303 # LB 17 304 $B2cm $SP* $B2cm; 305 306 # 307 # LB 18 Break after spaces. 308 # 309 $LB18NonBreaks = [$LB8NonBreaks - [$SP]]; 310 $LB18Breaks = [$LB8Breaks $SP]; 311 312 313 # LB 19 314 # x QU 315 $LB18NonBreaks $CM* $QUcm; 316 $CM+ $QUcm; 317 318 # QU x 319 $QUcm .?; 320 $QUcm $LB18NonBreaks $CM*; # Don't let a combining mark go onto $CR, $BK, etc. 321 # TODO: I don't think this rule is needed. 322 323 324 # LB 20 325 # <break> $CB 326 # $CB <break> 327 328 $LB20NonBreaks = [$LB18NonBreaks - $CB]; 329 330 # LB 21 x (BA | HY | NS) 331 # BB x 332 # 333 $LB20NonBreaks $CM* ($BAcm | $HYcm | $NScm); 334 335 $BBcm [^$CB]; # $BB x 336 $BBcm $LB20NonBreaks $CM*; 337 338 # LB 22 339 $ALcm $INcm; 340 $CM+ $INcm; # by rule 10, any otherwise unattached CM behaves as AL 341 $IDcm $INcm; 342 $INcm $INcm; 343 $NUcm $INcm; 344 345 346 # $LB 23 347 $IDcm $POcm; 348 $ALcm $NUcm; # includes $LB19 349 $CM+ $NUcm; # Rule 10, any otherwise unattached CM behaves as AL 350 $NUcm $ALcm; 351 352 # 353 # LB 24 354 # 355 $PRcm $IDcm; 356 $PRcm $ALcm; 357 $POcm $ALcm; 358 359 # 360 # LB 25 Numbers. 361 # 362 ($PRcm | $POcm)? ($OPcm | $HYcm)? $NUcm ($NUcm | $SYcm | $IScm)* ($CLcm | $CPcm)? ($PRcm | $POcm)?; 363 364 # LB 26 Do not break a Korean syllable 365 # 366 $JLcm ($JLcm | $JVcm | $H2cm | $H3cm); 367 ($JVcm | $H2cm) ($JVcm | $JTcm); 368 ($JTcm | $H3cm) $JTcm; 369 370 # LB 27 Treat korean Syllable Block the same as ID (don't break it) 371 ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm) $INcm; 372 ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm) $POcm; 373 $PRcm ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm); 374 375 376 # LB 28 Do not break between alphabetics 377 # 378 $ALcm $ALcm; 379 $CM+ $ALcm; # The $CM+ is from rule 10, an unattached CM is treated as AL 380 381 # LB 29 382 $IScm $ALcm; 383 384 # LB 30 385 ($ALcm | $NUcm) $OPcm; 386 $CM+ $OPcm; # The $CM+ is from rule 10, an unattached CM is treated as AL. 387 $CPcm ($ALcm | $NUcm); 388 389 390 # 391 # Reverse Rules. 392 # 393 ## ------------------------------------------------- 394 395 !!reverse; 396 397 $CM+ $ALPlus; 398 $CM+ $BA; 399 $CM+ $BB; 400 $CM+ $B2; 401 $CM+ $CL; 402 $CM+ $CP; 403 $CM+ $EX; 404 $CM+ $GL; 405 $CM+ $HY; 406 $CM+ $H2; 407 $CM+ $H3; 408 $CM+ $ID; 409 $CM+ $IN; 410 $CM+ $IS; 411 $CM+ $JL; 412 $CM+ $JV; 413 $CM+ $JT; 414 $CM+ $NS; 415 $CM+ $NU; 416 $CM+ $OP; 417 $CM+ $PO; 418 $CM+ $PR; 419 $CM+ $QU; 420 $CM+ $SY; 421 $CM+ $WJ; 422 $CM+; 423 424 425 # 426 # Sequences of the form (shown forwards) 427 # [CANT_CM] <break> [CM] [whatever] 428 # The CM needs to behave as an AL 429 # 430 $AL_FOLLOW $CM+ / ( 431 [$BK $CR $LF $NL $ZW {eof}] | 432 $SP+ $CM+ $SP | 433 $SP+ $CM* ([^$OP $CM $SP] | [$AL {eof}])); # if LB 14 will match, need to surpress this break. 434 # LB14 says OP SP* x . 435 # becomes OP SP* x AL 436 # becomes OP SP* x CM+ AL_FOLLOW 437 # 438 # Further note: the $AL in [$AL {eof}] is only to work around 439 # a rule compiler bug which complains about 440 # empty sets otherwise. 441 442 # 443 # Sequences of the form (shown forwards) 444 # [CANT_CM] <break> [CM] <break> [PR] 445 # The CM needs to behave as an AL 446 # This rule is concerned about getting the second of the two <breaks> in place. 447 # 448 449 [$PR ] / $CM+ [$BK $CR $LF $NL $ZW $SP {eof}]; 450 451 452 453 # LB 4, 5, 5 454 455 $LB4Breaks [$LB4NonBreaks-$CM]; 456 $LB4Breaks $CM+ $CAN_CM; 457 $LF $CR; 458 459 460 # LB 7 x SP 461 # x ZW 462 [$SP $ZW] [$LB4NonBreaks-$CM]; 463 [$SP $ZW] $CM+ $CAN_CM; 464 465 # LB 8 ZW SP* <break> 466 # TODO: to implement this, we need more than one look-ahead hard break in play at a time. 467 # Requires an engine enhancement. 468 # / $SP* $ZW 469 470 # LB 9,10 Combining marks. 471 # X $CM needs to behave like X, where X is not $SP or controls. 472 # $CM not covered by the above needs to behave like $AL 473 # Stick together any combining sequences that don't match other rules. 474 $CM+ $CAN_CM; 475 476 477 # LB 11 478 $CM* $WJ $CM* $CAN_CM; 479 $CM* $WJ [$LB8NonBreaks-$CM]; 480 481 $CANT_CM $CM* $WJ; 482 $CM* $CAN_CM $CM* $WJ; 483 484 # LB 12a 485 # [^SP BA HY] x GL 486 # 487 $CM* $GL $CM* [$LB8NonBreaks-[$CM $SP $BA $HY]]; 488 489 # LB 12 490 # GL x 491 # 492 $CANT_CM $CM* $GL; 493 $CM* $CAN_CM $CM* $GL; 494 495 496 # LB 13 497 $CL $CM+ $CAN_CM; 498 $CP $CM+ $CAN_CM; 499 $EX $CM+ $CAN_CM; 500 $IS $CM+ $CAN_CM; 501 $SY $CM+ $CAN_CM; 502 503 $CL [$LB8NonBreaks-$CM]; 504 $CP [$LB8NonBreaks-$CM]; 505 $EX [$LB8NonBreaks-$CM]; 506 $IS [$LB8NonBreaks-$CM]; 507 $SY [$LB8NonBreaks-$CM]; 508 509 # Rule 13 & 14 taken together for an edge case. 510 # Match this, shown forward 511 # OP SP+ ($CM+ behaving as $AL) (CL | CP | EX | IS | IY) 512 # This really wants to chain at the $CM+ (which is acting as an $AL) 513 # except for $CM chaining being disabled. 514 [$CL $CP $EX $IS $SY] $CM+ $SP+ $CM* $OP; 515 516 # LB 14 OP SP* x 517 # 518 $CM* $CAN_CM $SP* $CM* $OP; 519 $CANT_CM $SP* $CM* $OP; 520 $AL_FOLLOW? $CM+ $SP $SP* $CM* $OP; # by LB 10, behaves like $AL_FOLLOW? $AL $SP* $CM* $OP 521 522 $AL_FOLLOW_NOCM $CM+ $SP+ $CM* $OP; 523 $CM* $AL_FOLLOW_CM $CM+ $SP+ $CM* $OP; 524 $SY $CM $SP+ $OP; # TODO: Experiment. Remove. 525 526 527 528 # LB 15 529 $CM* $OP $SP* $CM* $QU; 530 531 # LB 16 532 $CM* $NS $SP* $CM* ($CL | $CP); 533 534 # LB 17 535 $CM* $B2 $SP* $CM* $B2; 536 537 # LB 18 break after spaces 538 # Nothing explicit needed here. 539 540 541 # 542 # LB 19 543 # 544 $CM* $QU $CM* $CAN_CM; # . x QU 545 $CM* $QU $LB18NonBreaks; 546 547 548 $CM* $CAN_CM $CM* $QU; # QU x . 549 $CANT_CM $CM* $QU; 550 551 # 552 # LB 20 Break before and after CB. 553 # nothing needed here. 554 # 555 556 # LB 21 557 $CM* ($BA | $HY | $NS) $CM* [$LB20NonBreaks-$CM]; # . x (BA | HY | NS) 558 559 $CM* [$LB20NonBreaks-$CM] $CM* $BB; # BB x . 560 [^$CB] $CM* $BB; # 561 562 563 564 # LB 22 565 $CM* $IN $CM* $ALPlus; 566 $CM* $IN $CM* $ID; 567 $CM* $IN $CM* $IN; 568 $CM* $IN $CM* $NU; 569 570 # LB 23 571 $CM* $PO $CM* $ID; 572 $CM* $NU $CM* $ALPlus; 573 $CM* $ALPlus $CM* $NU; 574 575 # LB 24 576 $CM* $ID $CM* $PR; 577 $CM* $ALPlus $CM* $PR; 578 $CM* $ALPlus $CM* $PO; 579 580 581 # LB 25 582 ($CM* ($PR | $PO))? ($CM* ($CL | $CP))? ($CM* ($NU | $IS | $SY))* $CM* $NU ($CM* ($OP | $HY))? ($CM* ($PR | $PO))?; 583 584 # LB 26 585 $CM* ($H3 | $H2 | $JV | $JL) $CM* $JL; 586 $CM* ($JT | $JV) $CM* ($H2 | $JV); 587 $CM* $JT $CM* ($H3 | $JT); 588 589 # LB 27 590 $CM* $IN $CM* ($H3 | $H2 | $JT | $JV | $JL); 591 $CM* $PO $CM* ($H3 | $H2 | $JT | $JV | $JL); 592 $CM* ($H3 | $H2 | $JT | $JV | $JL) $CM* $PR; 593 594 # LB 28 595 $CM* $ALPlus $CM* $ALPlus; 596 597 598 # LB 29 599 $CM* $ALPlus $CM* $IS; 600 601 # LB 30 602 $CM* $OP $CM* ($ALPlus | $NU); 603 $CM* ($ALPlus | $NU) $CM* $CP; 604 605 606 ## ------------------------------------------------- 607 608 !!safe_reverse; 609 610 # LB 9 611 $CM+ [^$CM $BK $CR $LF $NL $ZW $SP]; 612 $CM+ $SP / .; 613 614 # LB 14 615 $SP+ $CM* $OP; 616 617 # LB 15 618 $SP+ $CM* $QU; 619 620 # LB 16 621 $SP+ $CM* ($CL | $CP); 622 623 # LB 17 624 $SP+ $CM* $B2; 625 626 # LB 25 627 ($CM* ($IS | $SY))+ $CM* $NU; 628 ($CL | $CP) $CM* ($NU | $IS | $SY); 629 630 # For dictionary-based break 631 $dictionary $dictionary; 632 633 ## ------------------------------------------------- 634 635 !!safe_forward; 636 637 # Skip forward over all character classes that are involved in 638 # rules containing patterns with possibly more than one char 639 # of context. 640 # 641 # It might be slightly more efficient to have specific rules 642 # instead of one generic one, but only if we could 643 # turn off rule chaining. We don't want to move more 644 # than necessary. 645 # 646 [$CM $OP $QU $CL $CP $B2 $PR $HY $SP $dictionary]+ [^$CM $OP $QU $CL $CP $B2 $PR $HY $dictionary]; 647 $dictionary $dictionary; 648 649
