icuSources/data/brkitr/line_fi.txt

   1 # Copyright (c) 2002-2015  International Business Machines Corporation and
   2 # others. All Rights Reserved.
   3 #
   4 #  file:  line_fi.txt
   5 #
   6 #         Line Breaking Rules
   7 #         Implement default line breaking as defined by
   8 #         Unicode Standard Annex #14 Revision 29 for Unicode 6.2
   9 #         http://www.unicode.org/reports/tr14/
  10 #         tailored as noted in 2nd paragraph below..
  11 #
  12 #         TODO:  Rule LB 8 remains as it was in Unicode 5.2
  13 #         This is only because of a limitation of ICU break engine implementation,
  14 #         not because the older behavior is desirable.
  15 #
  16 #         This tailors the line break behavior for Finnish, while otherwise behaving
  17 #         per UAX 14 which corresponds to CSS line-break=strict (BCP47 -u-lb-strict).
  18 #         It sets characters of class CJ to behave like NS.
  19
  20 #
  21 #  Character Classes defined by TR 14.
  22 #
  23
  24 !!chain;
  25 !!LBCMNoChain;
  26
  27
  28 !!lookAheadHardBreak;
  29 #
  30 #  !!lookAheadHardBreak    Described here because it is (as yet) undocumented elsewhere
  31 #                          and only used for the line break rules.
  32 #
  33 #           It is used in the implementation of rule LB 10
  34 #           which says to treat any combining mark that is not attached to a base
  35 #           character as if it were of class AL  (alphabetic).
  36 #
  37 #           The problem occurs in the reverse rules.
  38 #
  39 #           Consider a sequence like, with correct breaks as shown
  40 #               LF  ID  CM  AL  AL
  41 #                  ^       ^       ^
  42 #           Then consider the sequence without the initial ID (ideographic)
  43 #                 LF  CM  AL  AL
  44 #                    ^           ^
  45 #           Our CM, which in the first example was attached to the ideograph,
  46 #           is now unattached, becomes an alpha, and joins in with the other
  47 #           alphas.
  48 #
  49 #           When iterating forwards, these sequences do not present any problems
  50 #           When iterating backwards, we need to look ahead when encountering
  51 #           a CM to see whether it attaches to something further on or not.
  52 #           (Look-ahead in a reverse rule is looking towards the start)
  53 #
  54 #           If the CM is unattached, we need to force a break.
  55 #
  56 #           !!lookAheadHardBreak forces the run time state machine to
  57 #           stop immediately when a look ahead rule ( '/' operator) matches,
  58 #           and set the match position to that of the look-ahead operator,
  59 #           no matter what other rules may be in play at the time.
  60 #
  61 #           See rule LB 19 for an example.
  62 #
  63
  64 $AI = [:LineBreak =  Ambiguous:];
  65 $AL = [:LineBreak =  Alphabetic:];
  66 $BA = [[:LineBreak =  Break_After:] - [\u2010]];
  67 $HH = [\u2010];
  68 $BB = [:LineBreak =  Break_Before:];
  69 $BK = [:LineBreak =  Mandatory_Break:];
  70 $B2 = [:LineBreak =  Break_Both:];
  71 $CB = [:LineBreak =  Contingent_Break:];
  72 $CJ = [:LineBreak =  Conditional_Japanese_Starter:];
  73 $CL = [:LineBreak =  Close_Punctuation:];
  74 $CM = [:LineBreak =  Combining_Mark:];
  75 $CP = [:LineBreak =  Close_Parenthesis:];
  76 $CR = [:LineBreak =  Carriage_Return:];
  77 $EX = [:LineBreak =  Exclamation:];
  78 $GL = [:LineBreak =  Glue:];
  79 $HL = [:LineBreak =  Hebrew_Letter:];
  80 $HY = [:LineBreak =  Hyphen:];
  81 $H2 = [:LineBreak =  H2:];
  82 $H3 = [:LineBreak =  H3:];
  83 $ID = [:LineBreak =  Ideographic:];
  84 $IN = [:LineBreak =  Inseperable:];
  85 $IS = [:LineBreak =  Infix_Numeric:];
  86 $JL = [:LineBreak =  JL:];
  87 $JV = [:LineBreak =  JV:];
  88 $JT = [:LineBreak =  JT:];
  89 $LF = [:LineBreak =  Line_Feed:];
  90 $NL = [:LineBreak =  Next_Line:];
  91 $NS = [[:LineBreak =  Nonstarter:] $CJ];
  92 $NU = [:LineBreak =  Numeric:];
  93 $OP = [:LineBreak =  Open_Punctuation:];
  94 $PO = [:LineBreak =  Postfix_Numeric:];
  95 $PR = [:LineBreak =  Prefix_Numeric:];
  96 $QU = [:LineBreak =  Quotation:];
  97 $RI = [:LineBreak =  Regional_Indicator:];
  98 $SA = [:LineBreak =  Complex_Context:];
  99 $SG = [:LineBreak =  Surrogate:];
 100 $SP = [:LineBreak =  Space:];
 101 $SY = [:LineBreak =  Break_Symbols:];
 102 $WJ = [:LineBreak =  Word_Joiner:];
 103 $XX = [:LineBreak =  Unknown:];
 104 $ZW = [:LineBreak =  ZWSpace:];
 105
 106 # Special character classes for people & body part emoji:
 107 # Subsets of $CM:
 108 $ZWJ = \u200D;
 109 $EmojiVar = \uFE0F;
 110 # The following are subsets of $ID
 111 $EmojiForSeqs = [\u2764 \U0001F441 \U0001F466-\U0001F469 \U0001F48B \U0001F5E8];
 112 $EmojiForMods = [\u261D \u26F9 \u270A-\u270D \U0001F385 \U0001F3C3-\U0001F3C4 \U0001F3CA-\U0001F3CB \U0001F442-\U0001F443 \U0001F446-\U0001F450 \U0001F466-\U0001F469 \U0001F46E-\U0001F478 \U0001F47C \U0001F481-\U0001F483 \U0001F485-\U0001F487 \U0001F4AA \U0001F575 \U0001F590 \U0001F595 \U0001F596 \U0001F645-\U0001F647 \U0001F64B-\U0001F64F \U0001F6A3 \U0001F6B4-\U0001F6B6 \U0001F6C0 \U0001F918];
 113 $EmojiMods = [\U0001F3FB-\U0001F3FF];
 114
 115 #   Dictionary character set, for triggering language-based break engines. Currently
 116 #   limited to LineBreak=Complex_Context. Note that this set only works in Unicode
 117 #   5.0 or later as the definition of Complex_Context was corrected to include all
 118 #   characters requiring dictionary break.
 119
 120 $dictionary = [:LineBreak = Complex_Context:];
 121
 122 #
 123 #  Rule LB1.  By default, treat AI  (characters with ambiguous east Asian width),
 124 #                               SA  (South East Asian: Thai, Lao, Khmer)
 125 #                               SG  (Unpaired Surrogates)
 126 #                               XX  (Unknown, unassigned)
 127 #                         as $AL  (Alphabetic)
 128 #
 129 $ALPlus = [$AL $AI $SA $SG $XX];
 130
 131 #
 132 #  Combining Marks.   X $CM*  behaves as if it were X.  Rule LB6.
 133 #
 134 $ALcm = $ALPlus $CM*;
 135 $BAcm = $BA $CM*;
 136 $HHcm = $HH $CM*;
 137 $BBcm = $BB $CM*;
 138 $B2cm = $B2 $CM*;
 139 $CLcm = $CL $CM*;
 140 $CPcm = $CP $CM*;
 141 $EXcm = $EX $CM*;
 142 $GLcm = $GL $CM*;
 143 $HLcm = $HL $CM*;
 144 $HYcm = $HY $CM*;
 145 $H2cm = $H2 $CM*;
 146 $H3cm = $H3 $CM*;
 147 $IDcm = $ID $CM*;
 148 $INcm = $IN $CM*;
 149 $IScm = $IS $CM*;
 150 $JLcm = $JL $CM*;
 151 $JVcm = $JV $CM*;
 152 $JTcm = $JT $CM*;
 153 $NScm = $NS $CM*;
 154 $NUcm = $NU $CM*;
 155 $OPcm = $OP $CM*;
 156 $POcm = $PO $CM*;
 157 $PRcm = $PR $CM*;
 158 $QUcm = $QU $CM*;
 159 $RIcm = $RI $CM*;
 160 $SYcm = $SY $CM*;
 161 $WJcm = $WJ $CM*;
 162
 163 ## -------------------------------------------------
 164
 165 !!forward;
 166
 167 #
 168 #  Each class of character can stand by itself as an unbroken token, with trailing combining stuff
 169 #
 170 $ALPlus $CM+;
 171 $BA $CM+;
 172 $HH $CM+;
 173 $BB $CM+;
 174 $B2 $CM+;
 175 $CL $CM+;
 176 $CP $CM+;
 177 $EX $CM+;
 178 $GL $CM+;
 179 $HL $CM+;
 180 $HY $CM+;
 181 $H2 $CM+;
 182 $H3 $CM+;
 183 $ID $CM+;
 184 $IN $CM+;
 185 $IS $CM+;
 186 $JL $CM+;
 187 $JV $CM+;
 188 $JT $CM+;
 189 $NS $CM+;
 190 $NU $CM+;
 191 $OP $CM+;
 192 $PO $CM+;
 193 $PR $CM+;
 194 $QU $CM+;
 195 $RI $CM+;
 196 $SY $CM+;
 197 $WJ $CM+;
 198
 199 #
 200 # CAN_CM  is the set of characters that may combine with CM combining chars.
 201 #         Note that Linebreak UAX 14's concept of a combining char and the rules
 202 #         for what they can combine with are _very_ different from the rest of Unicode.
 203 #
 204 #         Note that $CM itself is left out of this set.  If CM is needed as a base
 205 #         it must be listed separately in the rule.
 206 #
 207 $CAN_CM  = [^$SP $BK $CR $LF $NL $ZW $CM];       # Bases that can   take CMs
 208 $CANT_CM = [ $SP $BK $CR $LF $NL $ZW $CM];       # Bases that can't take CMs
 209
 210 #
 211 # AL_FOLLOW  set of chars that can unconditionally follow an AL
 212 #            Needed in rules where stand-alone $CM s are treated as AL.
 213 #            Chaining is disabled with CM because it causes other failures,
 214 #            so for this one case we need to manually list out longer sequences.
 215 #
 216 $AL_FOLLOW_NOCM = [$BK $CR $LF $NL $ZW $SP];
 217 $AL_FOLLOW_CM   = [$CL $CP $EX $HL $IS $SY $WJ $GL $OP $QU $BA $HH $HY $NS $IN $NU $ALPlus];
 218 $AL_FOLLOW      = [$AL_FOLLOW_NOCM $AL_FOLLOW_CM];
 219
 220
 221 #
 222 #  Rule LB 4, 5    Mandatory (Hard) breaks.
 223 #
 224 $LB4Breaks    = [$BK $CR $LF $NL];
 225 $LB4NonBreaks = [^$BK $CR $LF $NL];
 226 $CR $LF {100};
 227
 228 #
 229 #  LB 6    Do not break before hard line breaks.
 230 #
 231 $LB4NonBreaks?  $LB4Breaks {100};    # LB 5  do not break before hard breaks.
 232 $CAN_CM $CM*    $LB4Breaks {100};
 233 $CM+            $LB4Breaks {100};
 234
 235 # LB 7         x SP
 236 #              x ZW
 237 $LB4NonBreaks [$SP $ZW];
 238 $CAN_CM $CM*  [$SP $ZW];
 239 $CM+          [$SP $ZW];
 240
 241 #
 242 # LB 8         Break after zero width space
 243 #              TODO:  ZW SP* <break>
 244 #              An engine change is required to write the reverse rule for this.
 245 #              For now, leave the Unicode 5.2 rule, ZW <break>
 246 #
 247 $LB8Breaks    = [$LB4Breaks $ZW];
 248 $LB8NonBreaks = [[$LB4NonBreaks] - [$ZW]];
 249
 250
 251 # Special forward rule for people & body part emoji:
 252 # don't break $ZWJ from subsequent $EmojiForSeqs
 253 $EmojiForSeqs $EmojiVar? $EmojiMods? $ZWJ $EmojiForSeqs;
 254
 255 # LB 9     Combining marks.      X   $CM needs to behave like X, where X is not $SP, $BK $CR $LF $NL
 256 #                                $CM not covered by the above needs to behave like $AL
 257 #                                See definition of $CAN_CM.
 258
 259 $CAN_CM $CM+;                   #  Stick together any combining sequences that don't match other rules.
 260 $CM+;
 261
 262 #
 263 # LB 11  Do not break before or after WORD JOINER & related characters.
 264 #
 265 $CAN_CM $CM*  $WJcm;
 266 $LB8NonBreaks $WJcm;
 267 $CM+          $WJcm;
 268
 269 $WJcm $CANT_CM;
 270 $WJcm $CAN_CM $CM*;
 271
 272 #
 273 # LB 12  Do not break after NBSP and related characters.
 274 #         GL  x
 275 #
 276 $GLcm $CAN_CM $CM*;
 277 $GLcm $CANT_CM;
 278
 279 #
 280 # LB 12a  Do not break before NBSP and related characters ...
 281 #            [^SP BA HY] x GL
 282 #
 283 [[$LB8NonBreaks] - [$SP $BA $HH $HY]] $CM* $GLcm;
 284 $CM+ GLcm;
 285
 286
 287
 288 #
 289 # LB 13   Don't break before ']' or '!' or ';' or '/', even after spaces.
 290 #
 291 $LB8NonBreaks $CL;
 292 $CAN_CM $CM*  $CL;
 293 $CM+          $CL;              # by rule 10, stand-alone CM behaves as AL
 294
 295 $LB8NonBreaks $CP;
 296 $CAN_CM $CM*  $CP;
 297 $CM+          $CP;              # by rule 10, stand-alone CM behaves as AL
 298
 299 $LB8NonBreaks $EX;
 300 $CAN_CM $CM*  $EX;
 301 $CM+          $EX;              # by rule 10, stand-alone CM behaves as AL
 302
 303 $LB8NonBreaks $IS;
 304 $CAN_CM $CM*  $IS;
 305 $CM+          $IS;              # by rule 10, stand-alone CM behaves as AL
 306
 307 $LB8NonBreaks $SY;
 308 $CAN_CM $CM*  $SY;
 309 $CM+          $SY;              # by rule 10, stand-alone CM behaves as AL
 310
 311
 312 #
 313 # LB 14  Do not break after OP, even after spaces
 314 #
 315 $OPcm $SP* $CAN_CM $CM*;
 316 $OPcm $SP* $CANT_CM;
 317
 318 $OPcm $SP+ $CM+ $AL_FOLLOW?;    # by rule 10, stand-alone CM behaves as AL
 319
 320 # LB 15
 321 $QUcm $SP* $OPcm;
 322
 323 # LB 16
 324 ($CLcm | $CPcm) $SP* $NScm;
 325
 326 # LB 17
 327 $B2cm $SP* $B2cm;
 328
 329 #
 330 # LB 18  Break after spaces.
 331 #
 332 $LB18NonBreaks = [$LB8NonBreaks - [$SP]];
 333 $LB18Breaks    = [$LB8Breaks $SP];
 334
 335
 336 # LB 19
 337 #         x QU
 338 $LB18NonBreaks $CM* $QUcm;
 339 $CM+                $QUcm;
 340
 341 #         QU  x
 342 $QUcm .?;
 343 $QUcm $LB18NonBreaks $CM*;    # Don't let a combining mark go onto $CR, $BK, etc.
 344                               #  TODO:  I don't think this rule is needed.
 345
 346
 347 # LB 20
 348 #        <break>  $CB
 349 #        $CB   <break>
 350
 351 $LB20NonBreaks = [$LB18NonBreaks - $CB];
 352
 353 # LB 20.09 added rule for Finnish tailoring
 354 # LB 21        x   (BA | HY | NS)
 355 #           BB x
 356 #
 357 $LB20NonBreaks $CM* ($BAcm | $HHcm | $HYcm | $NScm) / $AL;
 358 $LB20NonBreaks $CM* ($BAcm | $HHcm | $HYcm | $NScm);
 359 ($HY | $HH) $AL;
 360
 361 $BBcm [^$CB];                                  #  $BB  x
 362 $BBcm $LB20NonBreaks $CM*;
 363
 364 # LB 21a Don't break after Hebrew + Hyphen
 365 #   HL (HY | BA) x
 366 #
 367 $HLcm ($HYcm | $BAcm | $HHcm) [^$CB]?;
 368
 369 # LB 21b (forward) Don't break between SY and HL
 370 # (break between HL and SY already disallowed by LB 13 above)
 371 $SYcm $HLcm;
 372
 373 # LB 22
 374 ($ALcm | $HLcm) $INcm;
 375 $CM+     $INcm;     #  by rule 10, any otherwise unattached CM behaves as AL
 376 $IDcm    $INcm;
 377 $INcm    $INcm;
 378 $NUcm    $INcm;
 379
 380
 381 # $LB 23
 382 $IDcm  $POcm;
 383 $ALcm  $NUcm;       # includes $LB19
 384 $HLcm  $NUcm;
 385 $CM+   $NUcm;       # Rule 10, any otherwise unattached CM behaves as AL
 386 $NUcm  $ALcm;
 387 $NUcm  $HLcm;
 388
 389 #
 390 # LB 24
 391 #
 392 $PRcm $IDcm;
 393 $PRcm ($ALcm | $HLcm);
 394 $POcm ($ALcm | $HLcm);
 395
 396 #
 397 # LB 25   Numbers.
 398 #
 399 ($PRcm | $POcm)? ($OPcm | $HYcm)? $NUcm ($NUcm | $SYcm | $IScm)* ($CLcm | $CPcm)? ($PRcm | $POcm)?;
 400
 401 # LB 26  Do not break a Korean syllable
 402 #
 403 $JLcm ($JLcm | $JVcm | $H2cm | $H3cm);
 404 ($JVcm | $H2cm) ($JVcm | $JTcm);
 405 ($JTcm | $H3cm) $JTcm;
 406
 407 # LB 27  Treat korean Syllable Block the same as ID  (don't break it)
 408 ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm) $INcm;
 409 ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm) $POcm;
 410 $PRcm ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm);
 411
 412
 413 # LB 28   Do not break between alphabetics
 414 #
 415 ($ALcm | $HLcm) ($ALcm | $HLcm);
 416 $CM+ ($ALcm | $HLcm);      # The $CM+ is from rule 10, an unattached CM is treated as AL
 417
 418 # LB 29
 419 $IScm ($ALcm | $HLcm);
 420
 421 # LB 30
 422 ($ALcm | $HLcm | $NUcm) $OPcm;
 423 $CM+ $OPcm;         # The $CM+ is from rule 10, an unattached CM is treated as AL.
 424 $CPcm ($ALcm | $HLcm | $NUcm);
 425
 426 # LB 30a  Do not break between regional indicators.
 427 $RIcm $RIcm;
 428
 429 # Special forward rule for people & body part emoji:
 430 # don't break between relevant emoji and $EmojiMods
 431 $EmojiForMods $EmojiVar? $EmojiMods;
 432
 433 #
 434 #  Reverse Rules.
 435 #
 436 ## -------------------------------------------------
 437
 438 !!reverse;
 439
 440 $CM+ $ALPlus;
 441 $CM+ $BA;
 442 $CM+ $HH;
 443 $CM+ $BB;
 444 $CM+ $B2;
 445 $CM+ $CL;
 446 $CM+ $CP;
 447 $CM+ $EX;
 448 $CM+ $GL;
 449 $CM+ $HL;
 450 $CM+ $HY;
 451 $CM+ $H2;
 452 $CM+ $H3;
 453 $CM+ $ID;
 454 $CM+ $IN;
 455 $CM+ $IS;
 456 $CM+ $JL;
 457 $CM+ $JV;
 458 $CM+ $JT;
 459 $CM+ $NS;
 460 $CM+ $NU;
 461 $CM+ $OP;
 462 $CM+ $PO;
 463 $CM+ $PR;
 464 $CM+ $QU;
 465 $CM+ $RI;
 466 $CM+ $SY;
 467 $CM+ $WJ;
 468 $CM+;
 469
 470
 471 #
 472 #  Sequences of the form  (shown forwards)
 473 #      [CANT_CM]  <break>  [CM]  [whatever]
 474 #  The CM needs to behave as an AL
 475 #
 476 $AL_FOLLOW $CM+ / (
 477           [$BK $CR $LF $NL $ZW {eof}] |
 478           $SP+ $CM+ $SP |
 479           $SP+ $CM* ([^$OP $CM $SP] | [$AL {eof}]));   # if LB 14 will match, need to surpress this break.
 480                                                #  LB14 says    OP SP* x .
 481                                                #    becomes    OP SP* x AL
 482                                                #    becomes    OP SP* x CM+ AL_FOLLOW
 483                                                #
 484                                                # Further note:  the $AL in [$AL {eof}] is only to work around
 485                                                #                a rule compiler bug which complains about
 486                                                #                empty sets otherwise.
 487
 488 #
 489 #  Sequences of the form  (shown forwards)
 490 #      [CANT_CM]  <break> [CM]  <break>  [PR]
 491 #  The CM needs to behave as an AL
 492 #  This rule is concerned about getting the second of the two <breaks> in place.
 493 #
 494
 495 [$PR   ] / $CM+ [$BK $CR $LF $NL $ZW $SP {eof}];
 496
 497
 498
 499 # LB 4, 5, 5
 500
 501 $LB4Breaks [$LB4NonBreaks-$CM];
 502 $LB4Breaks $CM+ $CAN_CM;
 503 $LF $CR;
 504
 505
 506 # LB 7         x SP
 507 #              x ZW
 508 [$SP $ZW] [$LB4NonBreaks-$CM];
 509 [$SP $ZW] $CM+ $CAN_CM;
 510
 511 # LB 8 ZW SP* <break>
 512 #     TODO: to implement this, we need more than one look-ahead hard break in play at a time.
 513 #           Requires an engine enhancement.
 514 #   / $SP* $ZW
 515
 516 # Special reverse rule for people & body part emoji:
 517 # don't break $ZWJ from subsequent $EmojiForSeqs
 518 $EmojiForSeqs $ZWJ $EmojiMods? $EmojiVar? $EmojiForSeqs;
 519
 520 # LB 9,10  Combining marks.
 521 #    X   $CM needs to behave like X, where X is not $SP or controls.
 522 #    $CM not covered by the above needs to behave like $AL
 523 # Stick together any combining sequences that don't match other rules.
 524 $CM+ $CAN_CM;
 525
 526
 527 # LB 11
 528 $CM* $WJ $CM* $CAN_CM;
 529 $CM* $WJ      [$LB8NonBreaks-$CM];
 530
 531      $CANT_CM $CM* $WJ;
 532 $CM* $CAN_CM  $CM* $WJ;
 533
 534 # LB 12a
 535 #      [^SP BA HY] x GL
 536 #
 537 $CM* $GL $CM* [$LB8NonBreaks-[$CM $SP $BA $HH $HY]];
 538
 539 # LB 12
 540 #     GL  x
 541 #
 542 $CANT_CM $CM* $GL;
 543 $CM* $CAN_CM $CM* $GL;
 544
 545
 546 # LB 13
 547 $CL $CM+ $CAN_CM;
 548 $CP $CM+ $CAN_CM;
 549 $EX $CM+ $CAN_CM;
 550 $IS $CM+ $CAN_CM;
 551 $SY $CM+ $CAN_CM;
 552
 553 $CL [$LB8NonBreaks-$CM];
 554 $CP [$LB8NonBreaks-$CM];
 555 $EX [$LB8NonBreaks-$CM];
 556 $IS [$LB8NonBreaks-$CM];
 557 $SY [$LB8NonBreaks-$CM];
 558
 559 # Rule 13 & 14 taken together for an edge case.
 560 #   Match this, shown forward
 561 #     OP SP+  ($CM+ behaving as $AL) (CL | CP | EX | IS | IY)
 562 #   This really wants to chain at the $CM+ (which is acting as an $AL)
 563 #   except for $CM chaining being disabled.
 564 [$CL $CP $EX $IS $SY] $CM+ $SP+ $CM* $OP;
 565
 566 # LB 14    OP SP* x
 567 #
 568 $CM* $CAN_CM    $SP* $CM* $OP;
 569      $CANT_CM   $SP* $CM* $OP;
 570 $AL_FOLLOW? $CM+  $SP $SP* $CM* $OP;     #  by LB 10, behaves like $AL_FOLLOW? $AL $SP* $CM* $OP
 571
 572      $AL_FOLLOW_NOCM $CM+ $SP+ $CM* $OP;
 573 $CM* $AL_FOLLOW_CM   $CM+ $SP+ $CM* $OP;
 574 $SY $CM $SP+ $OP;   # TODO:  Experiment.  Remove.
 575
 576
 577
 578 # LB 15
 579 $CM* $OP $SP* $CM* $QU;
 580
 581 # LB 16
 582 $CM* $NS $SP* $CM* ($CL | $CP);
 583
 584 # LB 17
 585 $CM* $B2 $SP* $CM* $B2;
 586
 587 # LB 18  break after spaces
 588 #        Nothing explicit needed here.
 589
 590
 591 #
 592 # LB 19
 593 #
 594 $CM* $QU $CM* $CAN_CM;                                #   . x QU
 595 $CM* $QU      $LB18NonBreaks;
 596
 597
 598 $CM* $CAN_CM  $CM* $QU;                               #   QU x .
 599      $CANT_CM $CM* $QU;
 600
 601 #
 602 #  LB 20  Break before and after CB.
 603 #         nothing needed here.
 604 #
 605
 606 # LB 20.09 added rule for Finnish tailoring
 607 $AL ($HY | $HH) / $SP;
 608
 609 # LB 21
 610 $CM* ($BA | $HH | $HY | $NS) $CM* [$LB20NonBreaks-$CM];     #  . x (BA | HY | NS)
 611
 612 $CM* [$LB20NonBreaks-$CM] $CM* $BB;                   #  BB x .
 613 [^$CB] $CM* $BB;                                      #
 614
 615 # LB21a
 616 [^$CB] $CM* ($HY | $BA | $HH) $CM* $HL;
 617
 618 # LB21b (reverse)
 619 $CM* $HL $CM* $SY;
 620
 621 # LB 22
 622 $CM* $IN $CM* ($ALPlus | $HL);
 623 $CM* $IN $CM* $ID;
 624 $CM* $IN $CM* $IN;
 625 $CM* $IN $CM* $NU;
 626
 627 # LB 23
 628 $CM* $PO $CM* $ID;
 629 $CM* $NU $CM* ($ALPlus | $HL);
 630 $CM* ($ALPlus | $HL) $CM* $NU;
 631
 632 # LB 24
 633 $CM* $ID $CM* $PR;
 634 $CM* ($ALPlus | $HL) $CM* $PR;
 635 $CM* ($ALPlus | $HL) $CM* $PO;
 636
 637
 638 # LB 25
 639 ($CM* ($PR | $PO))? ($CM* ($CL | $CP))? ($CM* ($NU | $IS | $SY))* $CM* $NU ($CM* ($OP | $HY))? ($CM* ($PR | $PO))?;
 640
 641 # LB 26
 642 $CM* ($H3 | $H2 | $JV | $JL) $CM* $JL;
 643 $CM* ($JT | $JV) $CM* ($H2 | $JV);
 644 $CM* $JT $CM* ($H3 | $JT);
 645
 646 # LB 27
 647 $CM* $IN $CM* ($H3 | $H2 | $JT | $JV | $JL);
 648 $CM* $PO $CM* ($H3 | $H2 | $JT | $JV | $JL);
 649 $CM* ($H3 | $H2 | $JT | $JV | $JL) $CM* $PR;
 650
 651 # LB 28
 652 $CM* ($ALPlus | $HL) $CM* ($ALPlus | $HL);
 653
 654
 655 # LB 29
 656 $CM* ($ALPlus | $HL) $CM* $IS;
 657
 658 # LB 30
 659 $CM* $OP $CM* ($ALPlus | $HL | $NU);
 660 $CM* ($ALPlus | $HL | $NU) $CM* $CP;
 661
 662 # LB 30a
 663 $CM* $RI $CM* $RI;
 664
 665 # Special reverse rule for people & body part emoji:
 666 # don't break between relevant emoji and $EmojiMods
 667 $EmojiMods $EmojiVar? $EmojiForMods;
 668
 669 ## -------------------------------------------------
 670
 671 !!safe_reverse;
 672
 673 # LB 9
 674 $CM+ [^$CM $BK $CR $LF $NL $ZW $SP];
 675 $CM+ $SP / .;
 676
 677 # LB 14
 678 $SP+ $CM* $OP;
 679
 680 # LB 15
 681 $SP+ $CM* $QU;
 682
 683 # LB 16
 684 $SP+ $CM* ($CL | $CP);
 685
 686 # LB 17
 687 $SP+ $CM* $B2;
 688
 689 # LB 21
 690 $CM* ($HY | $BA | $HH) $CM* $HL;
 691
 692 # LB 25
 693 ($CM* ($IS | $SY))+ $CM* $NU;
 694 ($CL | $CP) $CM* ($NU | $IS | $SY);
 695
 696 # For dictionary-based break
 697 $dictionary $dictionary;
 698
 699 ## -------------------------------------------------
 700
 701 !!safe_forward;
 702
 703 # Skip forward over all character classes that are involved in
 704 #   rules containing patterns with possibly more than one char
 705 #   of context.
 706 #
 707 #  It might be slightly more efficient to have specific rules
 708 #  instead of one generic one, but only if we could
 709 #  turn off rule chaining.  We don't want to move more
 710 #  than necessary.
 711 #
 712 [$CM $OP $QU $CL $CP $B2 $PR $HY $BA $SP $dictionary]+ [^$CM $OP $QU $CL $CP $B2 $PR $HY $BA $dictionary];
 713 $dictionary $dictionary;
 714