icuSources/data/brkitr/line_normal_fi.txt

   1 # Copyright (c) 2002-2015  International Business Machines Corporation and
   2 # others. All Rights Reserved.
   3 #
   4 #  file:  line_normal_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 both for Finnish and to correpond 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.
  20
  21 #
  22 #  Character Classes defined by TR 14.
  23 #
  24
  25 !!chain;
  26 !!LBCMNoChain;
  27
  28
  29 !!lookAheadHardBreak;
  30 #
  31 #  !!lookAheadHardBreak    Described here because it is (as yet) undocumented elsewhere
  32 #                          and only used for the line break rules.
  33 #
  34 #           It is used in the implementation of rule LB 10
  35 #           which says to treat any combining mark that is not attached to a base
  36 #           character as if it were of class AL  (alphabetic).
  37 #
  38 #           The problem occurs in the reverse rules.
  39 #
  40 #           Consider a sequence like, with correct breaks as shown
  41 #               LF  ID  CM  AL  AL
  42 #                  ^       ^       ^
  43 #           Then consider the sequence without the initial ID (ideographic)
  44 #                 LF  CM  AL  AL
  45 #                    ^           ^
  46 #           Our CM, which in the first example was attached to the ideograph,
  47 #           is now unattached, becomes an alpha, and joins in with the other
  48 #           alphas.
  49 #
  50 #           When iterating forwards, these sequences do not present any problems
  51 #           When iterating backwards, we need to look ahead when encountering
  52 #           a CM to see whether it attaches to something further on or not.
  53 #           (Look-ahead in a reverse rule is looking towards the start)
  54 #
  55 #           If the CM is unattached, we need to force a break.
  56 #
  57 #           !!lookAheadHardBreak forces the run time state machine to
  58 #           stop immediately when a look ahead rule ( '/' operator) matches,
  59 #           and set the match position to that of the look-ahead operator,
  60 #           no matter what other rules may be in play at the time.
  61 #
  62 #           See rule LB 19 for an example.
  63 #
  64
  65 $AI = [:LineBreak =  Ambiguous:];
  66 $AL = [:LineBreak =  Alphabetic:];
  67 $BA = [[:LineBreak =  Break_After:] - [\u2010]];
  68 $HH = [\u2010];
  69 $BB = [:LineBreak =  Break_Before:];
  70 $BK = [:LineBreak =  Mandatory_Break:];
  71 $B2 = [:LineBreak =  Break_Both:];
  72 $CB = [:LineBreak =  Contingent_Break:];
  73 $CJ = [:LineBreak =  Conditional_Japanese_Starter:];
  74 $CL = [:LineBreak =  Close_Punctuation:];
  75 $CM = [:LineBreak =  Combining_Mark:];
  76 $CP = [:LineBreak =  Close_Parenthesis:];
  77 $CR = [:LineBreak =  Carriage_Return:];
  78 $EX = [:LineBreak =  Exclamation:];
  79 $GL = [:LineBreak =  Glue:];
  80 $HL = [:LineBreak =  Hebrew_Letter:];
  81 $HY = [:LineBreak =  Hyphen:];
  82 $H2 = [:LineBreak =  H2:];
  83 $H3 = [:LineBreak =  H3:];
  84 $ID = [[:LineBreak =  Ideographic:] $CJ];
  85 $IN = [:LineBreak =  Inseperable:];
  86 $IS = [:LineBreak =  Infix_Numeric:];
  87 $JL = [:LineBreak =  JL:];
  88 $JV = [:LineBreak =  JV:];
  89 $JT = [:LineBreak =  JT:];
  90 $LF = [:LineBreak =  Line_Feed:];
  91 $NL = [:LineBreak =  Next_Line:];
  92 $NS = [:LineBreak =  Nonstarter:];
  93 $NU = [:LineBreak =  Numeric:];
  94 $OP = [:LineBreak =  Open_Punctuation:];
  95 $PO = [:LineBreak =  Postfix_Numeric:];
  96 $PR = [:LineBreak =  Prefix_Numeric:];
  97 $QU = [:LineBreak =  Quotation:];
  98 $RI = [:LineBreak =  Regional_Indicator:];
  99 $SA = [:LineBreak =  Complex_Context:];
 100 $SG = [:LineBreak =  Surrogate:];
 101 $SP = [:LineBreak =  Space:];
 102 $SY = [:LineBreak =  Break_Symbols:];
 103 $WJ = [:LineBreak =  Word_Joiner:];
 104 $XX = [:LineBreak =  Unknown:];
 105 $ZW = [:LineBreak =  ZWSpace:];
 106
 107 # Special character classes for people & body part emoji:
 108 # Subsets of $CM:
 109 $ZWJ = \u200D;
 110 $EmojiVar = \uFE0F;
 111 # The following are subsets of $ID
 112 $EmojiForSeqs = [\u2764 \U0001F441 \U0001F466-\U0001F469 \U0001F48B \U0001F5E8];
 113 $EmojiForMods = [\u261D \u26F9 \u270A-\u270D \U0001F385 \U0001F3C3-\U0001F3C4 \U0001F3C7 \U0001F3CA-\U0001F3CB \U0001F442-\U0001F443 \U0001F446-\U0001F450 \U0001F466-\U0001F469 \U0001F46E-\U0001F478 \U0001F47C \U0001F481-\U0001F483 \U0001F485-\U0001F487 \U0001F4AA \U0001F590 \U0001F595 \U0001F596 \U0001F645-\U0001F647 \U0001F64B-\U0001F64F \U0001F6A3 \U0001F6B4-\U0001F6B6 \U0001F6C0 \U0001F918];
 114 $EmojiMods = [\U0001F3FB-\U0001F3FF];
 115
 116 #   Dictionary character set, for triggering language-based break engines. Currently
 117 #   limited to LineBreak=Complex_Context. Note that this set only works in Unicode
 118 #   5.0 or later as the definition of Complex_Context was corrected to include all
 119 #   characters requiring dictionary break.
 120
 121 $dictionary = [:LineBreak = Complex_Context:];
 122
 123 #
 124 #  Rule LB1.  By default, treat AI  (characters with ambiguous east Asian width),
 125 #                               SA  (South East Asian: Thai, Lao, Khmer)
 126 #                               SG  (Unpaired Surrogates)
 127 #                               XX  (Unknown, unassigned)
 128 #                         as $AL  (Alphabetic)
 129 #
 130 $ALPlus = [$AL $AI $SA $SG $XX];
 131
 132 #
 133 #  Combining Marks.   X $CM*  behaves as if it were X.  Rule LB6.
 134 #
 135 $ALcm = $ALPlus $CM*;
 136 $BAcm = $BA $CM*;
 137 $HHcm = $HH $CM*;
 138 $BBcm = $BB $CM*;
 139 $B2cm = $B2 $CM*;
 140 $CLcm = $CL $CM*;
 141 $CPcm = $CP $CM*;
 142 $EXcm = $EX $CM*;
 143 $GLcm = $GL $CM*;
 144 $HLcm = $HL $CM*;
 145 $HYcm = $HY $CM*;
 146 $H2cm = $H2 $CM*;
 147 $H3cm = $H3 $CM*;
 148 $IDcm = $ID $CM*;
 149 $INcm = $IN $CM*;
 150 $IScm = $IS $CM*;
 151 $JLcm = $JL $CM*;
 152 $JVcm = $JV $CM*;
 153 $JTcm = $JT $CM*;
 154 $NScm = $NS $CM*;
 155 $NUcm = $NU $CM*;
 156 $OPcm = $OP $CM*;
 157 $POcm = $PO $CM*;
 158 $PRcm = $PR $CM*;
 159 $QUcm = $QU $CM*;
 160 $RIcm = $RI $CM*;
 161 $SYcm = $SY $CM*;
 162 $WJcm = $WJ $CM*;
 163
 164 ## -------------------------------------------------
 165
 166 !!forward;
 167
 168 #
 169 #  Each class of character can stand by itself as an unbroken token, with trailing combining stuff
 170 #
 171 $ALPlus $CM+;
 172 $BA $CM+;
 173 $HH $CM+;
 174 $BB $CM+;
 175 $B2 $CM+;
 176 $CL $CM+;
 177 $CP $CM+;
 178 $EX $CM+;
 179 $GL $CM+;
 180 $HL $CM+;
 181 $HY $CM+;
 182 $H2 $CM+;
 183 $H3 $CM+;
 184 $ID $CM+;
 185 $IN $CM+;
 186 $IS $CM+;
 187 $JL $CM+;
 188 $JV $CM+;
 189 $JT $CM+;
 190 $NS $CM+;
 191 $NU $CM+;
 192 $OP $CM+;
 193 $PO $CM+;
 194 $PR $CM+;
 195 $QU $CM+;
 196 $RI $CM+;
 197 $SY $CM+;
 198 $WJ $CM+;
 199
 200 #
 201 # CAN_CM  is the set of characters that may combine with CM combining chars.
 202 #         Note that Linebreak UAX 14's concept of a combining char and the rules
 203 #         for what they can combine with are _very_ different from the rest of Unicode.
 204 #
 205 #         Note that $CM itself is left out of this set.  If CM is needed as a base
 206 #         it must be listed separately in the rule.
 207 #
 208 $CAN_CM  = [^$SP $BK $CR $LF $NL $ZW $CM];       # Bases that can   take CMs
 209 $CANT_CM = [ $SP $BK $CR $LF $NL $ZW $CM];       # Bases that can't take CMs
 210
 211 #
 212 # AL_FOLLOW  set of chars that can unconditionally follow an AL
 213 #            Needed in rules where stand-alone $CM s are treated as AL.
 214 #            Chaining is disabled with CM because it causes other failures,
 215 #            so for this one case we need to manually list out longer sequences.
 216 #
 217 $AL_FOLLOW_NOCM = [$BK $CR $LF $NL $ZW $SP];
 218 $AL_FOLLOW_CM   = [$CL $CP $EX $HL $IS $SY $WJ $GL $OP $QU $BA $HH $HY $NS $IN $NU $ALPlus];
 219 $AL_FOLLOW      = [$AL_FOLLOW_NOCM $AL_FOLLOW_CM];
 220
 221
 222 #
 223 #  Rule LB 4, 5    Mandatory (Hard) breaks.
 224 #
 225 $LB4Breaks    = [$BK $CR $LF $NL];
 226 $LB4NonBreaks = [^$BK $CR $LF $NL];
 227 $CR $LF {100};
 228
 229 #
 230 #  LB 6    Do not break before hard line breaks.
 231 #
 232 $LB4NonBreaks?  $LB4Breaks {100};    # LB 5  do not break before hard breaks.
 233 $CAN_CM $CM*    $LB4Breaks {100};
 234 $CM+            $LB4Breaks {100};
 235
 236 # LB 7         x SP
 237 #              x ZW
 238 $LB4NonBreaks [$SP $ZW];
 239 $CAN_CM $CM*  [$SP $ZW];
 240 $CM+          [$SP $ZW];
 241
 242 #
 243 # LB 8         Break after zero width space
 244 #              TODO:  ZW SP* <break>
 245 #              An engine change is required to write the reverse rule for this.
 246 #              For now, leave the Unicode 5.2 rule, ZW <break>
 247 #
 248 $LB8Breaks    = [$LB4Breaks $ZW];
 249 $LB8NonBreaks = [[$LB4NonBreaks] - [$ZW]];
 250
 251
 252 # Special forward rule for people & body part emoji:
 253 # don't break $ZWJ from subsequent $EmojiForSeqs
 254 $EmojiForSeqs $EmojiVar? $EmojiMods? $ZWJ $EmojiForSeqs;
 255
 256 # LB 9     Combining marks.      X   $CM needs to behave like X, where X is not $SP, $BK $CR $LF $NL
 257 #                                $CM not covered by the above needs to behave like $AL
 258 #                                See definition of $CAN_CM.
 259
 260 $CAN_CM $CM+;                   #  Stick together any combining sequences that don't match other rules.
 261 $CM+;
 262
 263 #
 264 # LB 11  Do not break before or after WORD JOINER & related characters.
 265 #
 266 $CAN_CM $CM*  $WJcm;
 267 $LB8NonBreaks $WJcm;
 268 $CM+          $WJcm;
 269
 270 $WJcm $CANT_CM;
 271 $WJcm $CAN_CM $CM*;
 272
 273 #
 274 # LB 12  Do not break after NBSP and related characters.
 275 #         GL  x
 276 #
 277 $GLcm $CAN_CM $CM*;
 278 $GLcm $CANT_CM;
 279
 280 #
 281 # LB 12a  Do not break before NBSP and related characters ...
 282 #            [^SP BA HY] x GL
 283 #
 284 [[$LB8NonBreaks] - [$SP $BA $HH $HY]] $CM* $GLcm;
 285 $CM+ GLcm;
 286
 287
 288
 289 #
 290 # LB 13   Don't break before ']' or '!' or ';' or '/', even after spaces.
 291 #
 292 $LB8NonBreaks $CL;
 293 $CAN_CM $CM*  $CL;
 294 $CM+          $CL;              # by rule 10, stand-alone CM behaves as AL
 295
 296 $LB8NonBreaks $CP;
 297 $CAN_CM $CM*  $CP;
 298 $CM+          $CP;              # by rule 10, stand-alone CM behaves as AL
 299
 300 $LB8NonBreaks $EX;
 301 $CAN_CM $CM*  $EX;
 302 $CM+          $EX;              # by rule 10, stand-alone CM behaves as AL
 303
 304 $LB8NonBreaks $IS;
 305 $CAN_CM $CM*  $IS;
 306 $CM+          $IS;              # by rule 10, stand-alone CM behaves as AL
 307
 308 $LB8NonBreaks $SY;
 309 $CAN_CM $CM*  $SY;
 310 $CM+          $SY;              # by rule 10, stand-alone CM behaves as AL
 311
 312
 313 #
 314 # LB 14  Do not break after OP, even after spaces
 315 #
 316 $OPcm $SP* $CAN_CM $CM*;
 317 $OPcm $SP* $CANT_CM;
 318
 319 $OPcm $SP+ $CM+ $AL_FOLLOW?;    # by rule 10, stand-alone CM behaves as AL
 320
 321 # LB 15
 322 $QUcm $SP* $OPcm;
 323
 324 # LB 16
 325 ($CLcm | $CPcm) $SP* $NScm;
 326
 327 # LB 17
 328 $B2cm $SP* $B2cm;
 329
 330 #
 331 # LB 18  Break after spaces.
 332 #
 333 $LB18NonBreaks = [$LB8NonBreaks - [$SP]];
 334 $LB18Breaks    = [$LB8Breaks $SP];
 335
 336
 337 # LB 19
 338 #         x QU
 339 $LB18NonBreaks $CM* $QUcm;
 340 $CM+                $QUcm;
 341
 342 #         QU  x
 343 $QUcm .?;
 344 $QUcm $LB18NonBreaks $CM*;    # Don't let a combining mark go onto $CR, $BK, etc.
 345                               #  TODO:  I don't think this rule is needed.
 346
 347
 348 # LB 20
 349 #        <break>  $CB
 350 #        $CB   <break>
 351
 352 $LB20NonBreaks = [$LB18NonBreaks - $CB];
 353
 354 # LB 20.09 added rule for Finnish tailoring
 355 # LB 21        x   (BA | HY | NS)
 356 #           BB x
 357 #
 358 $LB20NonBreaks $CM* ($BAcm | $HHcm | $HYcm | $NScm) / $AL;
 359 $LB20NonBreaks $CM* ($BAcm | $HHcm | $HYcm | $NScm);
 360 ($HY | $HH) $AL;
 361
 362 $BBcm [^$CB];                                  #  $BB  x
 363 $BBcm $LB20NonBreaks $CM*;
 364
 365 # LB 21a Don't break after Hebrew + Hyphen
 366 #   HL (HY | BA) x
 367 #
 368 $HLcm ($HYcm | $BAcm | $HHcm) [^$CB]?;
 369
 370 # LB 21b (forward) Don't break between SY and HL
 371 # (break between HL and SY already disallowed by LB 13 above)
 372 $SYcm $HLcm;
 373
 374 # LB 22
 375 ($ALcm | $HLcm) $INcm;
 376 $CM+     $INcm;     #  by rule 10, any otherwise unattached CM behaves as AL
 377 $IDcm    $INcm;
 378 $INcm    $INcm;
 379 $NUcm    $INcm;
 380
 381
 382 # $LB 23
 383 $IDcm  $POcm;
 384 $ALcm  $NUcm;       # includes $LB19
 385 $HLcm  $NUcm;
 386 $CM+   $NUcm;       # Rule 10, any otherwise unattached CM behaves as AL
 387 $NUcm  $ALcm;
 388 $NUcm  $HLcm;
 389
 390 #
 391 # LB 24
 392 #
 393 $PRcm $IDcm;
 394 $PRcm ($ALcm | $HLcm);
 395 $POcm ($ALcm | $HLcm);
 396
 397 #
 398 # LB 25   Numbers.
 399 #
 400 ($PRcm | $POcm)? ($OPcm | $HYcm)? $NUcm ($NUcm | $SYcm | $IScm)* ($CLcm | $CPcm)? ($PRcm | $POcm)?;
 401
 402 # LB 26  Do not break a Korean syllable
 403 #
 404 $JLcm ($JLcm | $JVcm | $H2cm | $H3cm);
 405 ($JVcm | $H2cm) ($JVcm | $JTcm);
 406 ($JTcm | $H3cm) $JTcm;
 407
 408 # LB 27  Treat korean Syllable Block the same as ID  (don't break it)
 409 ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm) $INcm;
 410 ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm) $POcm;
 411 $PRcm ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm);
 412
 413
 414 # LB 28   Do not break between alphabetics
 415 #
 416 ($ALcm | $HLcm) ($ALcm | $HLcm);
 417 $CM+ ($ALcm | $HLcm);      # The $CM+ is from rule 10, an unattached CM is treated as AL
 418
 419 # LB 29
 420 $IScm ($ALcm | $HLcm);
 421
 422 # LB 30
 423 ($ALcm | $HLcm | $NUcm) $OPcm;
 424 $CM+ $OPcm;         # The $CM+ is from rule 10, an unattached CM is treated as AL.
 425 $CPcm ($ALcm | $HLcm | $NUcm);
 426
 427 # LB 30a  Do not break between regional indicators.
 428 $RIcm $RIcm;
 429
 430 # Special forward rule for people & body part emoji:
 431 # don't break between relevant emoji and $EmojiMods
 432 $EmojiForMods $EmojiVar? $EmojiMods;
 433
 434 #
 435 #  Reverse Rules.
 436 #
 437 ## -------------------------------------------------
 438
 439 !!reverse;
 440
 441 $CM+ $ALPlus;
 442 $CM+ $BA;
 443 $CM+ $HH;
 444 $CM+ $BB;
 445 $CM+ $B2;
 446 $CM+ $CL;
 447 $CM+ $CP;
 448 $CM+ $EX;
 449 $CM+ $GL;
 450 $CM+ $HL;
 451 $CM+ $HY;
 452 $CM+ $H2;
 453 $CM+ $H3;
 454 $CM+ $ID;
 455 $CM+ $IN;
 456 $CM+ $IS;
 457 $CM+ $JL;
 458 $CM+ $JV;
 459 $CM+ $JT;
 460 $CM+ $NS;
 461 $CM+ $NU;
 462 $CM+ $OP;
 463 $CM+ $PO;
 464 $CM+ $PR;
 465 $CM+ $QU;
 466 $CM+ $RI;
 467 $CM+ $SY;
 468 $CM+ $WJ;
 469 $CM+;
 470
 471
 472 #
 473 #  Sequences of the form  (shown forwards)
 474 #      [CANT_CM]  <break>  [CM]  [whatever]
 475 #  The CM needs to behave as an AL
 476 #
 477 $AL_FOLLOW $CM+ / (
 478           [$BK $CR $LF $NL $ZW {eof}] |
 479           $SP+ $CM+ $SP |
 480           $SP+ $CM* ([^$OP $CM $SP] | [$AL {eof}]));   # if LB 14 will match, need to surpress this break.
 481                                                #  LB14 says    OP SP* x .
 482                                                #    becomes    OP SP* x AL
 483                                                #    becomes    OP SP* x CM+ AL_FOLLOW
 484                                                #
 485                                                # Further note:  the $AL in [$AL {eof}] is only to work around
 486                                                #                a rule compiler bug which complains about
 487                                                #                empty sets otherwise.
 488
 489 #
 490 #  Sequences of the form  (shown forwards)
 491 #      [CANT_CM]  <break> [CM]  <break>  [PR]
 492 #  The CM needs to behave as an AL
 493 #  This rule is concerned about getting the second of the two <breaks> in place.
 494 #
 495
 496 [$PR   ] / $CM+ [$BK $CR $LF $NL $ZW $SP {eof}];
 497
 498
 499
 500 # LB 4, 5, 5
 501
 502 $LB4Breaks [$LB4NonBreaks-$CM];
 503 $LB4Breaks $CM+ $CAN_CM;
 504 $LF $CR;
 505
 506
 507 # LB 7         x SP
 508 #              x ZW
 509 [$SP $ZW] [$LB4NonBreaks-$CM];
 510 [$SP $ZW] $CM+ $CAN_CM;
 511
 512 # LB 8 ZW SP* <break>
 513 #     TODO: to implement this, we need more than one look-ahead hard break in play at a time.
 514 #           Requires an engine enhancement.
 515 #   / $SP* $ZW
 516
 517 # Special reverse rule for people & body part emoji:
 518 # don't break $ZWJ from subsequent $EmojiForSeqs
 519 $EmojiForSeqs $ZWJ $EmojiMods? $EmojiVar? $EmojiForSeqs;
 520
 521 # LB 9,10  Combining marks.
 522 #    X   $CM needs to behave like X, where X is not $SP or controls.
 523 #    $CM not covered by the above needs to behave like $AL
 524 # Stick together any combining sequences that don't match other rules.
 525 $CM+ $CAN_CM;
 526
 527
 528 # LB 11
 529 $CM* $WJ $CM* $CAN_CM;
 530 $CM* $WJ      [$LB8NonBreaks-$CM];
 531
 532      $CANT_CM $CM* $WJ;
 533 $CM* $CAN_CM  $CM* $WJ;
 534
 535 # LB 12a
 536 #      [^SP BA HY] x GL
 537 #
 538 $CM* $GL $CM* [$LB8NonBreaks-[$CM $SP $BA $HH $HY]];
 539
 540 # LB 12
 541 #     GL  x
 542 #
 543 $CANT_CM $CM* $GL;
 544 $CM* $CAN_CM $CM* $GL;
 545
 546
 547 # LB 13
 548 $CL $CM+ $CAN_CM;
 549 $CP $CM+ $CAN_CM;
 550 $EX $CM+ $CAN_CM;
 551 $IS $CM+ $CAN_CM;
 552 $SY $CM+ $CAN_CM;
 553
 554 $CL [$LB8NonBreaks-$CM];
 555 $CP [$LB8NonBreaks-$CM];
 556 $EX [$LB8NonBreaks-$CM];
 557 $IS [$LB8NonBreaks-$CM];
 558 $SY [$LB8NonBreaks-$CM];
 559
 560 # Rule 13 & 14 taken together for an edge case.
 561 #   Match this, shown forward
 562 #     OP SP+  ($CM+ behaving as $AL) (CL | CP | EX | IS | IY)
 563 #   This really wants to chain at the $CM+ (which is acting as an $AL)
 564 #   except for $CM chaining being disabled.
 565 [$CL $CP $EX $IS $SY] $CM+ $SP+ $CM* $OP;
 566
 567 # LB 14    OP SP* x
 568 #
 569 $CM* $CAN_CM    $SP* $CM* $OP;
 570      $CANT_CM   $SP* $CM* $OP;
 571 $AL_FOLLOW? $CM+  $SP $SP* $CM* $OP;     #  by LB 10, behaves like $AL_FOLLOW? $AL $SP* $CM* $OP
 572
 573      $AL_FOLLOW_NOCM $CM+ $SP+ $CM* $OP;
 574 $CM* $AL_FOLLOW_CM   $CM+ $SP+ $CM* $OP;
 575 $SY $CM $SP+ $OP;   # TODO:  Experiment.  Remove.
 576
 577
 578
 579 # LB 15
 580 $CM* $OP $SP* $CM* $QU;
 581
 582 # LB 16
 583 $CM* $NS $SP* $CM* ($CL | $CP);
 584
 585 # LB 17
 586 $CM* $B2 $SP* $CM* $B2;
 587
 588 # LB 18  break after spaces
 589 #        Nothing explicit needed here.
 590
 591
 592 #
 593 # LB 19
 594 #
 595 $CM* $QU $CM* $CAN_CM;                                #   . x QU
 596 $CM* $QU      $LB18NonBreaks;
 597
 598
 599 $CM* $CAN_CM  $CM* $QU;                               #   QU x .
 600      $CANT_CM $CM* $QU;
 601
 602 #
 603 #  LB 20  Break before and after CB.
 604 #         nothing needed here.
 605 #
 606
 607 # LB 20.09 added rule for Finnish tailoring
 608 $AL ($HY | $HH) / $SP;
 609
 610 # LB 21
 611 $CM* ($BA | $HH | $HY | $NS) $CM* [$LB20NonBreaks-$CM];     #  . x (BA | HY | NS)
 612
 613 $CM* [$LB20NonBreaks-$CM] $CM* $BB;                   #  BB x .
 614 [^$CB] $CM* $BB;                                      #
 615
 616 # LB21a
 617 [^$CB] $CM* ($HY | $BA | $HH) $CM* $HL;
 618
 619 # LB21b (reverse)
 620 $CM* $HL $CM* $SY;
 621
 622 # LB 22
 623 $CM* $IN $CM* ($ALPlus | $HL);
 624 $CM* $IN $CM* $ID;
 625 $CM* $IN $CM* $IN;
 626 $CM* $IN $CM* $NU;
 627
 628 # LB 23
 629 $CM* $PO $CM* $ID;
 630 $CM* $NU $CM* ($ALPlus | $HL);
 631 $CM* ($ALPlus | $HL) $CM* $NU;
 632
 633 # LB 24
 634 $CM* $ID $CM* $PR;
 635 $CM* ($ALPlus | $HL) $CM* $PR;
 636 $CM* ($ALPlus | $HL) $CM* $PO;
 637
 638
 639 # LB 25
 640 ($CM* ($PR | $PO))? ($CM* ($CL | $CP))? ($CM* ($NU | $IS | $SY))* $CM* $NU ($CM* ($OP | $HY))? ($CM* ($PR | $PO))?;
 641
 642 # LB 26
 643 $CM* ($H3 | $H2 | $JV | $JL) $CM* $JL;
 644 $CM* ($JT | $JV) $CM* ($H2 | $JV);
 645 $CM* $JT $CM* ($H3 | $JT);
 646
 647 # LB 27
 648 $CM* $IN $CM* ($H3 | $H2 | $JT | $JV | $JL);
 649 $CM* $PO $CM* ($H3 | $H2 | $JT | $JV | $JL);
 650 $CM* ($H3 | $H2 | $JT | $JV | $JL) $CM* $PR;
 651
 652 # LB 28
 653 $CM* ($ALPlus | $HL) $CM* ($ALPlus | $HL);
 654
 655
 656 # LB 29
 657 $CM* ($ALPlus | $HL) $CM* $IS;
 658
 659 # LB 30
 660 $CM* $OP $CM* ($ALPlus | $HL | $NU);
 661 $CM* ($ALPlus | $HL | $NU) $CM* $CP;
 662
 663 # LB 30a
 664 $CM* $RI $CM* $RI;
 665
 666 # Special reverse rule for people & body part emoji:
 667 # don't break between relevant emoji and $EmojiMods
 668 $EmojiMods $EmojiVar? $EmojiForMods;
 669
 670 ## -------------------------------------------------
 671
 672 !!safe_reverse;
 673
 674 # LB 9
 675 $CM+ [^$CM $BK $CR $LF $NL $ZW $SP];
 676 $CM+ $SP / .;
 677
 678 # LB 14
 679 $SP+ $CM* $OP;
 680
 681 # LB 15
 682 $SP+ $CM* $QU;
 683
 684 # LB 16
 685 $SP+ $CM* ($CL | $CP);
 686
 687 # LB 17
 688 $SP+ $CM* $B2;
 689
 690 # LB 21
 691 $CM* ($HY | $BA | $HH) $CM* $HL;
 692
 693 # LB 25
 694 ($CM* ($IS | $SY))+ $CM* $NU;
 695 ($CL | $CP) $CM* ($NU | $IS | $SY);
 696
 697 # For dictionary-based break
 698 $dictionary $dictionary;
 699
 700 ## -------------------------------------------------
 701
 702 !!safe_forward;
 703
 704 # Skip forward over all character classes that are involved in
 705 #   rules containing patterns with possibly more than one char
 706 #   of context.
 707 #
 708 #  It might be slightly more efficient to have specific rules
 709 #  instead of one generic one, but only if we could
 710 #  turn off rule chaining.  We don't want to move more
 711 #  than necessary.
 712 #
 713 [$CM $OP $QU $CL $CP $B2 $PR $HY $BA $SP $dictionary]+ [^$CM $OP $QU $CL $CP $B2 $PR $HY $BA $dictionary];
 714 $dictionary $dictionary;
 715