icuSources/data/brkitr/line_loose_fi.txt

   1 # Copyright (c) 2002-2015  International Business Machines Corporation and
   2 # others. All Rights Reserved.
   3 #
   4 #  file:  line_loose_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=loose (BCP47 -u-lb-loose) as defined for languages other than
  18 #         Chinese & Japanese.
  19 #         It sets characters of class CJ to behave like ID.
  20 #         In addition, it allows breaks before 3005, 303B, 309D, 309E, 30FD, 30FE (all NS).
  21
  22 #
  23 #  Character Classes defined by TR 14.
  24 #
  25
  26 !!chain;
  27 !!LBCMNoChain;
  28
  29
  30 !!lookAheadHardBreak;
  31 #
  32 #  !!lookAheadHardBreak    Described here because it is (as yet) undocumented elsewhere
  33 #                          and only used for the line break rules.
  34 #
  35 #           It is used in the implementation of rule LB 10
  36 #           which says to treat any combining mark that is not attached to a base
  37 #           character as if it were of class AL  (alphabetic).
  38 #
  39 #           The problem occurs in the reverse rules.
  40 #
  41 #           Consider a sequence like, with correct breaks as shown
  42 #               LF  ID  CM  AL  AL
  43 #                  ^       ^       ^
  44 #           Then consider the sequence without the initial ID (ideographic)
  45 #                 LF  CM  AL  AL
  46 #                    ^           ^
  47 #           Our CM, which in the first example was attached to the ideograph,
  48 #           is now unattached, becomes an alpha, and joins in with the other
  49 #           alphas.
  50 #
  51 #           When iterating forwards, these sequences do not present any problems
  52 #           When iterating backwards, we need to look ahead when encountering
  53 #           a CM to see whether it attaches to something further on or not.
  54 #           (Look-ahead in a reverse rule is looking towards the start)
  55 #
  56 #           If the CM is unattached, we need to force a break.
  57 #
  58 #           !!lookAheadHardBreak forces the run time state machine to
  59 #           stop immediately when a look ahead rule ( '/' operator) matches,
  60 #           and set the match position to that of the look-ahead operator,
  61 #           no matter what other rules may be in play at the time.
  62 #
  63 #           See rule LB 19 for an example.
  64 #
  65
  66 $AI = [:LineBreak =  Ambiguous:];
  67 $AL = [:LineBreak =  Alphabetic:];
  68 $BA = [[:LineBreak =  Break_After:] - [\u2010]];
  69 $HH = [\u2010];
  70 $BB = [:LineBreak =  Break_Before:];
  71 $BK = [:LineBreak =  Mandatory_Break:];
  72 $B2 = [:LineBreak =  Break_Both:];
  73 $CB = [:LineBreak =  Contingent_Break:];
  74 $CJ = [:LineBreak =  Conditional_Japanese_Starter:];
  75 $CL = [:LineBreak =  Close_Punctuation:];
  76 $CM = [:LineBreak =  Combining_Mark:];
  77 $CP = [:LineBreak =  Close_Parenthesis:];
  78 $CR = [:LineBreak =  Carriage_Return:];
  79 $EX = [:LineBreak =  Exclamation:];
  80 $GL = [:LineBreak =  Glue:];
  81 $HL = [:LineBreak =  Hebrew_Letter:];
  82 $HY = [:LineBreak =  Hyphen:];
  83 $H2 = [:LineBreak =  H2:];
  84 $H3 = [:LineBreak =  H3:];
  85 $ID = [[:LineBreak =  Ideographic:] $CJ];
  86 $IN = [:LineBreak =  Inseperable:];
  87 $IS = [:LineBreak =  Infix_Numeric:];
  88 $JL = [:LineBreak =  JL:];
  89 $JV = [:LineBreak =  JV:];
  90 $JT = [:LineBreak =  JT:];
  91 $LF = [:LineBreak =  Line_Feed:];
  92 $NL = [:LineBreak =  Next_Line:];
  93 $NSX = [\u3005 \u303B \u309D \u309E \u30FD \u30FE];
  94 $NS = [[:LineBreak =  Nonstarter:] - $NSX];
  95 $NU = [:LineBreak =  Numeric:];
  96 $OP = [:LineBreak =  Open_Punctuation:];
  97 $PO = [:LineBreak =  Postfix_Numeric:];
  98 $PR = [:LineBreak =  Prefix_Numeric:];
  99 $QU = [:LineBreak =  Quotation:];
 100 $RI = [:LineBreak =  Regional_Indicator:];
 101 $SA = [:LineBreak =  Complex_Context:];
 102 $SG = [:LineBreak =  Surrogate:];
 103 $SP = [:LineBreak =  Space:];
 104 $SY = [:LineBreak =  Break_Symbols:];
 105 $WJ = [:LineBreak =  Word_Joiner:];
 106 $XX = [:LineBreak =  Unknown:];
 107 $ZW = [:LineBreak =  ZWSpace:];
 108
 109 # Special character classes for people & body part emoji:
 110 # Subsets of $CM:
 111 $ZWJ = \u200D;
 112 $EmojiVar = \uFE0F;
 113 # The following are subsets of $ID
 114 $EmojiForSeqs = [\u2764 \U0001F466-\U0001F469 \U0001F48B];
 115 $EmojiForMods = [\u261D \u270A-\u270C \U0001F385 \U0001F3C3-\U0001F3C4 \U0001F3C7 \U0001F3CA \U0001F442-\U0001F443 \U0001F446-\U0001F450 \U0001F466-\U0001F469 \U0001F46E-\U0001F478 \U0001F47C \U0001F481-\U0001F483 \U0001F485-\U0001F487 \U0001F4AA \U0001F596 \U0001F645-\U0001F647 \U0001F64B-\U0001F64F \U0001F6A3 \U0001F6B4-\U0001F6B6 \U0001F6C0];
 116 $EmojiMods = [\U0001F3FB-\U0001F3FF];
 117
 118 #   Dictionary character set, for triggering language-based break engines. Currently
 119 #   limited to LineBreak=Complex_Context. Note that this set only works in Unicode
 120 #   5.0 or later as the definition of Complex_Context was corrected to include all
 121 #   characters requiring dictionary break.
 122
 123 $dictionary = [:LineBreak = Complex_Context:];
 124
 125 #
 126 #  Rule LB1.  By default, treat AI  (characters with ambiguous east Asian width),
 127 #                               SA  (South East Asian: Thai, Lao, Khmer)
 128 #                               SG  (Unpaired Surrogates)
 129 #                               XX  (Unknown, unassigned)
 130 #                         as $AL  (Alphabetic)
 131 #
 132 $ALPlus = [$AL $AI $SA $SG $XX];
 133
 134 #
 135 #  Combining Marks.   X $CM*  behaves as if it were X.  Rule LB6.
 136 #
 137 $ALcm = $ALPlus $CM*;
 138 $BAcm = $BA $CM*;
 139 $HHcm = $HH $CM*;
 140 $BBcm = $BB $CM*;
 141 $B2cm = $B2 $CM*;
 142 $CLcm = $CL $CM*;
 143 $CPcm = $CP $CM*;
 144 $EXcm = $EX $CM*;
 145 $GLcm = $GL $CM*;
 146 $HLcm = $HL $CM*;
 147 $HYcm = $HY $CM*;
 148 $H2cm = $H2 $CM*;
 149 $H3cm = $H3 $CM*;
 150 $IDcm = $ID $CM*;
 151 $INcm = $IN $CM*;
 152 $IScm = $IS $CM*;
 153 $JLcm = $JL $CM*;
 154 $JVcm = $JV $CM*;
 155 $JTcm = $JT $CM*;
 156 $NScm = $NS $CM*;
 157 $NSXcm = $NSX $CM*;
 158 $NUcm = $NU $CM*;
 159 $OPcm = $OP $CM*;
 160 $POcm = $PO $CM*;
 161 $PRcm = $PR $CM*;
 162 $QUcm = $QU $CM*;
 163 $RIcm = $RI $CM*;
 164 $SYcm = $SY $CM*;
 165 $WJcm = $WJ $CM*;
 166
 167 ## -------------------------------------------------
 168
 169 !!forward;
 170
 171 #
 172 #  Each class of character can stand by itself as an unbroken token, with trailing combining stuff
 173 #
 174 $ALPlus $CM+;
 175 $BA $CM+;
 176 $HH $CM+;
 177 $BB $CM+;
 178 $B2 $CM+;
 179 $CL $CM+;
 180 $CP $CM+;
 181 $EX $CM+;
 182 $GL $CM+;
 183 $HL $CM+;
 184 $HY $CM+;
 185 $H2 $CM+;
 186 $H3 $CM+;
 187 $ID $CM+;
 188 $IN $CM+;
 189 $IS $CM+;
 190 $JL $CM+;
 191 $JV $CM+;
 192 $JT $CM+;
 193 $NS $CM+;
 194 $NSX $CM+;
 195 $NU $CM+;
 196 $OP $CM+;
 197 $PO $CM+;
 198 $PR $CM+;
 199 $QU $CM+;
 200 $RI $CM+;
 201 $SY $CM+;
 202 $WJ $CM+;
 203
 204 #
 205 # CAN_CM  is the set of characters that may combine with CM combining chars.
 206 #         Note that Linebreak UAX 14's concept of a combining char and the rules
 207 #         for what they can combine with are _very_ different from the rest of Unicode.
 208 #
 209 #         Note that $CM itself is left out of this set.  If CM is needed as a base
 210 #         it must be listed separately in the rule.
 211 #
 212 $CAN_CM  = [^$SP $BK $CR $LF $NL $ZW $CM];       # Bases that can   take CMs
 213 $CANT_CM = [ $SP $BK $CR $LF $NL $ZW $CM];       # Bases that can't take CMs
 214
 215 #
 216 # AL_FOLLOW  set of chars that can unconditionally follow an AL
 217 #            Needed in rules where stand-alone $CM s are treated as AL.
 218 #            Chaining is disabled with CM because it causes other failures,
 219 #            so for this one case we need to manually list out longer sequences.
 220 #
 221 $AL_FOLLOW_NOCM = [$BK $CR $LF $NL $ZW $SP];
 222 $AL_FOLLOW_CM   = [$CL $CP $EX $HL $IS $SY $WJ $GL $OP $QU $BA $HH $HY $NS $NSX $IN $NU $ALPlus];
 223 $AL_FOLLOW      = [$AL_FOLLOW_NOCM $AL_FOLLOW_CM];
 224
 225
 226 #
 227 #  Rule LB 4, 5    Mandatory (Hard) breaks.
 228 #
 229 $LB4Breaks    = [$BK $CR $LF $NL];
 230 $LB4NonBreaks = [^$BK $CR $LF $NL];
 231 $CR $LF {100};
 232
 233 #
 234 #  LB 6    Do not break before hard line breaks.
 235 #
 236 $LB4NonBreaks?  $LB4Breaks {100};    # LB 5  do not break before hard breaks.
 237 $CAN_CM $CM*    $LB4Breaks {100};
 238 $CM+            $LB4Breaks {100};
 239
 240 # LB 7         x SP
 241 #              x ZW
 242 $LB4NonBreaks [$SP $ZW];
 243 $CAN_CM $CM*  [$SP $ZW];
 244 $CM+          [$SP $ZW];
 245
 246 #
 247 # LB 8         Break after zero width space
 248 #              TODO:  ZW SP* <break>
 249 #              An engine change is required to write the reverse rule for this.
 250 #              For now, leave the Unicode 5.2 rule, ZW <break>
 251 #
 252 $LB8Breaks    = [$LB4Breaks $ZW];
 253 $LB8NonBreaks = [[$LB4NonBreaks] - [$ZW]];
 254
 255
 256 # Special forward rule for people & body part emoji:
 257 # don't break $ZWJ from subsequent $EmojiForSeqs
 258 $EmojiForSeqs $EmojiVar? $EmojiMods? $ZWJ $EmojiForSeqs;
 259
 260 # LB 9     Combining marks.      X   $CM needs to behave like X, where X is not $SP, $BK $CR $LF $NL
 261 #                                $CM not covered by the above needs to behave like $AL
 262 #                                See definition of $CAN_CM.
 263
 264 $CAN_CM $CM+;                   #  Stick together any combining sequences that don't match other rules.
 265 $CM+;
 266
 267 #
 268 # LB 11  Do not break before or after WORD JOINER & related characters.
 269 #
 270 $CAN_CM $CM*  $WJcm;
 271 $LB8NonBreaks $WJcm;
 272 $CM+          $WJcm;
 273
 274 $WJcm $CANT_CM;
 275 $WJcm $CAN_CM $CM*;
 276
 277 #
 278 # LB 12  Do not break after NBSP and related characters.
 279 #         GL  x
 280 #
 281 $GLcm $CAN_CM $CM*;
 282 $GLcm $CANT_CM;
 283
 284 #
 285 # LB 12a  Do not break before NBSP and related characters ...
 286 #            [^SP BA HY] x GL
 287 #
 288 [[$LB8NonBreaks] - [$SP $BA $HH $HY]] $CM* $GLcm;
 289 $CM+ GLcm;
 290
 291
 292
 293 #
 294 # LB 13   Don't break before ']' or '!' or ';' or '/', even after spaces.
 295 #
 296 $LB8NonBreaks $CL;
 297 $CAN_CM $CM*  $CL;
 298 $CM+          $CL;              # by rule 10, stand-alone CM behaves as AL
 299
 300 $LB8NonBreaks $CP;
 301 $CAN_CM $CM*  $CP;
 302 $CM+          $CP;              # by rule 10, stand-alone CM behaves as AL
 303
 304 $LB8NonBreaks $EX;
 305 $CAN_CM $CM*  $EX;
 306 $CM+          $EX;              # by rule 10, stand-alone CM behaves as AL
 307
 308 $LB8NonBreaks $IS;
 309 $CAN_CM $CM*  $IS;
 310 $CM+          $IS;              # by rule 10, stand-alone CM behaves as AL
 311
 312 $LB8NonBreaks $SY;
 313 $CAN_CM $CM*  $SY;
 314 $CM+          $SY;              # by rule 10, stand-alone CM behaves as AL
 315
 316
 317 #
 318 # LB 14  Do not break after OP, even after spaces
 319 #
 320 $OPcm $SP* $CAN_CM $CM*;
 321 $OPcm $SP* $CANT_CM;
 322
 323 $OPcm $SP+ $CM+ $AL_FOLLOW?;    # by rule 10, stand-alone CM behaves as AL
 324
 325 # LB 15
 326 $QUcm $SP* $OPcm;
 327
 328 # LB 16
 329 # Do not break between closing punctuation and $NS, even with intervening spaces
 330 # But DO allow a break between closing punctuation and $NSX, don't include it here
 331 ($CLcm | $CPcm) $SP* $NScm;
 332
 333 # LB 17
 334 $B2cm $SP* $B2cm;
 335
 336 #
 337 # LB 18  Break after spaces.
 338 #
 339 $LB18NonBreaks = [$LB8NonBreaks - [$SP]];
 340 $LB18Breaks    = [$LB8Breaks $SP];
 341
 342
 343 # LB 19
 344 #         x QU
 345 $LB18NonBreaks $CM* $QUcm;
 346 $CM+                $QUcm;
 347
 348 #         QU  x
 349 $QUcm .?;
 350 $QUcm $LB18NonBreaks $CM*;    # Don't let a combining mark go onto $CR, $BK, etc.
 351                               #  TODO:  I don't think this rule is needed.
 352
 353
 354 # LB 20
 355 #        <break>  $CB
 356 #        $CB   <break>
 357
 358 $LB20NonBreaks = [$LB18NonBreaks - $CB];
 359
 360 # LB 20.09 added rule for Finnish tailoring
 361 # LB 21        x   (BA | HY | NS)
 362 #           BB x
 363 #
 364 # DO allow breaks here before NSXcm, so don't include it
 365 $LB20NonBreaks $CM* ($BAcm | $HHcm | $HYcm | $NScm) / $AL;
 366 $LB20NonBreaks $CM* ($BAcm | $HHcm | $HYcm | $NScm);
 367 ($HY | $HH) $AL;
 368
 369 $BBcm [^$CB];                                  #  $BB  x
 370 $BBcm $LB20NonBreaks $CM*;
 371
 372 # LB 21a Don't break after Hebrew + Hyphen
 373 #   HL (HY | BA) x
 374 #
 375 $HLcm ($HYcm | $BAcm | $HHcm) [^$CB]?;
 376
 377 # LB 21b (forward) Don't break between SY and HL
 378 # (break between HL and SY already disallowed by LB 13 above)
 379 $SYcm $HLcm;
 380
 381 # LB 22
 382 ($ALcm | $HLcm) $INcm;
 383 $CM+     $INcm;     #  by rule 10, any otherwise unattached CM behaves as AL
 384 $IDcm    $INcm;
 385 $INcm    $INcm;
 386 $NUcm    $INcm;
 387
 388
 389 # $LB 23
 390 $IDcm  $POcm;
 391 $ALcm  $NUcm;       # includes $LB19
 392 $HLcm  $NUcm;
 393 $CM+   $NUcm;       # Rule 10, any otherwise unattached CM behaves as AL
 394 $NUcm  $ALcm;
 395 $NUcm  $HLcm;
 396
 397 #
 398 # LB 24
 399 #
 400 $PRcm $IDcm;
 401 $PRcm ($ALcm | $HLcm);
 402 $POcm ($ALcm | $HLcm);
 403
 404 #
 405 # LB 25   Numbers.
 406 #
 407 ($PRcm | $POcm)? ($OPcm | $HYcm)? $NUcm ($NUcm | $SYcm | $IScm)* ($CLcm | $CPcm)? ($PRcm | $POcm)?;
 408
 409 # LB 26  Do not break a Korean syllable
 410 #
 411 $JLcm ($JLcm | $JVcm | $H2cm | $H3cm);
 412 ($JVcm | $H2cm) ($JVcm | $JTcm);
 413 ($JTcm | $H3cm) $JTcm;
 414
 415 # LB 27  Treat korean Syllable Block the same as ID  (don't break it)
 416 ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm) $INcm;
 417 ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm) $POcm;
 418 $PRcm ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm);
 419
 420
 421 # LB 28   Do not break between alphabetics
 422 #
 423 ($ALcm | $HLcm) ($ALcm | $HLcm);
 424 $CM+ ($ALcm | $HLcm);      # The $CM+ is from rule 10, an unattached CM is treated as AL
 425
 426 # LB 29
 427 $IScm ($ALcm | $HLcm);
 428
 429 # LB 30
 430 ($ALcm | $HLcm | $NUcm) $OPcm;
 431 $CM+ $OPcm;         # The $CM+ is from rule 10, an unattached CM is treated as AL.
 432 $CPcm ($ALcm | $HLcm | $NUcm);
 433
 434 # LB 30a  Do not break between regional indicators.
 435 $RIcm $RIcm;
 436
 437 # Special forward rule for people & body part emoji:
 438 # don't break between relevant emoji and $EmojiMods
 439 $EmojiForMods $EmojiVar? $EmojiMods;
 440
 441 #
 442 #  Reverse Rules.
 443 #
 444 ## -------------------------------------------------
 445
 446 !!reverse;
 447
 448 $CM+ $ALPlus;
 449 $CM+ $BA;
 450 $CM+ $HH;
 451 $CM+ $BB;
 452 $CM+ $B2;
 453 $CM+ $CL;
 454 $CM+ $CP;
 455 $CM+ $EX;
 456 $CM+ $GL;
 457 $CM+ $HL;
 458 $CM+ $HY;
 459 $CM+ $H2;
 460 $CM+ $H3;
 461 $CM+ $ID;
 462 $CM+ $IN;
 463 $CM+ $IS;
 464 $CM+ $JL;
 465 $CM+ $JV;
 466 $CM+ $JT;
 467 $CM+ $NS;
 468 $CM+ $NSX;
 469 $CM+ $NU;
 470 $CM+ $OP;
 471 $CM+ $PO;
 472 $CM+ $PR;
 473 $CM+ $QU;
 474 $CM+ $RI;
 475 $CM+ $SY;
 476 $CM+ $WJ;
 477 $CM+;
 478
 479
 480 #
 481 #  Sequences of the form  (shown forwards)
 482 #      [CANT_CM]  <break>  [CM]  [whatever]
 483 #  The CM needs to behave as an AL
 484 #
 485 $AL_FOLLOW $CM+ / (
 486           [$BK $CR $LF $NL $ZW {eof}] |
 487           $SP+ $CM+ $SP |
 488           $SP+ $CM* ([^$OP $CM $SP] | [$AL {eof}]));   # if LB 14 will match, need to surpress this break.
 489                                                #  LB14 says    OP SP* x .
 490                                                #    becomes    OP SP* x AL
 491                                                #    becomes    OP SP* x CM+ AL_FOLLOW
 492                                                #
 493                                                # Further note:  the $AL in [$AL {eof}] is only to work around
 494                                                #                a rule compiler bug which complains about
 495                                                #                empty sets otherwise.
 496
 497 #
 498 #  Sequences of the form  (shown forwards)
 499 #      [CANT_CM]  <break> [CM]  <break>  [PR]
 500 #  The CM needs to behave as an AL
 501 #  This rule is concerned about getting the second of the two <breaks> in place.
 502 #
 503
 504 [$PR   ] / $CM+ [$BK $CR $LF $NL $ZW $SP {eof}];
 505
 506
 507
 508 # LB 4, 5, 5
 509
 510 $LB4Breaks [$LB4NonBreaks-$CM];
 511 $LB4Breaks $CM+ $CAN_CM;
 512 $LF $CR;
 513
 514
 515 # LB 7         x SP
 516 #              x ZW
 517 [$SP $ZW] [$LB4NonBreaks-$CM];
 518 [$SP $ZW] $CM+ $CAN_CM;
 519
 520 # LB 8 ZW SP* <break>
 521 #     TODO: to implement this, we need more than one look-ahead hard break in play at a time.
 522 #           Requires an engine enhancement.
 523 #   / $SP* $ZW
 524
 525 # Special reverse rule for people & body part emoji:
 526 # don't break $ZWJ from subsequent $EmojiForSeqs
 527 $EmojiForSeqs $ZWJ $EmojiMods? $EmojiVar? $EmojiForSeqs;
 528
 529 # LB 9,10  Combining marks.
 530 #    X   $CM needs to behave like X, where X is not $SP or controls.
 531 #    $CM not covered by the above needs to behave like $AL
 532 # Stick together any combining sequences that don't match other rules.
 533 $CM+ $CAN_CM;
 534
 535
 536 # LB 11
 537 $CM* $WJ $CM* $CAN_CM;
 538 $CM* $WJ      [$LB8NonBreaks-$CM];
 539
 540      $CANT_CM $CM* $WJ;
 541 $CM* $CAN_CM  $CM* $WJ;
 542
 543 # LB 12a
 544 #      [^SP BA HY] x GL
 545 #
 546 $CM* $GL $CM* [$LB8NonBreaks-[$CM $SP $BA $HH $HY]];
 547
 548 # LB 12
 549 #     GL  x
 550 #
 551 $CANT_CM $CM* $GL;
 552 $CM* $CAN_CM $CM* $GL;
 553
 554
 555 # LB 13
 556 $CL $CM+ $CAN_CM;
 557 $CP $CM+ $CAN_CM;
 558 $EX $CM+ $CAN_CM;
 559 $IS $CM+ $CAN_CM;
 560 $SY $CM+ $CAN_CM;
 561
 562 $CL [$LB8NonBreaks-$CM];
 563 $CP [$LB8NonBreaks-$CM];
 564 $EX [$LB8NonBreaks-$CM];
 565 $IS [$LB8NonBreaks-$CM];
 566 $SY [$LB8NonBreaks-$CM];
 567
 568 # Rule 13 & 14 taken together for an edge case.
 569 #   Match this, shown forward
 570 #     OP SP+  ($CM+ behaving as $AL) (CL | CP | EX | IS | IY)
 571 #   This really wants to chain at the $CM+ (which is acting as an $AL)
 572 #   except for $CM chaining being disabled.
 573 [$CL $CP $EX $IS $SY] $CM+ $SP+ $CM* $OP;
 574
 575 # LB 14    OP SP* x
 576 #
 577 $CM* $CAN_CM    $SP* $CM* $OP;
 578      $CANT_CM   $SP* $CM* $OP;
 579 $AL_FOLLOW? $CM+  $SP $SP* $CM* $OP;     #  by LB 10, behaves like $AL_FOLLOW? $AL $SP* $CM* $OP
 580
 581      $AL_FOLLOW_NOCM $CM+ $SP+ $CM* $OP;
 582 $CM* $AL_FOLLOW_CM   $CM+ $SP+ $CM* $OP;
 583 $SY $CM $SP+ $OP;   # TODO:  Experiment.  Remove.
 584
 585
 586
 587 # LB 15
 588 $CM* $OP $SP* $CM* $QU;
 589
 590 # LB 16
 591 # Don't include $NSX here
 592 $CM* $NS $SP* $CM* ($CL | $CP);
 593
 594 # LB 17
 595 $CM* $B2 $SP* $CM* $B2;
 596
 597 # LB 18  break after spaces
 598 #        Nothing explicit needed here.
 599
 600
 601 #
 602 # LB 19
 603 #
 604 $CM* $QU $CM* $CAN_CM;                                #   . x QU
 605 $CM* $QU      $LB18NonBreaks;
 606
 607
 608 $CM* $CAN_CM  $CM* $QU;                               #   QU x .
 609      $CANT_CM $CM* $QU;
 610
 611 #
 612 #  LB 20  Break before and after CB.
 613 #         nothing needed here.
 614 #
 615
 616 # LB 20.09 added rule for Finnish tailoring
 617 $AL ($HY | $HH) / $SP;
 618
 619 # LB 21
 620 # Don't include $NSX here
 621 $CM* ($BA | $HH | $HY | $NS) $CM* [$LB20NonBreaks-$CM];     #  . x (BA | HY | NS)
 622
 623 $CM* [$LB20NonBreaks-$CM] $CM* $BB;                   #  BB x .
 624 [^$CB] $CM* $BB;                                      #
 625
 626 # LB21a
 627 [^$CB] $CM* ($HY | $BA | $HH) $CM* $HL;
 628
 629 # LB21b (reverse)
 630 $CM* $HL $CM* $SY;
 631
 632 # LB 22
 633 $CM* $IN $CM* ($ALPlus | $HL);
 634 $CM* $IN $CM* $ID;
 635 $CM* $IN $CM* $IN;
 636 $CM* $IN $CM* $NU;
 637
 638 # LB 23
 639 $CM* $PO $CM* $ID;
 640 $CM* $NU $CM* ($ALPlus | $HL);
 641 $CM* ($ALPlus | $HL) $CM* $NU;
 642
 643 # LB 24
 644 $CM* $ID $CM* $PR;
 645 $CM* ($ALPlus | $HL) $CM* $PR;
 646 $CM* ($ALPlus | $HL) $CM* $PO;
 647
 648
 649 # LB 25
 650 ($CM* ($PR | $PO))? ($CM* ($CL | $CP))? ($CM* ($NU | $IS | $SY))* $CM* $NU ($CM* ($OP | $HY))? ($CM* ($PR | $PO))?;
 651
 652 # LB 26
 653 $CM* ($H3 | $H2 | $JV | $JL) $CM* $JL;
 654 $CM* ($JT | $JV) $CM* ($H2 | $JV);
 655 $CM* $JT $CM* ($H3 | $JT);
 656
 657 # LB 27
 658 $CM* $IN $CM* ($H3 | $H2 | $JT | $JV | $JL);
 659 $CM* $PO $CM* ($H3 | $H2 | $JT | $JV | $JL);
 660 $CM* ($H3 | $H2 | $JT | $JV | $JL) $CM* $PR;
 661
 662 # LB 28
 663 $CM* ($ALPlus | $HL) $CM* ($ALPlus | $HL);
 664
 665
 666 # LB 29
 667 $CM* ($ALPlus | $HL) $CM* $IS;
 668
 669 # LB 30
 670 $CM* $OP $CM* ($ALPlus | $HL | $NU);
 671 $CM* ($ALPlus | $HL | $NU) $CM* $CP;
 672
 673 # LB 30a
 674 $CM* $RI $CM* $RI;
 675
 676 # Special reverse rule for people & body part emoji:
 677 # don't break between relevant emoji and $EmojiMods
 678 $EmojiMods $EmojiVar? $EmojiForMods;
 679
 680 ## -------------------------------------------------
 681
 682 !!safe_reverse;
 683
 684 # LB 9
 685 $CM+ [^$CM $BK $CR $LF $NL $ZW $SP];
 686 $CM+ $SP / .;
 687
 688 # LB 14
 689 $SP+ $CM* $OP;
 690
 691 # LB 15
 692 $SP+ $CM* $QU;
 693
 694 # LB 16
 695 $SP+ $CM* ($CL | $CP);
 696
 697 # LB 17
 698 $SP+ $CM* $B2;
 699
 700 # LB 21
 701 $CM* ($HY | $BA | $HH) $CM* $HL;
 702
 703 # LB 25
 704 ($CM* ($IS | $SY))+ $CM* $NU;
 705 ($CL | $CP) $CM* ($NU | $IS | $SY);
 706
 707 # For dictionary-based break
 708 $dictionary $dictionary;
 709
 710 ## -------------------------------------------------
 711
 712 !!safe_forward;
 713
 714 # Skip forward over all character classes that are involved in
 715 #   rules containing patterns with possibly more than one char
 716 #   of context.
 717 #
 718 #  It might be slightly more efficient to have specific rules
 719 #  instead of one generic one, but only if we could
 720 #  turn off rule chaining.  We don't want to move more
 721 #  than necessary.
 722 #
 723 [$CM $OP $QU $CL $CP $B2 $PR $HY $BA $SP $dictionary]+ [^$CM $OP $QU $CL $CP $B2 $PR $HY $BA $dictionary];
 724 $dictionary $dictionary;
 725