• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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 34 for Unicode 8.0
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#   Dictionary character set, for triggering language-based break engines. Currently
107#   limited to LineBreak=Complex_Context. Note that this set only works in Unicode
108#   5.0 or later as the definition of Complex_Context was corrected to include all
109#   characters requiring dictionary break.
110
111$dictionary = [:LineBreak = Complex_Context:];
112
113#
114#  Rule LB1.  By default, treat AI  (characters with ambiguous east Asian width),
115#                               SA  (South East Asian: Thai, Lao, Khmer)
116#                               SG  (Unpaired Surrogates)
117#                               XX  (Unknown, unassigned)
118#                         as $AL  (Alphabetic)
119#
120$ALPlus = [$AL $AI $SA $SG $XX];
121
122#
123#  Combining Marks.   X $CM*  behaves as if it were X.  Rule LB6.
124#
125$ALcm = $ALPlus $CM*;
126$BAcm = $BA $CM*;
127$HHcm = $HH $CM*;
128$BBcm = $BB $CM*;
129$B2cm = $B2 $CM*;
130$CLcm = $CL $CM*;
131$CPcm = $CP $CM*;
132$EXcm = $EX $CM*;
133$GLcm = $GL $CM*;
134$HLcm = $HL $CM*;
135$HYcm = $HY $CM*;
136$H2cm = $H2 $CM*;
137$H3cm = $H3 $CM*;
138$IDcm = $ID $CM*;
139$INcm = $IN $CM*;
140$IScm = $IS $CM*;
141$JLcm = $JL $CM*;
142$JVcm = $JV $CM*;
143$JTcm = $JT $CM*;
144$NScm = $NS $CM*;
145$NUcm = $NU $CM*;
146$OPcm = $OP $CM*;
147$POcm = $PO $CM*;
148$PRcm = $PR $CM*;
149$QUcm = $QU $CM*;
150$RIcm = $RI $CM*;
151$SYcm = $SY $CM*;
152$WJcm = $WJ $CM*;
153
154## -------------------------------------------------
155
156!!forward;
157
158#
159#  Each class of character can stand by itself as an unbroken token, with trailing combining stuff
160#
161$ALPlus $CM+;
162$BA $CM+;
163$HH $CM+;
164$BB $CM+;
165$B2 $CM+;
166$CL $CM+;
167$CP $CM+;
168$EX $CM+;
169$GL $CM+;
170$HL $CM+;
171$HY $CM+;
172$H2 $CM+;
173$H3 $CM+;
174$ID $CM+;
175$IN $CM+;
176$IS $CM+;
177$JL $CM+;
178$JV $CM+;
179$JT $CM+;
180$NS $CM+;
181$NU $CM+;
182$OP $CM+;
183$PO $CM+;
184$PR $CM+;
185$QU $CM+;
186$RI $CM+;
187$SY $CM+;
188$WJ $CM+;
189
190#
191# CAN_CM  is the set of characters that may combine with CM combining chars.
192#         Note that Linebreak UAX 14's concept of a combining char and the rules
193#         for what they can combine with are _very_ different from the rest of Unicode.
194#
195#         Note that $CM itself is left out of this set.  If CM is needed as a base
196#         it must be listed separately in the rule.
197#
198$CAN_CM  = [^$SP $BK $CR $LF $NL $ZW $CM];       # Bases that can   take CMs
199$CANT_CM = [ $SP $BK $CR $LF $NL $ZW $CM];       # Bases that can't take CMs
200
201#
202# AL_FOLLOW  set of chars that can unconditionally follow an AL
203#            Needed in rules where stand-alone $CM s are treated as AL.
204#            Chaining is disabled with CM because it causes other failures,
205#            so for this one case we need to manually list out longer sequences.
206#
207$AL_FOLLOW_NOCM = [$BK $CR $LF $NL $ZW $SP];
208$AL_FOLLOW_CM   = [$CL $CP $EX $HL $IS $SY $WJ $GL $OP $QU $BA $HH $HY $NS $IN $NU $ALPlus];
209$AL_FOLLOW      = [$AL_FOLLOW_NOCM $AL_FOLLOW_CM];
210
211
212#
213#  Rule LB 4, 5    Mandatory (Hard) breaks.
214#
215$LB4Breaks    = [$BK $CR $LF $NL];
216$LB4NonBreaks = [^$BK $CR $LF $NL];
217$CR $LF {100};
218
219#
220#  LB 6    Do not break before hard line breaks.
221#
222$LB4NonBreaks?  $LB4Breaks {100};    # LB 5  do not break before hard breaks.
223$CAN_CM $CM*    $LB4Breaks {100};
224$CM+            $LB4Breaks {100};
225
226# LB 7         x SP
227#              x ZW
228$LB4NonBreaks [$SP $ZW];
229$CAN_CM $CM*  [$SP $ZW];
230$CM+          [$SP $ZW];
231
232#
233# LB 8         Break after zero width space
234#              TODO:  ZW SP* <break>
235#              An engine change is required to write the reverse rule for this.
236#              For now, leave the Unicode 5.2 rule, ZW <break>
237#
238$LB8Breaks    = [$LB4Breaks $ZW];
239$LB8NonBreaks = [[$LB4NonBreaks] - [$ZW]];
240
241
242# LB 9     Combining marks.      X   $CM needs to behave like X, where X is not $SP, $BK $CR $LF $NL
243#                                $CM not covered by the above needs to behave like $AL
244#                                See definition of $CAN_CM.
245
246$CAN_CM $CM+;                   #  Stick together any combining sequences that don't match other rules.
247$CM+;
248
249#
250# LB 11  Do not break before or after WORD JOINER & related characters.
251#
252$CAN_CM $CM*  $WJcm;
253$LB8NonBreaks $WJcm;
254$CM+          $WJcm;
255
256$WJcm $CANT_CM;
257$WJcm $CAN_CM $CM*;
258
259#
260# LB 12  Do not break after NBSP and related characters.
261#         GL  x
262#
263$GLcm $CAN_CM $CM*;
264$GLcm $CANT_CM;
265
266#
267# LB 12a  Do not break before NBSP and related characters ...
268#            [^SP BA HY] x GL
269#
270[[$LB8NonBreaks] - [$SP $BA $HH $HY]] $CM* $GLcm;
271$CM+ GLcm;
272
273
274
275#
276# LB 13   Don't break before ']' or '!' or ';' or '/', even after spaces.
277#
278$LB8NonBreaks $CL;
279$CAN_CM $CM*  $CL;
280$CM+          $CL;              # by rule 10, stand-alone CM behaves as AL
281
282$LB8NonBreaks $CP;
283$CAN_CM $CM*  $CP;
284$CM+          $CP;              # by rule 10, stand-alone CM behaves as AL
285
286$LB8NonBreaks $EX;
287$CAN_CM $CM*  $EX;
288$CM+          $EX;              # by rule 10, stand-alone CM behaves as AL
289
290$LB8NonBreaks $IS;
291$CAN_CM $CM*  $IS;
292$CM+          $IS;              # by rule 10, stand-alone CM behaves as AL
293
294$LB8NonBreaks $SY;
295$CAN_CM $CM*  $SY;
296$CM+          $SY;              # by rule 10, stand-alone CM behaves as AL
297
298
299#
300# LB 14  Do not break after OP, even after spaces
301#
302$OPcm $SP* $CAN_CM $CM*;
303$OPcm $SP* $CANT_CM;
304
305$OPcm $SP+ $CM+ $AL_FOLLOW?;    # by rule 10, stand-alone CM behaves as AL
306
307# LB 15
308$QUcm $SP* $OPcm;
309
310# LB 16
311($CLcm | $CPcm) $SP* $NScm;
312
313# LB 17
314$B2cm $SP* $B2cm;
315
316#
317# LB 18  Break after spaces.
318#
319$LB18NonBreaks = [$LB8NonBreaks - [$SP]];
320$LB18Breaks    = [$LB8Breaks $SP];
321
322
323# LB 19
324#         x QU
325$LB18NonBreaks $CM* $QUcm;
326$CM+                $QUcm;
327
328#         QU  x
329$QUcm .?;
330$QUcm $LB18NonBreaks $CM*;    # Don't let a combining mark go onto $CR, $BK, etc.
331                              #  TODO:  I don't think this rule is needed.
332
333
334# LB 20
335#        <break>  $CB
336#        $CB   <break>
337
338$LB20NonBreaks = [$LB18NonBreaks - $CB];
339
340# LB 20.09 added rule for Finnish tailoring
341# LB 21        x   (BA | HY | NS)
342#           BB x
343#
344$LB20NonBreaks $CM* ($BAcm | $HHcm | $HYcm | $NScm) / $AL;
345$LB20NonBreaks $CM* ($BAcm | $HHcm | $HYcm | $NScm);
346($HY | $HH) $AL;
347
348$BBcm [^$CB];                                  #  $BB  x
349$BBcm $LB20NonBreaks $CM*;
350
351# LB 21a Don't break after Hebrew + Hyphen
352#   HL (HY | BA) x
353#
354$HLcm ($HYcm | $BAcm | $HHcm) [^$CB]?;
355
356# LB 21b (forward) Don't break between SY and HL
357# (break between HL and SY already disallowed by LB 13 above)
358$SYcm $HLcm;
359
360# LB 22
361($ALcm | $HLcm) $INcm;
362$CM+     $INcm;     #  by rule 10, any otherwise unattached CM behaves as AL
363$EXcm    $INcm;
364$IDcm    $INcm;
365$INcm    $INcm;
366$NUcm    $INcm;
367
368
369# $LB 23
370$IDcm  $POcm;
371$ALcm  $NUcm;       # includes $LB19
372$HLcm  $NUcm;
373$CM+   $NUcm;       # Rule 10, any otherwise unattached CM behaves as AL
374$NUcm  $ALcm;
375$NUcm  $HLcm;
376
377#
378# LB 24
379#
380$PRcm $IDcm;
381$PRcm ($ALcm | $HLcm);
382$POcm ($ALcm | $HLcm);
383
384#
385# LB 25   Numbers.
386#
387($PRcm | $POcm)? ($OPcm | $HYcm)? $NUcm ($NUcm | $SYcm | $IScm)* ($CLcm | $CPcm)? ($PRcm | $POcm)?;
388
389# LB 26  Do not break a Korean syllable
390#
391$JLcm ($JLcm | $JVcm | $H2cm | $H3cm);
392($JVcm | $H2cm) ($JVcm | $JTcm);
393($JTcm | $H3cm) $JTcm;
394
395# LB 27  Treat korean Syllable Block the same as ID  (don't break it)
396($JLcm | $JVcm | $JTcm | $H2cm | $H3cm) $INcm;
397($JLcm | $JVcm | $JTcm | $H2cm | $H3cm) $POcm;
398$PRcm ($JLcm | $JVcm | $JTcm | $H2cm | $H3cm);
399
400
401# LB 28   Do not break between alphabetics
402#
403($ALcm | $HLcm) ($ALcm | $HLcm);
404$CM+ ($ALcm | $HLcm);      # The $CM+ is from rule 10, an unattached CM is treated as AL
405
406# LB 29
407$IScm ($ALcm | $HLcm);
408
409# LB 30
410($ALcm | $HLcm | $NUcm) $OPcm;
411$CM+ $OPcm;         # The $CM+ is from rule 10, an unattached CM is treated as AL.
412$CPcm ($ALcm | $HLcm | $NUcm);
413
414# LB 30a  Do not break between regional indicators.
415$RIcm $RIcm;
416
417#
418#  Reverse Rules.
419#
420## -------------------------------------------------
421
422!!reverse;
423
424$CM+ $ALPlus;
425$CM+ $BA;
426$CM+ $HH;
427$CM+ $BB;
428$CM+ $B2;
429$CM+ $CL;
430$CM+ $CP;
431$CM+ $EX;
432$CM+ $GL;
433$CM+ $HL;
434$CM+ $HY;
435$CM+ $H2;
436$CM+ $H3;
437$CM+ $ID;
438$CM+ $IN;
439$CM+ $IS;
440$CM+ $JL;
441$CM+ $JV;
442$CM+ $JT;
443$CM+ $NS;
444$CM+ $NU;
445$CM+ $OP;
446$CM+ $PO;
447$CM+ $PR;
448$CM+ $QU;
449$CM+ $RI;
450$CM+ $SY;
451$CM+ $WJ;
452$CM+;
453
454
455#
456#  Sequences of the form  (shown forwards)
457#      [CANT_CM]  <break>  [CM]  [whatever]
458#  The CM needs to behave as an AL
459#
460$AL_FOLLOW $CM+ / (
461          [$BK $CR $LF $NL $ZW {eof}] |
462          $SP+ $CM+ $SP |
463          $SP+ $CM* ([^$OP $CM $SP] | [$AL {eof}]));   # if LB 14 will match, need to surpress this break.
464                                               #  LB14 says    OP SP* x .
465                                               #    becomes    OP SP* x AL
466                                               #    becomes    OP SP* x CM+ AL_FOLLOW
467                                               #
468                                               # Further note:  the $AL in [$AL {eof}] is only to work around
469                                               #                a rule compiler bug which complains about
470                                               #                empty sets otherwise.
471
472#
473#  Sequences of the form  (shown forwards)
474#      [CANT_CM]  <break> [CM]  <break>  [PR]
475#  The CM needs to behave as an AL
476#  This rule is concerned about getting the second of the two <breaks> in place.
477#
478
479[$PR   ] / $CM+ [$BK $CR $LF $NL $ZW $SP {eof}];
480
481
482
483# LB 4, 5, 5
484
485$LB4Breaks [$LB4NonBreaks-$CM];
486$LB4Breaks $CM+ $CAN_CM;
487$LF $CR;
488
489
490# LB 7         x SP
491#              x ZW
492[$SP $ZW] [$LB4NonBreaks-$CM];
493[$SP $ZW] $CM+ $CAN_CM;
494
495# LB 8 ZW SP* <break>
496#     TODO: to implement this, we need more than one look-ahead hard break in play at a time.
497#           Requires an engine enhancement.
498#   / $SP* $ZW
499
500# LB 9,10  Combining marks.
501#    X   $CM needs to behave like X, where X is not $SP or controls.
502#    $CM not covered by the above needs to behave like $AL
503# Stick together any combining sequences that don't match other rules.
504$CM+ $CAN_CM;
505
506
507# LB 11
508$CM* $WJ $CM* $CAN_CM;
509$CM* $WJ      [$LB8NonBreaks-$CM];
510
511     $CANT_CM $CM* $WJ;
512$CM* $CAN_CM  $CM* $WJ;
513
514# LB 12a
515#      [^SP BA HY] x GL
516#
517$CM* $GL $CM* [$LB8NonBreaks-[$CM $SP $BA $HH $HY]];
518
519# LB 12
520#     GL  x
521#
522$CANT_CM $CM* $GL;
523$CM* $CAN_CM $CM* $GL;
524
525
526# LB 13
527$CL $CM+ $CAN_CM;
528$CP $CM+ $CAN_CM;
529$EX $CM+ $CAN_CM;
530$IS $CM+ $CAN_CM;
531$SY $CM+ $CAN_CM;
532
533$CL [$LB8NonBreaks-$CM];
534$CP [$LB8NonBreaks-$CM];
535$EX [$LB8NonBreaks-$CM];
536$IS [$LB8NonBreaks-$CM];
537$SY [$LB8NonBreaks-$CM];
538
539# Rule 13 & 14 taken together for an edge case.
540#   Match this, shown forward
541#     OP SP+  ($CM+ behaving as $AL) (CL | CP | EX | IS | IY)
542#   This really wants to chain at the $CM+ (which is acting as an $AL)
543#   except for $CM chaining being disabled.
544[$CL $CP $EX $IS $SY] $CM+ $SP+ $CM* $OP;
545
546# LB 14    OP SP* x
547#
548$CM* $CAN_CM    $SP* $CM* $OP;
549     $CANT_CM   $SP* $CM* $OP;
550$AL_FOLLOW? $CM+  $SP $SP* $CM* $OP;     #  by LB 10, behaves like $AL_FOLLOW? $AL $SP* $CM* $OP
551
552     $AL_FOLLOW_NOCM $CM+ $SP+ $CM* $OP;
553$CM* $AL_FOLLOW_CM   $CM+ $SP+ $CM* $OP;
554$SY $CM $SP+ $OP;   # TODO:  Experiment.  Remove.
555
556
557
558# LB 15
559$CM* $OP $SP* $CM* $QU;
560
561# LB 16
562$CM* $NS $SP* $CM* ($CL | $CP);
563
564# LB 17
565$CM* $B2 $SP* $CM* $B2;
566
567# LB 18  break after spaces
568#        Nothing explicit needed here.
569
570
571#
572# LB 19
573#
574$CM* $QU $CM* $CAN_CM;                                #   . x QU
575$CM* $QU      $LB18NonBreaks;
576
577
578$CM* $CAN_CM  $CM* $QU;                               #   QU x .
579     $CANT_CM $CM* $QU;
580
581#
582#  LB 20  Break before and after CB.
583#         nothing needed here.
584#
585
586# LB 20.09 added rule for Finnish tailoring
587$AL ($HY | $HH) / $SP;
588
589# LB 21
590$CM* ($BA | $HH | $HY | $NS) $CM* [$LB20NonBreaks-$CM];     #  . x (BA | HY | NS)
591
592$CM* [$LB20NonBreaks-$CM] $CM* $BB;                   #  BB x .
593[^$CB] $CM* $BB;                                      #
594
595# LB21a
596[^$CB] $CM* ($HY | $BA | $HH) $CM* $HL;
597
598# LB21b (reverse)
599$CM* $HL $CM* $SY;
600
601# LB 22
602$CM* $IN $CM* ($ALPlus | $HL);
603$CM* $IN $CM* $EX;
604$CM* $IN $CM* $ID;
605$CM* $IN $CM* $IN;
606$CM* $IN $CM* $NU;
607
608# LB 23
609$CM* $PO $CM* $ID;
610$CM* $NU $CM* ($ALPlus | $HL);
611$CM* ($ALPlus | $HL) $CM* $NU;
612
613# LB 24
614$CM* $ID $CM* $PR;
615$CM* ($ALPlus | $HL) $CM* $PR;
616$CM* ($ALPlus | $HL) $CM* $PO;
617
618
619# LB 25
620($CM* ($PR | $PO))? ($CM* ($CL | $CP))? ($CM* ($NU | $IS | $SY))* $CM* $NU ($CM* ($OP | $HY))? ($CM* ($PR | $PO))?;
621
622# LB 26
623$CM* ($H3 | $H2 | $JV | $JL) $CM* $JL;
624$CM* ($JT | $JV) $CM* ($H2 | $JV);
625$CM* $JT $CM* ($H3 | $JT);
626
627# LB 27
628$CM* $IN $CM* ($H3 | $H2 | $JT | $JV | $JL);
629$CM* $PO $CM* ($H3 | $H2 | $JT | $JV | $JL);
630$CM* ($H3 | $H2 | $JT | $JV | $JL) $CM* $PR;
631
632# LB 28
633$CM* ($ALPlus | $HL) $CM* ($ALPlus | $HL);
634
635
636# LB 29
637$CM* ($ALPlus | $HL) $CM* $IS;
638
639# LB 30
640$CM* $OP $CM* ($ALPlus | $HL | $NU);
641$CM* ($ALPlus | $HL | $NU) $CM* $CP;
642
643# LB 30a
644$CM* $RI $CM* $RI;
645
646## -------------------------------------------------
647
648!!safe_reverse;
649
650# LB 9
651$CM+ [^$CM $BK $CR $LF $NL $ZW $SP];
652$CM+ $SP / .;
653
654# LB 14
655$SP+ $CM* $OP;
656
657# LB 15
658$SP+ $CM* $QU;
659
660# LB 16
661$SP+ $CM* ($CL | $CP);
662
663# LB 17
664$SP+ $CM* $B2;
665
666# LB 21
667$CM* ($HY | $BA | $HH) $CM* $HL;
668
669# LB 25
670($CM* ($IS | $SY))+ $CM* $NU;
671($CL | $CP) $CM* ($NU | $IS | $SY);
672
673# For dictionary-based break
674$dictionary $dictionary;
675
676## -------------------------------------------------
677
678!!safe_forward;
679
680# Skip forward over all character classes that are involved in
681#   rules containing patterns with possibly more than one char
682#   of context.
683#
684#  It might be slightly more efficient to have specific rules
685#  instead of one generic one, but only if we could
686#  turn off rule chaining.  We don't want to move more
687#  than necessary.
688#
689[$CM $OP $QU $CL $CP $B2 $PR $HY $BA $SP $dictionary]+ [^$CM $OP $QU $CL $CP $B2 $PR $HY $BA $dictionary];
690$dictionary $dictionary;
691
692