Bugzilla – Attachment 172720 Details for
Bug 114160
ZWJ shouldn't be treated as breaking character
Home
|
New
|
Browse
|
Search
|
[?]
|
Reports
|
Help
|
New Account
|
Log In
[x]
|
Forgot Password
Login:
[x]
Latest ICU line break iterator rules
line.txt (text/plain), 10.77 KB, created by
martin_hosken
on 2021-06-09 05:05:42 UTC
(
hide
)
Description:
Latest ICU line break iterator rules
Filename:
MIME Type:
Creator:
martin_hosken
Created:
2021-06-09 05:05:42 UTC
Size:
10.77 KB
patch
obsolete
># Copyright (C) 2016 and later: Unicode, Inc. and others. ># License & terms of use: http://www.unicode.org/copyright.html ># Copyright (c) 2002-2016 International Business Machines Corporation and ># others. All Rights Reserved. ># ># file: line.txt ># ># Line Breaking Rules ># Implement default line breaking as defined by ># Unicode Standard Annex #14 Revision 44 for Unicode 13.0 ># http://www.unicode.org/reports/tr14/, with the following modification: ># ># Boundaries between hyphens and following letters are suppressed when ># there is a boundary preceding the hyphen. See rule 20.9 ># ># This corresponds to CSS line-break=strict (BCP47 -u-lb-strict). ># It sets characters of class CJ to behave like NS. > ># ># Character Classes defined by TR 14. ># > >!!chain; >!!quoted_literals_only; > >$AI = [:LineBreak = Ambiguous:]; >$AL = [:LineBreak = Alphabetic:]; >$BA = [:LineBreak = Break_After:]; >$HH = [\u2010]; # \u2010 is HYPHEN, default line break is BA. >$BB = [:LineBreak = Break_Before:]; >$BK = [:LineBreak = Mandatory_Break:]; >$B2 = [:LineBreak = Break_Both:]; >$CB = [:LineBreak = Contingent_Break:]; >$CJ = [:LineBreak = Conditional_Japanese_Starter:]; >$CL = [:LineBreak = Close_Punctuation:]; ># $CM = [:LineBreak = Combining_Mark:]; >$CP = [:LineBreak = Close_Parenthesis:]; >$CR = [:LineBreak = Carriage_Return:]; >$EB = [:LineBreak = EB:]; >$EM = [:LineBreak = EM:]; >$EX = [:LineBreak = Exclamation:]; >$GL = [:LineBreak = Glue:]; >$HL = [:LineBreak = Hebrew_Letter:]; >$HY = [:LineBreak = Hyphen:]; >$H2 = [:LineBreak = H2:]; >$H3 = [:LineBreak = H3:]; >$ID = [:LineBreak = Ideographic:]; >$IN = [:LineBreak = Inseperable:]; >$IS = [:LineBreak = Infix_Numeric:]; >$JL = [:LineBreak = JL:]; >$JV = [:LineBreak = JV:]; >$JT = [:LineBreak = JT:]; >$LF = [:LineBreak = Line_Feed:]; >$NL = [:LineBreak = Next_Line:]; ># NS includes CJ for CSS strict line breaking. >$NS = [[:LineBreak = Nonstarter:] $CJ]; >$NU = [:LineBreak = Numeric:]; >$OP = [:LineBreak = Open_Punctuation:]; >$PO = [:LineBreak = Postfix_Numeric:]; >$PR = [:LineBreak = Prefix_Numeric:]; >$QU = [:LineBreak = Quotation:]; >$RI = [:LineBreak = Regional_Indicator:]; >$SA = [:LineBreak = Complex_Context:]; >$SG = [:LineBreak = Surrogate:]; >$SP = [:LineBreak = Space:]; >$SY = [:LineBreak = Break_Symbols:]; >$WJ = [:LineBreak = Word_Joiner:]; >$XX = [:LineBreak = Unknown:]; >$ZW = [:LineBreak = ZWSpace:]; >$ZWJ = [:LineBreak = ZWJ:]; > ># OP30 and CP30 are variants of OP and CP that appear in-line in rule LB30 from UAX 14, ># without a formal name. Because ICU rules require multiple uses of the expressions, ># give them a single definition with a name > >$OP30 = [$OP - [\p{ea=F}\p{ea=W}\p{ea=H}]]; >$CP30 = [$CP - [\p{ea=F}\p{ea=W}\p{ea=H}]]; > ># By LB9, a ZWJ also behaves as a CM. Including it in the definition of CM avoids having to explicitly ># list it in the numerous rules that use CM. ># By LB1, SA characters with general categor of Mn or Mc also resolve to CM. > >$CM = [[:LineBreak = Combining_Mark:] $ZWJ [$SA & [[:Mn:][:Mc:]]]]; >$CMX = [[$CM] - [$ZWJ]]; > ># Dictionary character set, for triggering language-based break engines. Currently ># limited to LineBreak=Complex_Context (SA). > >$dictionary = [$SA]; > ># ># Rule LB1. By default, treat AI (characters with ambiguous east Asian width), ># SA (Dictionary chars, excluding Mn and Mc) ># SG (Unpaired Surrogates) ># XX (Unknown, unassigned) ># as $AL (Alphabetic) ># >$ALPlus = [$AL $AI $SG $XX [$SA-[[:Mn:][:Mc:]]]]; > > >## ------------------------------------------------- > ># ># CAN_CM is the set of characters that may combine with CM combining chars. ># Note that Linebreak UAX 14's concept of a combining char and the rules ># for what they can combine with are _very_ different from the rest of Unicode. ># ># Note that $CM itself is left out of this set. If CM is needed as a base ># it must be listed separately in the rule. ># >$CAN_CM = [^$SP $BK $CR $LF $NL $ZW $CM]; # Bases that can take CMs >$CANT_CM = [ $SP $BK $CR $LF $NL $ZW $CM]; # Bases that can't take CMs > ># ># AL_FOLLOW set of chars that can unconditionally follow an AL ># Needed in rules where stand-alone $CM s are treated as AL. ># >$AL_FOLLOW = [$BK $CR $LF $NL $ZW $SP $CL $CP $EX $HL $IS $SY $WJ $GL $OP30 $QU $BA $HY $NS $IN $NU $PR $PO $ALPlus]; > > ># ># Rule LB 4, 5 Mandatory (Hard) breaks. ># >$LB4Breaks = [$BK $CR $LF $NL]; >$LB4NonBreaks = [^$BK $CR $LF $NL $CM]; >$CR $LF {100}; > ># ># LB 6 Do not break before hard line breaks. ># >$LB4NonBreaks? $LB4Breaks {100}; # LB 5 do not break before hard breaks. >$CAN_CM $CM* $LB4Breaks {100}; >^$CM+ $LB4Breaks {100}; > ># LB 7 x SP ># x ZW >$LB4NonBreaks [$SP $ZW]; >$CAN_CM $CM* [$SP $ZW]; >^$CM+ [$SP $ZW]; > ># ># LB 8 Break after zero width space ># ZW SP* ÷ ># >$LB8Breaks = [$LB4Breaks $ZW]; >$LB8NonBreaks = [[$LB4NonBreaks] - [$ZW]]; >$ZW $SP* / [^$SP $ZW $LB4Breaks]; > ># LB 8a ZWJ x Do not break Emoji ZWJ sequences. ># >$ZWJ [^$CM]; > ># LB 9 Combining marks. X $CM needs to behave like X, where X is not $SP, $BK $CR $LF $NL ># $CM not covered by the above needs to behave like $AL ># See definition of $CAN_CM. > >$CAN_CM $CM+; # Stick together any combining sequences that don't match other rules. >^$CM+; > ># ># LB 11 Do not break before or after WORD JOINER & related characters. ># >$CAN_CM $CM* $WJ; >$LB8NonBreaks $WJ; >^$CM+ $WJ; > >$WJ $CM* .; > ># ># LB 12 Do not break after NBSP and related characters. ># GL x ># >$GL $CM* .; > ># ># LB 12a Do not break before NBSP and related characters ... ># [^SP BA HY] x GL ># >[[$LB8NonBreaks] - [$SP $BA $HY]] $CM* $GL; >^$CM+ $GL; > > > > ># LB 13 Don't break before ']' or '!' or or '/', even after spaces. ># >$LB8NonBreaks $CL; >$CAN_CM $CM* $CL; >^$CM+ $CL; # by rule 10, stand-alone CM behaves as AL > >$LB8NonBreaks $CP; >$CAN_CM $CM* $CP; >^$CM+ $CP; # by rule 10, stand-alone CM behaves as AL > >$LB8NonBreaks $EX; >$CAN_CM $CM* $EX; >^$CM+ $EX; # by rule 10, stand-alone CM behaves as AL > >$LB8NonBreaks $SY; >$CAN_CM $CM* $SY; >^$CM+ $SY; # by rule 10, stand-alone CM behaves as AL > > ># ># LB 14 Do not break after OP, even after spaces ># Note subtle interaction with "SP IS /" rules in LB14a. ># This rule consumes the SP, chaining happens on the IS, effectivley overriding the SP IS rules, ># which is the desired behavior. ># >$OP $CM* $SP* .; > >$OP $CM* $SP+ $CM+ $AL_FOLLOW?; # by rule 10, stand-alone CM behaves as AL > # by rule 8, CM following a SP is stand-alone. > > ># LB 14a Force a break before start of a number with a leading decimal pt, e.g. " .23" ># Note: would be simpler to express as "$SP / $IS $CM* $NU;", but ICU rules have limitations. ># See issue ICU-20303 > > >$CanFollowIS = [$BK $CR $LF $NL $SP $ZW $WJ $GL $CL $CP $EX $IS $SY $QU $BA $HY $NS $ALPlus $HL $IN]; >$SP $IS / [^ $CanFollowIS $NU $CM]; >$SP $IS $CM* $CMX / [^ $CanFollowIS $NU $CM]; > ># ># LB 14b Do not break before numeric separators (IS), even after spaces. > >[$LB8NonBreaks - $SP] $IS; >$SP $IS $CM* [$CanFollowIS {eof}]; >$SP $IS $CM* $ZWJ [^$CM $NU]; > >$CAN_CM $CM* $IS; >^$CM+ $IS; # by rule 10, stand-alone CM behaves as AL > > ># LB 15 >$QU $CM* $SP* $OP; > ># LB 16 >($CL | $CP) $CM* $SP* $NS; > ># LB 17 >$B2 $CM* $SP* $B2; > ># ># LB 18 Break after spaces. ># >$LB18NonBreaks = [$LB8NonBreaks - [$SP]]; >$LB18Breaks = [$LB8Breaks $SP]; > > ># LB 19 ># x QU >$LB18NonBreaks $CM* $QU; >^$CM+ $QU; > ># QU x >$QU $CM* .; > ># LB 20 ># <break> $CB ># $CB <break> ># >$LB20NonBreaks = [$LB18NonBreaks - $CB]; > ># LB 20.09 Don't break between Hyphens and Letters when there is a break preceding the hyphen. ># Originally added as a Finnish tailoring, now promoted to default ICU behavior. ># Note: this is not default UAX-14 behaviour. See issue ICU-8151. ># >^($HY | $HH) $CM* $ALPlus; > ># LB 21 x (BA | HY | NS) ># BB x ># >$LB20NonBreaks $CM* ($BA | $HY | $NS); > > >^$CM+ ($BA | $HY | $NS); > >$BB $CM* [^$CB]; # $BB x >$BB $CM* $LB20NonBreaks; > ># LB 21a Don't break after Hebrew + Hyphen ># HL (HY | BA) x ># >$HL $CM* ($HY | $BA) $CM* [^$CB]?; > ># LB 21b (forward) Don't break between SY and HL ># (break between HL and SY already disallowed by LB 13 above) >$SY $CM* $HL; > ># LB 22 Do not break before ellipses ># >$LB20NonBreaks $CM* $IN; >^$CM+ $IN; > > ># LB 23 ># >($ALPlus | $HL) $CM* $NU; >^$CM+ $NU; # Rule 10, any otherwise unattached CM behaves as AL >$NU $CM* ($ALPlus | $HL); > ># LB 23a ># >$PR $CM* ($ID | $EB | $EM); >($ID | $EB | $EM) $CM* $PO; > > ># ># LB 24 ># >($PR | $PO) $CM* ($ALPlus | $HL); >($ALPlus | $HL) $CM* ($PR | $PO); >^$CM+ ($PR | $PO); # Rule 10, any otherwise unattached CM behaves as AL > ># ># LB 25 Numbers. ># >(($PR | $PO) $CM*)? (($OP | $HY) $CM*)? ($IS $CM*)? $NU ($CM* ($NU | $SY | $IS))* > ($CM* ($CL | $CP))? ($CM* ($PR | $PO))?; > ># LB 26 Do not break a Korean syllable ># >$JL $CM* ($JL | $JV | $H2 | $H3); >($JV | $H2) $CM* ($JV | $JT); >($JT | $H3) $CM* $JT; > ># LB 27 Treat korean Syllable Block the same as ID (don't break it) >($JL | $JV | $JT | $H2 | $H3) $CM* $IN; >($JL | $JV | $JT | $H2 | $H3) $CM* $PO; >$PR $CM* ($JL | $JV | $JT | $H2 | $H3); > > ># LB 28 Do not break between alphabetics ># >($ALPlus | $HL) $CM* ($ALPlus | $HL); >^$CM+ ($ALPlus | $HL); # The $CM+ is from rule 10, an unattached CM is treated as AL > ># LB 29 >$IS $CM* ($ALPlus | $HL); > ># LB 30 >($ALPlus | $HL | $NU) $CM* $OP30; >^$CM+ $OP30; # The $CM+ is from rule 10, an unattached CM is treated as AL. >$CP30 $CM* ($ALPlus | $HL | $NU); > ># LB 30a Do not break between regional indicators. Break after pairs of them. ># Tricky interaction with LB8a: ZWJ x . together with ZWJ acting like a CM. >$RI $CM* $RI / [[^$BK $CR $LF $NL $SP $ZW $WJ $CL $CP $EX $IS $SY $GL $QU $BA $HY $NS $IN $CM]]; >$RI $CM* $RI $CM* [$CM-$ZWJ] / [[^$BK $CR $LF $NL $SP $ZW $WJ $CL $CP $EX $IS $SY $GL $QU $BA $HY $NS $IN $CM]]; >$RI $CM* $RI $CM* [$BK $CR $LF $NL $SP $ZW $WJ $CL $CP $EX $IS $SY $GL $QU $BA $HY $NS $IN $ZWJ {eof}]; ># note: the preceding rule includes {eof} rather than having the last [set] term qualified with '?' ># because of the chain-out behavior difference. The rule must chain out only from the [set characters], ># not from the preceding $RI or $CM, which it would be able to do if the set were optional. > ># LB 30b Do not break between an Emoji Base and an Emoji Modifier >$EB $CM* $EM; > ># LB 31 Break everywhere else. ># Match a single code point if no other rule applies. >.;
<b>You cannot view the attachment while viewing its details because your browser does not support IFRAMEs. <a href="attachment.cgi?id=172720">View the attachment on a separate page</a>.</b>
View Attachment As Raw
Actions:
View
Attachments on
bug 114160
:
138103
|
138104
|
141753
| 172720