xref: /external/icu/icu4c/source/data/brkitr/rules/line.txt

# 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 37 for Unicode 9.0
#         http://www.unicode.org/reports/tr14/
#
#         Includes extensions to the handling of emoji ZWJ sequences from
#         https://goo.gl/cluFCn
#
#         TODO:  Rule LB 8 remains as it was in Unicode 5.2
#         This is only because of a limitation of ICU break engine implementation,
#         not because the older behavior is desirable.
#
#         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;

$AI = [:LineBreak =  Ambiguous:];
$AL = [:LineBreak =  Alphabetic:];
$BA = [:LineBreak =  Break_After:];
$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:]\U0001F3C2\U0001F3C7\U0001F3CC\U0001F46A-\U0001F46D\U0001F46F\U0001F574\U0001F6CC];
$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:];

$EmojiNRK = [[\p{Emoji}] - [$RI \u002a\u00230-9©®™〰〽]];
# Data for Extended Pictographic scraped from CLDR common/properties/ExtendedPictographic.txt, r12773
$Extended_Pict = [\U0001F774-\U0001F77F\u2700-\u2701\u2703-\u2704\u270E\u2710-\u2711\u2765-\u2767\U0001F030-\U0001F093\U0001F094-\U0001F09F\U0001F10D-\U0001F10F\U0001F12F\U0001F16C-\U0001F16F\U0001F1AD-\U0001F1E5\U0001F203-\U0001F20F\U0001F23C-\U0001F23F\U0001F249-\U0001F24F\U0001F252-\U0001F2FF\U0001F7D5-\U0001F7FF\U0001F000-\U0001F003\U0001F005-\U0001F02B\U0001F02C-\U0001F02F\U0001F322-\U0001F323\U0001F394-\U0001F395\U0001F398\U0001F39C-\U0001F39D\U0001F3F1-\U0001F3F2\U0001F3F6\U0001F4FE\U0001F53E-\U0001F548\U0001F54F\U0001F568-\U0001F56E\U0001F571-\U0001F572\U0001F57B-\U0001F586\U0001F588-\U0001F589\U0001F58E-\U0001F58F\U0001F591-\U0001F594\U0001F597-\U0001F5A3\U0001F5A6-\U0001F5A7\U0001F5A9-\U0001F5B0\U0001F5B3-\U0001F5BB\U0001F5BD-\U0001F5C1\U0001F5C5-\U0001F5D0\U0001F5D4-\U0001F5DB\U0001F5DF-\U0001F5E0\U0001F5E2\U0001F5E4-\U0001F5E7\U0001F5E9-\U0001F5EE\U0001F5F0-\U0001F5F2\U0001F5F4-\U0001F5F9\u2605\u2607-\u260D\u260F-\u2610\u2612\u2616-\u2617\u2619-\u261C\u261E-\u261F\u2621\u2624-\u2625\u2627-\u2629\u262B-\u262D\u2630-\u2637\u263B-\u2647\u2654-\u265F\u2661-\u2662\u2664\u2667\u2669-\u267A\u267C-\u267E\u2680-\u2691\u2695\u2698\u269A\u269D-\u269F\u26A2-\u26A9\u26AC-\u26AF\u26B2-\u26BC\u26BF-\u26C3\u26C6-\u26C7\u26C9-\u26CD\u26D0\u26D2\u26D5-\u26E8\u26EB-\u26EF\u26F6\u26FB-\u26FC\u26FE-\u26FF\u2388\U0001FA00-\U0001FFFD\U0001F0A0-\U0001F0AE\U0001F0B1-\U0001F0BF\U0001F0C1-\U0001F0CF\U0001F0D1-\U0001F0F5\U0001F0AF-\U0001F0B0\U0001F0C0\U0001F0D0\U0001F0F6-\U0001F0FF\U0001F80C-\U0001F80F\U0001F848-\U0001F84F\U0001F85A-\U0001F85F\U0001F888-\U0001F88F\U0001F8AE-\U0001F8FF\U0001F900-\U0001F90F\U0001F91F\U0001F928-\U0001F92F\U0001F931-\U0001F932\U0001F93F\U0001F94C-\U0001F94F\U0001F95F-\U0001F97F\U0001F992-\U0001F9BF\U0001F9C1-\U0001F9FF\U0001F6C6-\U0001F6CA\U0001F6E6-\U0001F6E8\U0001F6EA\U0001F6F1-\U0001F6F2\U0001F6D3-\U0001F6DF\U0001F6ED-\U0001F6EF\U0001F6F7-\U0001F6FF];

# 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:]]]];

#   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:]]]];


## -------------------------------------------------

!!forward;

#
# 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 $OP $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
#              TODO:  ZW SP* <break>
#              An engine change is required to write the reverse rule for this.
#              For now, leave the Unicode 5.2 rule, ZW <break>
#
$LB8Breaks    = [$LB4Breaks $ZW];
$LB8NonBreaks = [[$LB4NonBreaks] - [$ZW]];

# LB 8a        ZWJ x (ID | Extended_Pict | EmojiNRK)   Emoji ZWJ sequences.
#
$ZWJ ($ID | $Extended_Pict | $EmojiNRK);

# 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 $IS;
$CAN_CM $CM*  $IS;
^$CM+         $IS;              # 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
#
$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 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 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
($ALPlus | $HL) $CM* $IN;
^$CM+    $IN;     #  by rule 10, any otherwise unattached CM behaves as AL
$EX $CM*    $IN;
($ID | $EB | $EM) $CM*  $IN;
$IN $CM*    $IN;
$NU $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*)? $NU ($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* $OP;
^$CM+ $OP;         # The $CM+ is from rule 10, an unattached CM is treated as AL.
$CP $CM* ($ALPlus | $HL | $NU);

# LB 30a  Do not break between regional indicators. Break after pairs of them.
#         Tricky interaction with LB8a: ZWJ x ID
$RI $CM* $RI                / [[^$BK $CR $LF $NL $SP $ZW $WJ $CL $CP $EX $IS $SY $GL $QU $BA $HY $NS $CM] {eof}];
$RI $CM* $RI $CM*  $ZWJ     / [[^$BK $CR $LF $NL $SP $ZW $WJ $CL $CP $EX $IS $SY $GL $QU $BA $HY $NS $CM $ID $Extended_Pict $EmojiNRK] {eof}];
$RI $CM* $RI $CM* [$CM-$ZWJ] / [[^$BK $CR $LF $NL $SP $ZW $WJ $CL $CP $EX $IS $SY $GL $QU $BA $HY $NS $CM] {eof}];

$RI $CM* $RI $CM* [$BK $CR $LF $NL $SP $ZW $WJ $CL $CP $EX $IS $SY $GL $QU $BA $HY $NS {eof}];
$RI $CM* $RI $CM* $ZWJ ($ID | $Extended_Pict | $EmojiNRK);

# LB 30b Do not break between an Emoji Base and an Emoji Modifier
$EB $CM* $EM;

#
#  Reverse Rules.
#
## -------------------------------------------------

!!reverse;

#  LB 9 Combining Marks.
#  Stick together any combining sequences that don't match other rules.

^$CM+ $CAN_CM?;

#
#  Sequences of the form  (shown forwards)
#      [CANT_CM]  <break>  [CM]  [whatever]
#  The CM needs to behave as an AL
#
$AL_FOLLOW $CM+ / (
          [$BK $CR $LF $NL $ZW {eof}] |
          $SP+ $CM+ $SP |
          $SP+ $CM* ([^$OP $CM $SP] | [$AL {eof}]));   # if LB 14 will match, need to surpress this break.
                                               #  LB14 says    OP SP* x .
                                               #    becomes    OP SP* x AL
                                               #    becomes    OP SP* x CM+ AL_FOLLOW
                                               #
                                               # Further note:  the $AL in [$AL {eof}] is only to work around
                                               #                a rule compiler bug which complains about
                                               #                empty sets otherwise.


# LB 4, 5, 6

$LB4Breaks [$LB4NonBreaks-$CM];
$LB4Breaks $CM+ $CAN_CM;
$LF $CR;


# LB 7         x SP
#              x ZW
[$SP $ZW] [$LB4NonBreaks-$CM];
[$SP $ZW] $CM+ $CAN_CM;

# LB 8 ZW SP* <break>
#     TODO: to implement this, we need more than one look-ahead hard break in play at a time.
#           Requires an engine enhancement.
#   / $SP* $ZW

# LB 8a        ZWJ x (ID | Extended_Pict | EmojiNRK)
#
($ID | $Extended_Pict | $EmojiNRK) $ZWJ $CM* $CAN_CM?;


# LB 9,10  Combining marks.
#    X   $CM needs to behave like X, where X is not $SP or controls.
#    $CM not covered by the above needs to behave like $AL
# Stick together any combining sequences that don't match other rules.
^$CM+ $CAN_CM;


# LB 11
#
$WJ $CM* $CAN_CM;
$WJ      [$LB8NonBreaks-$CM];

     $CANT_CM $CM* $WJ;
$CAN_CM  $CM* $WJ;

# LB 12a
#      [^SP BA HY] x GL
#
$GL $CM* [$LB8NonBreaks-[$CM $SP $BA $HY]];

# LB 12
#     GL  x
#
$CANT_CM $CM* $GL;
$CAN_CM $CM* $GL;


# LB 13
$CL $CM+ $CAN_CM;
$CP $CM+ $CAN_CM;
$EX $CM+ $CAN_CM;
$IS $CM+ $CAN_CM;
$SY $CM+ $CAN_CM;

$CL [$LB8NonBreaks-$CM];
$CP [$LB8NonBreaks-$CM];
$EX [$LB8NonBreaks-$CM];
$IS [$LB8NonBreaks-$CM];
$SY [$LB8NonBreaks-$CM];


# LB 14    OP SP* x
#
.   $SP* $CM* $OP;
$AL_FOLLOW? $CM+ $SP+ $CM* $OP;     #  by LB 10, behaves like $AL_FOLLOW? $AL $SP+ $CM* $OP


# LB 15
$OP $SP* $CM* $QU;

# LB 16
$NS $SP* $CM* ($CL | $CP);

# LB 17
$B2 $SP* $CM* $B2;

# LB 18  break after spaces
#        Nothing explicit needed here.


#
# LB 19
#
$QU $CM* $CAN_CM;                                #   . x QU
$QU      $LB18NonBreaks;


$CAN_CM  $CM* $QU;                               #   QU x .
     $CANT_CM $CM* $QU;

#
#  LB 20  Break before and after CB.
#         nothing needed here.
#

# LB 21
($BA | $HY | $NS) $CM* [$LB20NonBreaks-$CM];     #  . x (BA | HY | NS)

[$LB20NonBreaks-$CM] $CM* $BB;                   #  BB x .
[^$CB] $CM* $BB;                                      #

# LB21a
[^$CB] $CM* ($HY | $BA) $CM* $HL;

# LB21b (reverse)
$HL $CM* $SY;

# LB 22
$IN $CM* ($ALPlus | $HL);
$IN $CM* $EX;
$IN $CM* ($ID | $EB | $EM);
$IN $CM* $IN;
$IN $CM* $NU;

# LB 23
$NU $CM* ($ALPlus | $HL);
($ALPlus | $HL) $CM* $NU;

# LB23a
($ID | $EB | $EM) $CM* $PR;
$PO $CM* ($ID | $EB | $EM);

# LB 24
($ALPlus | $HL) $CM* ($PR | $PO);
($PR | $PO) $CM* ($ALPlus | $HL);


# LB 25
($CM* ($PR | $PO))? ($CM* ($CL | $CP))? ($CM* ($NU | $IS | $SY))* $CM* $NU ($CM* ($OP | $HY))? ($CM* ($PR | $PO))?;

# LB 26
($H3 | $H2 | $JV | $JL) $CM* $JL;
($JT | $JV) $CM* ($H2 | $JV);
$JT $CM* ($H3 | $JT);

# LB 27
$IN $CM* ($H3 | $H2 | $JT | $JV | $JL);
$PO $CM* ($H3 | $H2 | $JT | $JV | $JL);
 ($H3 | $H2 | $JT | $JV | $JL) $CM* $PR;

# LB 28
($ALPlus | $HL) $CM* ($ALPlus | $HL);


# LB 29
($ALPlus | $HL) $CM* $IS;

# LB 30
$OP $CM* ($ALPlus | $HL | $NU);
($ALPlus | $HL | $NU) $CM* $CP;

# LB 30a
#    Pairs of Regional Indicators.
#    The following two rules are nearly identical. The first matches only sequences with an odd number of adjacent RIs,
#    the second with an even number. Stripping away the cruft they look like
#         [^RI] RI / (RI RI)+ ^RI;
#         [^RI] RI RI / (RI RI)+ ^RI;
#
[{bof} $NS $HY $BA $QU $CL $CP $EX $IS $SY $WJ $GL $ZW $SP $BK $CR $LF $NL $ZWJ] $CM* $RI / ($CM* $RI $CM* $RI)+ $CM* [{eof}[^$RI $CM]];
[{bof} $NS $HY $BA $QU $CL $CP $EX $IS $SY $WJ $GL $ZW $SP $BK $CR $LF $NL $ZWJ] $CM* $RI $CM* $RI / ($CM* $RI $CM* $RI)+ $CM* [{eof}[^$RI $CM]];

# In general, adjacent RIs stay together. The hard-break rules, above, overide this, forcing in the boundaries between pairs.
$RI $CM* $RI;

#    WJ, GL, QU, etc. are classes with rules like "WJ x "   which includes "WJ x RI".
$RI $CM* ([$WJ $GL $QU $BB] |  (($HY | $BA)$CM* $HL));


# LB 30b Do not break between an Emoji Base and an Emoji Modifier
$EM $CM* $EB;


## -------------------------------------------------

!!safe_reverse;

# LB 9
^$CM+ [^$CM $BK $CR $LF $NL $ZW $SP];
^$CM+ $SP / .;

# LB 14
$SP+ $CM* $OP;

# LB 15
$SP+ $CM* $QU;

# LB 16
$SP+ $CM* ($CL | $CP);

# LB 17
$SP+ $CM* $B2;

# LB 21
$CM* ($HY | $BA) $CM* $HL;

# LB 25
($CM* ($IS | $SY))+ $CM* $NU;
($CL | $CP) $CM* ($NU | $IS | $SY);

#  LB 30
($CM* $RI)+;

# For dictionary-based break
$dictionary $dictionary;

## -------------------------------------------------

!!safe_forward;

# Skip forward over all character classes that are involved in
#   rules containing patterns with possibly more than one char
#   of context.
#
#  It might be slightly more efficient to have specific rules
#  instead of one generic one, but only if we could
#  turn off rule chaining.  We don't want to move more
#  than necessary.
#
^[$CM $OP $QU $CL $CP $B2 $PR $HY $BA $SP $RI $ZWJ $dictionary]+ [^$CM $OP $QU $CL $CP $B2 $PR $HY $BA $RI $ZWJ $dictionary];
$dictionary $dictionary;