2 # Copyright (C) 2016 and later: Unicode, Inc. and others.
3 # License & terms of use: http://www.unicode.org/copyright.html#License
5 # Copyright (C) 2002-2015, International Business Machines Corporation and others.
10 # ICU Sentence Break Rules
11 # See Unicode Standard Annex #29.
12 # These rules are based on UAX #29 Revision 26 for Unicode Version 8.0
17 # Character categories as defined in TR 29
19 $CR = [\p{Sentence_Break = CR}];
20 $LF = [\p{Sentence_Break = LF}];
21 $Extend = [\p{Sentence_Break = Extend}];
22 $Sep = [\p{Sentence_Break = Sep}];
23 $Format = [\p{Sentence_Break = Format}];
24 $Sp = [\p{Sentence_Break = Sp}];
25 $Lower = [\p{Sentence_Break = Lower}];
26 $Upper = [\p{Sentence_Break = Upper}];
27 $OLetter = [\p{Sentence_Break = OLetter}];
28 $Numeric = [\p{Sentence_Break = Numeric}];
29 $ATerm = [\p{Sentence_Break = ATerm}];
30 $SContinue = [\p{Sentence_Break = SContinue}];
31 $STerm = [\p{Sentence_Break = STerm} [\u003B \u037E]];
32 $Close = [\p{Sentence_Break = Close}];
35 # Define extended forms of the character classes,
36 # incorporate trailing Extend or Format chars.
39 $SpEx = $Sp ($Extend | $Format)*;
40 $LowerEx = $Lower ($Extend | $Format)*;
41 $UpperEx = $Upper ($Extend | $Format)*;
42 $OLetterEx = $OLetter ($Extend | $Format)*;
43 $NumericEx = $Numeric ($Extend | $Format)*;
44 $ATermEx = $ATerm ($Extend | $Format)*;
45 $SContinueEx= $SContinue ($Extend | $Format)*;
46 $STermEx = $STerm ($Extend | $Format)*;
47 $CloseEx = $Close ($Extend | $Format)*;
50 ## -------------------------------------------------
55 # Rule 3 - break after separators. Keep CR/LF together.
60 # Rule 4 - Break after $Sep.
61 # Rule 5 - Ignore $Format and $Extend
63 [^$Sep $CR $LF]? ($Extend | $Format)*;
70 ($UpperEx | $LowerEx) $ATermEx $UpperEx;
73 $NotLettersEx = [^$OLetter $Upper $Lower $Sep $CR $LF $ATerm $STerm] ($Extend | $Format)*;
74 $ATermEx $CloseEx* $SpEx* $NotLettersEx* $Lower;
77 ($STermEx | $ATermEx) $CloseEx* $SpEx* ($SContinueEx | $STermEx | $ATermEx);
80 ($STermEx | $ATermEx) $CloseEx* $SpEx* ($Sep | $CR | $LF)?;
83 [[^$STerm $ATerm $Close $Sp $Sep $LF $CR $Format $Extend]{bof}] ($Extend | $Format | $Close | $Sp)* .;
84 [[^$STerm $ATerm $Close $Sp $Sep $LF $CR $Format $Extend]{bof}] ($Extend | $Format | $Close | $Sp)* ([$Sep $LF $CR {eof}] | $CR $LF){100};
86 ## -------------------------------------------------
90 $SpEx_R = ($Extend | $Format)* $Sp;
91 $ATermEx_R = ($Extend | $Format)* $ATerm;
92 $STermEx_R = ($Extend | $Format)* $STerm;
93 $CloseEx_R = ($Extend | $Format)* $Close;
97 # For now, use the old style inexact reverse rules, which are easier
98 # to write, but less efficient.
99 # TODO: exact reverse rules. It appears that exact reverse rules
100 # may require improving support for look-ahead breaks in the
101 # builder. Needs more investigation.
104 [{bof}] (.? | $LF $CR) [^$Sep $CR $LF]* [$Sep $CR $LF {eof}] ($SpEx_R* $CloseEx_R* ($STermEx_R | $ATermEx_R))*;
107 # Explanation for this rule:
109 # It needs to back over
110 # The $Sep at which we probably begin
111 # All of the non $Sep chars leading to the preceding $Sep
112 # The preceding $Sep, which will be the second one that the rule matches.
113 # Any immediately preceding STerm or ATerm sequences. We need to see these
114 # to get the correct rule status when moving forwards again.
116 # [{bof}] inhibit rule chaining. Without this, rule would loop on itself and match
119 # (.? | $LF $CR) Match one $Sep instance. Use .? rather than $Sep because position might be
120 # at the beginning of the string at this point, and we don't want to fail.
121 # Can only use {eof} once, and it is used later.