• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1#
2#   Copyright (C) 2016 and later: Unicode, Inc. and others.
3#   License & terms of use: http://www.unicode.org/copyright.html#License
4#
5#   Copyright (C) 2002-2015, International Business Machines Corporation and others.
6#       All Rights Reserved.
7#
8#   file:  sent.txt
9#
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
13#
14
15
16#
17# Character categories as defined in TR 29
18#
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$STerm     = [\p{Sentence_Break = STerm}];
32$Close     = [\p{Sentence_Break = Close}];
33
34#
35# Define extended forms of the character classes,
36#   incorporate trailing Extend or Format chars.
37#   Rules 4 and 5.
38
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 | $Format)*;
48
49
50## -------------------------------------------------
51
52!!chain;
53!!forward;
54
55# Rule 3 - break after separators.  Keep CR/LF together.
56#
57$CR $LF;
58
59
60# Rule 4 - Break after $Sep.
61# Rule 5 - Ignore $Format and $Extend
62#
63[^$Sep $CR $LF]? ($Extend | $Format)*;
64
65
66# Rule 6
67$ATermEx $NumericEx;
68
69# Rule 7
70($UpperEx | $LowerEx) $ATermEx $UpperEx;
71
72#Rule 8
73$NotLettersEx = [^$OLetter $Upper $Lower $Sep $CR $LF $ATerm $STerm] ($Extend | $Format)*;
74$ATermEx $CloseEx* $SpEx* $NotLettersEx* $Lower;
75
76# Rule 8a
77($STermEx | $ATermEx) $CloseEx* $SpEx* ($SContinueEx | $STermEx | $ATermEx);
78
79#Rule 9, 10, 11
80($STermEx | $ATermEx) $CloseEx* $SpEx* ($Sep | $CR | $LF)?;
81
82#Rule 12
83[[^$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 $LF){100};
85
86## -------------------------------------------------
87
88!!reverse;
89
90$SpEx_R       = ($Extend | $Format)* $Sp;
91$ATermEx_R    = ($Extend | $Format)* $ATerm;
92$STermEx_R    = ($Extend | $Format)* $STerm;
93$CloseEx_R    = ($Extend | $Format)* $Close;
94
95#
96#  Reverse rules.
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.
102#
103
104[{bof}] (.? | $LF $CR) [^$Sep $CR $LF]* [$Sep $CR $LF {eof}] ($SpEx_R* $CloseEx_R* ($STermEx_R | $ATermEx_R))*;
105#.*;
106
107# Explanation for this rule:
108#
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.
115#
116# [{bof}]           inhibit rule chaining.  Without this, rule would loop on itself and match
117#                   the entire string.
118#
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.
122#
123