line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package ShiftJIS::Regexp::Const; |
2
|
6
|
|
|
6
|
|
32
|
use strict; |
|
6
|
|
|
|
|
46
|
|
|
6
|
|
|
|
|
201
|
|
3
|
6
|
|
|
6
|
|
29
|
use Carp; |
|
6
|
|
|
|
|
12
|
|
|
6
|
|
|
|
|
433
|
|
4
|
|
|
|
|
|
|
|
5
|
6
|
|
|
6
|
|
35
|
use vars qw($VERSION @ISA @EXPORT @EXPORT_OK); |
|
6
|
|
|
|
|
10
|
|
|
6
|
|
|
|
|
457
|
|
6
|
6
|
|
|
6
|
|
48
|
use vars qw(%AbbrevProp %Re %Err $Char $Trail $PadA $PadG $PadGA); |
|
6
|
|
|
|
|
27
|
|
|
6
|
|
|
|
|
9260
|
|
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
$VERSION = '1.03'; |
9
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
require Exporter; |
11
|
|
|
|
|
|
|
@ISA = qw(Exporter); |
12
|
|
|
|
|
|
|
@EXPORT = qw(); |
13
|
|
|
|
|
|
|
@EXPORT_OK = qw(%AbbrevProp %Re %Err $Char $Trail $PadA $PadG $PadGA); |
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
my $PACKAGE = 'ShiftJIS::Regexp'; |
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
my $Open = 5.005 > $] ? '(?:' : '(?-i:'; |
18
|
|
|
|
|
|
|
my $Close = ')'; |
19
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
my $SBC = '[\x00-\x7F\xA1-\xDF]'; |
21
|
|
|
|
|
|
|
my $DBC = '[\x81-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]'; |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
$Char = "(?:$SBC|$DBC)"; |
24
|
|
|
|
|
|
|
$Trail = '[\x40-\x7E\x80-\xFC]'; |
25
|
|
|
|
|
|
|
|
26
|
|
|
|
|
|
|
$PadA = '(?:\A|[\x00-\x80\xA0-\xDF])(?:[\x81-\x9F\xE0-\xFC]{2})*?'; |
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
$PadG = '\G(?:|[\x00-\xFF]*?'. |
29
|
|
|
|
|
|
|
'[\x00-\x80\xA0-\xDF])(?:[\x81-\x9F\xE0-\xFC]{2})*?'; |
30
|
|
|
|
|
|
|
#$PadG = '(?:\G|[\x00-\x80\xA0-\xDF])(?:[\x81-\x9F\xE0-\xFC]{2})*?'; |
31
|
|
|
|
|
|
|
|
32
|
|
|
|
|
|
|
#$PadGA = '(?:\G\A|\G(?:[\x81-\x9F\xE0-\xFC]{2})+?|'. |
33
|
|
|
|
|
|
|
# '[\x00-\x80\xA0-\xDF](?:[\x81-\x9F\xE0-\xFC]{2})*?)'; |
34
|
|
|
|
|
|
|
$PadGA = '\G(?:\A|(?:[\x81-\x9F\xE0-\xFC]{2})+?|[\x00-\xFF]*?'. |
35
|
|
|
|
|
|
|
'[\x00-\x80\xA0-\xDF](?:[\x81-\x9F\xE0-\xFC]{2})*?)'; |
36
|
|
|
|
|
|
|
|
37
|
|
|
|
|
|
|
%Err = ( |
38
|
|
|
|
|
|
|
'Code' => $PACKAGE.' Sequence (?{...}) not terminated or not {}-balanced', |
39
|
|
|
|
|
|
|
'Undef' => $PACKAGE.' "%s" is not defined', |
40
|
|
|
|
|
|
|
'notTermin' => $PACKAGE.' %s is not terminated ("%s" missing)', |
41
|
|
|
|
|
|
|
'notASCII' => $PACKAGE.' "%s" is not followed by an ASCII, [\x21-\x7e]', |
42
|
|
|
|
|
|
|
'notAlnum' => $PACKAGE.' "%s" is not followed by an Alnum, [0-9A-Za-z]', |
43
|
|
|
|
|
|
|
'notBrace' => $PACKAGE.' "%s" is not followed by a brace', |
44
|
|
|
|
|
|
|
'backtips' => $PACKAGE.' Trailing \ in regexp', |
45
|
|
|
|
|
|
|
'oddTrail' => $PACKAGE.' "\\x%02x" is not followed by trail byte', |
46
|
|
|
|
|
|
|
'revRange' => $PACKAGE.' Invalid [] range (reverse) %d > %d', |
47
|
|
|
|
|
|
|
'invalRng' => $PACKAGE.' Invalid [] range "%s"', |
48
|
|
|
|
|
|
|
'invalMch' => $PACKAGE.' Invalid Metacharacter "%s"', |
49
|
|
|
|
|
|
|
'invalHex' => $PACKAGE.' Invalid Hexadecimal %s following "\x"', |
50
|
|
|
|
|
|
|
'invalFlw' => $PACKAGE.' Invalid byte "\\x%02x" following "%s" (only "%s" allowed)', |
51
|
|
|
|
|
|
|
); |
52
|
|
|
|
|
|
|
|
53
|
|
|
|
|
|
|
%Re = ( |
54
|
|
|
|
|
|
|
'\R{pada}' => $PadA, |
55
|
|
|
|
|
|
|
'\R{padg}' => $PadG, |
56
|
|
|
|
|
|
|
'\R{padga}' => $PadGA, |
57
|
|
|
|
|
|
|
|
58
|
|
|
|
|
|
|
'\C' => '[\x00-\xFF]', |
59
|
|
|
|
|
|
|
'\j' => $Char, |
60
|
|
|
|
|
|
|
'\J' => "(?:(?!\\n)$Char)", |
61
|
|
|
|
|
|
|
'\d' => '[0-9]', |
62
|
|
|
|
|
|
|
'\D' => '(?:[\x00-\x2F\x3A-\x7F\xA1-\xDF]|' . $DBC . ')', |
63
|
|
|
|
|
|
|
'\w' => '[0-9A-Za-z\x5F]', |
64
|
|
|
|
|
|
|
'\W' => '(?:[\x00-\x2F\x3A-\x40\x5B-\x5E\x60\x7B-\x7F\xA1-\xDF]|'.$DBC.')', |
65
|
|
|
|
|
|
|
'\s' => '[\x09\x0A\x0C\x0D\x20]', |
66
|
|
|
|
|
|
|
'\S' => '(?:[\x00-\x08\x0B\x0E-\x1F\x21-\x7F\xA1-\xDF]|' . $DBC . ')', |
67
|
|
|
|
|
|
|
|
68
|
|
|
|
|
|
|
'\p{xdigit}' => '[0-9A-Fa-f]', |
69
|
|
|
|
|
|
|
'\P{xdigit}' => $Open.'[\x00-\x2F\x3A-\x40\x47-\x60\x67-\x7F\xA1-\xDF]|' |
70
|
|
|
|
|
|
|
. $DBC .$Close, |
71
|
|
|
|
|
|
|
|
72
|
|
|
|
|
|
|
'\p{digit}' => $Open.'[\x30-\x39]|\x82[\x4F-\x58]'.$Close, |
73
|
|
|
|
|
|
|
'\P{digit}' => $Open.'[\x00-\x2F\x3A-\x7F\xA1-\xDF]' |
74
|
|
|
|
|
|
|
. '|[\x81\x83-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]' |
75
|
|
|
|
|
|
|
. '|\x82[\x40-\x4E\x59-\x7E\x80-\xFC]' |
76
|
|
|
|
|
|
|
. $Close, |
77
|
|
|
|
|
|
|
|
78
|
|
|
|
|
|
|
'\p{upper}' => $Open.'[\x41-\x5A]|\x82[\x60-\x79]'.$Close, |
79
|
|
|
|
|
|
|
'\P{upper}' => $Open.'[\x00-\x40\x5B-\x7F\xA1-\xDF]' |
80
|
|
|
|
|
|
|
. '|[\x81\x83-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]' |
81
|
|
|
|
|
|
|
. '|\x82[\x40-\x5F\x7A-\x7E\x80-\xFC]' |
82
|
|
|
|
|
|
|
. $Close, |
83
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
'\p{lower}' => $Open.'[\x61-\x7A]|\x82[\x81-\x9A]'.$Close, |
85
|
|
|
|
|
|
|
'\P{lower}' => $Open.'[\x00-\x60\x7B-\x7F\xA1-\xDF]' |
86
|
|
|
|
|
|
|
. '|[\x81\x83-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]' |
87
|
|
|
|
|
|
|
. '|\x82[\x40-\x7E\x80\x9B-\xFC]'. $Close, |
88
|
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
'\p{alpha}' => $Open.'[\x41-\x5A\x61-\x7A]|\x82[\x60-\x79\x81-\x9A]'.$Close, |
90
|
|
|
|
|
|
|
'\P{alpha}' => $Open.'[\x00-\x40\x5B-\x60\x7B-\x7F\xA1-\xDF]' |
91
|
|
|
|
|
|
|
. '|[\x81\x83-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]' |
92
|
|
|
|
|
|
|
. '|\x82[\x40-\x5F\x7A-\x7E\x80\x9B-\xFC]'. $Close, |
93
|
|
|
|
|
|
|
|
94
|
|
|
|
|
|
|
'\p{alnum}' => $Open.'[0-9A-Za-z]|\x82[\x4F-\x58\x60-\x79\x81-\x9A]'.$Close, |
95
|
|
|
|
|
|
|
'\P{alnum}' => $Open.'[\x00-\x2F\x3A-\x40\x5B-\x60\x7B-\x7F\xA1-\xDF]' |
96
|
|
|
|
|
|
|
. '|[\x81\x83-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]' |
97
|
|
|
|
|
|
|
. '|\x82[\x40-\x4E\x59-\x5F\x7A-\x7E\x80\x9B-\xFC]'. $Close, |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
'\p{blank}' => $Open.'[\x09\x20]|\x81\x40'.$Close, |
100
|
|
|
|
|
|
|
'\P{blank}' => $Open.'[\x00-\x08\x0A-\x1F\x21-\x7F\xA1-\xDF]' |
101
|
|
|
|
|
|
|
. '|\x81[\x41-\x7E\x80-\xFC]' |
102
|
|
|
|
|
|
|
. '|[\x82-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]'. $Close, |
103
|
|
|
|
|
|
|
|
104
|
|
|
|
|
|
|
'\p{space}' => $Open.'[\x09-\x0D\x20]|\x81\x40'.$Close, |
105
|
|
|
|
|
|
|
'\P{space}' => $Open.'[\x00-\x08\x0E-\x1F\x21-\x7F\xA1-\xDF]' |
106
|
|
|
|
|
|
|
. '|\x81[\x41-\x7E\x80-\xFC]' |
107
|
|
|
|
|
|
|
. '|[\x82-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]'. $Close, |
108
|
|
|
|
|
|
|
|
109
|
|
|
|
|
|
|
'\p{punct}' => $Open.'[\x21-\x2F\x3A-\x40\x5B-\x60\x7B-\x7E\xA1-\xA5]' |
110
|
|
|
|
|
|
|
. '|\x81[\x41-\x49\x4C-\x51\x5C-\x7E\x80-\xAC\xB8-\xBF' |
111
|
|
|
|
|
|
|
. '\xC8-\xCE\xDA-\xE8\xF0-\xF7\xFC]|\x84[\x9F-\xBE]'. $Close, |
112
|
|
|
|
|
|
|
'\P{punct}' => $Open.'[\x00-\x20\x30-\x39\x41-\x5A\x61-\x7A\x7F\xA6-\xDF]' |
113
|
|
|
|
|
|
|
. '|\x81[\x40\x4A\x4B\x52-\x5B\xAD-\xB7\xC0-\xC7\xCF-\xD9' |
114
|
|
|
|
|
|
|
. '\xE9-\xEF\xF8-\xFB]|\x84[\x40-\x7E\x80-\x9E\xBF-\xFC]' |
115
|
|
|
|
|
|
|
. '|[\x82\x83\x85-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]'. $Close, |
116
|
|
|
|
|
|
|
|
117
|
|
|
|
|
|
|
'\p{graph}' => $Open.'[\x21-\x7E\xA1-\xDF]|\x81[\x41-\x7E' |
118
|
|
|
|
|
|
|
. '\x80-\xAC\xB8-\xBF\xC8-\xCE\xDA-\xE8\xF0-\xF7\xFC]|' |
119
|
|
|
|
|
|
|
. '\x82[\x4F-\x58\x60-\x79\x81-\x9A\x9F-\xF1]|' |
120
|
|
|
|
|
|
|
. '\x83[\x40-\x7E\x80-\x96\x9F-\xB6\xBF-\xD6]|' |
121
|
|
|
|
|
|
|
. '\x84[\x40-\x60\x70-\x7E\x80-\x91\x9F-\xBE]|' |
122
|
|
|
|
|
|
|
. '\x88[\x9F-\xFC]|\x98[\x40-\x72\x9F-\xFC]|' |
123
|
|
|
|
|
|
|
. '[\x89-\x97\x99-\x9F\xE0-\xE9][\x40-\x7E\x80-\xFC]|' |
124
|
|
|
|
|
|
|
. '\xEA[\x40-\x7E\x80-\xA4]' |
125
|
|
|
|
|
|
|
. $Close, |
126
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
'\P{graph}' => $Open.'[\x00-\x20\x7F]|' |
128
|
|
|
|
|
|
|
. '\x81[\x40\xAD-\xB7\xC0-\xC7\xCF-\xD9\xE9-\xEF\xF8-\xFB]|' |
129
|
|
|
|
|
|
|
. '\x82[\x40-\x4E\x59-\x5F\x7A-\x7E\x80\x9B-\x9E\xF2-\xFC]|' |
130
|
|
|
|
|
|
|
. '\x83[\x97-\x9E\xB7-\xBE\xD7-\xFC]|' |
131
|
|
|
|
|
|
|
. '\x84[\x61-\x6F\x92-\x9E\xBF-\xFC]|' |
132
|
|
|
|
|
|
|
. '[\x85-\x87\xEB-\xFC][\x40-\x7E\x80-\xFC]|' |
133
|
|
|
|
|
|
|
. '\x88[\x40-\x7E\x80-\x9E]|\x98[\x73-\x7E\x80-\x9E]|' |
134
|
|
|
|
|
|
|
. '\xEA[\xA5-\xFC]' |
135
|
|
|
|
|
|
|
.$Close, |
136
|
|
|
|
|
|
|
|
137
|
|
|
|
|
|
|
'\p{print}' => $Open.'[\x20-\x7E\xA1-\xDF]|\x81[\x40-\x7E' |
138
|
|
|
|
|
|
|
. '\x80-\xAC\xB8-\xBF\xC8-\xCE\xDA-\xE8\xF0-\xF7\xFC]|' |
139
|
|
|
|
|
|
|
. '\x82[\x4F-\x58\x60-\x79\x81-\x9A\x9F-\xF1]|' |
140
|
|
|
|
|
|
|
. '\x83[\x40-\x7E\x80-\x96\x9F-\xB6\xBF-\xD6]|' |
141
|
|
|
|
|
|
|
. '\x84[\x40-\x60\x70-\x7E\x80-\x91\x9F-\xBE]|' |
142
|
|
|
|
|
|
|
. '\x88[\x9F-\xFC]|\x98[\x40-\x72\x9F-\xFC]|' |
143
|
|
|
|
|
|
|
. '[\x89-\x97\x99-\x9F\xE0-\xE9][\x40-\x7E\x80-\xFC]|' |
144
|
|
|
|
|
|
|
. '\xEA[\x40-\x7E\x80-\xA4]' |
145
|
|
|
|
|
|
|
. $Close, |
146
|
|
|
|
|
|
|
|
147
|
|
|
|
|
|
|
'\P{print}' => $Open.'[\x00-\x1F\x7F]|' |
148
|
|
|
|
|
|
|
. '\x81[\xAD-\xB7\xC0-\xC7\xCF-\xD9\xE9-\xEF\xF8-\xFB]|' |
149
|
|
|
|
|
|
|
. '\x82[\x40-\x4E\x59-\x5F\x7A-\x7E\x80\x9B-\x9E\xF2-\xFC]|' |
150
|
|
|
|
|
|
|
. '\x83[\x97-\x9E\xB7-\xBE\xD7-\xFC]|' |
151
|
|
|
|
|
|
|
. '\x84[\x61-\x6F\x92-\x9E\xBF-\xFC]|' |
152
|
|
|
|
|
|
|
. '[\x85-\x87\xEB-\xFC][\x40-\x7E\x80-\xFC]|' |
153
|
|
|
|
|
|
|
. '\x88[\x40-\x7E\x80-\x9E]|\x98[\x73-\x7E\x80-\x9E]|' |
154
|
|
|
|
|
|
|
. '\xEA[\xA5-\xFC]' |
155
|
|
|
|
|
|
|
. $Close, |
156
|
|
|
|
|
|
|
|
157
|
|
|
|
|
|
|
'\p{cntrl}' => $Open.'[\x00-\x1F\x7F]'.$Close, |
158
|
|
|
|
|
|
|
'\P{cntrl}' => $Open.'[\x20-\x7E\xA1-\xDF]|'.$DBC.$Close, |
159
|
|
|
|
|
|
|
|
160
|
|
|
|
|
|
|
'\p{ascii}' => $Open.'[\x00-\x7F]'.$Close, |
161
|
|
|
|
|
|
|
'\P{ascii}' => $Open.'[\xA1-\xDF]|'.$DBC.$Close, |
162
|
|
|
|
|
|
|
|
163
|
|
|
|
|
|
|
'\p{roman}' => $Open.'[\x21-\x7E]'.$Close, |
164
|
|
|
|
|
|
|
'\P{roman}' => $Open.'[\x00-\x20\x7F\xA1-\xDF]|'.$DBC.$Close, |
165
|
|
|
|
|
|
|
|
166
|
|
|
|
|
|
|
'\p{word}' => $Open |
167
|
|
|
|
|
|
|
. '[0-9A-Za-z\x5F\xA6-\xDF]|\x81[\x4A\x4B\x52-\x5B]|' |
168
|
|
|
|
|
|
|
. '\x82[\x4F-\x58\x60-\x79\x81-\x9A\x9F-\xF1]|' |
169
|
|
|
|
|
|
|
. '\x83[\x40-\x7E\x80-\x96\x9F-\xB6\xBF-\xD6]|' |
170
|
|
|
|
|
|
|
. '\x84[\x40-\x60\x70-\x7E\x80-\x91]|\x88[\x9F-\xFC]|' |
171
|
|
|
|
|
|
|
. '[\x89-\x97\x99-\x9F\xE0-\xE9][\x40-\x7E\x80-\xFC]|' |
172
|
|
|
|
|
|
|
. '\x98[\x40-\x72\x9F-\xFC]|\xEA[\x40-\x7E\x80-\xA4]' |
173
|
|
|
|
|
|
|
. $Close, |
174
|
|
|
|
|
|
|
|
175
|
|
|
|
|
|
|
'\P{word}' => $Open |
176
|
|
|
|
|
|
|
. '[\x00-\x2F\x3A-\x40\x5B-\x5E\x60\x7B-\x7F\xA1-\xA5]|' |
177
|
|
|
|
|
|
|
. '\x81[\x40-\x49\x4C-\x51\x5C-\x7E\x80-\xFC]|' |
178
|
|
|
|
|
|
|
. '\x82[\x40-\x4E\x59-\x5F\x7A-\x7E\x80\x9B-\x9E\xF2-\xFC]|' |
179
|
|
|
|
|
|
|
. '\x83[\x97-\x9E\xB7-\xBE\xD7-\xFC]|' |
180
|
|
|
|
|
|
|
. '\x84[\x61-\x6F\x92-\xFC]|' |
181
|
|
|
|
|
|
|
. '[\x85-\x87\xEB-\xFC][\x40-\x7E\x80-\xFC]|' |
182
|
|
|
|
|
|
|
. '\x88[\x40-\x7E\x80-\x9E]|\x98[\x73-\x7E\x80-\x9E]|' |
183
|
|
|
|
|
|
|
. '\xEA[\xA5-\xFC]'.$Close, |
184
|
|
|
|
|
|
|
|
185
|
|
|
|
|
|
|
'\p{halfwidth}' => '[\x21\x23-\x26\x28-\x2C\x2E-\x7E]', |
186
|
|
|
|
|
|
|
'\P{halfwidth}' => $Open.'[\x00-\x20\x22\x27\x2D\x7F\xA1-\xDF]|' |
187
|
|
|
|
|
|
|
. $DBC.$Close, |
188
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
'\p{fullwidth}' => $Open.'\x81[\x43\x44\x46-\x49\x4d\x4f\x50\x51\x5e' |
190
|
|
|
|
|
|
|
. '\x62\x69\x6a\x6d-\x70\x7b\x81\x83\x84\x8f\x90\x93-\x97]|' |
191
|
|
|
|
|
|
|
. '\x82[\x4f-\x58\x60-\x79\x81-\x9a]'.$Close, |
192
|
|
|
|
|
|
|
|
193
|
|
|
|
|
|
|
'\P{fullwidth}' => $Open.$SBC |
194
|
|
|
|
|
|
|
. '|\x81[\x40-\x42\x45\x4a-\x4c\x4e\x52-\x5d\x5f-\x61' |
195
|
|
|
|
|
|
|
. '\x63-\x68\x6b\x6c\x71-\x7a\x7c-\x7e\x80\x82\x85-\x8e' |
196
|
|
|
|
|
|
|
. '\x91\x92\x98-\xfc]' |
197
|
|
|
|
|
|
|
. '|\x82[\x40-\x4e\x59-\x5f\x7a-\x7e\x80\x9b-\xfc]' |
198
|
|
|
|
|
|
|
. '|[\x83-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]'. $Close, |
199
|
|
|
|
|
|
|
|
200
|
|
|
|
|
|
|
'\p{hankaku}' => $Open.'[\xA1-\xDF]'.$Close, |
201
|
|
|
|
|
|
|
'\P{hankaku}' => $Open.'[\x00-\x7F]|'.$DBC.$Close, |
202
|
|
|
|
|
|
|
'\p{zenkaku}' => "$Open$DBC$Close", |
203
|
|
|
|
|
|
|
'\P{zenkaku}' => "$Open$SBC$Close", |
204
|
|
|
|
|
|
|
|
205
|
|
|
|
|
|
|
'\p{x0201}' => $Open.'[\x20-\x7F\xA1-\xDF]'.$Close, |
206
|
|
|
|
|
|
|
'\P{x0201}' => $Open.'[\x00-\x1F]|'.$DBC.$Close, |
207
|
|
|
|
|
|
|
|
208
|
|
|
|
|
|
|
'\p{x0211}' => $Open.'[\x00-\x1F]'.$Close, |
209
|
|
|
|
|
|
|
'\P{x0211}' => $Open.'[\x20-\x7F\xA1-\xDF]|'.$DBC.$Close, |
210
|
|
|
|
|
|
|
|
211
|
|
|
|
|
|
|
'\p{x0208}' => $Open.'\x81[\x40-\x7E' |
212
|
|
|
|
|
|
|
. '\x80-\xAC\xB8-\xBF\xC8-\xCE\xDA-\xE8\xF0-\xF7\xFC]|' |
213
|
|
|
|
|
|
|
. '\x82[\x4F-\x58\x60-\x79\x81-\x9A\x9F-\xF1]|' |
214
|
|
|
|
|
|
|
. '\x83[\x40-\x7E\x80-\x96\x9F-\xB6\xBF-\xD6]|' |
215
|
|
|
|
|
|
|
. '\x84[\x40-\x60\x70-\x7E\x80-\x91\x9F-\xBE]|' |
216
|
|
|
|
|
|
|
. '\x88[\x9F-\xFC]|\x98[\x40-\x72\x9F-\xFC]|' |
217
|
|
|
|
|
|
|
. '[\x89-\x97\x99-\x9F\xE0-\xE9][\x40-\x7E\x80-\xFC]|' |
218
|
|
|
|
|
|
|
. '\xEA[\x40-\x7E\x80-\xA4]' |
219
|
|
|
|
|
|
|
.$Close, |
220
|
|
|
|
|
|
|
|
221
|
|
|
|
|
|
|
'\P{x0208}' => $Open.$SBC |
222
|
|
|
|
|
|
|
. '|\x81[\xAD-\xB7\xC0-\xC7\xCF-\xD9\xE9-\xEF\xF8-\xFB]' |
223
|
|
|
|
|
|
|
. '|\x82[\x40-\x4E\x59-\x5F\x7A-\x7E\x80\x9B-\x9E\xF2-\xFC]' |
224
|
|
|
|
|
|
|
. '|\x83[\x97-\x9E\xB7-\xBE\xD7-\xFC]' |
225
|
|
|
|
|
|
|
. '|\x84[\x61-\x6F\x92-\x9E\xBF-\xFC]' |
226
|
|
|
|
|
|
|
. '|[\x85-\x87\xEB-\xFC][\x40-\x7E\x80-\xFC]' |
227
|
|
|
|
|
|
|
. '|\x88[\x40-\x7E\x80-\x9E]|\x98[\x73-\x7E\x80-\x9E]' |
228
|
|
|
|
|
|
|
. '|\xEA[\xA5-\xFC]'.$Close, |
229
|
|
|
|
|
|
|
|
230
|
|
|
|
|
|
|
'\p{jis}' => $Open.$SBC.'|\x81[\x40-\x7E' |
231
|
|
|
|
|
|
|
. '\x80-\xAC\xB8-\xBF\xC8-\xCE\xDA-\xE8\xF0-\xF7\xFC]|' |
232
|
|
|
|
|
|
|
. '\x82[\x4F-\x58\x60-\x79\x81-\x9A\x9F-\xF1]|' |
233
|
|
|
|
|
|
|
. '\x83[\x40-\x7E\x80-\x96\x9F-\xB6\xBF-\xD6]|' |
234
|
|
|
|
|
|
|
. '\x84[\x40-\x60\x70-\x7E\x80-\x91\x9F-\xBE]|' |
235
|
|
|
|
|
|
|
. '\x88[\x9F-\xFC]|\x98[\x40-\x72\x9F-\xFC]|' |
236
|
|
|
|
|
|
|
. '[\x89-\x97\x99-\x9F\xE0-\xE9][\x40-\x7E\x80-\xFC]|' |
237
|
|
|
|
|
|
|
. '\xEA[\x40-\x7E\x80-\xA4]'. $Close, |
238
|
|
|
|
|
|
|
|
239
|
|
|
|
|
|
|
'\P{jis}' => $Open |
240
|
|
|
|
|
|
|
. '\x81[\xAD-\xB7\xC0-\xC7\xCF-\xD9\xE9-\xEF\xF8-\xFB]|' |
241
|
|
|
|
|
|
|
. '\x82[\x40-\x4E\x59-\x5F\x7A-\x7E\x80\x9B-\x9E\xF2-\xFC]|' |
242
|
|
|
|
|
|
|
. '\x83[\x97-\x9E\xB7-\xBE\xD7-\xFC]|' |
243
|
|
|
|
|
|
|
. '\x84[\x61-\x6F\x92-\x9E\xBF-\xFC]|' |
244
|
|
|
|
|
|
|
. '[\x85-\x87\xEB-\xFC][\x40-\x7E\x80-\xFC]|' |
245
|
|
|
|
|
|
|
. '\x88[\x40-\x7E\x80-\x9E]|\x98[\x73-\x7E\x80-\x9E]|' |
246
|
|
|
|
|
|
|
. '\xEA[\xA5-\xFC]' |
247
|
|
|
|
|
|
|
.$Close, |
248
|
|
|
|
|
|
|
|
249
|
|
|
|
|
|
|
'\p{latin}' => $Open.'[\x41-\x5A\x61-\x7A]'.$Close, |
250
|
|
|
|
|
|
|
'\P{latin}' => $Open.'[\x00-\x40\x5B-\x60\x7B-\x7F\xA1-\xDF]|'.$DBC.$Close, |
251
|
|
|
|
|
|
|
|
252
|
|
|
|
|
|
|
'\p{fulllatin}' => $Open.'\x82[\x60-\x79\x81-\x9A]'.$Close, |
253
|
|
|
|
|
|
|
'\P{fulllatin}' => $Open.$SBC.'|\x82[\x40-\x5F\x7A-\x7E\x80\x9B-\xFC]|' |
254
|
|
|
|
|
|
|
. '[\x81\x83-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]'. $Close, |
255
|
|
|
|
|
|
|
|
256
|
|
|
|
|
|
|
'\p{greek}' => $Open.'\x83[\x9f-\xb6\xbf-\xd6]'.$Close, |
257
|
|
|
|
|
|
|
'\P{greek}' => $Open.$SBC.'|\x83[\x40-\x7E\x80-\x9e\xb7-\xbe\xd7-\xFC]|' |
258
|
|
|
|
|
|
|
. '[\x81\x82\x84-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]'. $Close, |
259
|
|
|
|
|
|
|
|
260
|
|
|
|
|
|
|
'\p{cyrillic}' => $Open.'\x84[\x40-\x60\x70-\x7E\x80-\x91]'.$Close, |
261
|
|
|
|
|
|
|
'\P{cyrillic}' => $Open.$SBC.'|\x84[\x61-\x6f\x92-\xFC]|' |
262
|
|
|
|
|
|
|
. '[\x81-\x83\x85-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]'. $Close, |
263
|
|
|
|
|
|
|
|
264
|
|
|
|
|
|
|
'\p{european}' => $Open.'[\x41-\x5A\x61-\x7A]|\x82[\x60-\x79\x81-\x9A]|' |
265
|
|
|
|
|
|
|
. '\x83[\x9f-\xb6\xbf-\xd6]|\x84[\x40-\x60\x70-\x7E\x80-\x91]' |
266
|
|
|
|
|
|
|
. $Close, |
267
|
|
|
|
|
|
|
|
268
|
|
|
|
|
|
|
'\P{european}' => $Open.'[\x00-\x40\x5B-\x60\x7B-\x7F\xA1-\xDF]|' |
269
|
|
|
|
|
|
|
. '[\x81\x85-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]|' |
270
|
|
|
|
|
|
|
. '\x82[\x40-\x5F\x7A-\x7E\x80\x9B-\xFC]|' |
271
|
|
|
|
|
|
|
. '\x83[\x40-\x7E\x80-\x9e\xb7-\xbe\xd7-\xFC]|' |
272
|
|
|
|
|
|
|
. '\x84[\x61-\x6f\x92-\xFC]'. $Close, |
273
|
|
|
|
|
|
|
|
274
|
|
|
|
|
|
|
'\p{halfkana}' => $Open.'[\xA6-\xDF]'.$Close, |
275
|
|
|
|
|
|
|
'\P{halfkana}' => $Open.'[\x00-\x7F\xA1-\xA5]|' .$DBC.$Close, |
276
|
|
|
|
|
|
|
|
277
|
|
|
|
|
|
|
'\p{hiragana}' => $Open.'\x82[\x9F-\xF1]|\x81[\x4A\x4B\x54\x55]'.$Close, |
278
|
|
|
|
|
|
|
'\P{hiragana}' => $Open.$SBC.'|[\x83-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]' |
279
|
|
|
|
|
|
|
. '|\x82[\x40-\x7E\x80-\x9E\xF2-\xFC]' |
280
|
|
|
|
|
|
|
. '|\x81[\x40-\x49\x4C-\x53\x56-\x7E\x80-\xFC]'. $Close, |
281
|
|
|
|
|
|
|
|
282
|
|
|
|
|
|
|
'\p{katakana}' => $Open.'\x83[\x40-\x7E\x80-\x96]|\x81[\x52\x53\x5B]'.$Close, |
283
|
|
|
|
|
|
|
'\P{katakana}' => $Open.$SBC |
284
|
|
|
|
|
|
|
. '|[\x82\x84-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]' |
285
|
|
|
|
|
|
|
. '|\x83[\x97-\xFC]|\x81[\x40-\x51\x54-\x5A\x5C-\x7E\x80-\xFC]' |
286
|
|
|
|
|
|
|
. $Close, |
287
|
|
|
|
|
|
|
|
288
|
|
|
|
|
|
|
'\p{fullkana}' => $Open.'\x82[\x9F-\xF1]|\x83[\x40-\x7E\x80-\x96]|' |
289
|
|
|
|
|
|
|
. '\x81[\x4A\x4B\x5B\x52-\x55]'.$Close, |
290
|
|
|
|
|
|
|
'\P{fullkana}' => $Open.$SBC.'|[\x84-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]' |
291
|
|
|
|
|
|
|
. '|\x82[\x40-\x7E\x80-\x9E\xF2-\xFC]|\x83[\x97-\xFC]' |
292
|
|
|
|
|
|
|
. '|\x81[\x40-\x49\x4C-\x51\x56-\x5A\x5C-\x7E\x80-\xFC]' |
293
|
|
|
|
|
|
|
. $Close, |
294
|
|
|
|
|
|
|
|
295
|
|
|
|
|
|
|
'\p{kana}' => $Open.'[\xA6-\xDF]|\x82[\x9F-\xF1]|\x83[\x40-\x7E\x80-\x96]|' |
296
|
|
|
|
|
|
|
. '\x81[\x4A\x4B\x5B\x52-\x55]'.$Close, |
297
|
|
|
|
|
|
|
'\P{kana}' => $Open.'[\x00-\x7F\xA1-\xA5]|' |
298
|
|
|
|
|
|
|
. '[\x84-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]|' |
299
|
|
|
|
|
|
|
. '\x82[\x40-\x7E\x80-\x9E\xF2-\xFC]|\x83[\x97-\xFC]|' |
300
|
|
|
|
|
|
|
. '\x81[\x40-\x49\x4C-\x51\x56-\x5A\x5C-\x7E\x80-\xFC]' |
301
|
|
|
|
|
|
|
. $Close, |
302
|
|
|
|
|
|
|
|
303
|
|
|
|
|
|
|
'\p{kanji0}' => $Open.'\x81[\x56-\x5A]'.$Close, |
304
|
|
|
|
|
|
|
'\P{kanji0}' => $Open.$SBC.'|\x81[\x40-\x55\x5b-\x7E\x80-\xFC]' |
305
|
|
|
|
|
|
|
. '|[\x82-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]'. $Close, |
306
|
|
|
|
|
|
|
|
307
|
|
|
|
|
|
|
'\p{kanji1}' => $Open.'\x88[\x9F-\xFC]|\x98[\x40-\x72]|' |
308
|
|
|
|
|
|
|
. '[\x89-\x97][\x40-\x7E\x80-\xFC]'.$Close, |
309
|
|
|
|
|
|
|
'\P{kanji1}' => $Open.$SBC.'|\x88[\x40-\x7E\x80-\x9E]|' |
310
|
|
|
|
|
|
|
. '\x98[\x73-\x7E\x80-\xFC]|' |
311
|
|
|
|
|
|
|
. '[\x81-\x87\x99-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]' |
312
|
|
|
|
|
|
|
. $Close, |
313
|
|
|
|
|
|
|
|
314
|
|
|
|
|
|
|
'\p{kanji2}' => $Open.'\x98[\x9F-\xFC]|[\x99-\x9F\xE0-\xE9]' |
315
|
|
|
|
|
|
|
. '[\x40-\x7E\x80-\xFC]|\xEA[\x40-\x7E\x80-\xA4]' |
316
|
|
|
|
|
|
|
. $Close, |
317
|
|
|
|
|
|
|
'\P{kanji2}' => $Open.$SBC.'|\x98[\x40-\x7E\x80-\x9E]|' |
318
|
|
|
|
|
|
|
. '[\x81-\x97\xEB-\xFC][\x40-\x7E\x80-\xFC]|\xEA[\xA5-\xFC]' |
319
|
|
|
|
|
|
|
. $Close, |
320
|
|
|
|
|
|
|
|
321
|
|
|
|
|
|
|
'\p{kanji}' => $Open.'\x81[\x56-\x5A]|\x88[\x9F-\xFC]' |
322
|
|
|
|
|
|
|
. '|[\x89-\x97\x99-\x9F\xE0-\xE9][\x40-\x7E\x80-\xFC]' |
323
|
|
|
|
|
|
|
. '|\x98[\x40-\x72\x9F-\xFC]|\xEA[\x40-\x7E\x80-\xA4]' |
324
|
|
|
|
|
|
|
. $Close, |
325
|
|
|
|
|
|
|
'\P{kanji}' => $Open.$SBC |
326
|
|
|
|
|
|
|
. '|\x81[\x40-\x55\x5b-\x7E\x80-\xFC]' |
327
|
|
|
|
|
|
|
. '|\x88[\x40-\x7E\x80-\x9E]|\x98[\x73-\x7E\x80-\x9E]' |
328
|
|
|
|
|
|
|
. '|[\x82-\x87\xEB-\xFC][\x40-\x7E\x80-\xFC]' |
329
|
|
|
|
|
|
|
. '|\xEA[\xA5-\xFC]'. $Close, |
330
|
|
|
|
|
|
|
|
331
|
|
|
|
|
|
|
'\p{boxdrawing}' => $Open.'\x84[\x9F-\xBE]'.$Close, |
332
|
|
|
|
|
|
|
'\P{boxdrawing}' => $Open.$SBC |
333
|
|
|
|
|
|
|
. '|[\x81-\x83\x85-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC]' |
334
|
|
|
|
|
|
|
. '|\x84[\x40-\x7E\x80-\x9E\xBF-\xFC]'. $Close, |
335
|
|
|
|
|
|
|
|
336
|
|
|
|
|
|
|
'\p{nec}' => $Open. '\x87[\x40-\x5d\x5f-\x75\x7e\x80-\x9c]' |
337
|
|
|
|
|
|
|
. '|\xed[\x40-\x7e\x80-\xfc]|\xee[\x40-\x7e\x80-\xec\xef-\xfc]' |
338
|
|
|
|
|
|
|
. $Close, |
339
|
|
|
|
|
|
|
|
340
|
|
|
|
|
|
|
'\p{ibm}' => $Open.'[\xfa-\xfb][\x40-\x7e\x80-\xfc]|\xfc[\x40-\x4b]'.$Close, |
341
|
|
|
|
|
|
|
|
342
|
|
|
|
|
|
|
'\p{vendor}' => $Open. '\x87[\x40-\x5d\x5f-\x75\x7e\x80-\x9c]' |
343
|
|
|
|
|
|
|
. '|[\xed\xfa-\xfb][\x40-\x7e\x80-\xfc]' |
344
|
|
|
|
|
|
|
. '|\xee[\x40-\x7e\x80-\xec\xef-\xfc]|\xfc[\x40-\x4b]' |
345
|
|
|
|
|
|
|
. $Close, |
346
|
|
|
|
|
|
|
|
347
|
|
|
|
|
|
|
'\p{mswin}' => $Open.'[\x00-\x7f\xa1-\xdf]|' |
348
|
|
|
|
|
|
|
. '\x81[\x40-\x7e\x80-\xac\xb8-\xbf\xc8-\xce\xda-\xe8\xf0-\xf7\xfc]|' |
349
|
|
|
|
|
|
|
. '\x82[\x4f-\x58\x60-\x79\x81-\x9a\x9f-\xf1]|' |
350
|
|
|
|
|
|
|
. '\x83[\x40-\x7e\x80-\x96\x9f-\xb6\xbf-\xd6]|' |
351
|
|
|
|
|
|
|
. '\x84[\x40-\x60\x70-\x7e\x80-\x91\x9f-\xbe]|' |
352
|
|
|
|
|
|
|
. '\x88[\x9f-\xfc]|\x98[\x40-\x72\x9f-\xfc]|\xea[\x40-\x7e\x80-\xa4]|' |
353
|
|
|
|
|
|
|
. '[\x89-\x97\x99-\x9f\xe0-\xe9][\x40-\x7e\x80-\xfc]|' |
354
|
|
|
|
|
|
|
. '\x87[\x40-\x5d\x5f-\x75\x7e\x80-\x9c]|' |
355
|
|
|
|
|
|
|
. '\xed[\x40-\x7e\x80-\xfc]|\xee[\x40-\x7e\x80-\xec\xef-\xfc]|' |
356
|
|
|
|
|
|
|
. '[\xfa\xfb][\x40-\x7e\x80-\xfc]|\xfc[\x40-\x4b]' |
357
|
|
|
|
|
|
|
. $Close, |
358
|
|
|
|
|
|
|
); |
359
|
|
|
|
|
|
|
|
360
|
|
|
|
|
|
|
for (qw/ nec ibm mswin vendor /) { |
361
|
|
|
|
|
|
|
$Re{"\\P{$_}"} = $Open.'(?!'. $Re{ "\\p{$_}" } .')'. $Char.$Close; |
362
|
|
|
|
|
|
|
} |
363
|
|
|
|
|
|
|
|
364
|
|
|
|
|
|
|
%AbbrevProp = qw( |
365
|
|
|
|
|
|
|
X xdigit |
366
|
|
|
|
|
|
|
D digit |
367
|
|
|
|
|
|
|
U upper |
368
|
|
|
|
|
|
|
L lower |
369
|
|
|
|
|
|
|
A alpha |
370
|
|
|
|
|
|
|
Q alnum |
371
|
|
|
|
|
|
|
W word |
372
|
|
|
|
|
|
|
P punct |
373
|
|
|
|
|
|
|
G graph |
374
|
|
|
|
|
|
|
T print |
375
|
|
|
|
|
|
|
S space |
376
|
|
|
|
|
|
|
B blank |
377
|
|
|
|
|
|
|
C cntrl |
378
|
|
|
|
|
|
|
R roman |
379
|
|
|
|
|
|
|
Y hankaku |
380
|
|
|
|
|
|
|
Z zenkaku |
381
|
|
|
|
|
|
|
F fullwidth |
382
|
|
|
|
|
|
|
J jis |
383
|
|
|
|
|
|
|
N nec |
384
|
|
|
|
|
|
|
I ibm |
385
|
|
|
|
|
|
|
V vendor |
386
|
|
|
|
|
|
|
M mswin |
387
|
|
|
|
|
|
|
E european |
388
|
|
|
|
|
|
|
H hiragana |
389
|
|
|
|
|
|
|
K katakana |
390
|
|
|
|
|
|
|
0 kanji0 |
391
|
|
|
|
|
|
|
1 kanji1 |
392
|
|
|
|
|
|
|
2 kanji2 |
393
|
|
|
|
|
|
|
); |
394
|
|
|
|
|
|
|
|
395
|
|
|
|
|
|
|
1; |
396
|
|
|
|
|
|
|
__END__ |