line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Data::Kanji::Tomoe; |
2
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
our $VERSION = '0.05'; |
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
# Read in the data file from Tomoe |
6
|
|
|
|
|
|
|
|
7
|
1
|
|
|
1
|
|
15110
|
use warnings; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
31
|
|
8
|
1
|
|
|
1
|
|
3
|
use strict; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
19
|
|
9
|
1
|
|
|
1
|
|
198
|
use XML::Parser; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
use utf8; |
11
|
|
|
|
|
|
|
use Carp; |
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
sub point() |
14
|
|
|
|
|
|
|
{ |
15
|
|
|
|
|
|
|
my ($tomoe) = @_; |
16
|
|
|
|
|
|
|
my $stroke_points = $$tomoe{stroke_points}; |
17
|
|
|
|
|
|
|
my $attrval = $$tomoe{attrval}; |
18
|
|
|
|
|
|
|
my $x = $$attrval{x}; |
19
|
|
|
|
|
|
|
my $y = $$attrval{y}; |
20
|
|
|
|
|
|
|
# print $x, $y; |
21
|
|
|
|
|
|
|
push @$stroke_points, [$x, $y]; |
22
|
|
|
|
|
|
|
} |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
sub stroke() |
25
|
|
|
|
|
|
|
{ |
26
|
|
|
|
|
|
|
my ($tomoe) = @_; |
27
|
|
|
|
|
|
|
$$tomoe{stroke_points} = []; |
28
|
|
|
|
|
|
|
$$tomoe{stroke_count}++; |
29
|
|
|
|
|
|
|
} |
30
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
sub strokes() |
32
|
|
|
|
|
|
|
{ |
33
|
|
|
|
|
|
|
my ($tomoe) = @_; |
34
|
|
|
|
|
|
|
$$tomoe{stroke_count} = 0; |
35
|
|
|
|
|
|
|
} |
36
|
|
|
|
|
|
|
|
37
|
|
|
|
|
|
|
sub character() |
38
|
|
|
|
|
|
|
{ |
39
|
|
|
|
|
|
|
my ($tomoe) = @_; |
40
|
|
|
|
|
|
|
# Kill old character. |
41
|
|
|
|
|
|
|
$tomoe->{character} = {}; |
42
|
|
|
|
|
|
|
} |
43
|
|
|
|
|
|
|
|
44
|
|
|
|
|
|
|
sub utf8() |
45
|
|
|
|
|
|
|
{ |
46
|
|
|
|
|
|
|
my ($tomoe) = @_; |
47
|
|
|
|
|
|
|
} |
48
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
sub dictionary() |
50
|
|
|
|
|
|
|
{ |
51
|
|
|
|
|
|
|
my ($tomoe) = @_; |
52
|
|
|
|
|
|
|
|
53
|
|
|
|
|
|
|
} |
54
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
sub point_end() |
56
|
|
|
|
|
|
|
{ |
57
|
|
|
|
|
|
|
my ($tomoe) = @_; |
58
|
|
|
|
|
|
|
} |
59
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
sub stroke_end() |
61
|
|
|
|
|
|
|
{ |
62
|
|
|
|
|
|
|
my ($tomoe) = @_; |
63
|
|
|
|
|
|
|
my $stroke_points = $$tomoe{stroke_points}; |
64
|
|
|
|
|
|
|
my $character = $tomoe->{character}; |
65
|
|
|
|
|
|
|
push @{$character->{strokes}}, $stroke_points; |
66
|
|
|
|
|
|
|
# print "Number of points is ", scalar @$stroke_points, "\n"; |
67
|
|
|
|
|
|
|
# my $stroke_data = StrokeRecognition::identify($stroke_points); |
68
|
|
|
|
|
|
|
# print $stroke_data; |
69
|
|
|
|
|
|
|
} |
70
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
sub strokes_end() |
72
|
|
|
|
|
|
|
{ |
73
|
|
|
|
|
|
|
my ($tomoe) = @_; |
74
|
|
|
|
|
|
|
# print " ", $$tomoe{stroke_count}, "\n"; |
75
|
|
|
|
|
|
|
} |
76
|
|
|
|
|
|
|
|
77
|
|
|
|
|
|
|
sub character_end() |
78
|
|
|
|
|
|
|
{ |
79
|
|
|
|
|
|
|
my ($tomoe) = @_; |
80
|
|
|
|
|
|
|
my $character_callback = $tomoe->{character_callback}; |
81
|
|
|
|
|
|
|
if ($character_callback) { |
82
|
|
|
|
|
|
|
&{$character_callback} ($tomoe, $tomoe->{character}); |
83
|
|
|
|
|
|
|
} |
84
|
|
|
|
|
|
|
} |
85
|
|
|
|
|
|
|
|
86
|
|
|
|
|
|
|
sub utf8_end() |
87
|
|
|
|
|
|
|
{ |
88
|
|
|
|
|
|
|
my ($tomoe, $string) = @_; |
89
|
|
|
|
|
|
|
$tomoe->{character}->{utf8} = $string; |
90
|
|
|
|
|
|
|
# print $string, " "; |
91
|
|
|
|
|
|
|
} |
92
|
|
|
|
|
|
|
|
93
|
|
|
|
|
|
|
sub dictionary_end() |
94
|
|
|
|
|
|
|
{ |
95
|
|
|
|
|
|
|
my ($tomoe) = @_; |
96
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
} |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
# Subroutines which handle the various tags in the Tomoe data. |
100
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
my %start_handlers = ( |
102
|
|
|
|
|
|
|
point => \& point, |
103
|
|
|
|
|
|
|
stroke => \& stroke, |
104
|
|
|
|
|
|
|
strokes => \& strokes, |
105
|
|
|
|
|
|
|
character => \& character, |
106
|
|
|
|
|
|
|
utf8 => \& utf8, |
107
|
|
|
|
|
|
|
dictionary => \& dictionary, |
108
|
|
|
|
|
|
|
); |
109
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
# Subroutines which handle the various tags in the Tomoe data. |
111
|
|
|
|
|
|
|
|
112
|
|
|
|
|
|
|
my %end_handlers = ( |
113
|
|
|
|
|
|
|
point => \& point_end, |
114
|
|
|
|
|
|
|
stroke => \& stroke_end, |
115
|
|
|
|
|
|
|
strokes => \& strokes_end, |
116
|
|
|
|
|
|
|
character => \& character_end, |
117
|
|
|
|
|
|
|
utf8 => \& utf8_end, |
118
|
|
|
|
|
|
|
dictionary => \& dictionary_end, |
119
|
|
|
|
|
|
|
); |
120
|
|
|
|
|
|
|
|
121
|
|
|
|
|
|
|
# Handle an XML starting tag |
122
|
|
|
|
|
|
|
|
123
|
|
|
|
|
|
|
sub handle_start |
124
|
|
|
|
|
|
|
{ |
125
|
|
|
|
|
|
|
my ($tomoe, $parsexml, $element, %attrval) = @_; |
126
|
|
|
|
|
|
|
$$tomoe{element} = $element; |
127
|
|
|
|
|
|
|
$$tomoe{attrval} = \%attrval; |
128
|
|
|
|
|
|
|
if ($start_handlers{$element}) { |
129
|
|
|
|
|
|
|
&{$start_handlers{$element}}($tomoe); |
130
|
|
|
|
|
|
|
} else { |
131
|
|
|
|
|
|
|
print STDERR "No element handler for element `$element' at line $.\n"; |
132
|
|
|
|
|
|
|
} |
133
|
|
|
|
|
|
|
} |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
# Handle an XML ending tag |
136
|
|
|
|
|
|
|
|
137
|
|
|
|
|
|
|
sub handle_end |
138
|
|
|
|
|
|
|
{ |
139
|
|
|
|
|
|
|
my ($tomoe, $parsexml, $element) = @_; |
140
|
|
|
|
|
|
|
my $values = $$tomoe{values}; |
141
|
|
|
|
|
|
|
my $string = $$values{$element}; |
142
|
|
|
|
|
|
|
$string =~ s/^\s+|\s+$//g if $string; |
143
|
|
|
|
|
|
|
if ($end_handlers{$element}) { |
144
|
|
|
|
|
|
|
&{$end_handlers{$element}}($tomoe, $string) |
145
|
|
|
|
|
|
|
} |
146
|
|
|
|
|
|
|
else { |
147
|
|
|
|
|
|
|
print STDERR "No end handler for element `$element' at line $.\n"; |
148
|
|
|
|
|
|
|
} |
149
|
|
|
|
|
|
|
undef $$values{$element}; |
150
|
|
|
|
|
|
|
} |
151
|
|
|
|
|
|
|
|
152
|
|
|
|
|
|
|
# Handle XML character data |
153
|
|
|
|
|
|
|
|
154
|
|
|
|
|
|
|
sub handle_char |
155
|
|
|
|
|
|
|
{ |
156
|
|
|
|
|
|
|
my ($tomoe, $parsexml, $string) = @_; |
157
|
|
|
|
|
|
|
my $values = $$tomoe{values}; |
158
|
|
|
|
|
|
|
$$values{$$tomoe{element}} .= $string; |
159
|
|
|
|
|
|
|
} |
160
|
|
|
|
|
|
|
|
161
|
|
|
|
|
|
|
sub parse |
162
|
|
|
|
|
|
|
{ |
163
|
|
|
|
|
|
|
my ($parser) = @_; |
164
|
|
|
|
|
|
|
$$parser{xmlparser}->parse($$parser{file}); |
165
|
|
|
|
|
|
|
} |
166
|
|
|
|
|
|
|
|
167
|
|
|
|
|
|
|
sub parsefile |
168
|
|
|
|
|
|
|
{ |
169
|
|
|
|
|
|
|
my ($parser, $tomoe_data_file) = @_; |
170
|
|
|
|
|
|
|
open_file ($parser, $tomoe_data_file); |
171
|
|
|
|
|
|
|
$parser->parse ($parser->{file}); |
172
|
|
|
|
|
|
|
close $parser->{file} or croak $!; |
173
|
|
|
|
|
|
|
} |
174
|
|
|
|
|
|
|
|
175
|
|
|
|
|
|
|
sub open_file |
176
|
|
|
|
|
|
|
{ |
177
|
|
|
|
|
|
|
my ($parser, $tomoe_data_file) = @_; |
178
|
|
|
|
|
|
|
open my $tomoe_data, "<:encoding(utf8)", $tomoe_data_file |
179
|
|
|
|
|
|
|
or die "Can't open $tomoe_data_file: $!"; |
180
|
|
|
|
|
|
|
$$parser{file_name} = $tomoe_data_file; |
181
|
|
|
|
|
|
|
$$parser{file} = $tomoe_data; |
182
|
|
|
|
|
|
|
} |
183
|
|
|
|
|
|
|
|
184
|
|
|
|
|
|
|
sub new |
185
|
|
|
|
|
|
|
{ |
186
|
|
|
|
|
|
|
my ($package, %inputs) = @_; |
187
|
|
|
|
|
|
|
my $parser = \%inputs; |
188
|
|
|
|
|
|
|
$parser->{xmlparser} = new XML::Parser( |
189
|
|
|
|
|
|
|
Handlers => { |
190
|
|
|
|
|
|
|
Start => sub { handle_start ($parser, @_); }, |
191
|
|
|
|
|
|
|
End => sub { handle_end ($parser, @_); }, |
192
|
|
|
|
|
|
|
Char => sub { handle_char ($parser, @_); }, |
193
|
|
|
|
|
|
|
}, |
194
|
|
|
|
|
|
|
); |
195
|
|
|
|
|
|
|
$$parser{values} = {}; |
196
|
|
|
|
|
|
|
bless $parser; |
197
|
|
|
|
|
|
|
if ($inputs{tomoe_data_file}) { |
198
|
|
|
|
|
|
|
open_file ($parser, $inputs{tomoe_data_file}); |
199
|
|
|
|
|
|
|
} |
200
|
|
|
|
|
|
|
return $parser; |
201
|
|
|
|
|
|
|
} |
202
|
|
|
|
|
|
|
|
203
|
|
|
|
|
|
|
1; |