line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Lingua::YaTeA::ChunkingDataSet; |
2
|
5
|
|
|
5
|
|
36
|
use strict; |
|
5
|
|
|
|
|
38
|
|
|
5
|
|
|
|
|
142
|
|
3
|
5
|
|
|
5
|
|
25
|
use warnings; |
|
5
|
|
|
|
|
11
|
|
|
5
|
|
|
|
|
120
|
|
4
|
5
|
|
|
5
|
|
2081
|
use Lingua::YaTeA::ChunkingDataSubset; |
|
5
|
|
|
|
|
45
|
|
|
5
|
|
|
|
|
55
|
|
5
|
|
|
|
|
|
|
|
6
|
|
|
|
|
|
|
our $VERSION=$Lingua::YaTeA::VERSION; |
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
sub new |
10
|
|
|
|
|
|
|
{ |
11
|
4
|
|
|
4
|
1
|
17
|
my ($class,$file_set) = @_; |
12
|
4
|
|
|
|
|
13
|
my $this = {}; |
13
|
4
|
|
|
|
|
13
|
bless ($this,$class); |
14
|
4
|
|
|
|
|
40
|
$this->{ChunkingFrontiers} = Lingua::YaTeA::ChunkingDataSubset->new; |
15
|
4
|
|
|
|
|
18
|
$this->{ChunkingExceptions} = Lingua::YaTeA::ChunkingDataSubset->new; |
16
|
4
|
|
|
|
|
18
|
$this->{CleaningFrontiers} = Lingua::YaTeA::ChunkingDataSubset->new; |
17
|
4
|
|
|
|
|
18
|
$this->{CleaningExceptions} = Lingua::YaTeA::ChunkingDataSubset->new; |
18
|
|
|
|
|
|
|
|
19
|
4
|
|
|
|
|
22
|
$this->loadData($file_set); |
20
|
4
|
|
|
|
|
179
|
return $this; |
21
|
|
|
|
|
|
|
} |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
sub loadData |
25
|
|
|
|
|
|
|
{ |
26
|
4
|
|
|
4
|
1
|
13
|
my ($this,$file_set) = @_; |
27
|
4
|
|
|
|
|
35
|
my $name; |
28
|
|
|
|
|
|
|
my $path; |
29
|
4
|
|
|
|
|
0
|
my $subset; |
30
|
4
|
|
|
|
|
0
|
my $line; |
31
|
4
|
|
|
|
|
0
|
my $type; |
32
|
4
|
|
|
|
|
0
|
my $value; |
33
|
4
|
|
|
|
|
20
|
my @file_names = ("ChunkingFrontiers","ChunkingExceptions","CleaningFrontiers","CleaningExceptions"); |
34
|
4
|
|
|
|
|
13
|
foreach $name (@file_names) |
35
|
|
|
|
|
|
|
{ |
36
|
16
|
|
|
|
|
572
|
$path = $file_set->getFile($name)->getPath; |
37
|
|
|
|
|
|
|
|
38
|
16
|
|
|
|
|
46
|
$subset = $this->getSubset($name); |
39
|
|
|
|
|
|
|
|
40
|
16
|
|
|
|
|
125
|
my $fh = FileHandle->new("<$path"); |
41
|
|
|
|
|
|
|
|
42
|
16
|
|
|
|
|
1926
|
while ($line= $fh->getline) |
43
|
|
|
|
|
|
|
{ |
44
|
276
|
50
|
66
|
|
|
7772
|
if(($line !~ /^#/)&&($line !~ /^\s*$/)) |
45
|
|
|
|
|
|
|
{ |
46
|
244
|
|
|
|
|
462
|
$line =~ s/\r//g; |
47
|
244
|
|
|
|
|
353
|
chomp $line; |
48
|
244
|
|
|
|
|
686
|
$line =~ /([^\t]+)\t([^\t]+)/; |
49
|
244
|
|
|
|
|
482
|
$type= $1; |
50
|
244
|
|
|
|
|
383
|
$value = $2; |
51
|
244
|
|
|
|
|
602
|
$subset->addSome($type,$value); |
52
|
|
|
|
|
|
|
} |
53
|
|
|
|
|
|
|
} |
54
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
} |
56
|
|
|
|
|
|
|
} |
57
|
|
|
|
|
|
|
|
58
|
|
|
|
|
|
|
sub getSubset |
59
|
|
|
|
|
|
|
{ |
60
|
21640
|
|
|
21640
|
1
|
34866
|
my ($this,$name) = @_; |
61
|
21640
|
|
|
|
|
49584
|
return $this->{$name}; |
62
|
|
|
|
|
|
|
} |
63
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
|
65
|
|
|
|
|
|
|
sub existData |
66
|
|
|
|
|
|
|
{ |
67
|
21624
|
|
|
21624
|
1
|
41847
|
my ($this,$set,$type,$data) = @_; |
68
|
21624
|
|
|
|
|
37702
|
return $this->getSubset($set)->existData($type,$data); |
69
|
|
|
|
|
|
|
} |
70
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
1; |
72
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
__END__ |