| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package EBook::Ishmael::EBook::Text; |
|
2
|
17
|
|
|
17
|
|
355
|
use 5.016; |
|
|
17
|
|
|
|
|
68
|
|
|
3
|
|
|
|
|
|
|
our $VERSION = '2.03'; |
|
4
|
17
|
|
|
17
|
|
134
|
use strict; |
|
|
17
|
|
|
|
|
35
|
|
|
|
17
|
|
|
|
|
658
|
|
|
5
|
17
|
|
|
17
|
|
94
|
use warnings; |
|
|
17
|
|
|
|
|
47
|
|
|
|
17
|
|
|
|
|
1151
|
|
|
6
|
|
|
|
|
|
|
|
|
7
|
17
|
|
|
17
|
|
120
|
use Encode qw(decode); |
|
|
17
|
|
|
|
|
42
|
|
|
|
17
|
|
|
|
|
1129
|
|
|
8
|
17
|
|
|
17
|
|
116
|
use File::Basename; |
|
|
17
|
|
|
|
|
35
|
|
|
|
17
|
|
|
|
|
1297
|
|
|
9
|
17
|
|
|
17
|
|
101
|
use File::Spec; |
|
|
17
|
|
|
|
|
45
|
|
|
|
17
|
|
|
|
|
478
|
|
|
10
|
|
|
|
|
|
|
|
|
11
|
17
|
|
|
17
|
|
105
|
use EBook::Ishmael::CharDet; |
|
|
17
|
|
|
|
|
71
|
|
|
|
17
|
|
|
|
|
1032
|
|
|
12
|
17
|
|
|
17
|
|
106
|
use EBook::Ishmael::EBook::Metadata; |
|
|
17
|
|
|
|
|
48
|
|
|
|
17
|
|
|
|
|
486
|
|
|
13
|
17
|
|
|
17
|
|
84
|
use EBook::Ishmael::TextToHtml; |
|
|
17
|
|
|
|
|
55
|
|
|
|
17
|
|
|
|
|
13011
|
|
|
14
|
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
# Use -T to check if file is a text file. EBook.pm's ebook_id() makes sure |
|
16
|
|
|
|
|
|
|
# to use the Text heuristic last, so that it doesn't incorrectly identify other |
|
17
|
|
|
|
|
|
|
# plain text ebook formats as just text. |
|
18
|
|
|
|
|
|
|
sub heuristic { |
|
19
|
|
|
|
|
|
|
|
|
20
|
12
|
|
|
12
|
0
|
28
|
my $class = shift; |
|
21
|
12
|
|
|
|
|
25
|
my $file = shift; |
|
22
|
|
|
|
|
|
|
|
|
23
|
12
|
|
|
|
|
763
|
return -T $file; |
|
24
|
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
} |
|
26
|
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
sub new { |
|
28
|
|
|
|
|
|
|
|
|
29
|
11
|
|
|
11
|
0
|
22
|
my $class = shift; |
|
30
|
11
|
|
|
|
|
22
|
my $file = shift; |
|
31
|
11
|
|
|
|
|
24
|
my $enc = shift; |
|
32
|
|
|
|
|
|
|
|
|
33
|
11
|
|
|
|
|
114
|
my $self = { |
|
34
|
|
|
|
|
|
|
Source => undef, |
|
35
|
|
|
|
|
|
|
Metadata => EBook::Ishmael::EBook::Metadata->new, |
|
36
|
|
|
|
|
|
|
Encode => $enc, |
|
37
|
|
|
|
|
|
|
}; |
|
38
|
|
|
|
|
|
|
|
|
39
|
11
|
|
|
|
|
26
|
bless $self, $class; |
|
40
|
|
|
|
|
|
|
|
|
41
|
11
|
|
|
|
|
425
|
$self->{Source} = File::Spec->rel2abs($file); |
|
42
|
|
|
|
|
|
|
|
|
43
|
11
|
|
|
|
|
691
|
$self->{Metadata}->set_title(basename($self->{Source})); |
|
44
|
11
|
|
|
|
|
376
|
$self->{Metadata}->set_modified((stat $self->{Source})[9]); |
|
45
|
11
|
|
|
|
|
50
|
$self->{Metadata}->set_format('Text'); |
|
46
|
|
|
|
|
|
|
|
|
47
|
11
|
|
|
|
|
48
|
return $self; |
|
48
|
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
} |
|
50
|
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
sub html { |
|
52
|
|
|
|
|
|
|
|
|
53
|
3
|
|
|
3
|
0
|
9
|
my $self = shift; |
|
54
|
3
|
|
|
|
|
9
|
my $out = shift; |
|
55
|
|
|
|
|
|
|
|
|
56
|
3
|
|
|
|
|
16
|
my $html = text2html($self->raw); |
|
57
|
|
|
|
|
|
|
|
|
58
|
3
|
50
|
|
|
|
21
|
if (defined $out) { |
|
59
|
0
|
0
|
|
|
|
0
|
open my $wh, '>', $out |
|
60
|
|
|
|
|
|
|
or die "Failed to open $out for writing: $!\n"; |
|
61
|
0
|
|
|
|
|
0
|
binmode $wh, ':utf8'; |
|
62
|
0
|
|
|
|
|
0
|
print { $wh } $html; |
|
|
0
|
|
|
|
|
0
|
|
|
63
|
0
|
|
|
|
|
0
|
close $wh; |
|
64
|
0
|
|
|
|
|
0
|
return $out; |
|
65
|
|
|
|
|
|
|
} else { |
|
66
|
3
|
|
|
|
|
38
|
return $html; |
|
67
|
|
|
|
|
|
|
} |
|
68
|
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
} |
|
70
|
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
sub raw { |
|
72
|
|
|
|
|
|
|
|
|
73
|
6
|
|
|
6
|
0
|
14
|
my $self = shift; |
|
74
|
6
|
|
|
|
|
15
|
my $out = shift; |
|
75
|
|
|
|
|
|
|
|
|
76
|
|
|
|
|
|
|
open my $rh, '<', $self->{Source} |
|
77
|
6
|
50
|
|
|
|
449
|
or die "Failed to open $self->{Source} for reading: $!\n"; |
|
78
|
6
|
|
|
|
|
28
|
binmode $rh; |
|
79
|
6
|
|
|
|
|
13
|
my $raw = do { local $/; <$rh> }; |
|
|
6
|
|
|
|
|
247
|
|
|
|
6
|
|
|
|
|
372
|
|
|
80
|
6
|
|
|
|
|
78
|
close $rh; |
|
81
|
|
|
|
|
|
|
|
|
82
|
6
|
100
|
|
|
|
27
|
if (not defined $self->{Encode}) { |
|
83
|
5
|
|
50
|
|
|
29
|
$self->{Encode} = chardet($raw) // 'ASCII'; |
|
84
|
|
|
|
|
|
|
} |
|
85
|
6
|
|
|
|
|
222
|
say $self->{Encode}; |
|
86
|
6
|
|
|
|
|
103
|
$raw = decode($self->{Encode}, $raw); |
|
87
|
|
|
|
|
|
|
|
|
88
|
6
|
50
|
|
|
|
2946
|
if (defined $out) { |
|
89
|
0
|
0
|
|
|
|
0
|
open my $wh, '>', $out |
|
90
|
|
|
|
|
|
|
or die "Failed to open $out for writing: $!\n"; |
|
91
|
0
|
|
|
|
|
0
|
binmode $wh, ':utf8'; |
|
92
|
0
|
|
|
|
|
0
|
print { $wh } $raw; |
|
|
0
|
|
|
|
|
0
|
|
|
93
|
0
|
|
|
|
|
0
|
close $wh; |
|
94
|
0
|
|
|
|
|
0
|
return $out; |
|
95
|
|
|
|
|
|
|
} else { |
|
96
|
6
|
|
|
|
|
93
|
return $raw; |
|
97
|
|
|
|
|
|
|
} |
|
98
|
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
} |
|
100
|
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
sub metadata { |
|
102
|
|
|
|
|
|
|
|
|
103
|
6
|
|
|
6
|
0
|
1525
|
my $self = shift; |
|
104
|
|
|
|
|
|
|
|
|
105
|
6
|
|
|
|
|
32
|
return $self->{Metadata}; |
|
106
|
|
|
|
|
|
|
|
|
107
|
|
|
|
|
|
|
} |
|
108
|
|
|
|
|
|
|
|
|
109
|
2
|
|
|
2
|
0
|
11
|
sub has_cover { 0 } |
|
110
|
|
|
|
|
|
|
|
|
111
|
1
|
|
|
1
|
0
|
6
|
sub cover { (undef, undef) } |
|
112
|
|
|
|
|
|
|
|
|
113
|
2
|
|
|
2
|
0
|
9
|
sub image_num { 0 } |
|
114
|
|
|
|
|
|
|
|
|
115
|
1
|
|
|
1
|
0
|
5
|
sub image { (undef, undef) } |
|
116
|
|
|
|
|
|
|
|
|
117
|
|
|
|
|
|
|
1; |