| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package MARC::Detrans::Rule; |
|
2
|
|
|
|
|
|
|
|
|
3
|
9
|
|
|
9
|
|
3333
|
use strict; |
|
|
9
|
|
|
|
|
21
|
|
|
|
9
|
|
|
|
|
326
|
|
|
4
|
9
|
|
|
9
|
|
51
|
use warnings; |
|
|
9
|
|
|
|
|
16
|
|
|
|
9
|
|
|
|
|
452
|
|
|
5
|
9
|
|
|
9
|
|
51
|
use Carp qw( croak ); |
|
|
9
|
|
|
|
|
16
|
|
|
|
9
|
|
|
|
|
3975
|
|
|
6
|
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
=head1 NAME |
|
8
|
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
MARC::Detrans::Rule |
|
10
|
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
=head1 SYNOPSIS |
|
12
|
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
use MARC::Detrans::Rule; |
|
14
|
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
my $rule = MARC::Detrans::Rule->new( |
|
16
|
|
|
|
|
|
|
from => 'b', |
|
17
|
|
|
|
|
|
|
to => 'B', |
|
18
|
|
|
|
|
|
|
escape => '(B' |
|
19
|
|
|
|
|
|
|
); |
|
20
|
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
=head1 DESCRIPTION |
|
22
|
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
It's unlikely that you'll want to use MARC::Detrans::Rule directly since |
|
24
|
|
|
|
|
|
|
other modules wrap access to it. Each detransliteration rule is represented |
|
25
|
|
|
|
|
|
|
as a MARC::Detrans::Rule object, which basically provides the Romanized text |
|
26
|
|
|
|
|
|
|
and the corresponding MARC-8 or UTF-8 text, along with an escape character |
|
27
|
|
|
|
|
|
|
(for MARC-8) rules. |
|
28
|
|
|
|
|
|
|
|
|
29
|
|
|
|
|
|
|
=head1 METHODS |
|
30
|
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
=head2 new() |
|
32
|
|
|
|
|
|
|
|
|
33
|
|
|
|
|
|
|
Pass in the C and c parameters which define the original text |
|
34
|
|
|
|
|
|
|
and what to translate to; these parameters are not limited to single |
|
35
|
|
|
|
|
|
|
characters. In addition an C parameter can be passed in to |
|
36
|
|
|
|
|
|
|
indicate a MARC-8 escape sequence to use. Also a C parameter |
|
37
|
|
|
|
|
|
|
can be set to C, C or C if the rule applies only |
|
38
|
|
|
|
|
|
|
when the character is found at or within a particular word boundary. |
|
39
|
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
=cut |
|
41
|
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
sub new { |
|
43
|
465
|
|
|
465
|
1
|
3638
|
my ( $class, %opts ) = @_; |
|
44
|
465
|
50
|
|
|
|
1376
|
croak( "must supply 'from' parameter" ) if ! exists( $opts{from} ); |
|
45
|
465
|
50
|
|
|
|
1046
|
croak( "must supply 'to' parameter" ) if ! exists( $opts{to} ); |
|
46
|
465
|
|
|
|
|
872
|
$opts{to} =~ s/\^ESC/\x1B/g; |
|
47
|
465
|
|
33
|
|
|
4335
|
return bless \%opts, ref($class) || $class; |
|
48
|
|
|
|
|
|
|
} |
|
49
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
=head2 from() |
|
51
|
|
|
|
|
|
|
|
|
52
|
|
|
|
|
|
|
Returns the Romanized text that this rule refers to. |
|
53
|
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
=cut |
|
55
|
|
|
|
|
|
|
|
|
56
|
|
|
|
|
|
|
sub from { |
|
57
|
3711
|
|
|
3711
|
1
|
13228
|
return shift->{from}; |
|
58
|
|
|
|
|
|
|
} |
|
59
|
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
=head2 to() |
|
61
|
|
|
|
|
|
|
|
|
62
|
|
|
|
|
|
|
Returns the MARC-8 or UTF-8 text that the corresponding Romanized text should |
|
63
|
|
|
|
|
|
|
be converted to. |
|
64
|
|
|
|
|
|
|
|
|
65
|
|
|
|
|
|
|
=cut |
|
66
|
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
sub to { |
|
68
|
1115
|
|
|
1115
|
1
|
5341
|
return shift->{to}; |
|
69
|
|
|
|
|
|
|
} |
|
70
|
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
=head2 escape() |
|
72
|
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
Returns a MARC-8 character set escape sequence to be used, or undef if the rule |
|
74
|
|
|
|
|
|
|
is for an UTF-8 mapping. |
|
75
|
|
|
|
|
|
|
|
|
76
|
|
|
|
|
|
|
=cut |
|
77
|
|
|
|
|
|
|
|
|
78
|
|
|
|
|
|
|
sub escape { |
|
79
|
2365
|
|
|
2365
|
1
|
28044
|
return shift->{escape}; |
|
80
|
|
|
|
|
|
|
} |
|
81
|
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
=head2 position() |
|
83
|
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
Returns a position specification for the rule mapping which can be |
|
85
|
|
|
|
|
|
|
initial, medial, final or the empty string if there is no positional |
|
86
|
|
|
|
|
|
|
qualification for the rule. |
|
87
|
|
|
|
|
|
|
|
|
88
|
|
|
|
|
|
|
=cut |
|
89
|
|
|
|
|
|
|
|
|
90
|
|
|
|
|
|
|
sub position { |
|
91
|
1746
|
|
|
1746
|
1
|
10238
|
my $p = shift->{position}; |
|
92
|
1746
|
100
|
|
|
|
3895
|
return $p if defined($p); |
|
93
|
1696
|
|
|
|
|
9016
|
return ''; |
|
94
|
|
|
|
|
|
|
} |
|
95
|
|
|
|
|
|
|
|
|
96
|
|
|
|
|
|
|
=head1 AUTHORS |
|
97
|
|
|
|
|
|
|
|
|
98
|
|
|
|
|
|
|
=over 4 |
|
99
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
=item * Ed Summers |
|
101
|
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
=back |
|
103
|
|
|
|
|
|
|
|
|
104
|
|
|
|
|
|
|
=cut |
|
105
|
|
|
|
|
|
|
|
|
106
|
|
|
|
|
|
|
1; |