1 |
package WebPAC::Normalize; |
package WebPAC::Normalize; |
2 |
use Exporter 'import'; |
use Exporter 'import'; |
3 |
@EXPORT = qw/ |
@EXPORT = qw/ |
4 |
set_rec set_lookup |
_set_rec _set_lookup |
5 |
get_ds clean_ds |
_get_ds _clean_ds |
6 |
|
|
7 |
tag search display |
tag search display |
8 |
|
marc21 |
9 |
|
|
10 |
rec1 rec2 rec |
rec1 rec2 rec |
11 |
regex prefix suffix surround |
regex prefix suffix surround |
12 |
first lookup join_with |
first lookup join_with |
17 |
|
|
18 |
#use base qw/WebPAC::Common/; |
#use base qw/WebPAC::Common/; |
19 |
use Data::Dumper; |
use Data::Dumper; |
20 |
|
use Encode qw/from_to/; |
21 |
|
|
22 |
=head1 NAME |
=head1 NAME |
23 |
|
|
25 |
|
|
26 |
=head1 VERSION |
=head1 VERSION |
27 |
|
|
28 |
Version 0.04 |
Version 0.06 |
29 |
|
|
30 |
=cut |
=cut |
31 |
|
|
32 |
our $VERSION = '0.04'; |
our $VERSION = '0.06'; |
33 |
|
|
34 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
35 |
|
|
42 |
C<perl -c normalize.pl>. |
C<perl -c normalize.pl>. |
43 |
|
|
44 |
Normalisation can generate multiple output normalized data. For now, supported output |
Normalisation can generate multiple output normalized data. For now, supported output |
45 |
types (on the left side of definition) are: C<tag>, C<display> and C<search>. |
types (on the left side of definition) are: C<tag>, C<display>, C<search> and |
46 |
|
C<marc21>. |
47 |
|
|
48 |
=head1 FUNCTIONS |
=head1 FUNCTIONS |
49 |
|
|
50 |
|
Functions which start with C<_> are private and used by WebPAC internally. |
51 |
|
All other functions are available for use within normalisation rules. |
52 |
|
|
53 |
=head2 data_structure |
=head2 data_structure |
54 |
|
|
55 |
Return data structure |
Return data structure |
56 |
|
|
57 |
my $ds = WebPAC::Normalize( |
my $ds = WebPAC::Normalize::data_structure( |
58 |
lookup => $lookup->lookup_hash, |
lookup => $lookup->lookup_hash, |
59 |
row => $row, |
row => $row, |
60 |
rules => $normalize_pl_config, |
rules => $normalize_pl_config, |
61 |
|
marc_encoding => 'utf-8', |
62 |
); |
); |
63 |
|
|
64 |
|
Options C<lookup>, C<row>, C<rules> and C<log> are mandatory while all |
65 |
|
other are optional. |
66 |
|
|
67 |
This function will B<die> if normalizastion can't be evaled. |
This function will B<die> if normalizastion can't be evaled. |
68 |
|
|
69 |
|
Since this function isn't exported you have to call it with |
70 |
|
C<WebPAC::Normalize::data_structure>. |
71 |
|
|
72 |
=cut |
=cut |
73 |
|
|
74 |
sub data_structure { |
sub data_structure { |
78 |
die "need normalisation argument" unless ($arg->{rules}); |
die "need normalisation argument" unless ($arg->{rules}); |
79 |
|
|
80 |
no strict 'subs'; |
no strict 'subs'; |
81 |
set_lookup( $arg->{lookup} ); |
_set_lookup( $arg->{lookup} ); |
82 |
set_rec( $arg->{row} ); |
_set_rec( $arg->{row} ); |
83 |
clean_ds(); |
_clean_ds( %{ $arg } ); |
84 |
eval "$arg->{rules}"; |
eval "$arg->{rules}"; |
85 |
die "error evaling $arg->{rules}: $@\n" if ($@); |
die "error evaling $arg->{rules}: $@\n" if ($@); |
86 |
return get_ds(); |
|
87 |
|
return _get_ds(); |
88 |
} |
} |
89 |
|
|
90 |
=head2 set_rec |
=head2 _set_rec |
91 |
|
|
92 |
Set current record hash |
Set current record hash |
93 |
|
|
94 |
set_rec( $rec ); |
_set_rec( $rec ); |
95 |
|
|
96 |
=cut |
=cut |
97 |
|
|
98 |
my $rec; |
my $rec; |
99 |
|
|
100 |
sub set_rec { |
sub _set_rec { |
101 |
$rec = shift or die "no record hash"; |
$rec = shift or die "no record hash"; |
102 |
} |
} |
103 |
|
|
104 |
|
=head2 _get_ds |
105 |
|
|
106 |
|
Return hash formatted as data structure |
107 |
|
|
108 |
|
my $ds = _get_ds(); |
109 |
|
|
110 |
|
=cut |
111 |
|
|
112 |
|
my $out; |
113 |
|
my $marc21; |
114 |
|
my $marc_encoding; |
115 |
|
|
116 |
|
sub _get_ds { |
117 |
|
return $out; |
118 |
|
} |
119 |
|
|
120 |
|
=head2 _clean_ds |
121 |
|
|
122 |
|
Clean data structure hash for next record |
123 |
|
|
124 |
|
_clean_ds(); |
125 |
|
|
126 |
|
=cut |
127 |
|
|
128 |
|
sub _clean_ds { |
129 |
|
my $a = {@_}; |
130 |
|
$out = undef; |
131 |
|
$marc21 = undef; |
132 |
|
$marc_encoding = $a->{marc_encoding}; |
133 |
|
} |
134 |
|
|
135 |
|
=head2 _set_lookup |
136 |
|
|
137 |
|
Set current lookup hash |
138 |
|
|
139 |
|
_set_lookup( $lookup ); |
140 |
|
|
141 |
|
=cut |
142 |
|
|
143 |
|
my $lookup; |
144 |
|
|
145 |
|
sub _set_lookup { |
146 |
|
$lookup = shift; |
147 |
|
} |
148 |
|
|
149 |
|
=head2 _get_marc21_fields |
150 |
|
|
151 |
|
Get all fields defined by calls to C<marc21> |
152 |
|
|
153 |
|
$marc->add_fields( WebPAC::Normalize:_get_marc21_fields() ); |
154 |
|
|
155 |
|
=cut |
156 |
|
|
157 |
|
sub _get_marc21_fields { |
158 |
|
return @{$marc21}; |
159 |
|
} |
160 |
|
|
161 |
|
=head1 Functions to create C<data_structure> |
162 |
|
|
163 |
|
Those functions generally have to first in your normalization file. |
164 |
|
|
165 |
=head2 tag |
=head2 tag |
166 |
|
|
167 |
Define new tag for I<search> and I<display>. |
Define new tag for I<search> and I<display>. |
171 |
|
|
172 |
=cut |
=cut |
173 |
|
|
|
my $out; |
|
|
|
|
174 |
sub tag { |
sub tag { |
175 |
my $name = shift or die "tag needs name as first argument"; |
my $name = shift or die "tag needs name as first argument"; |
176 |
my @o = grep { defined($_) && $_ ne '' } @_; |
my @o = grep { defined($_) && $_ ne '' } @_; |
212 |
$out->{$name}->{search} = \@o; |
$out->{$name}->{search} = \@o; |
213 |
} |
} |
214 |
|
|
215 |
=head2 get_ds |
=head2 marc21 |
216 |
|
|
217 |
Return hash formatted as data structure |
Save value for MARC field |
218 |
|
|
219 |
my $ds = get_ds(); |
marc21('900','a', rec('200','a') ); |
220 |
|
|
221 |
=cut |
=cut |
222 |
|
|
223 |
sub get_ds { |
sub marc21 { |
224 |
return $out; |
my $f = shift or die "marc21 needs field"; |
225 |
} |
die "marc21 field must be numer" unless ($f =~ /^\d+$/); |
|
|
|
|
=head2 clean_ds |
|
|
|
|
|
Clean data structure hash for next record |
|
226 |
|
|
227 |
clean_ds(); |
my $sf = shift or die "marc21 needs subfield"; |
228 |
|
|
229 |
=cut |
foreach (@_) { |
230 |
|
my $v = $_; # make var read-write for Encode |
231 |
sub clean_ds { |
next unless (defined($v) && $v !~ /^\s+$/); |
232 |
$out = undef; |
from_to($v, 'iso-8859-2', $marc_encoding) if ($marc_encoding); |
233 |
|
push @{ $marc21 }, [ $f, ' ', ' ', $sf => $v ]; |
234 |
|
} |
235 |
} |
} |
236 |
|
|
237 |
=head2 set_lookup |
=head1 Functions to extract data from input |
|
|
|
|
Set current lookup hash |
|
|
|
|
|
set_lookup( $lookup ); |
|
|
|
|
|
=cut |
|
238 |
|
|
239 |
my $lookup; |
This function should be used inside functions to create C<data_structure> described |
240 |
|
above. |
|
sub set_lookup { |
|
|
$lookup = shift; |
|
|
} |
|
241 |
|
|
242 |
=head2 rec1 |
=head2 rec1 |
243 |
|
|