1 |
package WebPAC::Normalize::MARC; |
2 |
use Exporter 'import'; |
3 |
@EXPORT = qw/ |
4 |
marc_template |
5 |
/; |
6 |
|
7 |
use strict; |
8 |
use warnings; |
9 |
|
10 |
use Data::Dump qw/dump/; |
11 |
use Carp qw/confess/; |
12 |
|
13 |
use WebPAC::Normalize; |
14 |
|
15 |
my $debug = 0; |
16 |
|
17 |
=head1 NAME |
18 |
|
19 |
WebPAC::Normalize::MARC - create MARC/ISO2709 records |
20 |
|
21 |
=cut |
22 |
|
23 |
=head1 FUNCTIONS |
24 |
|
25 |
=head2 marc_template |
26 |
|
27 |
marc_template( |
28 |
from => 225, to => 440, |
29 |
subfields_rename => [ |
30 |
'a' => 'a', |
31 |
'x' => 'x', |
32 |
'v' => 'v', |
33 |
'h' => 'n', |
34 |
'i' => 'p', |
35 |
'w' => 'v', |
36 |
], |
37 |
isis_template => [ |
38 |
'a ; |v. |i', |
39 |
'a. |i ; |w', |
40 |
], |
41 |
marc_template => [ |
42 |
'a, |x ; |v. |n, |p ; |v', |
43 |
'a ; |v. |p ; |v', |
44 |
], |
45 |
); |
46 |
|
47 |
Returns number of records produced. |
48 |
|
49 |
=cut |
50 |
|
51 |
sub marc_template { |
52 |
my $args = {@_}; |
53 |
warn "## marc_template(",dump($args),")",$/ if $debug; |
54 |
|
55 |
foreach ( qw/subfields_rename isis_template marc_template/ ) { |
56 |
# warn "ref($_) = ",ref($args->{$_}) if $debug; |
57 |
die "$_ not ARRAY" if defined($args->{$_}) && ref($args->{$_}) ne 'ARRAY'; |
58 |
} |
59 |
|
60 |
die "marc_template needs isis_template or marc_template" |
61 |
if ! defined $args->{isis_template} && ! defined $args->{marc_template}; |
62 |
|
63 |
my $rec = WebPAC::Normalize::_get_rec() || die "_get_rec?"; |
64 |
my $r = $rec->{ $args->{from} } || return; |
65 |
die "record field ", $args->{from}, " isn't array ",dump( $rec ) unless (ref($r) eq 'ARRAY'); |
66 |
|
67 |
my @subfields_rename = @{ $args->{subfields_rename} }; |
68 |
# warn "### subfields_rename [$#subfields_rename] = ",dump( @subfields_rename ) if $debug; |
69 |
|
70 |
confess "need mapping in pairs for subfields_rename" |
71 |
if $#subfields_rename % 2 != 1; |
72 |
|
73 |
my ( $subfields_rename, $from_subfields, $to_subfields ); |
74 |
while ( my ( $from, $to ) = splice(@subfields_rename, 0, 2) ) { |
75 |
my ( $f, $t ) = ( |
76 |
$from_subfields->{ $from }++, |
77 |
$to_subfields->{ $to }++ |
78 |
); |
79 |
$subfields_rename->{ $from }->[ $f ] = [ $to => $t ]; |
80 |
} |
81 |
warn "### subfields_rename = ",dump( $subfields_rename ),$/ if $debug; |
82 |
warn "### from_subfields = ", dump( $from_subfields ),$/ if $debug; |
83 |
warn "### to_subfields = ", dump( $to_subfields ),$/ if $debug; |
84 |
|
85 |
our $_template; |
86 |
|
87 |
$_template->{isis}->{fields_re} = join('|', keys %$from_subfields ); |
88 |
$_template->{marc}->{fields_re} = join('|', keys %$to_subfields ); |
89 |
|
90 |
my @marc_out; |
91 |
|
92 |
sub _parse_template { |
93 |
my ( $name, $templates ) = @_; |
94 |
|
95 |
my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} ); |
96 |
|
97 |
foreach my $template ( @{ $templates } ) { |
98 |
our $count = {}; |
99 |
our @order = (); |
100 |
sub my_count { |
101 |
my $sf = shift; |
102 |
my $nr = $count->{$sf}++; |
103 |
push @order, [ $sf, $nr ]; |
104 |
return $sf . $nr; |
105 |
} |
106 |
my $pos_template = $template; |
107 |
$pos_template =~ s/($fields_re)/my_count($1)/ge; |
108 |
my $count_key = dump( $count ); |
109 |
warn "### template: |$template| -> |$pos_template| count = $count_key order = ",dump( @order ),$/ if $debug; |
110 |
$_template->{$name}->{pos}->{ $count_key } = $pos_template; |
111 |
$_template->{$name}->{order}->{ $pos_template } = [ @order ]; |
112 |
} |
113 |
warn "### from ",dump( $templates ), " using $fields_re created ", dump( $_template ),$/ if $debug; |
114 |
} |
115 |
|
116 |
_parse_template( 'marc', $args->{marc_template} ); |
117 |
_parse_template( 'isis', $args->{isis_template} ); |
118 |
warn "### _template = ",dump( $_template ),$/ if $debug; |
119 |
|
120 |
my $m; |
121 |
|
122 |
foreach my $r ( @{ $rec->{ $args->{from} } } ) { |
123 |
|
124 |
my $i1 = $r->{i1} || ' '; |
125 |
my $i2 = $r->{i2} || ' '; |
126 |
$m = [ $args->{to}, $i1, $i2 ]; |
127 |
|
128 |
warn "### r = ",dump( $r ),$/ if $debug; |
129 |
|
130 |
my ( $from_mapping, $to_mapping, $from_count, $to_count ); |
131 |
foreach my $sf ( keys %{$r} ) { |
132 |
# skip everything which isn't one char subfield (e.g. 'subfields') |
133 |
next unless $sf =~ m/^\w$/; |
134 |
my $nr = $from_count->{$sf}++; |
135 |
my $rename_to = $subfields_rename->{ $sf } || |
136 |
die "can't find subfield rename for $sf/$nr in ", dump( $subfields_rename ); |
137 |
warn "### rename $sf/$nr to ", dump( $rename_to->[$nr] ),$/ if $debug; |
138 |
my ( $to_sf, $to_nr ) = @{ $rename_to->[$nr] }; |
139 |
$from_mapping->{ $sf }->[ $nr ] = [ $to_sf => $to_nr ]; |
140 |
$to_mapping->{ $to_sf }->[ $to_nr ] = [ $sf => $nr ]; |
141 |
|
142 |
$to_count->{ $to_sf }++; |
143 |
} |
144 |
|
145 |
warn "### to_mapping = ",dump( $to_mapping ),$/ if $debug; |
146 |
|
147 |
my $count_key = { |
148 |
from => dump( $from_count ), |
149 |
to => dump( $to_count), |
150 |
}; |
151 |
|
152 |
warn "### count_key = ",dump( $count_key ),$/ if $debug; |
153 |
|
154 |
my $processed_templates = 0; |
155 |
|
156 |
# this defines order of traversal |
157 |
foreach ( qw/isis:from marc:to/ ) { |
158 |
my ($name,$count_name) = split(/:/); |
159 |
|
160 |
my $ckey = $count_key->{$count_name} || die "can't find count_key $count_name in ",dump( $count_key ); |
161 |
|
162 |
my $template = $_template->{$name}->{pos}->{ $ckey } || next; |
163 |
$processed_templates++; |
164 |
|
165 |
warn "### traverse $name $count_name selected template: |$template|\n",$/ if $debug; |
166 |
|
167 |
our $fill_in = {}; |
168 |
|
169 |
my @templates = split(/\|/, $template ); |
170 |
@templates = ( $template ); |
171 |
|
172 |
foreach my $sf ( @templates ) { |
173 |
sub fill_in { |
174 |
my ( $name, $r, $sf, $nr ) = @_; |
175 |
my ( $from_sf, $from_nr ); |
176 |
if ( $name eq 'marc' ) { |
177 |
( $from_sf, $from_nr ) = @{ $to_mapping->{$sf}->[$nr] }; |
178 |
} else { |
179 |
( $from_sf, $from_nr ) = ( $sf, $nr ); |
180 |
} |
181 |
my $v = $r->{ $from_sf }; # || die "no $from_sf/$from_nr"; |
182 |
warn "#### fill_in( $sf, $nr ) = $from_sf/$from_nr >>>> ",dump( $v ),$/ if $debug; |
183 |
if ( ref( $v ) eq 'ARRAY' ) { |
184 |
$fill_in->{$sf}->[$nr] = $v->[$from_nr]; |
185 |
return $v->[$from_nr]; |
186 |
} elsif ( $from_nr == 0 ) { |
187 |
$fill_in->{$sf}->[$nr] = $v; |
188 |
return $v; |
189 |
} else { |
190 |
die "requested subfield $from_sf/$from_nr but it's ",dump( $v ); |
191 |
} |
192 |
} |
193 |
my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} ); |
194 |
warn "#### $sf <<<< $fields_re\n",$/ if $debug; |
195 |
$sf =~ s/($fields_re)(\d+)/fill_in($name,$r,$1,$2)/ge; |
196 |
warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/ if $debug; |
197 |
} |
198 |
|
199 |
warn "## template: |$template|\n## _template->$name = ",dump( $_template->{$name} ),$/ if $debug; |
200 |
|
201 |
foreach my $sf ( @{ $_template->{$name}->{order}->{$template} } ) { |
202 |
my ( $sf, $nr ) = @$sf; |
203 |
my $v = $fill_in->{$sf}->[$nr] || die "can't find fill_in $sf/$nr"; |
204 |
if ( $name eq 'isis') { |
205 |
( $sf, $nr ) = @{ $from_mapping->{$sf}->[$nr] }; |
206 |
} |
207 |
warn "++ $sf/$nr |$v|\n" if $debug; |
208 |
push @$m, ( $sf, $v ); |
209 |
} |
210 |
|
211 |
warn "#### >>>> created MARC record: ", dump( $m ),$/ if $debug; |
212 |
|
213 |
push @marc_out, $m; |
214 |
} |
215 |
|
216 |
die "I don't have template for fields ",dump( $count_key ), "\n## available templates\n", dump( $_template ) unless $processed_templates; |
217 |
warn ">>> $processed_templates templates applied to data\n",$/ if $debug; |
218 |
} |
219 |
|
220 |
|
221 |
my $recs = 0; |
222 |
|
223 |
foreach my $marc ( @marc_out ) { |
224 |
warn "+++ ",dump( $marc ),$/ if $debug; |
225 |
WebPAC::Normalize::_marc_push( $marc ); |
226 |
$recs++; |
227 |
} |
228 |
|
229 |
warn "### marc_template produced $recs MARC records: ",dump( @marc_out ),$/ if $debug; |
230 |
|
231 |
return $recs; |
232 |
} |
233 |
|
234 |
1; |