/[webpac2]/trunk/lib/WebPAC/Normalize/MARC.pm
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Contents of /trunk/lib/WebPAC/Normalize/MARC.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1035 - (show annotations)
Mon Nov 12 10:16:16 2007 UTC (16 years, 4 months ago) by dpavlin
File size: 7310 byte(s)
 r1608@llin:  dpavlin | 2007-11-12 11:16:13 +0100
 fix debug output

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 );
74 our $to_subfields = {};
75 while ( my ( $from, $to ) = splice(@subfields_rename, 0, 2) ) {
76 my ( $f, $t ) = (
77 $from_subfields->{ $from }++,
78 $to_subfields->{ $to }++
79 );
80 $subfields_rename->{ $from }->[ $f ] = [ $to => $t ];
81 }
82 warn "### subfields_rename = ",dump( $subfields_rename ),$/ if $debug;
83 warn "### from_subfields = ", dump( $from_subfields ),$/ if $debug;
84 warn "### to_subfields = ", dump( $to_subfields ),$/ if $debug;
85
86 our $_template;
87
88 $_template->{isis}->{fields_re} = join('|', keys %$from_subfields );
89 $_template->{marc}->{fields_re} = join('|', keys %$to_subfields );
90
91 my @marc_out;
92
93 sub _parse_template {
94 my ( $name, $templates ) = @_;
95
96 my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
97
98 foreach my $template ( @{ $templates } ) {
99 our $count = {};
100 our @order = ();
101 sub my_count {
102 my $sf = shift;
103 my $nr = $count->{$sf}++;
104 push @order, [ $sf, $nr ];
105 return $sf . $nr;
106 }
107 my $pos_template = $template;
108 $pos_template =~ s/($fields_re)/my_count($1)/ge;
109 my $count_key = dump( $count );
110 warn "### template: |$template| -> |$pos_template| count = $count_key order = ",dump( @order ),$/ if $debug;
111 $_template->{$name}->{pos}->{ $count_key } = $pos_template;
112 $_template->{$name}->{order}->{ $pos_template } = [ @order ];
113 }
114 warn "### from ",dump( $templates ), " using $fields_re created ", dump( $_template ),$/ if $debug;
115 }
116
117 _parse_template( 'marc', $args->{marc_template} );
118 _parse_template( 'isis', $args->{isis_template} );
119 warn "### _template = ",dump( $_template ),$/ if $debug;
120
121 my $m;
122
123 foreach my $r ( @{ $rec->{ $args->{from} } } ) {
124
125 my $i1 = $r->{i1} || ' ';
126 my $i2 = $r->{i2} || ' ';
127 $m = [ $args->{to}, $i1, $i2 ];
128
129 warn "### r = ",dump( $r ),$/ if $debug;
130
131 my ( $from_mapping, $to_mapping, $from_count, $to_count );
132 foreach my $from_sf ( keys %{$r} ) {
133 # skip everything which isn't one char subfield (e.g. 'subfields')
134 next unless $from_sf =~ m/^\w$/;
135 my $from_nr = $from_count->{$from_sf}++;
136 my $rename_to = $subfields_rename->{ $from_sf } ||
137 die "can't find subfield rename for $from_sf/$from_nr in ", dump( $subfields_rename );
138 my ( $to_sf, $to_nr ) = @{ $rename_to->[$from_nr] };
139 $to_mapping->{ $to_sf }->[ $to_nr ] = [ $from_sf => $from_nr ];
140
141 my $to_nr2 = $to_count->{ $to_sf }++;
142 $from_mapping->{ $from_sf }->[ $from_nr ] = [ $to_sf => $to_nr2 ];
143
144 warn "### from $from_sf/$from_nr -> $to_sf/$to_nr\tto $from_sf/$from_nr -> $to_sf/$to_nr2\n" if $debug;
145 }
146
147 warn "### from_mapping = ",dump( $from_mapping ), "\n### to_mapping = ",dump( $to_mapping ),$/ if $debug;
148
149 my $count_key = {
150 from => dump( $from_count ),
151 to => dump( $to_count),
152 };
153
154 warn "### count_key = ",dump( $count_key ),$/ if $debug;
155
156 my $processed_templates = 0;
157
158 # this defines order of traversal
159 foreach ( qw/isis:from marc:to/ ) {
160 my ($name,$count_name) = split(/:/);
161
162 my $ckey = $count_key->{$count_name} || die "can't find count_key $count_name in ",dump( $count_key );
163
164 my $template = $_template->{$name}->{pos}->{ $ckey } || next;
165 $processed_templates++;
166
167 warn "### traverse $name $count_name selected template: |$template|\n" if $debug;
168
169 our $fill_in = {};
170
171 my @templates = split(/\|/, $template );
172 @templates = ( $template ) unless @templates;
173
174 warn "### templates = ",dump( @templates ),$/ if $debug;
175
176 foreach my $sf ( @templates ) {
177 sub fill_in {
178 my ( $name, $r, $pre, $sf, $nr, $post ) = @_;
179 warn "#### fill_in( $name, r, '$pre', $sf, $nr, '$post' )\n" if $debug;
180 my ( $from_sf, $from_nr );
181 if ( $name eq 'marc' ) {
182 die "no $sf/$nr in to_mapping: ",dump( $to_mapping ), " form record ",dump( $r ) unless defined $to_mapping->{$sf}->[$nr];
183 ( $from_sf, $from_nr ) = @{ $to_mapping->{$sf}->[$nr] };
184 } else {
185 ( $from_sf, $from_nr ) = ( $sf, $nr );
186 }
187 my $v = $r->{ $from_sf }; # || die "no $from_sf/$from_nr";
188 if ( ref( $v ) eq 'ARRAY' ) {
189 $v = $pre . $v->[$from_nr] . $post;
190 } elsif ( $from_nr == 0 ) {
191 $v = $pre . $v . $post;
192 } else {
193 die "requested subfield $from_sf/$from_nr but it's ",dump( $v );
194 }
195 warn "#### fill_in( $sf, $nr ) = $from_sf/$from_nr >>>> ",dump( $v ),$/ if $debug;
196 $fill_in->{$sf}->[$nr] = $v;
197 return $v;
198 }
199 my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
200 warn "#### $sf <<<< $fields_re\n" if $debug;
201 $sf =~ s/^(.*?)($fields_re)(\d+)(.*?)$/fill_in($name,$r,$1,$2,$3,$4)/ge;
202 warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/ if $debug;
203 }
204
205 warn "## template: |$template|\n## _template->$name = ",dump( $_template->{$name} ),$/ if $debug;
206
207 foreach my $sf ( @{ $_template->{$name}->{order}->{$template} } ) {
208 my ( $sf, $nr ) = @$sf;
209 my $v = $fill_in->{$sf}->[$nr] || die "can't find fill_in $sf/$nr";
210 if ( $name eq 'isis') {
211 ( $sf, $nr ) = @{ $from_mapping->{$sf}->[$nr] };
212 }
213 warn "++ $sf/$nr |$v|\n" if $debug;
214 push @$m, ( $sf, $v );
215 }
216
217 warn "#### >>>> created MARC record: ", dump( $m ),$/ if $debug;
218
219 push @marc_out, $m;
220
221 last;
222 }
223
224 die "I don't have template for fields ",dump( $count_key ), "\n## available templates\n", dump( $_template ) unless $processed_templates;
225 warn ">>> $processed_templates templates applied to data\n",$/ if $debug;
226 }
227
228
229 my $recs = 0;
230
231 foreach my $marc ( @marc_out ) {
232 warn "+++ ",dump( $marc ),$/ if $debug;
233 WebPAC::Normalize::_marc_push( $marc );
234 $recs++;
235 }
236
237 warn "### marc_template produced $recs MARC records: ",dump( @marc_out ),$/ if $debug;
238
239 return $recs;
240 }
241
242 1;

  ViewVC Help
Powered by ViewVC 1.1.26