/[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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1035 - (hide annotations)
Mon Nov 12 10:16:16 2007 UTC (16 years, 5 months ago) by dpavlin
File size: 7310 byte(s)
 r1608@llin:  dpavlin | 2007-11-12 11:16:13 +0100
 fix debug output

1 dpavlin 1021 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 dpavlin 1031 my $debug = 0;
16 dpavlin 1028
17 dpavlin 1021 =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 dpavlin 1026 Returns number of records produced.
48    
49 dpavlin 1021 =cut
50    
51     sub marc_template {
52     my $args = {@_};
53 dpavlin 1028 warn "## marc_template(",dump($args),")",$/ if $debug;
54 dpavlin 1021
55     foreach ( qw/subfields_rename isis_template marc_template/ ) {
56 dpavlin 1028 # warn "ref($_) = ",ref($args->{$_}) if $debug;
57 dpavlin 1021 die "$_ 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 = @{ $args->{subfields_rename} };
68 dpavlin 1028 # warn "### subfields_rename [$#subfields_rename] = ",dump( @subfields_rename ) if $debug;
69 dpavlin 1021
70     confess "need mapping in pairs for subfields_rename"
71     if $#subfields_rename % 2 != 1;
72    
73 dpavlin 1029 my ( $subfields_rename, $from_subfields );
74     our $to_subfields = {};
75 dpavlin 1021 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 dpavlin 1028 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 dpavlin 1021
86 dpavlin 1025 our $_template;
87 dpavlin 1021
88 dpavlin 1026 $_template->{isis}->{fields_re} = join('|', keys %$from_subfields );
89     $_template->{marc}->{fields_re} = join('|', keys %$to_subfields );
90 dpavlin 1025
91 dpavlin 1021 my @marc_out;
92    
93 dpavlin 1023 sub _parse_template {
94 dpavlin 1024 my ( $name, $templates ) = @_;
95 dpavlin 1021
96 dpavlin 1026 my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
97 dpavlin 1025
98 dpavlin 1023 foreach my $template ( @{ $templates } ) {
99     our $count = {};
100 dpavlin 1025 our @order = ();
101 dpavlin 1023 sub my_count {
102     my $sf = shift;
103     my $nr = $count->{$sf}++;
104 dpavlin 1025 push @order, [ $sf, $nr ];
105 dpavlin 1023 return $sf . $nr;
106     }
107     my $pos_template = $template;
108     $pos_template =~ s/($fields_re)/my_count($1)/ge;
109     my $count_key = dump( $count );
110 dpavlin 1028 warn "### template: |$template| -> |$pos_template| count = $count_key order = ",dump( @order ),$/ if $debug;
111 dpavlin 1024 $_template->{$name}->{pos}->{ $count_key } = $pos_template;
112 dpavlin 1025 $_template->{$name}->{order}->{ $pos_template } = [ @order ];
113 dpavlin 1021 }
114 dpavlin 1028 warn "### from ",dump( $templates ), " using $fields_re created ", dump( $_template ),$/ if $debug;
115 dpavlin 1021 }
116    
117 dpavlin 1024 _parse_template( 'marc', $args->{marc_template} );
118     _parse_template( 'isis', $args->{isis_template} );
119 dpavlin 1028 warn "### _template = ",dump( $_template ),$/ if $debug;
120 dpavlin 1023
121 dpavlin 1021 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 dpavlin 1028 warn "### r = ",dump( $r ),$/ if $debug;
130 dpavlin 1021
131 dpavlin 1026 my ( $from_mapping, $to_mapping, $from_count, $to_count );
132 dpavlin 1029 foreach my $from_sf ( keys %{$r} ) {
133 dpavlin 1021 # skip everything which isn't one char subfield (e.g. 'subfields')
134 dpavlin 1029 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 dpavlin 1021
141 dpavlin 1029 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 dpavlin 1021 }
146    
147 dpavlin 1029 warn "### from_mapping = ",dump( $from_mapping ), "\n### to_mapping = ",dump( $to_mapping ),$/ if $debug;
148 dpavlin 1021
149 dpavlin 1025 my $count_key = {
150     from => dump( $from_count ),
151     to => dump( $to_count),
152     };
153 dpavlin 1021
154 dpavlin 1028 warn "### count_key = ",dump( $count_key ),$/ if $debug;
155 dpavlin 1021
156 dpavlin 1025 my $processed_templates = 0;
157 dpavlin 1021
158 dpavlin 1025 # this defines order of traversal
159     foreach ( qw/isis:from marc:to/ ) {
160     my ($name,$count_name) = split(/:/);
161 dpavlin 1021
162 dpavlin 1025 my $ckey = $count_key->{$count_name} || die "can't find count_key $count_name in ",dump( $count_key );
163 dpavlin 1021
164 dpavlin 1025 my $template = $_template->{$name}->{pos}->{ $ckey } || next;
165     $processed_templates++;
166 dpavlin 1021
167 dpavlin 1031 warn "### traverse $name $count_name selected template: |$template|\n" if $debug;
168 dpavlin 1025
169     our $fill_in = {};
170    
171     my @templates = split(/\|/, $template );
172 dpavlin 1031 @templates = ( $template ) unless @templates;
173 dpavlin 1025
174 dpavlin 1035 warn "### templates = ",dump( @templates ),$/ if $debug;
175 dpavlin 1031
176 dpavlin 1025 foreach my $sf ( @templates ) {
177     sub fill_in {
178 dpavlin 1031 my ( $name, $r, $pre, $sf, $nr, $post ) = @_;
179     warn "#### fill_in( $name, r, '$pre', $sf, $nr, '$post' )\n" if $debug;
180 dpavlin 1028 my ( $from_sf, $from_nr );
181 dpavlin 1025 if ( $name eq 'marc' ) {
182 dpavlin 1029 die "no $sf/$nr in to_mapping: ",dump( $to_mapping ), " form record ",dump( $r ) unless defined $to_mapping->{$sf}->[$nr];
183 dpavlin 1026 ( $from_sf, $from_nr ) = @{ $to_mapping->{$sf}->[$nr] };
184 dpavlin 1025 } 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 dpavlin 1031 $v = $pre . $v->[$from_nr] . $post;
190 dpavlin 1025 } elsif ( $from_nr == 0 ) {
191 dpavlin 1031 $v = $pre . $v . $post;
192 dpavlin 1025 } else {
193     die "requested subfield $from_sf/$from_nr but it's ",dump( $v );
194     }
195 dpavlin 1031 warn "#### fill_in( $sf, $nr ) = $from_sf/$from_nr >>>> ",dump( $v ),$/ if $debug;
196     $fill_in->{$sf}->[$nr] = $v;
197     return $v;
198 dpavlin 1021 }
199 dpavlin 1026 my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
200 dpavlin 1031 warn "#### $sf <<<< $fields_re\n" if $debug;
201     $sf =~ s/^(.*?)($fields_re)(\d+)(.*?)$/fill_in($name,$r,$1,$2,$3,$4)/ge;
202 dpavlin 1028 warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/ if $debug;
203 dpavlin 1021 }
204    
205 dpavlin 1028 warn "## template: |$template|\n## _template->$name = ",dump( $_template->{$name} ),$/ if $debug;
206 dpavlin 1021
207 dpavlin 1025 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 dpavlin 1026 if ( $name eq 'isis') {
211     ( $sf, $nr ) = @{ $from_mapping->{$sf}->[$nr] };
212     }
213 dpavlin 1028 warn "++ $sf/$nr |$v|\n" if $debug;
214 dpavlin 1025 push @$m, ( $sf, $v );
215     }
216 dpavlin 1021
217 dpavlin 1028 warn "#### >>>> created MARC record: ", dump( $m ),$/ if $debug;
218 dpavlin 1025
219     push @marc_out, $m;
220 dpavlin 1029
221     last;
222 dpavlin 1025 }
223    
224     die "I don't have template for fields ",dump( $count_key ), "\n## available templates\n", dump( $_template ) unless $processed_templates;
225 dpavlin 1028 warn ">>> $processed_templates templates applied to data\n",$/ if $debug;
226 dpavlin 1021 }
227    
228    
229 dpavlin 1026 my $recs = 0;
230 dpavlin 1025
231 dpavlin 1021 foreach my $marc ( @marc_out ) {
232 dpavlin 1028 warn "+++ ",dump( $marc ),$/ if $debug;
233 dpavlin 1021 WebPAC::Normalize::_marc_push( $marc );
234 dpavlin 1026 $recs++;
235 dpavlin 1021 }
236 dpavlin 1026
237 dpavlin 1028 warn "### marc_template produced $recs MARC records: ",dump( @marc_out ),$/ if $debug;
238    
239     return $recs;
240 dpavlin 1021 }
241    
242     1;

  ViewVC Help
Powered by ViewVC 1.1.26