12 |
|
|
13 |
use WebPAC::Normalize; |
use WebPAC::Normalize; |
14 |
|
|
15 |
|
my $debug = 0; |
16 |
|
|
17 |
=head1 NAME |
=head1 NAME |
18 |
|
|
19 |
WebPAC::Normalize::MARC - create MARC/ISO2709 records |
WebPAC::Normalize::MARC - create MARC/ISO2709 records |
44 |
], |
], |
45 |
); |
); |
46 |
|
|
47 |
|
Returns number of records produced. |
48 |
|
|
49 |
=cut |
=cut |
50 |
|
|
51 |
sub marc_template { |
sub marc_template { |
52 |
my $args = {@_}; |
my $args = {@_}; |
53 |
warn "## marc_template(",dump($args),")"; |
warn "## marc_template(",dump($args),")",$/ if $debug; |
54 |
|
|
55 |
foreach ( qw/subfields_rename isis_template marc_template/ ) { |
foreach ( qw/subfields_rename isis_template marc_template/ ) { |
56 |
# warn "ref($_) = ",ref($args->{$_}); |
# warn "ref($_) = ",ref($args->{$_}) if $debug; |
57 |
die "$_ not ARRAY" if defined($args->{$_}) && ref($args->{$_}) ne 'ARRAY'; |
die "$_ not ARRAY" if defined($args->{$_}) && ref($args->{$_}) ne 'ARRAY'; |
58 |
} |
} |
59 |
|
|
65 |
die "record field ", $args->{from}, " isn't array ",dump( $rec ) unless (ref($r) eq 'ARRAY'); |
die "record field ", $args->{from}, " isn't array ",dump( $rec ) unless (ref($r) eq 'ARRAY'); |
66 |
|
|
67 |
my @subfields_rename = @{ $args->{subfields_rename} }; |
my @subfields_rename = @{ $args->{subfields_rename} }; |
68 |
# warn "### subfields_rename [$#subfields_rename] = ",dump( @subfields_rename ); |
# warn "### subfields_rename [$#subfields_rename] = ",dump( @subfields_rename ) if $debug; |
69 |
|
|
70 |
confess "need mapping in pairs for subfields_rename" |
confess "need mapping in pairs for subfields_rename" |
71 |
if $#subfields_rename % 2 != 1; |
if $#subfields_rename % 2 != 1; |
72 |
|
|
73 |
my ( $subfields_rename, $from_subfields, $to_subfields ); |
my ( $subfields_rename, $from_subfields ); |
74 |
|
our $to_subfields = {}; |
75 |
while ( my ( $from, $to ) = splice(@subfields_rename, 0, 2) ) { |
while ( my ( $from, $to ) = splice(@subfields_rename, 0, 2) ) { |
76 |
my ( $f, $t ) = ( |
my ( $f, $t ) = ( |
77 |
$from_subfields->{ $from }++, |
$from_subfields->{ $from }++, |
79 |
); |
); |
80 |
$subfields_rename->{ $from }->[ $f ] = [ $to => $t ]; |
$subfields_rename->{ $from }->[ $f ] = [ $to => $t ]; |
81 |
} |
} |
82 |
warn "### subfields_rename = ",dump( $subfields_rename ),$/; |
warn "### subfields_rename = ",dump( $subfields_rename ),$/ if $debug; |
83 |
warn "### from_subfields = ", dump( $from_subfields ),$/; |
warn "### from_subfields = ", dump( $from_subfields ),$/ if $debug; |
84 |
warn "### to_subfields = ", dump( $to_subfields ),$/; |
warn "### to_subfields = ", dump( $to_subfields ),$/ if $debug; |
85 |
|
|
86 |
our $_template; |
our $_template; |
87 |
|
|
88 |
$_template->{fields_re} = { |
$_template->{isis}->{fields_re} = join('|', keys %$from_subfields ); |
89 |
isis => join('|', keys %$from_subfields ), |
$_template->{marc}->{fields_re} = join('|', keys %$to_subfields ); |
|
marc => join('|', keys %$to_subfields ), |
|
|
}; |
|
90 |
|
|
91 |
my @marc_out; |
my @marc_out; |
92 |
|
|
93 |
sub _parse_template { |
sub _parse_template { |
94 |
my ( $name, $templates ) = @_; |
my ( $name, $templates ) = @_; |
95 |
|
|
96 |
my $fields_re = $_template->{fields_re}->{ $name } || die "can't find $name in ",dump( $_template->{fields_re} ); |
my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} ); |
97 |
|
|
98 |
foreach my $template ( @{ $templates } ) { |
foreach my $template ( @{ $templates } ) { |
99 |
our $count = {}; |
our $count = {}; |
107 |
my $pos_template = $template; |
my $pos_template = $template; |
108 |
$pos_template =~ s/($fields_re)/my_count($1)/ge; |
$pos_template =~ s/($fields_re)/my_count($1)/ge; |
109 |
my $count_key = dump( $count ); |
my $count_key = dump( $count ); |
110 |
warn "### template: |$template| -> |$pos_template| count = $count_key order = ",dump( @order ),$/; |
warn "### template: |$template| -> |$pos_template| count = $count_key order = ",dump( @order ),$/ if $debug; |
111 |
$_template->{$name}->{pos}->{ $count_key } = $pos_template; |
$_template->{$name}->{pos}->{ $count_key } = $pos_template; |
112 |
$_template->{$name}->{order}->{ $pos_template } = [ @order ]; |
$_template->{$name}->{order}->{ $pos_template } = [ @order ]; |
113 |
} |
} |
114 |
warn "### from ",dump( $templates ), " using $fields_re created ", dump( $_template ); |
warn "### from ",dump( $templates ), " using $fields_re created ", dump( $_template ),$/ if $debug; |
115 |
} |
} |
116 |
|
|
117 |
_parse_template( 'marc', $args->{marc_template} ); |
_parse_template( 'marc', $args->{marc_template} ); |
118 |
_parse_template( 'isis', $args->{isis_template} ); |
_parse_template( 'isis', $args->{isis_template} ); |
119 |
warn "### _template = ",dump( $_template ); |
warn "### _template = ",dump( $_template ),$/ if $debug; |
120 |
|
|
121 |
my $m; |
my $m; |
122 |
|
|
126 |
my $i2 = $r->{i2} || ' '; |
my $i2 = $r->{i2} || ' '; |
127 |
$m = [ $args->{to}, $i1, $i2 ]; |
$m = [ $args->{to}, $i1, $i2 ]; |
128 |
|
|
129 |
warn "### r = ",dump( $r ); |
warn "### r = ",dump( $r ),$/ if $debug; |
130 |
|
|
131 |
my ( $new_r, $from_count, $to_count ); |
my ( $from_mapping, $to_mapping, $from_count, $to_count ); |
132 |
foreach my $sf ( keys %{$r} ) { |
foreach my $from_sf ( keys %{$r} ) { |
133 |
# skip everything which isn't one char subfield (e.g. 'subfields') |
# skip everything which isn't one char subfield (e.g. 'subfields') |
134 |
next unless $sf =~ m/^\w$/; |
next unless $from_sf =~ m/^\w$/; |
135 |
my $nr = $from_count->{$sf}++; |
my $from_nr = $from_count->{$from_sf}++; |
136 |
my $rename_to = $subfields_rename->{ $sf } || |
my $rename_to = $subfields_rename->{ $from_sf } || |
137 |
die "can't find subfield rename for $sf/$nr in ", dump( $subfields_rename ); |
die "can't find subfield rename for $from_sf/$from_nr in ", dump( $subfields_rename ); |
138 |
warn "### rename $sf/$nr to ", dump( $rename_to->[$nr] ), $/; |
my ( $to_sf, $to_nr ) = @{ $rename_to->[$from_nr] }; |
139 |
my ( $to_sf, $to_nr ) = @{ $rename_to->[$nr] }; |
$to_mapping->{ $to_sf }->[ $to_nr ] = [ $from_sf => $from_nr ]; |
140 |
$new_r->{ $to_sf }->[ $to_nr ] = [ $sf => $nr ]; |
|
141 |
|
my $to_nr2 = $to_count->{ $to_sf }++; |
142 |
|
$from_mapping->{ $from_sf }->[ $from_nr ] = [ $to_sf => $to_nr2 ]; |
143 |
|
|
144 |
$to_count->{ $to_sf }++; |
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 "### new_r = ",dump( $new_r ); |
warn "### from_mapping = ",dump( $from_mapping ), "\n### to_mapping = ",dump( $to_mapping ),$/ if $debug; |
148 |
|
|
149 |
my $count_key = { |
my $count_key = { |
150 |
from => dump( $from_count ), |
from => dump( $from_count ), |
151 |
to => dump( $to_count), |
to => dump( $to_count), |
152 |
}; |
}; |
153 |
|
|
154 |
warn "### count_key = ",dump( $count_key ), $/; |
warn "### count_key = ",dump( $count_key ),$/ if $debug; |
155 |
|
|
156 |
my $processed_templates = 0; |
my $processed_templates = 0; |
157 |
|
|
158 |
# this defines order of traversal |
# this defines order of traversal |
159 |
foreach ( qw/isis:from marc:to/ ) { |
foreach ( qw/isis:from marc:to/ ) { |
160 |
my ($name,$count_name) = split(/:/); |
my ($name,$count_name) = split(/:/); |
|
warn "## traverse $name $count_name\n"; |
|
161 |
|
|
162 |
my $ckey = $count_key->{$count_name} || die "can't find count_key $count_name in ",dump( $count_key ); |
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; |
my $template = $_template->{$name}->{pos}->{ $ckey } || next; |
165 |
$processed_templates++; |
$processed_templates++; |
166 |
|
|
167 |
warn "### selected template: |$template|\n"; |
warn "### traverse $name $count_name selected template: |$template|\n" if $debug; |
168 |
|
|
169 |
our $fill_in = {}; |
our $fill_in = {}; |
170 |
|
|
171 |
my @templates = split(/\|/, $template ); |
my @templates = split(/\|/, $template ); |
172 |
@templates = ( $template ); |
@templates = ( $template ) unless @templates; |
173 |
|
|
174 |
|
warn "### templates = ",dump( @templates ),$/ if $debug; |
175 |
|
|
176 |
foreach my $sf ( @templates ) { |
foreach my $sf ( @templates ) { |
177 |
sub fill_in { |
sub fill_in { |
178 |
my ( $name, $r, $sf, $nr ) = @_; |
my ( $name, $r, $pre, $sf, $nr, $post ) = @_; |
179 |
my ( $from_sf, $from_nr, $v ); |
warn "#### fill_in( $name, r, '$pre', $sf, $nr, '$post' )\n" if $debug; |
180 |
|
my ( $from_sf, $from_nr ); |
181 |
if ( $name eq 'marc' ) { |
if ( $name eq 'marc' ) { |
182 |
( $from_sf, $from_nr ) = @{ $new_r->{$sf}->[$nr] }; |
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 { |
} else { |
185 |
( $from_sf, $from_nr ) = ( $sf, $nr ); |
( $from_sf, $from_nr ) = ( $sf, $nr ); |
186 |
} |
} |
187 |
my $v = $r->{ $from_sf }; # || die "no $from_sf/$from_nr"; |
my $v = $r->{ $from_sf }; # || die "no $from_sf/$from_nr"; |
|
warn "#### fill_in( $sf, $nr ) = $from_sf/$from_nr >>>> ",dump( $v ), $/; |
|
188 |
if ( ref( $v ) eq 'ARRAY' ) { |
if ( ref( $v ) eq 'ARRAY' ) { |
189 |
$fill_in->{$sf}->[$nr] = $v->[$from_nr]; |
$v = $pre . $v->[$from_nr] . $post; |
|
return $v->[$from_nr]; |
|
190 |
} elsif ( $from_nr == 0 ) { |
} elsif ( $from_nr == 0 ) { |
191 |
$fill_in->{$sf}->[$nr] = $v; |
$v = $pre . $v . $post; |
|
return $v; |
|
192 |
} else { |
} else { |
193 |
die "requested subfield $from_sf/$from_nr but it's ",dump( $v ); |
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->{fields_re}->{ $name } || die "can't find $name in ",dump( $_template->{fields_re} ); |
my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} ); |
200 |
warn "#### $sf <<<< $fields_re\n"; |
warn "#### $sf <<<< $fields_re\n" if $debug; |
201 |
$sf =~ s/($fields_re)(\d+)/fill_in($name,$r,$1,$2)/ge; |
$sf =~ s/^(.*?)($fields_re)(\d+)(.*?)$/fill_in($name,$r,$1,$2,$3,$4)/ge; |
202 |
warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/; |
warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/ if $debug; |
203 |
} |
} |
204 |
|
|
205 |
warn "## template: |$template|\n## _template->$name = ",dump( $_template->{$name} ); |
warn "## template: |$template|\n## _template->$name = ",dump( $_template->{$name} ),$/ if $debug; |
|
|
|
|
$sf_pos = $#m; |
|
206 |
|
|
207 |
foreach my $sf ( @{ $_template->{$name}->{order}->{$template} } ) { |
foreach my $sf ( @{ $_template->{$name}->{order}->{$template} } ) { |
208 |
my ( $sf, $nr ) = @$sf; |
my ( $sf, $nr ) = @$sf; |
209 |
my $v = $fill_in->{$sf}->[$nr] || die "can't find fill_in $sf/$nr"; |
my $v = $fill_in->{$sf}->[$nr] || die "can't find fill_in $sf/$nr"; |
210 |
warn "++ $sf/$nr |$v|\n"; |
if ( $name eq 'isis') { |
211 |
|
( $sf, $nr ) = @{ $from_mapping->{$sf}->[$nr] }; |
212 |
|
} |
213 |
|
warn "++ $sf/$nr |$v|\n" if $debug; |
214 |
push @$m, ( $sf, $v ); |
push @$m, ( $sf, $v ); |
215 |
} |
} |
216 |
|
|
217 |
warn "#### >>>> created MARC record: ", dump( $m ); |
warn "#### >>>> created MARC record: ", dump( $m ),$/ if $debug; |
218 |
|
|
219 |
push @marc_out, $m; |
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; |
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"; |
warn ">>> $processed_templates templates applied to data\n",$/ if $debug; |
226 |
} |
} |
227 |
|
|
228 |
|
|
229 |
warn "### marc_template produced following MARC records: ",dump( @marc_out ); |
my $recs = 0; |
230 |
|
|
231 |
foreach my $marc ( @marc_out ) { |
foreach my $marc ( @marc_out ) { |
232 |
warn "+++ ",dump( $marc ); |
warn "+++ ",dump( $marc ),$/ if $debug; |
233 |
WebPAC::Normalize::_marc_push( $marc ); |
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; |
1; |