use warnings;
use strict;
-use WWW::Mechanize;
use MARC::Record;
+use Data::Dump qw/dump/;
-binmode STDOUT, ':utf8';
+use base 'Scraper';
my $cobiss_marc21 = {
'010' => { a => [ '020', 'a' ] },
},
205 => { a => [ 250 , 'a' ] },
210 => {
- a => [ 250 , 'a' ],
+ a => [ 260 , 'a' ],
c => [ 260 , 'b' ],
d => [ 260 , 'c' ],
},
},
};
-our $mech = WWW::Mechanize->new();
-our $hits;
-
sub diag {
print "# ", @_, $/;
}
# @attr 1=1007 standard-id
# @attr 1=1016 any
-our $usemap = {
+sub usemap {{
8 => 'BN', # FIXME check
7 => 'SN', # FIXME check
4 => 'TI',
# 1007 => '',
# 1016 => '',
-};
-
-sub usemap {
- my $f = shift || die;
- $usemap->{$f};
-}
+}};
sub search {
my ( $self, $query ) = @_;
diag "get $url";
+ my $mech = $self->{mech} || die "no mech?";
+
+ my $hits;
$mech->get( $url );
diag "got session";
diag "in COMARC format";
$mech->follow_link( url_regex => qr/fmt=13/ );
+
+ return $hits;
}
-sub fetch_marc {
- my ($self) = @_;
+sub next_marc {
+ my ($self,$format) = @_;
+
+ my $mech = $self->{mech} || die "no mech?";
+
+ $format ||= 'unimarc';
+
+ die "unknown format: $format" unless $format =~ m{(uni|us)marc};
my $comarc;
my $nr = $2;
my $id = $3;
-diag "fetch_marc $nr [$id]";
+diag "fetch_marc $nr [$id] $format";
$comarc =~ s{</?b>}{}gs;
$comarc =~ s{<font[^>]*>}{<s>}gs;
if ( $line !~ s{^(\d\d\d)([01 ])([01 ])}{} ) {
diag "SKIP: $line";
} else {
+ our @f = ( $1, $2, $3 );
$line .= "<eol>";
- our @f = ( $1, $2, $3 );
- sub sf { push @f, @_; }
- $line =~ s{<s>(\w)<e>([^<]+)\s*}{sf($1, $2)}ges;
- diag "f:", join('|', @f), " left: |$line|";
- $marc->add_fields( @f );
+ if ( $format eq 'unimarc' ) {
+
+ diag dump(@f), "line: $line";
+ sub sf_uni {
+ warn "sf ",dump(@_);
+ push @f, @_;
+ }
+ $line =~ s{<s>(\w)<e>([^<]+)\s*}{sf_uni($1, $2)}ges;
+ diag "f:", dump(@f), " left: |$line|";
+ $marc->add_fields( @f );
+
+ } elsif ( $format eq 'usmarc' ) {
+
+ my ( $f, $i1, $i2 ) = @f;
+
+ our $out = {};
+
+ sub sf_us {
+ my ($f,$sf,$v) = @_;
+ if ( my $m = $cobiss_marc21->{$f}->{$sf} ) {
+ push @{ $out->{ $m->[0] } }, ( $m->[1], $v );
+ }
+ return;
+ }
+ $line =~ s{<s>(\w)<e>([^<]+)\s*}{sf_us($f,$1, $2)}ges;
+
+ diag "converted marc21 ",dump( $out );
+
+ foreach my $f ( keys %$out ) {
+ $marc->add_fields( $f, $i1, $i2, @{ $out->{$f} } );
+ }
+ }
}
}
- open(my $out, '>:utf8', "marc/$id");
- print $out $marc->as_usmarc;
- close($out);
-
+ $self->save_marc( $id, $marc->as_usmarc );
diag $marc->as_formatted;
$nr++;