produce unimarc (without conversion) of usmarc (with conversion)

git-svn-id: svn+ssh://llin.lib/home/dpavlin/private/svn/Z3950-HTML-Scraper@9 ae73d1a6-5fa4-44a9-8f13-f281fb455051

diff --git a/COBISS.pm b/COBISS.pm
index f3c698d..e55eeaf 100644 (file)
--- a/COBISS.pm
+++ b/COBISS.pm
@@ -5,6 +5,7 @@ use strict;
 
 use WWW::Mechanize;
 use MARC::Record;
+use Data::Dump qw/dump/;
 
 binmode STDOUT, ':utf8';
 
@@ -17,8 +18,8 @@ my $cobiss_marc21 = {
         205  => { a => [  250 , 'a' ] },
         210  => {
                a => [  250 , 'a' ],
-               c => [  260 , 'b' ],
-               d => [  260 , 'c' ],
+               c => [  250 , 'b' ],
+               d => [  250 , 'c' ],
        },
        215 => {
                a => [  300 , 'a' ],
@@ -113,8 +114,12 @@ diag "in COMARC format";
 }
 
 
-sub fetch_marc {
-       my ($self) = @_;
+sub fetch_rec {
+       my ($self,$format) = @_;
+
+       $format ||= 'unimarc';
+
+       die "unknown format: $format" unless $format =~ m{(uni|us)marc};
 
        my $comarc;
 
@@ -124,7 +129,7 @@ sub fetch_marc {
                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;
@@ -143,20 +148,52 @@ diag "fetch_marc $nr [$id]";
                        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");
+               my $path = "marc/$id.$format";
+
+               open($out, '>:utf8', $path);
                print $out $marc->as_usmarc;
                close($out);
 
+               diag "created $path ", -s $path, " bytes";
+
                diag $marc->as_formatted;
 
                $nr++;

diff --git a/server.pl b/server.pl
index 6d26b06..1e17f2c 100755 (executable)
--- a/server.pl
+++ b/server.pl
@@ -103,7 +103,11 @@ diag Dumper( $this );
     }
     elsif ( $req_form eq &Net::Z3950::OID::unimarc ) { # FIXME convert to usmarc
         $this->{REP_FORM} = &Net::Z3950::OID::unimarc;
-        $this->{RECORD} = COBISS->fetch_marc;
+        $this->{RECORD} = COBISS->fetch_rec('unimarc');
+    }
+    elsif ( $req_form eq &Net::Z3950::OID::usmarc ) {  # FIXME convert to usmarc
+        $this->{REP_FORM} = &Net::Z3950::OID::usmarc;
+        $this->{RECORD} = COBISS->fetch_rec('usmarc');
     }
     else {    ## Unsupported record format
         $this->{ERR_CODE} = 239;