use Scraper module

[Biblio-Z3950.git] / Aleph.pm

diff --git a/Aleph.pm b/Aleph.pm
index 3ad6287..5c88d2e 100644 (file)
--- a/Aleph.pm
+++ b/Aleph.pm
@@ -6,6 +6,7 @@ use strict;
 use MARC::Record;
 use Data::Dump qw/dump/;
 
+use Scraper;
 use base 'Scraper';
 
 sub diag {
@@ -15,9 +16,9 @@ sub diag {
 # Koha Z39.50 query:
 #
 # Bib-1 @and @and @and @and @and @and @and @or
-# @attr 1=8 isbn-issn 
-# @attr 1=7 isbn-issn 
 # @attr 1=4 title 
+# @attr 1=7 isbn
+# @attr 1=8 issn 
 # @attr 1=1003 author 
 # @attr 1=16 dewey 
 # @attr 1=21 subject-holding 
@@ -50,13 +51,15 @@ sub diag {
 # WYR - Godina izdavanja
 
 sub usemap {{
-       4               => 'WTI',
-       1003    => 'WTI',
-       16              => 'CU',
-       21              => 'SU',
-#      12              => '',
+       4               => 'WTI=',
+       7               => 'ISBN=',
+       8               => 'ISSN=',
+       1003    => 'AUT=',
+       16              => 'DDC=',
+       21              => 'SUB=',
+       12              => 'LCN=',
 #      1007    => '',
-#      1016    => '',
+       1016    => 'WRD=',
 }};
 
 our $session_id;
@@ -68,7 +71,7 @@ sub search {
 
        $session_id ||= int rand(1000000000);
        # FIXME allocate session just once
-       my $url = 'http://161.53.240.197:8991/F?RN=' . $session_id;
+       my $url = 'http://katalog.nsk.hr/F?RN=' . $session_id;
        # fake JavaScript code on page which creates random session
 
 diag "get $url";
@@ -80,11 +83,13 @@ diag "advanced search";
 
        $mech->follow_link( url_regex => qr/find-c/ );
 
+       my $database = $self->{database};
+
        if ( $mech->content =~ m{Requested library is unavailable at the moment} ) {
-               warn "WARNING: default database not available";
+               warn "ERROR: default database not available, try to swith to $database\n";
                $self->save_content;
-               my $url = 'local_base=' . lc $self->{database};
-               $mech->follow_link( url_regex => qr/$url/ );
+               $mech->follow_link( url_regex => qr/local_base=$database/i );
+               diag "re-try advanced search";
                $mech->follow_link( url_regex => qr/find-c/ );
        }
 
@@ -99,9 +104,11 @@ diag "submit search [$query] on ", $self->{database};
        );
 
        my $hits = 0;
-       if ( $mech->content =~ m{ukupno\s+(\d+).*do\s+(\d+)}s ) {
+       if ( $mech->content =~ m{ukupno\s+(\d+).*do\s+(\d+)} ) { # FIXME Many results in Crotian
                $hits = $1;
                $hits = $2 if $2 && $2 < $1; # correct for max. results
+       } elsif ( $mech->content =~ m{(\d+)\s+od\s+(\d+)} ) { # FIXME single result in Croatian
+               $hits = $2;
        } else {
                diag "get't find results in ", $mech->content;
                return;
@@ -137,6 +144,7 @@ sub next_marc {
 warn "parse $nr";
 
                $marc = MARC::Record->new;
+               $marc->encoding('utf-8');
                $hash = {};
 
                my $html = $mech->content;
@@ -144,6 +152,7 @@ warn "parse $nr";
                sub field {
                        my ( $f, $v ) = @_;
                        $v =~ s/\Q&nbsp;\E/ /gs;
+                       $v =~ s/\s+$//gs;
 warn "## $f\t$v\n";
                        $hash->{$f} = $v;
 
@@ -165,6 +174,8 @@ warn "## ++ ", dump( $f, $v );
 
                        my ($i1,$i2) = (' ',' ');
                        ($i1,$i2) = ($2,$3) if $f =~ s/^(...)(.)?(.)?/$1/;
+                       $i1 ||= ' ';
+                       $i2 ||= ' ';
                        my @sf = split(/\|/, $v);
                        @sf = map { s/^(\w)\s+//; { $1 => $_ } } @sf;
 #warn "## sf = ", dump(@sf);
@@ -178,7 +189,7 @@ warn "## ++ ", dump( $f, $i1, $i2, @sf );
 
                my $id = $hash->{SYS} || die "no SYS";
 
-               $self->save_marc( $id, $marc->as_usmarc );
+               $self->save_marc( "$id.marc", $marc->as_usmarc );
 
                if ( $nr < $self->{hits} ) {
                        $nr++;
@@ -186,7 +197,7 @@ warn "## ++ ", dump( $f, $i1, $i2, @sf );
                        $mech->follow_link( url_regex => qr/set_entry=0*$nr/ );
                }
 
-               return $marc->as_usmarc;
+               return $id;
        } else {
                die "can't fetch COMARC format from ", $mech->content;
        }