X-Git-Url: http://git.rot13.org/?a=blobdiff_plain;f=Aleph.pm;h=99c054dbeb03c89c5c799220fe185436917dadf5;hb=0bba87180282519f469d7696c3efed63a38e955c;hp=7c4b76876fad2032761490d8d3d9669b58d44f9b;hpb=4e069bde46b3bf0f1a771b3d8249c3f0d8bf330c;p=Biblio-Z3950.git diff --git a/Aleph.pm b/Aleph.pm index 7c4b768..99c054d 100644 --- a/Aleph.pm +++ b/Aleph.pm @@ -15,9 +15,9 @@ 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 @@ -50,22 +50,27 @@ 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; sub search { my ( $self, $query ) = @_; die "need query" unless defined $query; - my $url = 'http://161.53.240.197:8991/F?RN=' . rand(1000000000); + $session_id ||= int rand(1000000000); + # FIXME allocate session just once + my $url = 'http://katalog.nsk.hr/F?RN=' . $session_id; # fake JavaScript code on page which creates random session diag "get $url"; @@ -77,18 +82,32 @@ diag "advanced search"; $mech->follow_link( url_regex => qr/find-c/ ); -diag "submit search $query"; + my $database = $self->{database}; + + if ( $mech->content =~ m{Requested library is unavailable at the moment} ) { + warn "ERROR: default database not available, try to swith to $database\n"; + $self->save_content; + $mech->follow_link( url_regex => qr/local_base=$database/i ); + diag "re-try advanced search"; + $mech->follow_link( url_regex => qr/find-c/ ); + } + +diag "submit search [$query] on ", $self->{database}; + $mech->submit_form( fields => { 'ccl_term' => $query, + 'local_base' => $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; @@ -102,7 +121,7 @@ diag "in MARC format"; $mech->follow_link( url_regex => qr/format=001/ ); - return $hits; + return $self->{hits} = $hits; } @@ -124,6 +143,7 @@ sub next_marc { warn "parse $nr"; $marc = MARC::Record->new; + $marc->encoding('utf-8'); $hash = {}; my $html = $mech->content; @@ -131,6 +151,7 @@ warn "parse $nr"; sub field { my ( $f, $v ) = @_; $v =~ s/\Q \E/ /gs; + $v =~ s/\s+$//gs; warn "## $f\t$v\n"; $hash->{$f} = $v; @@ -165,13 +186,15 @@ warn "## ++ ", dump( $f, $i1, $i2, @sf ); my $id = $hash->{SYS} || die "no SYS"; - $self->save_marc( $id, $marc->as_usmarc ); - - $nr++; + $self->save_marc( "$id.marc", $marc->as_usmarc ); - $mech->follow_link( url_regex => qr/set_entry=0*$nr/ ); + if ( $nr < $self->{hits} ) { + $nr++; + diag "follow link to next record $nr"; + $mech->follow_link( url_regex => qr/set_entry=0*$nr/ ); + } - return $marc->as_usmarc; + return $id; } else { die "can't fetch COMARC format from ", $mech->content; }