X-Git-Url: http://git.rot13.org/?a=blobdiff_plain;f=html.pl;h=b8f041fddbd885c3e145b66113337fde1382b682;hb=9179d80beebae7b2f9d6fdee92c9a446baa52d64;hp=9c4ec3297cd77300222df21656a3ef93294d03d7;hpb=ecc58c1936e543dc082d4df5604dc4be4cb65cf6;p=koha-bibliografija diff --git a/html.pl b/html.pl index 9c4ec32..b8f041f 100755 --- a/html.pl +++ b/html.pl @@ -21,7 +21,7 @@ my $dbh = C4::Context->dbh; sub debug { my ($title, $data) = @_; - print "# $title ",dump($data), $/; + print "# $title ",dump($data), $/ if $ENV{DEBUG}; } my $xslfilename = 'compact.xsl'; @@ -42,11 +42,15 @@ from auth_header $sth_auth->execute(); while( my $row = $sth_auth->fetchrow_hashref ) { + if ( $row->{department} !~ m/Filozofski fakultet u Zagrebu/ ) { + push @{ $skip->{nije_ffzg} }, $row; + next; + } $auth_header->{ $row->{authid} } = $row->{full_name}; - $row->{department} =~ s/, Filozofski fakultet u Zagrebu\s*// || next; + $row->{department} =~ s/, Filozofski fakultet u Zagrebu.*$//; $row->{department} =~ s/^.+\.\s*//; - push @{ $auth_department->{ $row->{department} } }, $row->{authid}; # warn dump( $row ); + push @{ $auth_department->{ $row->{department} } }, $row->{authid}; push @authors, $row; } @@ -86,13 +90,14 @@ my $xslt = XML::LibXSLT->new(); my $parsed = $xslt->parse_stylesheet($style_doc); my $biblio_html; +my $biblio_parsed; open(my $xml_fh, '>', '/tmp/bibliografija.xml') if $ENV{XML}; sub biblioitem_html { - my $biblionumber = shift; + my ($biblionumber, $parse_only) = @_; - return $biblio_html->{$biblionumber} if exists $biblio_html->{$biblionumber}; + return $biblio_html->{$biblionumber} if exists $biblio_html->{$biblionumber} && ! $parse_only; my $xmlrecord = $marcxml->{$biblionumber} || confess "missing $biblionumber marcxml"; @@ -105,10 +110,16 @@ sub biblioitem_html { return; } + if ( $parse_only ) { + $biblio_parsed->{$biblionumber} = $source; + return $source; + } + my $transformed = $parsed->transform($source); $biblio_html->{$biblionumber} = $parsed->output_string( $transformed ); - return ( $biblio_html->{$biblionumber}, $source ) if wantarray; + delete $biblio_parsed->{$biblionumber}; + return $biblio_html->{$biblionumber}; } @@ -120,9 +131,9 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) { $marcxml->{ $row->{biblionumber} } = $row->{marcxml}; - my ( undef, $doc ) = biblioitem_html( $row->{biblionumber} ); + my $doc = biblioitem_html( $row->{biblionumber}, 1 ); if ( ! $doc ) { - warn "ERROR can't parse MARCXML ", $row->{biblionumber}, " ", $row->{marcxml}, "\n"; +# warn "ERROR can't parse MARCXML ", $row->{biblionumber}, " ", $row->{marcxml}, "\n"; next; } @@ -151,7 +162,12 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) { if ($elt->localname eq 'controlfield') { if ( $tag eq '008' ) { - $biblio_year->{ $row->{biblionumber} } = $elt->textContent; + my $year = substr($elt->textContent, 7, 4 ); + if ( $year !~ m/^\d+$/ ) { + $year = 0; + push @{ $skip->{invalid_year} }, $row->{biblionumber}; + } + $biblio_year->{ $row->{biblionumber} } = $data->{year} = $year; } next; } elsif ($elt->localname eq 'datafield') { @@ -169,6 +185,14 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) { } } + if ( $data->{year} < 2008 ) { + push @{ $skip->{year_lt_2008} }, $row->{biblionumber}; + next; + } elsif ( $data->{year} > 2013 ) { + push @{ $skip->{year_gt_2013} }, $row->{biblionumber}; + next; + } + # warn "# ", $row->{biblionumber}, " data ",dump($data); my $category = $data->{942}->[0]->{'t'}; @@ -225,6 +249,7 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) { debug 'authors' => $authors; debug 'type_stats' => $type_stats; debug 'skip' => $skip; +debug 'biblio_year' => $biblio_year; my $category_label; my $sth_categories = $dbh->prepare(q{ @@ -303,7 +328,7 @@ foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) { print $fh qq|