X-Git-Url: http://git.rot13.org/?a=blobdiff_plain;f=html.pl;h=ce3f30f84097b6bfd443bcb900d48d688c420dcc;hb=0997992c1e7efe90d3a8854b101f89efa7b9de38;hp=dd395acf3595fb43f4b80874701bbeb807994e4d;hpb=45af0260bcfe70f46a8ab02594aa7637f41309ae;p=koha-bibliografija diff --git a/html.pl b/html.pl index dd395ac..ce3f30f 100755 --- a/html.pl +++ b/html.pl @@ -10,6 +10,7 @@ use Data::Dump qw(dump); use autodie; use locale; use Text::Unaccent; +use Carp qw(confess); use lib '/srv/koha_ffzg'; use C4::Context; @@ -20,7 +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'; @@ -41,11 +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; } @@ -54,7 +59,6 @@ debug 'auth_department' => $auth_department; my $authors; -my $author_count; my $marcxml; my $sth_select_authors = $dbh->prepare(q{ @@ -77,6 +81,7 @@ where =cut my $biblio_year; +my $type_stats; my $parser = XML::LibXML->new(); $parser->recover_silently(0); # don't die when you find &, >, etc @@ -85,29 +90,37 @@ my $xslt = XML::LibXSLT->new(); my $parsed = $xslt->parse_stylesheet($style_doc); my $biblio_html; +my $biblio_parsed; +my $biblio_in_database; 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} || die "missing $biblionumber marcxml"; + my $xmlrecord = $marcxml->{$biblionumber} || confess "missing $biblionumber marcxml"; print $xml_fh $xmlrecord if $ENV{XML}; my $source = eval { $parser->parse_string($xmlrecord) }; if ( $@ ) { - warn "SKIP $biblionumber corrupt XML"; +# warn "SKIP $biblionumber corrupt XML"; push @{ $skip->{XML_corrupt} }, $biblionumber; 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}; } @@ -119,9 +132,9 @@ 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; } @@ -138,7 +151,7 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) { '008' => undef, '100' => '9', '700' => '(9|4)', - '942' => 't' + '942' => '(t|r)' }; my $data; @@ -150,7 +163,12 @@ 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') { @@ -168,37 +186,64 @@ 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'}; if ( ! $category ) { - warn "# SKIP ", $row->{biblionumber}, " no category in ", dump($data); +# warn "# SKIP ", $row->{biblionumber}, " no category in ", dump($data); push @{ $skip->{no_category} }, $row->{biblionumber}; next; } + if ( my $in_database = $data->{942}->[0]->{'r'} ) { + $biblio_in_database->{ $row->{biblionumber} } = $in_database; + } + + my $have_100 = 1; if ( exists $data->{100} ) { my @first_author = map { $_->{'9'} } @{ $data->{100} }; foreach my $authid ( @first_author ) { - push @{ $authors->{$authid}->{ $category } }, $row->{biblionumber}; - $author_count->{aut}->{$authid}++; + push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber}; } + } else { + $have_100 = 0; } + my $have_edt; + if ( exists $data->{700} ) { foreach my $auth ( @{ $data->{700} } ) { my $authid = $auth->{9} || next; my $type = $auth->{4} || next; #die "no 4 in ",dump($data); - - push @{ $authors->{$authid}->{ $category } }, $row->{biblionumber}; - if ( $type =~ m/aut/ ) { - $author_count->{aut}->{ $authid }++; - } elsif ( $type =~ m/(edt|tr)/ ) { - $author_count->{$1}->{ $authid }++; + + $type_stats->{$type}++; + + if ( $type =~ m/(edt|trl|com|ctb)/ ) { + push @{ $authors->{$authid}->{sec}->{ $category } }, $row->{biblionumber}; + push @{ $authors->{$authid}->{$1}->{ $category } }, $row->{biblionumber}; + } elsif ( $type =~ m/aut/ ) { + if ( ! $have_100 ) { + $have_edt = grep { exists $_->{4} && $_->{4} =~ m/edt/ } @{ $data->{700} } if ! defined $have_edt; + if ( $have_edt ) { + $skip->{ have_700_edt }->{ $row->{biblionumber} }++; + } else { + push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber}; + } + } else { + push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber}; + } } else { - warn "# SKIP ", $row->{biblionumber}, ' no 700$4 in ', dump($data); - push @{ $skip->{ 'no_700$4' } }, $row->{biblionumber}; +# warn "# SKIP ", $row->{biblionumber}, ' no 700$4 in ', dump($data); + $skip->{ 'no_700$4' }->{ $row->{biblionumber} }++; } } } @@ -206,7 +251,10 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) { } debug 'authors' => $authors; -debug 'author_count' => $author_count; +debug 'type_stats' => $type_stats; +debug 'skip' => $skip; +debug 'biblio_year' => $biblio_year; +debug 'biblio_in_database' => $biblio_in_database; my $category_label; my $sth_categories = $dbh->prepare(q{ @@ -243,6 +291,32 @@ my $first_letter = ''; debug 'authors' => \@authors; +sub li_biblio { + my ($biblionumber) = @_; + return qq|
  • |, + qq|$biblionumber|, + biblioitem_html($biblionumber), + qq|edit|, + qq|
    • \n|; +} + +sub author_html { + my ( $fh, $authid, $type, $label ) = @_; + + return unless exists $authors->{$authid}->{$type}; + + print $fh qq|

    $label

    \n|; + + foreach my $category ( sort keys %{ $authors->{$authid}->{$type} } ) { + my $label = $category_label->{$category} || 'Bez kategorije'; + print $fh qq|

    $label

    \n\n|; + } +} + foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) { my $first = substr( $row->{full_name}, 0, 1 ); @@ -257,18 +331,10 @@ foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) { open(my $fh, '>:encoding(utf-8)', "$path.new"); print $fh html_title($row->{full_name}, "bibliografija"); print $fh qq|

    $row->{full_name} - bibliografija za razdoblje 2008-2013

    |; - foreach my $category ( sort keys %{ $authors->{ $row->{authid} } } ) { - my $label = $category_label->{$category} || 'Bez kategorije'; - print $fh qq|

    $label

    \n\n|; - } + + author_html( $fh, $row->{authid}, 'aut' => 'Primarno autorstvo' ); + author_html( $fh, $row->{authid}, 'sec' => 'Uredništva, prijevodi, krička izdanja' ); + print $fh html_end; close($fh); rename "$path.new", "$path.html"; @@ -285,7 +351,9 @@ debug 'auth_header' => $auth_header; my $department_category_author; foreach my $department ( sort keys %$auth_department ) { foreach my $authid ( sort @{ $auth_department->{$department} } ) { - foreach my $category ( sort keys %{ $authors->{$authid} } ) { + my @categories = keys %{ $authors->{$authid}->{aut} }; + push @categories, keys %{ $authors->{$authid}->{sec} }; + foreach my $category ( sort @categories ) { push @{ $department_category_author->{$department}->{$category} }, $authid; } } @@ -295,6 +363,13 @@ debug 'department_category_author' => $department_category_author; mkdir 'html/departments' unless -d 'html/departments'; +sub unique_biblionumber { + my @v = @_; + my $u; + $u->{$_}++ foreach @v; + return sort { $biblio_year->{$b} <=> $biblio_year->{$a} || $a <=> $b } keys %$u; +} + open(my $dep_fh, '>:encoding(utf-8)', 'html/departments/index.new'); print $dep_fh html_title('Odsijeci Filozofskog fakulteta u Zagrebu'), qq|