X-Git-Url: http://git.rot13.org/?a=blobdiff_plain;f=html.pl;h=f23e38be12cf7a7a0fe5563d236e5c16f61ccd17;hb=37f76cb9ed121b9507b040bf4530aa9d78d4465d;hp=79d56bbd2e94573d8006dd3f5a25c8d52d40caca;hpb=367c7ead2be9d483197dd5e383c1af5a862ed543;p=koha-bibliografija diff --git a/html.pl b/html.pl index 79d56bb..f23e38b 100755 --- a/html.pl +++ b/html.pl @@ -11,12 +11,30 @@ use autodie; use locale; use Text::Unaccent; use Carp qw(confess); +use utf8; +use JSON; +use POSIX qw(strftime); +use Storable; use lib '/srv/koha_ffzg'; use C4::Context; use XML::LibXML; use XML::LibXSLT; +my $pid_file = '/dev/shm/bibliografija.pid'; +{ + if ( -e $pid_file ) { + open(my $fh, '<', $pid_file); + my $pid = <$fh>; + no autodie; # it will die on kill + kill 0, $pid || die "$0 allready running as pid $pid"; + } + open(my $fh, '>', $pid_file); + print $fh $$; + close($fh); +} + + my $dbh = C4::Context->dbh; sub debug { @@ -26,9 +44,23 @@ sub debug { my $xslfilename = 'compact.xsl'; +my $azvo_group_title = { +'znanstveno nastavni' => qr/(profes|docent|znanstveni savjetnik|znanstveni suradnik)/i, +'lektori i predavaÄi' => qr/(lektor|predavaÄ)/i, +'asistenti i novaci' => qr/(asistent|novak)/i, +}; + +my $department_groups = { +'AAB_humanistiÄke' => qr/(anglistiku|arheologiju|antropologiju|filozofiju|fonetiku|germanistiku|hungarologiju|indologiju|slavenske|filologiju|komparativnu|kroatistiku|lingvistiku|povijest|romanistiku|talijanistiku)/i, +'AAC_druÅ¡tvene' => qr/(informacijske|pedagogiju|psihologiju|sociologiju)/i, +}; + my $auth_header; my $auth_department; +my $auth_group; my @authors; +my $department_in_sum; +my $department_in_group; my $skip; @@ -36,7 +68,8 @@ my $sth_auth = $dbh->prepare(q{ select authid, ExtractValue(marcxml,'//datafield[@tag="100"]/subfield[@code="a"]') as full_name, - ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="a"]') as department + ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="a"]') as department, + ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="i"]') as academic_title from auth_header }); @@ -49,13 +82,45 @@ while( my $row = $sth_auth->fetchrow_hashref ) { $auth_header->{ $row->{authid} } = $row->{full_name}; $row->{department} =~ s/, Filozofski fakultet u Zagrebu.*$//; $row->{department} =~ s/^.+\.\s*//; -# warn dump( $row ); + $row->{department} =~ s/\s+$//s; + my $group; + foreach my $title ( keys %$azvo_group_title ) { + if ( $row->{academic_title} =~ $azvo_group_title->{$title} ) { + $group = $title; + last; + } + } + if ( $group ) { + $row->{academic_group} = $group; + $auth_group->{ $row->{authid} } = $group; + $skip->{group_stat}->{$group}++; + } else { + push @{ $skip->{no_academic_group} }, $row; + } + +# warn "# ", dump( $row ); push @{ $auth_department->{ $row->{department} } }, $row->{authid}; push @authors, $row; + $department_in_sum->{ $row->{department} }++; + foreach my $name ( keys %$department_groups ) { + my $regex = $department_groups->{$name}; + if ( $row->{department} =~ $regex ) { + $department_in_group->{ $row->{department} } = $name; + last; + } + } +} + +debug 'department_in_group' => $department_in_group; +foreach my $department ( keys %$department_in_sum ) { +# $department_in_sum->{$department} = 0 unless $department =~ m/(centar|croaticum|katedra|odsjek)/i; } debug 'auth_department' => $auth_department; +store $auth_department, '/dev/shm/auth_department.storable'; +debug 'auth_group' => $auth_group; +debug 'department_in_sum' => $department_in_sum; my $authors; @@ -63,10 +128,11 @@ my $marcxml; my $sth_select_authors = $dbh->prepare(q{ select - biblionumber, + biblioitems.biblionumber, itemtype, - marcxml + metadata as marcxml from biblioitems +join biblio_metadata on (biblio_metadata.biblionumber = biblioitems.biblionumber) where agerestriction > 0 }); @@ -81,6 +147,7 @@ where =cut my $biblio_year; +my $biblio_full_name; my $type_stats; my $parser = XML::LibXML->new(); @@ -91,6 +158,8 @@ my $parsed = $xslt->parse_stylesheet($style_doc); my $biblio_html; my $biblio_parsed; +my $biblio_data; +my $biblio_author_external; open(my $xml_fh, '>', '/tmp/bibliografija.xml') if $ENV{XML}; @@ -148,9 +217,11 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) { my $extract = { '008' => undef, - '100' => '9', - '700' => '(9|4)', - '942' => 't' + '100' => '(9|a)', + '245' => 'a', + '680' => 'i', + '700' => '(9|4|a)', + '942' => '(t|r|v)' }; my $data; @@ -162,12 +233,14 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) { if ($elt->localname eq 'controlfield') { if ( $tag eq '008' ) { - my $year = substr($elt->textContent, 7, 4 ); + my $content = $elt->textContent; + my $year = substr($content, 7, 4 ); if ( $year !~ m/^\d+$/ ) { $year = 0; push @{ $skip->{invalid_year} }, $row->{biblionumber}; } $biblio_year->{ $row->{biblionumber} } = $data->{year} = $year; + $data->{'008'} = $content; } next; } elsif ($elt->localname eq 'datafield') { @@ -182,15 +255,19 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) { } } push @{ $data->{$tag} }, $sf_data if $sf_data; - } - } + } + } - if ( $data->{year} < 2008 ) { + if ( ! defined $data->{year} ) { + warn "MISSING year in ", $row->{biblionumber}; +=for remove-year-limit + } elsif ( $data->{year} < 2008 ) { push @{ $skip->{year_lt_2008} }, $row->{biblionumber}; next; } elsif ( $data->{year} > 2013 ) { push @{ $skip->{year_gt_2013} }, $row->{biblionumber}; next; +=cut } # warn "# ", $row->{biblionumber}, " data ",dump($data); @@ -202,54 +279,97 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) { next; } - my $have_100 = 1; if ( exists $data->{100} ) { - my @first_author = map { $_->{'9'} } @{ $data->{100} }; + my @first_author = + map { $_->{'9'} } + grep { + if ( ! exists $_->{9} ) { + $biblio_author_external->{ $row->{biblionumber} }++; + 0; + } elsif ( exists $auth_header->{ $_->{9} } ) { + 1; # from FFZXG + } else { + 0; + } + } + @{ $data->{100} }; foreach my $authid ( @first_author ) { push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber}; } + $biblio_full_name->{ $row->{biblionumber} } = $data->{100}->[0]->{a}; } else { $have_100 = 0; } + $biblio_full_name->{ $row->{biblionumber} } ||= $data->{245}->[0]->{a}; + my $have_edt; if ( exists $data->{700} ) { - foreach my $auth ( @{ $data->{700} } ) { + my @other_authors = + grep { + if ( ! exists $_->{9} ) { + $biblio_author_external->{ $row->{biblionumber} }++; + 0; + } elsif ( exists $auth_header->{ $_->{9} } ) { + 1; # from FFZXG + } else { + 0; + } + } + @{ $data->{700} }; + foreach my $auth ( @other_authors ) { my $authid = $auth->{9} || next; my $type = $auth->{4} || next; #die "no 4 in ",dump($data); $type_stats->{$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} }++; + my @types = split(/[\s\/]+/, $type); + + foreach my $type ( @types ) { + my $type = substr($type,0,3); + $type_stats->{_count_each_type}->{$type}++; + + if ( $type =~ m/(edt|trl|com|ctb)/ ) { + push @{ $authors->{$authid}->{__sec}->{ $category } }, $row->{biblionumber}; + push @{ $authors->{$authid}->{$type}->{ $category } }, $row->{biblionumber}; + $type =~ s/(com|ctb)/_ostalo/; + push @{ $authors->{$authid}->{$type}->{ $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 { - push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber}; +# warn "# SKIP ", $row->{biblionumber}, ' no 700$4 in ', dump($data); + $skip->{ 'no_700$4' }->{ $row->{biblionumber} }++; } - } else { -# warn "# SKIP ", $row->{biblionumber}, ' no 700$4 in ', dump($data); - $skip->{ 'no_700$4' }->{ $row->{biblionumber} }++; } } + delete $data->{700}; } + $biblio_data->{ $row->{biblionumber} } = $data; + } debug 'authors' => $authors; +store $authors, '/dev/shm/authors.storable'; debug 'type_stats' => $type_stats; debug 'skip' => $skip; debug 'biblio_year' => $biblio_year; +debug 'biblio_full_name' => $biblio_full_name; +debug 'biblio_data' => $biblio_data; +debug 'biblio_author_external' => $biblio_author_external; my $category_label; my $sth_categories = $dbh->prepare(q{ @@ -268,13 +388,18 @@ sub html_title {