use autodie;
use locale;
use Text::Unaccent;
+use Carp qw(confess);
use lib '/srv/koha_ffzg';
use C4::Context;
sub debug {
my ($title, $data) = @_;
- print "# $title ",dump($data), $/;
+ print "# $title ",dump($data), $/ if $ENV{DEBUG};
}
my $xslfilename = 'compact.xsl';
$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;
}
my $authors;
-my $author_count;
my $marcxml;
my $sth_select_authors = $dbh->prepare(q{
=cut
my $biblio_year;
+my $type_stats;
my $parser = XML::LibXML->new();
$parser->recover_silently(0); # don't die when you find &, >, etc
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} || 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};
}
$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}, " ", $row->{marcxml}, "\n";
next;
}
if ($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') {
}
}
+ 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;
}
+ 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} }++;
}
}
}
}
debug 'authors' => $authors;
-debug 'author_count' => $author_count;
+debug 'type_stats' => $type_stats;
+debug 'skip' => $skip;
+debug 'biblio_year' => $biblio_year;
my $category_label;
my $sth_categories = $dbh->prepare(q{
debug 'authors' => \@authors;
+sub li_biblio {
+ my ($biblionumber) = @_;
+ return qq|<li>|,
+ qq|<a href="https://koha.ffzg.hr/cgi-bin/koha/opac-detail.pl?biblionumber=$biblionumber">$biblionumber</a>|,
+ biblioitem_html($biblionumber),
+ qq|<a href="https://koha.ffzg.hr:8443/cgi-bin/koha/cataloguing/addbiblio.pl?biblionumber=$biblionumber">edit</a>|,
+ qq|</li>\n|;
+}
+
+sub author_html {
+ my ( $fh, $authid, $type, $label ) = @_;
+
+ return unless exists $authors->{$authid}->{$type};
+
+ print $fh qq|<h2>$label</h2>\n|;
+
+ foreach my $category ( sort keys %{ $authors->{$authid}->{$type} } ) {
+ my $label = $category_label->{$category} || 'Bez kategorije';
+ print $fh qq|<h3>$label</h3>\n<ul>\n|;
+ foreach my $biblionumber ( @{ $authors->{$authid}->{$type}->{$category} } ) {
+ print $fh li_biblio( $biblionumber );
+ }
+ print $fh qq|</ul>\n|;
+ }
+}
+
foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) {
my $first = substr( $row->{full_name}, 0, 1 );
open(my $fh, '>:encoding(utf-8)', "$path.new");
print $fh html_title($row->{full_name}, "bibliografija");
print $fh qq|<h1>$row->{full_name} - bibliografija za razdoblje 2008-2013</h1>|;
- foreach my $category ( sort keys %{ $authors->{ $row->{authid} } } ) {
- my $label = $category_label->{$category} || 'Bez kategorije';
- print $fh qq|<h2>$label</h2>\n<ul>\n|;
- foreach my $biblionumber ( @{ $authors->{ $row->{authid} }->{$category} } ) {
- print $fh qq|<li>|,
- qq|<a href="https://koha.ffzg.hr/cgi-bin/koha/opac-detail.pl?biblionumber=$biblionumber">$biblionumber</a>|,
- biblioitem_html($biblionumber),
- qq|<a href="https://koha.ffzg.hr:8443/cgi-bin/koha/cataloguing/addbiblio.pl?biblionumber=$biblionumber">edit</a>|,
- qq|</li>\n|;
- }
- print $fh qq|</ul>\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";
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;
}
}
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|<ul>\n|;
foreach my $department ( sort keys %$department_category_author ) {
my $dep_file = unac_string('utf-8',$dep);
print $dep_fh qq|<li><a href="$dep_file.html">$dep</a></li>\n|;
open(my $fh, '>:encoding(utf-8)', "html/departments/$dep_file.new");
+
print $fh html_title($department . ' bibliografija');
+ print $fh qq|<h1>$department bibliografija</h1>\n|;
+
+ print $fh qq|<h2>Primarno autorstvo</h2>\n|;
+
foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
+
+ my @authids = @{ $department_category_author->{$department}->{$category} };
+ next unless @authids;
+
+ my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{aut}->{$category} } } grep { exists $authors->{$_}->{aut}->{$category} } @authids;
+ my $unique;
+ $unique->{$_}++ foreach @biblionumber;
+
+ next unless @biblionumber;
+
my $label = $category_label->{$category} || 'Bez kategorije';
- print $fh qq|<h
1>$label</h1>\n<ul>\n|;
+ print $fh qq|<h
3>$label</h3>\n<ul>\n|;
- foreach my $authid ( @{ $department_category_author->{$department}->{$category} } ) {
- foreach my $biblionumber ( @{ $authors->{$authid}->{$category} } ) {
- print $fh qq|<li>|,
- qq|<a href="https://koha.ffzg.hr/cgi-bin/koha/opac-detail.pl?biblionumber=$biblionumber">$biblionumber</a>|,
- biblioitem_html($biblionumber),
- qq|<a href="https://koha.ffzg.hr:8443/cgi-bin/koha/cataloguing/addbiblio.pl?biblionumber=$biblionumber">edit</a>|,
- qq|</li>\n|;
- }
- }
+ print $fh li_biblio( $_ ) foreach @biblionumber;
print $fh qq|</ul>|;
}
+
+
+ print $fh qq|<h2>Sekundarno autorstvo</h2>\n|;
+
+ foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
+
+ my @authids = @{ $department_category_author->{$department}->{$category} };
+ next unless @authids;
+
+ my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{sec}->{$category} } } grep { exists $authors->{$_}->{sec}->{$category} } @authids;
+
+ next unless @biblionumber;
+
+ my $label = $category_label->{$category} || 'Bez kategorije';
+ print $fh qq|<h3>$label</h3>\n<ul>\n|;
+
+ print $fh li_biblio( $_ ) foreach @biblionumber;
+
+ print $fh qq|</ul>|;
+ }
+
+
print $fh html_end;
close($fh);
rename "html/departments/$dep_file.new", "html/departments/$dep_file.html";
close($dep_fh);
rename 'html/departments/index.new', 'html/departments/index.html';
-debug 'skip' => $skip;
+my $azvo_stat;
+
+foreach my $department ( sort keys %$department_category_author ) {
+ foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
+ foreach my $authid ( @{ $department_category_author->{$department}->{$category} } ) {
+ foreach my $type ( keys %{ $authors->{$authid} } ) {
+ next unless exists $authors->{$authid}->{$type}->{$category};
+ $azvo_stat->{ $department }->{ $category }->{ $type } += $#{ $authors->{$authid}->{$type}->{$category} } + 1;
+ }
+ }
+ }
+}
+
+debug 'azvo_stat' => $azvo_stat;
+
+=for later
+open(my $fh, '>', 'html/azvo.new');
+
+
+
+close($fh);
+rename 'html/azvo.new', 'html/azvo.html';
+=cut
projects / koha-bibliografija / blobdiff