--- trunk/all2xml.pl 2003/01/11 19:55:30 9 +++ trunk2/all2all.pl 2004/09/14 09:23:00 440 @@ -1,250 +1,438 @@ #!/usr/bin/perl -w +=head1 NAME + +all2all.pl - basic script for all WebPAC needs + +=cut + use strict; -use OpenIsis; -use Getopt::Std; use Data::Dumper; -use XML::Simple; -use Text::Unaccent 1.02; # 1.01 won't compile on my platform, -require Unicode::Map8; -use DBI; - -my $config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1); -my $dbh = DBI->connect("DBI:Pg:dbname=webpac","","") || die $DBI::errstr; # FIX -# FIX; select relname from pg_class where relname like 'index_%' ; -$dbh->begin_work || die $dbh->errstr(); - -$dbh->do("delete from index_author") || die $dbh->errstr(); -$dbh->do("delete from index_title") || die $dbh->errstr(); - -my %opts; - -# usage: -# -d directory name -# -m multiple directories -# -q quiet -# -s run swish - -getopts('d:m:qs', \%opts); - -my $db_dir = $opts{d} || "ps"; # FIX - -#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts); - -#print Dumper($config->{indexer}); -#print "-" x 70,"\n"; - -# how to convert isis code page to UTF8? -my $isis_map = Unicode::Map8->new($config->{isis_codepage}) || die; - -sub isis2xml { - - my $row = shift @_; - - my $xml; - $xml->{db_dir} = [ $db_dir ]; # FIX remove? - - sub isis_sf { - my $row = shift @_; - my $isis_id = shift @_; - my $subfield = shift @_; - if ($row->{$isis_id}->[0]) { - my $sf = OpenIsis::subfields($row->{$isis_id}->[0]); - if (! defined $subfield || length($subfield) == 0) { - # subfield list undef, empty or no defined subfields for this record - my $all_sf = $row->{$isis_id}->[0]; - $all_sf =~ s/\^./ /g; nuke definirions - return $all_sf; - } elsif ($sf->{$subfield}) { - return $sf->{$subfield}; - } - } +use Carp; +use Getopt::Long; + +use lib './lib'; +use WebPAC; +use WebPAC::jsFind; +use WebPAC::Index; + +# options which can be changed via command line +# +my $code_page = 'ISO-8859-2'; +my ($limit_mfn, $start_mfn, $debug, $low_mem); +my $index_path = './out/index'; + +my $result = GetOptions( + "code_page=s" => \$code_page, + "limit_mfn=i" => \$limit_mfn, + "start_mfn=i" => \$start_mfn, + "debug!" => \$debug, + "low_mem!" => \$low_mem, +); + +# create WebPAC object +# +my $webpac = new WebPAC( + code_page => $code_page, + limit_mfn => $limit_mfn, + start_mfn => $start_mfn, + debug => $debug, + low_mem => $low_mem, +) || die; + +my $log = $webpac->_get_logger() || die "can't get logger"; + +$log->debug("creating WebPAC::jsFind object"); + +my $index = new WebPAC::jsFind( + index_path => $index_path, + keys => 10, +) || die; + +my $thes; + +$|=1; + +my $maxmfn = $webpac->open_isis( + filename => shift @ARGV || '/data/hidra/THS/THS', + lookup => [ + { 'key' => 'd:v900', 'val' => 'v250^a' }, +# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, +# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, +# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, + { 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, + { 'key' => '900_mfn:v900', 'val' => 'v000' }, + # tree structure + { 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, + { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, + ], +); + +$log->debug("isis file ",$webpac->{'isis_filename'}," opened"); + +$log->info("rows: $maxmfn"); + +$webpac->open_import_xml(type => 'isis_hidra_ths'); + +if(1) { # XXX + +while (my $rec = $webpac->fetch_rec) { + + my @ds = $webpac->data_structure($rec); + + if (0 && $log->is_debug) { + $log->debug("rec = ",Dumper($rec)); + $log->debug("ds = ",Dumper(\@ds)); } - foreach my $field (keys %{$config->{indexer}}) { + next if (! @ds); - my $display_data = ""; - my $swish_data = ""; - my $index_data = ""; - - foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { - - my $display_tmp = ""; - my $swish_tmp = ""; - my $index_tmp = ""; - - my $format = $x->{content}; - my $s = 1; # swish only - my $d = 1; # display only - my $i = 0; # index only - $s = 0 if (lc($x->{type}) eq "display"); - $d = 0 if (lc($x->{type}) eq "swish"); - ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); -#print STDERR "## s: $s d: $d i: $i ## $format ##\n"; - # parse format - my $prefix = ""; - if ($format =~ s/^([^\d]+)//) { - $prefix = $1; - } - while ($format) { - if ($format =~ s/^(\d\d\d)(\w?)//) { - my $isis_tmp = isis_sf($row,$1,$2); - if ($isis_tmp) { -# $display_tmp .= $prefix . "/$1/$2/".$isis_tmp if ($d); - $display_tmp .= $prefix . $isis_tmp if ($d); - $swish_tmp .= $isis_tmp." " if ($s); - $index_tmp .= $prefix . $isis_tmp if ($i); -#print STDERR " $isis_tmp <--\n"; - } - $prefix = ""; - } elsif ($format =~ s/^([^\d]+)//) { - $prefix = $1; - } else { - print STDERR "WARNING: unparsed format '$format'\n"; - last; - }; - } - # add suffix - $display_tmp .= $prefix if ($display_tmp); - $index_tmp .= $prefix if ($index_tmp); - -# $display_data .= $display_tmp if ($display_tmp ne ""); -# $swish_data .= $swish_tmp if ($swish_tmp ne ""); - $display_data .= $display_tmp; - $swish_data .= $swish_tmp; - $index_data .= $index_tmp; + my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object"); - } -#print "--display:$display_data\n--swish:$swish_data\n"; - #$xml->{$field."_display"} = $isis_map->tou($display_data)->utf8 if ($display_data); - #$xml->{$field."_swish"} = unac_string($config->{isis_codepage},$swish_data) if ($swish_data); - $xml->{$field."_display" } = [ $isis_map->tou($display_data)->utf8 ] if ($display_data); - $xml->{$field."_swish"} = [ unac_string($config->{isis_codepage},$swish_data) ] if ($swish_data); - - # index - if ($index_data && $index_data ne "") { - my $sql = "select $field from index_$field where upper($field)=upper(?)"; - my $sth = $dbh->prepare($sql) || die $dbh->errstr(); - $sth->execute($index_data) || die "SQL: $sql; ".$dbh->errstr(); -#print STDERR "--->$index_data<---\n"; - if (! $sth->fetchrow_hashref) { - my $sql = "insert into index_$field values (?)"; - my $sth = $dbh->prepare($sql) || die $dbh->errstr(); -#print STDERR "$sql: $index_data + + + + search | + thesarus | + browse + + +
+Folding: +
+« default » +
+reset +all +none +
+debug + +outline +
+ + + + +}; + +close(HTML); + + +my $js_file = "out/$js_url"; +$log->info("creating '$js_file' with arrays of shown and hidden ids"); +open(JS, ">", $js_file) || $log->logdie("can't open '$js_file': $!"); +print JS "var show = ['",join("','",@show_ids),"'];\n"; +print JS "var hide = ['",join("','",@hide_ids),"'];\n"; +close(JS); -=head1 NAME - -isis2xml.pl - read isis file and dump XML - -=head1 DESCRIPTION - -This command will read ISIS data file using OpenIsis perl module and -create XML file for usage with I -indexer. Dispite it's name, this script B -from isis files (isis allready has something like that). Output of this -script is tailor-made for SWISH-E. - -=head1 AUTHOR - -Dobrica Pavlinusic +$log->info("closing index"); +$index->close; -=head1 COPYRIGHT - -GNU Public License (GPL) v2 or later - -=head1 SEE ALSO - -SWISH-E web site at http://www.swish-e.org - -=cut +$log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'}));