--- trunk/all2xml.pl 2002/11/24 20:52:11 1
+++ trunk2/all2all.pl 2004/09/13 15:32:55 437
@@ -1,66 +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 Carp;
+use Getopt::Long;
-my $config=XMLin();
+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));
+ $log->debug("ds = ",Dumper(\@ds));
+ }
-print Dumper($config);
+ next if (! @ds);
-my %opts;
+ my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
-getopts('d:m:q', \%opts);
+ if ($filename) {
+ $webpac->output_file(
+ file => $filename,
+ template => 'html.tt',
+ data => \@ds,
+ headline => $webpac->{'headline'},
+ );
+ } else {
+ print $webpac->output(
+ template => 'text.tt',
+ data => \@ds,
+ headline => $webpac->{'headline'},
+ );
+ }
-my $db_dir = $opts{d};
+ my $headline = $webpac->{'headline'};
-die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
+ my $f = $filename;
+ $f =~ s!out/!!;
-#--------------------------------------------------------------------
+ # save into index
+ foreach my $ds (@ds) {
+ next if (! $ds->{'swish'});
+
+ $index->insert(
+ index_name => $ds->{'tag'},
+ path => $f,
+ headline => $headline,
+ words => join(" ",@{$ds->{'swish'}})
+ );
+ }
-my $last_tell=0;
+ # save into sorted index (thesaurus)
+ foreach my $ds (@ds) {
+ next if (! $ds->{'index'});
+
+ $thes->{$ds->{'tag'}} ||= new WebPAC::Index;
+
+ foreach my $h (@{$ds->{'index'}}) {
+ $thes->{$ds->{'tag'}}->insert(
+ path => $f,
+ headline => $h,
+ );
+ }
+ }
-my @isis_dirs = ( '.' ); # use dirname as database name
+# print Dumper(\@ds);
-if ($opts{m}) {
- @isis_dirs = split(/,/,$opts{m});
}
-my @isis_dbs;
+foreach my $t (keys %{$thes}) {
-foreach (@isis_dirs) {
- if (-e "$common::isis_data/$db_dir/$_/LIBRI") {
- push @isis_dbs,"$common::isis_data/$db_dir/$_/LIBRI/LIBRI";
- }
- if (-e "$common::isis_data/$db_dir/$_/PERI") {
- push @isis_dbs,"$common::isis_data/$db_dir/$_/PERI/PERI";
- }
- if (-e "$common::isis_data/$db_dir/$_/AMS") {
- push @isis_dbs,"$common::isis_data/$db_dir/$_/AMS/AMS";
+ my @e = $thes->{$t}->elements;
+ if (! @e) {
+ $log->logwarn("no elements in sorted index $t?");
+ next;
}
- if (-e "$common::isis_data/$db_dir/$_/ARTI") {
-# push @isis_dbs,"$common::isis_data/$db_dir/$_/ARTI/ARTI";
- }
-}
-foreach my $isis_db (@isis_dbs) {
+ my $file = "./out/bfilter/$t.txt";
+ $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
- my $db = OpenIsis::open( "$isis_db" ) || warn "can't open '$isis_db'";
+ $webpac->output_file(
+ file => $file,
+ template => 'index.tt',
+ data => \@e,
+ index_name => $t,
+ );
+}
- my $max_rowid = OpenIsis::maxRowid( $db );
+if (0 && $log->is_debug) {
+ $log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
+ $log->debug("data hash: ",Dumper($webpac->{'data'}));
+ foreach my $t (keys %{$thes}) {
+ $log->debug("thesaurus $t hash: ",Dumper($thes->{$t}));
+ }
+}
- my $last_pcnt = 0;
+} # XXX if(0)
- for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
- my $row = OpenIsis::read( $db, $row_id );
+$log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
+
+#
+# tree dump implementation which shouldn't be here :-)
+#
+
+
+my $file = 'out/browse.html';
+my $js_url = 'tree-ids.js';
+
+$log->info("creating '$file' for tree html");
+open(HTML, "> $file") || $log->logdie("can't open '$file'");
+
+print HTML qq{
+
+
+ Browse
+
+
+
+
+
+
+
+
+
+ search |
+ thesarus |
+ browse
+
+
+
+
+
+};
+
+my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
+
+my @tree = ({
+ # level 0
+ code_arr => sub { sort keys %{$l} },
+ filter_code => sub {
+ my $t = shift;
+ return $t if ($t =~ s/root://);
+ },
+ lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} },
+ lookup_term => sub { shift @{$l->{"d:".$_[1]}} },
+ lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} },
+ have_children => sub { defined($l->{"a:".$_[0]."::"}) },
+ child_code => sub { return $_[0] },
+ style => 'display: none',
+ },{
+ # 1
+ code_arr => sub { @{$l->{"a:".$_[0]."::"}} },
+ filter_code => sub { shift }, # nop
+ lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
+ lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
+ lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
+ have_children => sub { defined($l->{"a:".$_[1].":"}) },
+ child_code => sub { return $_[1] },
+ style => 'display: none',
+ },{
+ # 2
+ code_arr => sub { @{$l->{"a:".$_[0].":"}} },
+ filter_code => sub { shift },
+ lookup_v900 => sub { shift },
+ lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
+ lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
+ have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) },
+ child_code => sub { return "a:".$_[2].":".$_[1] },
+ style => 'display: none',
+ },{
+ # 3 uži pojam
+ code_arr => sub { @{$l->{$_[0]}} },
+ filter_code => sub { shift },
+ lookup_v900 => sub {
+ my ($c,$p) = @_;
+ $p =~ s/^a:(..:....):.*$/$1/;
+ return "a:".$p.":".$c;
+ },
+ lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
+ lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
+ have_children => sub { defined($l->{$_[1]}) },
+ child_code => sub { return $_[1] },
+ },{
+ # 4
+ code_arr => sub { @{$l->{$_[0]}} },
+ filter_code => sub { shift },
+ lookup_v900 => sub {
+ my ($c,$p) = @_;
+ $p =~ s/^a:(..:....):.*$/$1/;
+ return "a:".$p.":".$c;
+ },
+ lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
+ lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
+ have_children => sub { defined($l->{$_[1]}) },
+ child_code => sub { return $_[1] },
+ },{
+ # 5
+ code_arr => sub { @{$l->{$_[0]}} },
+ filter_code => sub { shift },
+ lookup_v900 => sub {
+ my ($c,$p) = @_;
+ $p =~ s/^a:(..:....):.*$/$1/;
+ return "a:".$p.":".$c;
+ },
+ lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
+ lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
+ have_children => sub { defined($l->{$_[1]}) },
+ child_code => sub { return $_[1] },
+ },{
+ # 6
+ code_arr => sub { @{$l->{$_[0]}} },
+ filter_code => sub { shift },
+ lookup_v900 => sub {
+ my ($c,$p) = @_;
+ $p =~ s/^a:(..:....):.*$/$1/;
+ return "a:".$p.":".$c;
+ },
+ lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
+ lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
+ have_children => sub { defined($l->{$_[1]}) },
+ child_code => sub { return $_[1] },
+ },{
+ # 7
+ code_arr => sub { @{$l->{$_[0]}} },
+ filter_code => sub { shift },
+ lookup_v900 => sub {
+ my ($c,$p) = @_;
+ $p =~ s/^a:(..:....):.*$/$1/;
+ return "a:".$p.":".$c;
+ },
+ lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
+ lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
+ have_children => sub { defined($l->{$_[1]}) },
+ child_code => sub { return $_[1] },
+ },{
+ # 8
+ code_arr => sub { @{$l->{$_[0]}} },
+ filter_code => sub { shift },
+ lookup_v900 => sub {
+ my ($c,$p) = @_;
+ $p =~ s/^a:(..:....):.*$/$1/;
+ return "a:".$p.":".$c;
+ },
+ lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
+ lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
+# have_children => sub { defined($l->{$_[1]}) },
+# child_code => sub { return $_[1] },
+ have_children => sub { 0 },
+ child_code => sub { 0 },
+});
+
+my @show_ids;
+my @hide_ids;
+
+unroll(0,'');
+
+$log->debug("test filter: ",$tree[0]->{'filter_code'}->("root:99"));
+
+sub unroll {
+ my ($level,$start_code) = @_;
+
+ $log->logconfess("need level") unless (defined($level));
+
+ # all levels passed?
+ return if (! defined($tree[$level]));
+
+ $log->debug("unroll level $level, start code $start_code");
+
+ foreach my $code ($tree[$level]->{'code_arr'}->($start_code)) {
+
+ if ($code = $tree[$level]->{'filter_code'}->($code)) {
+
+ $log->debug("# $level filter passed code $code");
+
+ my $v900 = $tree[$level]->{'lookup_v900'}->($code,$start_code) || $log->warn("can't lookup_v900($code,$start_code)");
+ $log->debug("# $level lookup_v900($code,$start_code) = $v900");
+
+ my $term = $tree[$level]->{'lookup_term'}->($code,$v900) || $log->warn("can't lookup_term($code,$v900)");
+ $log->debug("# $level lookup_term($code,$v900) = $term");
+
+ my $mfn = $tree[$level]->{'lookup_mfn'}->($code,$v900) || $log->warn("can't lookup_mfn($code,$v900)");
+ $log->debug("# $level lookup_mfn($code,$v900) = $mfn");
+
+ $log->debug("$code -> $v900 : $term [$mfn]");
+
+ my ($link_start,$link_end) = ('','');
+
+ my $have_children = $tree[$level]->{'have_children'}->($code,$v900,$start_code);
+ if ($have_children) {
+ ($link_start,$link_end) = (qq{},qq{});
+ } else {
+ $log->debug("# $level doesn't have_children($code,$v900,$start_code)");
+ }
+
+ my $mfn_link = "thes/$mfn.html";
+ if (-e "out/$mfn_link") {
+ print HTML " " x $level .
+ qq{- ${link_start}${term}${link_end}}.
+ qq{ »
\n};
+ } else {
+ $log->warn("file 'out/$mfn_link' doesn't exist, skipping");
+ }
+
+ unless ($have_children) {
+ next;
+ }
+ my $style = $tree[$level]->{'style'};
+
+ print HTML " " x $level .
+ qq{\n \n};
+
+ if ($style) {
+ if ($style =~ m/display\s*:\s*none/i) {
+ push @hide_ids, "id$mfn";
+ } else {
+ push @show_ids, "id$mfn";
+ }
+ } else {
+ # default: show
+ push @show_ids, "id$mfn";
+ }
+
+ unroll($level+1, $tree[$level]->{'child_code'}->($code,$v900,$start_code));
+
+ print HTML " " x $level . qq{
\n};
- # output current process indicator
- my $pcnt = int($row->{mfn} * 100 / $max_rowid);
- if ($pcnt != $last_pcnt) {
- printf STDERR ("%5d / %5d -- %-2d %%\n",$row->{mfn},$max_rowid,$pcnt) if (! $opts{q});
- $last_pcnt = $pcnt;
}
}
}
+
+print HTML qq{
+
+
+
+};
+
+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': $!");
+print JS "var show = ['",join("','",@show_ids),"'];\n";
+print JS "var hide = ['",join("','",@hide_ids),"'];\n";
+close(JS);
+
+$log->info("closing index");
+$index->close;
+
+$log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'}));