--- trunk2/all2all.pl 2004/09/10 22:24:42 421
+++ trunk2/all2all.pl 2004/10/09 21:44:25 490
@@ -7,19 +7,39 @@
=cut
use strict;
-use Data::Dumper;
+use locale;
+use YAML;
use Carp;
+use Getopt::Long;
use lib './lib';
use WebPAC;
use WebPAC::jsFind;
use WebPAC::Index;
+use WebPAC::Tree;
+# 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 => 'ISO-8859-2',
-# limit_mfn => 100,
-# debug => 1,
- low_mem => 1,
+ 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";
@@ -27,7 +47,7 @@
$log->debug("creating WebPAC::jsFind object");
my $index = new WebPAC::jsFind(
- index_path => './out/index',
+ index_path => $index_path,
keys => 10,
) || die;
@@ -45,8 +65,9 @@
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
{ 'key' => '900_mfn:v900', 'val' => 'v000' },
# tree structure
- { 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' },
+ { 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' },
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
+ { 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' },
],
);
@@ -56,15 +77,15 @@
$webpac->open_import_xml(type => 'isis_hidra_ths');
-$log->debug("data: ",Dumper($webpac->{'data'}));
+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));
+ $log->debug("rec = ",Dump($rec));
+ $log->debug("ds = ",Dump(\@ds));
}
next if (! @ds);
@@ -95,11 +116,19 @@
foreach my $ds (@ds) {
next if (! $ds->{'swish'});
+ # strip all non word characters from beginning or end
+ # of word
+ my $words = join(" ",@{$ds->{'swish'}});
+ $words =~ s/^\W+//;
+ $words =~ s/\W*\s+\W*/ /g;
+ $words =~ s/\W+$//;
+
$index->insert(
index_name => $ds->{'tag'},
- path => $f,
+ #path => $f,
+ path => $webpac->mfn,
headline => $headline,
- words => join(" ",@{$ds->{'swish'}})
+ words => $words,
);
}
@@ -111,13 +140,13 @@
foreach my $h (@{$ds->{'index'}}) {
$thes->{$ds->{'tag'}}->insert(
- path => $f,
headline => $h,
+ mfn => $webpac->mfn,
);
}
}
-# print Dumper(\@ds);
+# print Dump(\@ds);
}
@@ -141,92 +170,132 @@
}
if (0 && $log->is_debug) {
- $log->debug("lookup hash: ",Dumper($webpac->{'lookup'}));
- $log->debug("data hash: ",Dumper($webpac->{'data'}));
+ $log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
+ $log->debug("data hash: ",Dump($webpac->{'data'}));
foreach my $t (keys %{$thes}) {
- $log->debug("thesaurus $t hash: ",Dumper($thes->{$t}));
+ $log->debug("thesaurus $t hash: ",Dump($thes->{$t}));
}
}
-$log->info("lookup hash: ",Dumper($webpac->{'lookup'}));
-
-#
-# tree dump implementation which shouldn't be here :-)
-#
-
-my $file = 'out/browse.html';
-open(HTML, "> $file") || $log->logdie("can't open '$file'");
-
-print HTML qq{
-
-
- Browse
-
-
-
-
-};
-
-foreach my $code (sort keys %{$webpac->{'lookup'}}) {
-
- my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
-
- if ($code =~ m/^root:/) {
- my $v900 = shift @{ $l->{$code} } || $log->logconfess("can't lookup '$code'");
- $code =~ s/^root:// || die;
-
- my $term = shift @{ $l->{"d:${v900}"} } || die;
- my $mfn = shift @{ $l->{"900_mfn:${v900}"} } || die;
-
- $log->debug("$code -> $v900 : $term [$mfn]");
+} # XXX if(0)
- print HTML qq{ - $term
\n} if (-e "out/thes/$mfn.html");
+$log->debug("lookup hash: ",Dump($webpac->{'lookup'}));
- unless ($l->{"a:${code}::"}) {
- $log->warn("can't find 'a:${code}::'");
- next;
- }
-
- print HTML qq{ \n};
-
- foreach my $l2_v900 (@{ $l->{"a:${code}::"} }) {
-
- my $l2_code = shift @{ $l->{"code:${l2_v900}"} } || die;
- my $l2_term = shift @{ $l->{"d:$l2_v900"} } || die;
- my $l2_mfn = shift @{ $l->{"900_mfn:${l2_v900}"} } || die;
-
- $log->debug("$l2_code -> $l2_v900 : $l2_term [$l2_mfn]");
-
- print HTML qq{ - $l2_term
\n} if (-e "out/thes/$l2_mfn.html");
-
- unless ($l->{"a:${l2_code}:"}) {
- $log->warn("can't find 'a:${l2_code}:'");
- next;
- }
-
- print HTML qq{ \n};
-
- foreach my $l3_v900 (@{ $l->{"a:${l2_code}:"} }) {
-
- my $l3_term = shift @{ $l->{"d:$l3_v900"} } || die;
- my $l3_mfn = shift @{ $l->{"900_mfn:${l3_v900}"} } || die;
+$log->info("creating tree");
- $log->debug(" -> $l3_v900 : $l3_term [$l3_mfn]");
+#
+# define tree structure
+#
- print HTML qq{ - $l3_term
\n} if (-e "out/thes/$l3_mfn.html");
- }
+my $l = $webpac->{'lookup'} || $log->logconfess("can't find lookup");
- print HTML qq{
\n};
- }
- print HTML qq{
\n};
- }
-}
-
-print HTML qq{
-
-
-
-};
+my @tree = ({
+ # level 0
+ code_arr => sub { sort keys %{$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 { return $l->{"a:".$_[0]."::"} },
+ have_children_at_level => sub {
+ return unless (defined($l->{"code:".$_[1]}));
+ my $code = shift @{$l->{"code:".$_[1]}};
+ print STDERR "## $_[1] -> $code\n";
+ return undef unless($code);
+ return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"}));
+ },
+ style => 'display: none',
+ },{
+ # 1
+ code_arr => sub { @{$_[0]} },
+ filter_code => sub { shift }, # nop
+ lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
+ lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
+ lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
+ have_children => sub { return $l->{"a:".$_[1].":"} },
+ style => 'display: none',
+ },{
+ # 2
+ code_arr => sub { @{$_[0]} },
+ filter_code => sub { shift },
+ lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} },
+ lookup_term => sub { shift @{$l->{"d:".$_[0]}} },
+ lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} },
+ have_children => sub { return $l->{"a:".$_[1].":".$_[0]} },
+ #style => 'display: none',
+ },{
+ # 3 uži pojam
+ code_arr => sub { @{$_[0]} },
+ filter_code => sub { shift },
+ 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 { return $l->{"a:".$_[1].":".$_[0]} },
+ },{
+ # 4
+ code_arr => sub { @{$_[0]} },
+ filter_code => sub { shift },
+ 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 { return $l->{"a:".$_[1].":".$_[0]} },
+ },{
+ # 5
+ code_arr => sub { @{$_[0]} },
+ filter_code => sub { shift },
+ 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 { return $l->{"a:".$_[1].":".$_[0]} },
+ },{
+ # 6
+ code_arr => sub { @{$_[0]} },
+ filter_code => sub { shift },
+ 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 { return $l->{"a:".$_[1].":".$_[0]} },
+ },{
+ # 7
+ code_arr => sub { @{$_[0]} },
+ filter_code => sub { shift },
+ 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 { return $l->{"a:".$_[1].":".$_[0]} },
+ },{
+ # 8
+ code_arr => sub { @{$_[0]} },
+ filter_code => sub { shift },
+ 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 { return $l->{"a:".$_[1].":".$_[0]} },
+ have_children => sub { 0 },
+ },{
+ # 9 - level which is never reached except explicitly
+ code_arr => sub { @{$_[0]} },
+ filter_code => sub { shift },
+ 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 { 0 },
+ have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) },
+ },{
+});
+
+my $tree = new WebPAC::Tree(
+ dir => './out',
+ html => 'browse.html',
+ template => './output_template/tree.tt',
+ js => 'tree-ids.js',
+ tree => \@tree,
+);
+$log->info("closing index");
$index->close;
+$log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'}));