--- trunk2/all2all.pl 2004/07/21 16:18:45 390
+++ trunk2/all2all.pl 2004/09/11 08:36:38 422
@@ -13,19 +13,26 @@
use lib './lib';
use WebPAC;
use WebPAC::jsFind;
+use WebPAC::Index;
my $webpac = new WebPAC(
code_page => 'ISO-8859-2',
- limit_mfn => 100,
+# limit_mfn => 100,
# debug => 1,
+ low_mem => 1,
) || die;
-my $log = $webpac->_get_logger();
+my $log = $webpac->_get_logger() || die "can't get logger";
+
+$log->debug("creating WebPAC::jsFind object");
my $index = new WebPAC::jsFind(
index_path => './out/index',
+ keys => 10,
) || die;
+my $thes;
+
$|=1;
my $maxmfn = $webpac->open_isis(
@@ -37,47 +44,52 @@
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' },
{ 'key' => '900_mfn:v900', 'val' => 'v000' },
+ # tree structure
+ { 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' },
+ { 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' },
],
);
+$log->debug("isis file ",$webpac->{'isis_filename'}," opened");
+
$log->info("rows: $maxmfn");
$webpac->open_import_xml(type => 'isis_hidra_ths');
+$log->debug("data: ",Dumper($webpac->{'data'}));
+
while (my $rec = $webpac->fetch_rec) {
my @ds = $webpac->data_structure($rec);
- if ($log->is_debug) {
+ if (0 && $log->is_debug) {
$log->debug("rec = ",Dumper($rec));
$log->debug("ds = ",Dumper(\@ds));
}
next if (! @ds);
- my $filename = $webpac->{'current_filename'};
+ my $filename = $webpac->{'current_filename'} || $log->logdie("no current_filename in webpac object");
if ($filename) {
- open(OUT,"> $filename") || $log->logdie("can't open output '$filename': $!");
- print OUT $webpac->output(
+ $webpac->output_file(
+ file => $filename,
template => 'html.tt',
data => \@ds,
+ headline => $webpac->{'headline'},
);
- close(OUT);
} else {
print $webpac->output(
template => 'text.tt',
data => \@ds,
+ headline => $webpac->{'headline'},
);
}
- my $headline;
- foreach my $ds (@ds) {
- if ($ds->{'tag'} eq 'headline') {
- $headline = join(" ",@{$ds->{'display'}});
- last;
- }
- }
+ my $headline = $webpac->{'headline'};
+
+ my $f = $filename;
+ $f =~ s!out/!!;
# save into index
foreach my $ds (@ds) {
@@ -85,20 +97,137 @@
$index->insert(
index_name => $ds->{'tag'},
- path => $filename,
+ path => $f,
headline => $headline,
words => join(" ",@{$ds->{'swish'}})
);
}
+ # 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 => $h,
+ );
+ }
+ }
+
# print Dumper(\@ds);
}
-if ($log->is_debug) {
+foreach my $t (keys %{$thes}) {
+
+ my @e = $thes->{$t}->elements;
+ if (! @e) {
+ $log->logwarn("no elements in sorted index $t?");
+ next;
+ }
+
+ my $file = "./out/bfilter/$t.txt";
+ $log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]");
+
+ $webpac->output_file(
+ file => $file,
+ template => 'index.tt',
+ data => \@e,
+ index_name => $t,
+ );
+}
+
+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->{$t}));
+ }
}
+$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]");
+
+ print HTML qq{ - $term
\n} if (-e "out/thes/$mfn.html");
+
+ 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->debug(" -> $l3_v900 : $l3_term [$l3_mfn]");
+
+ print HTML qq{ - $l3_term
\n} if (-e "out/thes/$l3_mfn.html");
+ }
+
+ print HTML qq{
\n};
+ }
+ print HTML qq{
\n};
+ }
+}
+
+print HTML qq{
+
+
+
+};
+
$index->close;
+$log->info("elapsed time: ",$webpac->fmt_time(time() - $webpac->{'start_t'}));