--- trunk2/all2all.pl 2004/09/12 19:15:54 430 +++ trunk2/all2all.pl 2004/09/23 23:17:42 466 @@ -7,19 +7,38 @@ =cut use strict; -use Data::Dumper; +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 => 500, -# 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 +46,7 @@ $log->debug("creating WebPAC::jsFind object"); my $index = new WebPAC::jsFind( - index_path => './out/index', + index_path => $index_path, keys => 10, ) || die; @@ -47,6 +66,7 @@ # tree structure { 'eval' => 'length("v251") == 2', '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' }, ], ); @@ -63,8 +83,8 @@ 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 +115,19 @@ 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 +139,13 @@ foreach my $h (@{$ds->{'index'}}) { $thes->{$ds->{'tag'}}->insert( - path => $f, headline => $h, + mfn => $webpac->mfn, ); } } -# print Dumper(\@ds); +# print Dump(\@ds); } @@ -141,57 +169,23 @@ } 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->{$t})); } } } # XXX if(0) -$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); +$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); + +$log->info("creating tree"); # -# tree dump implementation which shouldn't be here :-) +# define tree structure # - -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 - - - - - - - - -
-Folding: -
-« default » -
-reset -all -none -
-debug -
- - - - -}; - -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); +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;