--- trunk/run.pl 2005/12/05 17:47:04 210 +++ trunk/run.pl 2005/12/15 14:12:00 251 @@ -24,102 +24,98 @@ die "no databases in config file!\n" unless ($config->{databases}); -while (my ($database, $db_config) = each %{ $config->{databases} }) { - - my $type = lc($db_config->{input}->{type}); +my $total_rows = 0; - die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis'); +while (my ($database, $db_config) = each %{ $config->{databases} }) { my $abs_path = abs_path($0); $abs_path =~ s#/[^/]*$#/#; - my $lookup = new WebPAC::Lookup( - lookup_file => $db_config->{input}->{lookup}, - ); - my $db_path = $config->{webpac}->{db_path} . '/' . $database; - - my $log = $lookup->_get_logger; - $log->info("working on $database in $db_path"); - - my $isis = new WebPAC::Input::ISIS( - code_page => $config->{webpac}->{webpac_encoding}, - limit_mfn => $db_config->{input}->{limit}, - ); - - my $maxmfn = $isis->open( - filename => $db_config->{input}->{path}, - code_page => $db_config->{input}->{encoding}, # database encoding - ); - - my $path = './db/'; - my $db = new WebPAC::Store( path => $db_path, + database => $database, + debug => 1, ); - my $n = new WebPAC::Normalize::XML( - # filter => { 'foo' => sub { shift } }, - db => $db, - lookup_regex => $lookup->regex, - lookup => $lookup, - ); - - $n->open( - tag => $db_config->{normalize}->{tag}, - xml_file => $db_config->{normalize}->{path}, - ); - - my $out = new WebPAC::Output::TT( - include_path => $config->{webpac}->{template_path}, - filters => { foo => sub { shift } }, - ); + my $log = $db->_get_logger; + $log->info("working on $database in $db_path"); my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration"); $est_config->{database} = $database; - $log->info("using HyperEstraier URL $est_config->{url}"); + $log->info("using HyperEstraier URL $est_config->{masterurl}"); my $est = new WebPAC::Output::Estraier( %{ $est_config }, ); - my $total_rows = 0; + # + # now, iterate through input formats + # - for ( 0 ... $isis->size ) { + my @inputs; + if (ref($db_config->{input}) eq 'ARRAY') { + @inputs = @{ $db_config->{input} }; + } else { + push @inputs, $db_config->{input}; + } - my $row = $isis->fetch || next; + foreach my $input (@inputs) { - my $mfn = $row->{'000'}->[0] || die "can't find MFN"; + my $type = lc($input->{type}); - my $ds = $n->data_structure($row); + die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis'); - # print STDERR Dumper($row, $ds); + my $lookup = new WebPAC::Lookup( + lookup_file => $input->{lookup}, + ); - # my $html = $out->apply( - # template => 'html_ffzg.tt', - # data => $ds, - # ); - # - # # create test output - # - # my $file = sprintf('out/%02d.html', $mfn ); - # open(my $fh, '>', $file) or die "can't open $file: $!"; - # print $fh $html; - # close($fh); - # - # $html =~ s#\s*[\n\r]+\s*##gs; - # - # print STDERR $html; + $log->info("working on input $input->{path} [$input->{type}]"); - $est->add( - id => $mfn, - ds => $ds, - type => $config->{hyperestraier}->{type}, + my $isis = new WebPAC::Input::ISIS( + code_page => $config->{webpac}->{webpac_encoding}, + limit_mfn => $input->{limit}, + lookup => $lookup, ); - $total_rows++; + my $maxmfn = $isis->open( + filename => $input->{path}, + code_page => $input->{encoding}, # database encoding + ); + +$log->info( Dumper($lookup->{_lookup_data}) ); + + my $n = new WebPAC::Normalize::XML( + # filter => { 'foo' => sub { shift } }, + db => $db, + lookup_regex => $lookup->regex, + lookup => $lookup, + prefix => $input->{name}, + ); + + $n->open( + tag => $input->{normalize}->{tag}, + xml_file => $input->{normalize}->{path}, + ); + + for ( 0 ... $isis->size ) { + + my $row = $isis->fetch || next; + + my $mfn = $row->{'000'}->[0] || die "can't find MFN"; + + my $ds = $n->data_structure($row); + + $est->add( + id => $input->{name} . "#" . $mfn, + ds => $ds, + type => $config->{hyperestraier}->{type}, + ); + + $total_rows++; + } };