use Data::Dumper;
use lib './lib';
+use WebPAC::Common 0.02;
use WebPAC::Lookup;
-use WebPAC::Input::ISIS;
-use WebPAC::DB;
+use WebPAC::Input 0.03;
+use WebPAC::Store 0.03;
use WebPAC::Normalize::XML;
use WebPAC::Output::TT;
-use WebPAC::Output::Estraier;
+use WebPAC::Output::Estraier 0.08;
+use YAML qw/LoadFile/;
+use Getopt::Long;
+use File::Path;
-my $limit = shift @ARGV;
+=head1 NAME
-my $abs_path = abs_path($0);
-$abs_path =~ s#/[^/]*$#/#;
+run.pl - start WebPAC indexing
-my $isis_file = '/data/isis_data/ps/LIBRI/LIBRI';
+B<this command will probably go away. Don't get used to it!>
-my $lookup = new WebPAC::Lookup(
- lookup_file => "$abs_path/conf/lookup/isis.pm",
-);
+Options:
-my $isis = new WebPAC::Input::ISIS(
- code_page => 'ISO-8859-2', # application encoding
- limit_mfn => $limit,
-);
+=over 4
-my $maxmfn = $isis->open(
- filename => $isis_file,
- code_page => '852', # database encoding
-);
+=item --offset 42
-my $path = './db/';
+start loading (all) databases at offset 42
-my $db = new WebPAC::DB(
- path => $path,
-);
+=item --limit 100
-my $n = new WebPAC::Normalize::XML(
-# filter => { 'foo' => sub { shift } },
- db => $db,
- lookup_regex => $lookup->regex,
- lookup => $lookup,
-);
+limit loading to 100 records
-$n->open(
- tag => 'isis',
- xml_file => "$abs_path/conf/normalize/isis_ffzg.xml",
-);
+=item --clean
-my $out = new WebPAC::Output::TT(
- include_path => "$abs_path/conf/output/tt",
- filters => { foo => sub { shift } },
-);
+remove database and Hyper Estraier index before indexing
+
+=item --config conf/config.yml
+
+path to YAML configuration file
+
+=back
+
+=cut
-my $est = new WebPAC::Output::Estraier(
- url => 'http://localhost:1978/node/webpac2',
- user => 'admin',
- passwd => 'admin',
- database => 'ps',
+my $offset;
+my $limit;
+
+my $clean = 0;
+my $config = 'conf/config.yml';
+my $debug = 0;
+
+GetOptions(
+ "limit=i" => \$limit,
+ "offset=i" => \$offset,
+ "clean" => \$clean,
+ "config" => \$config,
+ "debug" => \$debug,
);
-while (my $row = $isis->fetch) {
+$config = LoadFile($config);
- my $mfn = $row->{'000'}->[0] || die "can't find MFN";
+print "config = ",Dumper($config) if ($debug);
- my $ds = $n->data_structure($row);
+die "no databases in config file!\n" unless ($config->{databases});
-# print STDERR Dumper($row, $ds);
+my $total_rows = 0;
- my $html = $out->apply(
- template => 'html_ffzg.tt',
- data => $ds,
- );
+while (my ($database, $db_config) = each %{ $config->{databases} }) {
+
+ my $log = _new WebPAC::Common()->_get_logger();
+
+ #
+ # open Hyper Estraier database
+ #
- # create test output
+ my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration");
+ $est_config->{database} = $database;
+ $est_config->{clean} = $clean;
- my $file = sprintf('out/%02d.html', $mfn );
- open(my $fh, '>', $file) or die "can't open $file: $!";
- print $fh $html;
- close($fh);
+ my $est = new WebPAC::Output::Estraier( %{ $est_config } );
- $html =~ s#\s*[\n\r]+\s*##gs;
+ #
+ # now WebPAC::Store
+ #
+ my $abs_path = abs_path($0);
+ $abs_path =~ s#/[^/]*$#/#;
-# print STDERR $html;
+ my $db_path = $config->{webpac}->{db_path} . '/' . $database;
- $est->add(
- id => $mfn,
- ds => $ds,
- type => 'search',
+ if ($clean) {
+ $log->info("creating new database $database in $db_path");
+ rmtree( $db_path ) || $log->warn("can't remove $db_path: $!");
+ } else {
+ $log->info("working on $database in $db_path");
+ }
+
+ my $db = new WebPAC::Store(
+ path => $db_path,
+ database => $database,
+ debug => $debug,
);
-};
+
+ #
+ # now, iterate through input formats
+ #
+
+ my @inputs;
+ if (ref($db_config->{input}) eq 'ARRAY') {
+ @inputs = @{ $db_config->{input} };
+ } elsif ($db_config->{input}) {
+ push @inputs, $db_config->{input};
+ } else {
+ $log->info("database $database doesn't have inputs defined");
+ }
+
+ my @supported_inputs = keys %{ $config->{webpac}->{inputs} };
+
+ foreach my $input (@inputs) {
+
+ my $type = lc($input->{type});
+
+ die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs));
+
+ my $lookup = new WebPAC::Lookup(
+ lookup_file => $input->{lookup},
+ );
+
+ my $input_module = $config->{webpac}->{inputs}->{$type};
+
+ $log->info("working on input $input->{path} [$input->{type}] using $input_module");
+
+ my $input_db = new WebPAC::Input(
+ module => $input_module,
+ code_page => $config->{webpac}->{webpac_encoding},
+ limit => $limit || $input->{limit},
+ offset => $offset,
+ lookup => $lookup,
+ );
+ $log->logdie("can't create input using $input_module") unless ($input);
+
+ my $maxmfn = $input_db->open(
+ path => $input->{path},
+ code_page => $input->{encoding}, # database encoding
+ );
+
+ my $n = new WebPAC::Normalize::XML(
+ # filter => { 'foo' => sub { shift } },
+ db => $db,
+ lookup_regex => $lookup->regex,
+ lookup => $lookup,
+ prefix => $input->{name},
+ );
+
+ my $normalize_path = $input->{normalize}->{path};
+
+ if ($normalize_path =~ m/\.xml$/i) {
+ $n->open(
+ tag => $input->{normalize}->{tag},
+ xml_file => $input->{normalize}->{path},
+ );
+ } elsif ($normalize_path =~ m/\.(?:yml|yaml)$/i) {
+ $n->open_yaml(
+ path => $normalize_path,
+ tag => $input->{normalize}->{tag},
+ );
+ }
+
+ foreach my $pos ( 0 ... $input_db->size ) {
+
+ my $row = $input_db->fetch || next;
+
+ my $mfn = $row->{'000'}->[0];
+
+ if (! $mfn || $mfn !~ m#^\d+$#) {
+ $log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos");
+ $mfn = $pos;
+ push @{ $row->{'000'} }, $pos;
+ }
+
+ my $ds = $n->data_structure($row);
+
+ $est->add(
+ id => $input->{name} . "/" . $mfn,
+ ds => $ds,
+ type => $config->{hyperestraier}->{type},
+ );
+
+ $total_rows++;
+ }
+
+ };
+
+ $log->info("$total_rows records indexed");
+
+ #
+ # add Hyper Estraier links to other databases
+ #
+ if (ref($db_config->{links}) eq 'ARRAY') {
+ foreach my $link (@{ $db_config->{links} }) {
+ $log->info("adding link $database -> $link->{to} [$link->{credit}]");
+ $est->add_link(
+ from => $database,
+ to => $link->{to},
+ credit => $link->{credit},
+ );
+ }
+ }
+
+}
+
projects / webpac2 / blobdiff