webpac2/trunk/run.pl

#!/usr/bin/perl -w

use strict;

use Cwd qw/abs_path/;
use File::Temp qw/tempdir/;
use Data::Dumper;
use lib './lib';

use WebPAC::Lookup;
use WebPAC::Input::ISIS;
use WebPAC::Store 0.03;
use WebPAC::Normalize::XML;
use WebPAC::Output::TT;
use WebPAC::Output::Estraier 0.02;
use YAML qw/LoadFile/;
use LWP::Simple;

my $limit = shift @ARGV;

my $config = LoadFile('conf/config.yml');

print "config = ",Dumper($config);

die "no databases in config file!\n" unless ($config->{databases});

my $total_rows = 0;

while (my ($database, $db_config) = each %{ $config->{databases} }) {

        my $abs_path = abs_path($0);
        $abs_path =~ s#/[^/]*$#/#;

        my $db_path = $config->{webpac}->{db_path} . '/' . $database;

        my $db = new WebPAC::Store(
                path => $db_path,
                database => $database,
                debug => 1,
        );

        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->{masterurl}");

        my $est = new WebPAC::Output::Estraier(
                %{ $est_config },
        );

        #
        # now, iterate through input formats
        #

        my @inputs;
        if (ref($db_config->{input}) eq 'ARRAY') {
                @inputs = @{ $db_config->{input} };
        } else {
                push @inputs, $db_config->{input};
        }

        foreach my $input (@inputs) {

                my $type = lc($input->{type});

                die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis');

                my $lookup = new WebPAC::Lookup(
                        lookup_file => $input->{lookup},
                );

                $log->info("working on input $input->{path} [$input->{type}]");

                my $isis = new WebPAC::Input::ISIS(
                        code_page => $config->{webpac}->{webpac_encoding},
                        limit_mfn => $input->{limit},
                );

                my $maxmfn = $isis->open(
                        filename => $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},
                );

                $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++;
                }

        };

        $log->info("$total_rows records indexed");
}

1	dpavlin	74	#!/usr/bin/perl -w
2
3			use strict;
4
5			use Cwd qw/abs_path/;
6			use File::Temp qw/tempdir/;
7			use Data::Dumper;
8			use lib './lib';
9
10			use WebPAC::Lookup;
11			use WebPAC::Input::ISIS;
12	dpavlin	209	use WebPAC::Store 0.03;
13	dpavlin	74	use WebPAC::Normalize::XML;
14			use WebPAC::Output::TT;
15	dpavlin	210	use WebPAC::Output::Estraier 0.02;
16	dpavlin	141	use YAML qw/LoadFile/;
17	dpavlin	210	use LWP::Simple;
18	dpavlin	74
19	dpavlin	76	my $limit = shift @ARGV;
20
21	dpavlin	141	my $config = LoadFile('conf/config.yml');
22
23			print "config = ",Dumper($config);
24
25	dpavlin	210	die "no databases in config file!\n" unless ($config->{databases});
26	dpavlin	141
27	dpavlin	213	my $total_rows = 0;
28
29	dpavlin	210	while (my ($database, $db_config) = each %{ $config->{databases} }) {
30	dpavlin	141
31	dpavlin	210	my $abs_path = abs_path($0);
32			$abs_path =~ s#/[^/]*$#/#;
33	dpavlin	141
34	dpavlin	210	my $db_path = $config->{webpac}->{db_path} . '/' . $database;
35	dpavlin	74
36	dpavlin	210	my $db = new WebPAC::Store(
37			path => $db_path,
38	dpavlin	217	database => $database,
39	dpavlin	221	debug => 1,
40	dpavlin	210	);
41	dpavlin	74
42	dpavlin	233	my $log = $db->_get_logger;
43			$log->info("working on $database in $db_path");
44
45	dpavlin	210	my $est_config = $config->{hyperestraier} \|\| $log->logdie("can't find 'hyperestraier' part in confguration");
46			$est_config->{database} = $database;
47	dpavlin	113
48	dpavlin	213	$log->info("using HyperEstraier URL $est_config->{masterurl}");
49	dpavlin	74
50	dpavlin	210	my $est = new WebPAC::Output::Estraier(
51			%{ $est_config },
52			);
53	dpavlin	74
54	dpavlin	213	#
55			# now, iterate through input formats
56			#
57	dpavlin	74
58	dpavlin	213	my @inputs;
59			if (ref($db_config->{input}) eq 'ARRAY') {
60			@inputs = @{ $db_config->{input} };
61			} else {
62			push @inputs, $db_config->{input};
63			}
64	dpavlin	74
65	dpavlin	213	foreach my $input (@inputs) {
66	dpavlin	233
67			my $type = lc($input->{type});
68
69			die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis');
70
71			my $lookup = new WebPAC::Lookup(
72			lookup_file => $input->{lookup},
73			);
74
75	dpavlin	213	$log->info("working on input $input->{path} [$input->{type}]");
76	dpavlin	74
77	dpavlin	213	my $isis = new WebPAC::Input::ISIS(
78			code_page => $config->{webpac}->{webpac_encoding},
79			limit_mfn => $input->{limit},
80			);
81	dpavlin	113
82	dpavlin	213	my $maxmfn = $isis->open(
83			filename => $input->{path},
84			code_page => $input->{encoding}, # database encoding
85			);
86	dpavlin	113
87	dpavlin	213	my $n = new WebPAC::Normalize::XML(
88			# filter => { 'foo' => sub { shift } },
89			db => $db,
90			lookup_regex => $lookup->regex,
91			lookup => $lookup,
92	dpavlin	221	prefix => $input->{name},
93	dpavlin	213	);
94	dpavlin	113
95	dpavlin	213	$n->open(
96			tag => $input->{normalize}->{tag},
97			xml_file => $input->{normalize}->{path},
98	dpavlin	210	);
99
100	dpavlin	213	for ( 0 ... $isis->size ) {
101	dpavlin	210
102	dpavlin	213	my $row = $isis->fetch \|\| next;
103
104			my $mfn = $row->{'000'}->[0] \|\| die "can't find MFN";
105
106			my $ds = $n->data_structure($row);
107
108			$est->add(
109			id => $input->{name} . "#" . $mfn,
110			ds => $ds,
111			type => $config->{hyperestraier}->{type},
112			);
113
114			$total_rows++;
115			}
116
117	dpavlin	210	};
118
119			$log->info("$total_rows records indexed");
120			}
121