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});

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

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

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

        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,
        );

        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 $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}");

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

        my $total_rows = 0;

        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);

        #       print STDERR Dumper($row, $ds);

        #       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;

                $est->add(
                        id => $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 ($config->{databases});
26	dpavlin	141
27	dpavlin	210	while (my ($database, $db_config) = each %{ $config->{databases} }) {
28	dpavlin	141
29	dpavlin	210	my $type = lc($db_config->{input}->{type});
30	dpavlin	74
31	dpavlin	210	die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis');
32	dpavlin	74
33	dpavlin	210	my $abs_path = abs_path($0);
34			$abs_path =~ s#/[^/]*$#/#;
35	dpavlin	141
36	dpavlin	210	my $lookup = new WebPAC::Lookup(
37			lookup_file => $db_config->{input}->{lookup},
38			);
39	dpavlin	141
40	dpavlin	210	my $db_path = $config->{webpac}->{db_path} . '/' . $database;
41	dpavlin	74
42
43	dpavlin	210	my $log = $lookup->_get_logger;
44			$log->info("working on $database in $db_path");
45	dpavlin	74
46	dpavlin	210	my $isis = new WebPAC::Input::ISIS(
47			code_page => $config->{webpac}->{webpac_encoding},
48			limit_mfn => $db_config->{input}->{limit},
49			);
50	dpavlin	74
51	dpavlin	210	my $maxmfn = $isis->open(
52			filename => $db_config->{input}->{path},
53			code_page => $db_config->{input}->{encoding}, # database encoding
54			);
55	dpavlin	74
56	dpavlin	210	my $path = './db/';
57	dpavlin	74
58	dpavlin	210	my $db = new WebPAC::Store(
59			path => $db_path,
60			);
61	dpavlin	74
62	dpavlin	210	my $n = new WebPAC::Normalize::XML(
63			# filter => { 'foo' => sub { shift } },
64			db => $db,
65			lookup_regex => $lookup->regex,
66			lookup => $lookup,
67			);
68	dpavlin	74
69	dpavlin	210	$n->open(
70			tag => $db_config->{normalize}->{tag},
71			xml_file => $db_config->{normalize}->{path},
72			);
73	dpavlin	74
74	dpavlin	210	my $out = new WebPAC::Output::TT(
75			include_path => $config->{webpac}->{template_path},
76			filters => { foo => sub { shift } },
77			);
78	dpavlin	113
79	dpavlin	210	my $est_config = $config->{hyperestraier} \|\| $log->logdie("can't find 'hyperestraier' part in confguration");
80			$est_config->{database} = $database;
81	dpavlin	113
82	dpavlin	210	$log->info("using HyperEstraier URL $est_config->{url}");
83	dpavlin	74
84	dpavlin	210	my $est = new WebPAC::Output::Estraier(
85			%{ $est_config },
86			);
87	dpavlin	74
88	dpavlin	210	my $total_rows = 0;
89	dpavlin	74
90	dpavlin	210	for ( 0 ... $isis->size ) {
91	dpavlin	74
92	dpavlin	210	my $row = $isis->fetch \|\| next;
93	dpavlin	74
94	dpavlin	210	my $mfn = $row->{'000'}->[0] \|\| die "can't find MFN";
95	dpavlin	113
96	dpavlin	210	my $ds = $n->data_structure($row);
97	dpavlin	113
98	dpavlin	210	# print STDERR Dumper($row, $ds);
99	dpavlin	113
100	dpavlin	210	# my $html = $out->apply(
101			# template => 'html_ffzg.tt',
102			# data => $ds,
103			# );
104			#
105			# # create test output
106			#
107			# my $file = sprintf('out/%02d.html', $mfn );
108			# open(my $fh, '>', $file) or die "can't open $file: $!";
109			# print $fh $html;
110			# close($fh);
111			#
112			# $html =~ s#\s[\n\r]+\s##gs;
113			#
114			# print STDERR $html;
115
116			$est->add(
117			id => $mfn,
118			ds => $ds,
119			type => $config->{hyperestraier}->{type},
120			);
121
122			$total_rows++;
123
124			};
125
126			$log->info("$total_rows records indexed");
127			}
128