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::DB;
use WebPAC::Normalize::XML;
use WebPAC::Output::TT;
use WebPAC::Output::Estraier;

my $limit = shift @ARGV;

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

my $isis_file = '/data/isis_data/ps/LIBRI/LIBRI';

my $lookup = new WebPAC::Lookup(
        lookup_file => "$abs_path/conf/lookup/isis.pm",
);

my $isis = new WebPAC::Input::ISIS(
        code_page => 'ISO-8859-2',      # application encoding
        limit_mfn => $limit,
);

my $maxmfn = $isis->open(
        filename => $isis_file,
        code_page => '852',             # database encoding
);

my $path = './db/';

my $db = new WebPAC::DB(
        path => $path,
);

my $n = new WebPAC::Normalize::XML(
#       filter => { 'foo' => sub { shift } },
        db => $db,
        lookup_regex => $lookup->regex,
        lookup => $lookup,
);

$n->open(
        tag => 'isis',
        xml_file => "$abs_path/conf/normalize/isis_ffzg.xml",
);

my $out = new WebPAC::Output::TT(
        include_path => "$abs_path/conf/output/tt",
        filters => { foo => sub { shift } },
);

my $est = new WebPAC::Output::Estraier(
        url => 'http://localhost:1978/node/webpac2',
        user => 'admin',
        passwd => 'admin',
        database => 'ps',
);

while (my $row = $isis->fetch) {

        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 => 'search',
        );

};
1	#!/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	use WebPAC::DB;
13	use WebPAC::Normalize::XML;
14	use WebPAC::Output::TT;
15	use WebPAC::Output::Estraier;
16
17	my $limit = shift @ARGV;
18
19	my $abs_path = abs_path($0);
20	$abs_path =~ s#/[^/]*$#/#;
21
22	my $isis_file = '/data/isis_data/ps/LIBRI/LIBRI';
23
24	my $lookup = new WebPAC::Lookup(
25	lookup_file => "$abs_path/conf/lookup/isis.pm",
26	);
27
28	my $isis = new WebPAC::Input::ISIS(
29	code_page => 'ISO-8859-2', # application encoding
30	limit_mfn => $limit,
31	);
32
33	my $maxmfn = $isis->open(
34	filename => $isis_file,
35	code_page => '852', # database encoding
36	);
37
38	my $path = './db/';
39
40	my $db = new WebPAC::DB(
41	path => $path,
42	);
43
44	my $n = new WebPAC::Normalize::XML(
45	# filter => { 'foo' => sub { shift } },
46	db => $db,
47	lookup_regex => $lookup->regex,
48	lookup => $lookup,
49	);
50
51	$n->open(
52	tag => 'isis',
53	xml_file => "$abs_path/conf/normalize/isis_ffzg.xml",
54	);
55
56	my $out = new WebPAC::Output::TT(
57	include_path => "$abs_path/conf/output/tt",
58	filters => { foo => sub { shift } },
59	);
60
61	my $est = new WebPAC::Output::Estraier(
62	url => 'http://localhost:1978/node/webpac2',
63	user => 'admin',
64	passwd => 'admin',
65	database => 'ps',
66	);
67
68	while (my $row = $isis->fetch) {
69
70	my $mfn = $row->{'000'}->[0] \|\| die "can't find MFN";
71
72	my $ds = $n->data_structure($row);
73
74	# print STDERR Dumper($row, $ds);
75
76	my $html = $out->apply(
77	template => 'html_ffzg.tt',
78	data => $ds,
79	);
80
81	# create test output
82
83	my $file = sprintf('out/%02d.html', $mfn );
84	open(my $fh, '>', $file) or die "can't open $file: $!";
85	print $fh $html;
86	close($fh);
87
88	$html =~ s#\s[\n\r]+\s##gs;
89
90	# print STDERR $html;
91
92	$est->add(
93	id => $mfn,
94	ds => $ds,
95	type => 'search',
96	);
97
98	};