trunk/scripts/estcp.pl

#!/usr/bin/perl -w

use strict;
use Search::Estraier;
use URI::Escape qw/uri_escape/;
use Time::HiRes;
use POSIX qw/strftime/;

=head1 NAME

estcp.pl - copy Hyper Estraier index from one node to another

=cut

my ($from,$to) = @ARGV;

die "usage: $0 http://localhost:1978/node/from http://remote.example.com:1978/node/to\n" unless ($from && $to);

my $debug = 0;
my $max = 256;

$max = 1024;

# create and configure node
my $from_n = new Search::Estraier::Node(
        url => $from,
        croak_on_error => 1,
        debug => $debug,
);
my $to_n = new Search::Estraier::Node(
        url => $to,
        croak_on_error => 1,
        debug => $debug,
);

print "Copy from ",$from_n->name," (",$from_n->label,") to ",$to_n->name," (",$to_n->label,") - ",$from_n->doc_num," documents (",$from_n->word_num," words, ",$from_n->size," bytes)\n";

my $doc_num = $from_n->doc_num || 1;

my $prev;
my $i = 0;
my $more = 1;

my $t = time();

while($more) {
        my $res;
        $from_n->shuttle_url( $from_n->{url} . '/list',
                'application/x-www-form-urlencoded',
                'max=' . $max . ( $prev ? '&prev=' . uri_escape( $prev ) : '' ),
                \$res,
        );
        if (! $res || $res eq '') {
                $more = 0;
                last;
        }
        foreach my $l (split(/\n/,$res)) {
                (my $id, $prev) = split(/\t/,$l, 2);
                $to_n->put_doc( $from_n->get_doc( $id ));
                $i++;
        }
        warn "$prev\n";

        my $rate = ( $i / (time() - $t) );
        printf("%d records, %1.2f%% [%1.2f rec/s] estimated finish: %s\n",
                $i,
                ($i * 100 / $doc_num),
                $rate, 
                strftime("%Y-%m-%d %H:%M:%S", localtime( time() + int(($doc_num-$i) / $rate))),
        );

}

print "Copy completed.\n";

1	dpavlin	80	#!/usr/bin/perl -w
2
3			use strict;
4			use Search::Estraier;
5			use URI::Escape qw/uri_escape/;
6	dpavlin	82	use Time::HiRes;
7			use POSIX qw/strftime/;
8	dpavlin	80
9			=head1 NAME
10
11			estcp.pl - copy Hyper Estraier index from one node to another
12
13			=cut
14
15			my ($from,$to) = @ARGV;
16
17			die "usage: $0 http://localhost:1978/node/from http://remote.example.com:1978/node/to\n" unless ($from && $to);
18
19			my $debug = 0;
20	dpavlin	83	my $max = 256;
21	dpavlin	80
22	dpavlin	83	$max = 1024;
23
24	dpavlin	80	# create and configure node
25			my $from_n = new Search::Estraier::Node(
26			url => $from,
27			croak_on_error => 1,
28			debug => $debug,
29			);
30			my $to_n = new Search::Estraier::Node(
31			url => $to,
32			croak_on_error => 1,
33			debug => $debug,
34			);
35
36			print "Copy from ",$from_n->name," (",$from_n->label,") to ",$to_n->name," (",$to_n->label,") - ",$from_n->doc_num," documents (",$from_n->word_num," words, ",$from_n->size," bytes)\n";
37
38	dpavlin	82	my $doc_num = $from_n->doc_num \|\| 1;
39	dpavlin	80
40			my $prev;
41	dpavlin	83	my $i = 0;
42			my $more = 1;
43	dpavlin	80
44	dpavlin	82	my $t = time();
45
46	dpavlin	83	while($more) {
47			my $res;
48	dpavlin	80	$from_n->shuttle_url( $from_n->{url} . '/list',
49			'application/x-www-form-urlencoded',
50	dpavlin	83	'max=' . $max . ( $prev ? '&prev=' . uri_escape( $prev ) : '' ),
51	dpavlin	80	\$res,
52			);
53	dpavlin	83	if (! $res \|\| $res eq '') {
54			$more = 0;
55			last;
56			}
57	dpavlin	80	foreach my $l (split(/\n/,$res)) {
58			(my $id, $prev) = split(/\t/,$l, 2);
59			$to_n->put_doc( $from_n->get_doc( $id ));
60	dpavlin	82	$i++;
61	dpavlin	80	}
62	dpavlin	82	warn "$prev\n";
63
64			my $rate = ( $i / (time() - $t) );
65			printf("%d records, %1.2f%% [%1.2f rec/s] estimated finish: %s\n",
66			$i,
67			($i * 100 / $doc_num),
68			$rate,
69			strftime("%Y-%m-%d %H:%M:%S", localtime( time() + int(($doc_num-$i) / $rate))),
70			);
71
72	dpavlin	83	}
73	dpavlin	80
74	dpavlin	83	print "Copy completed.\n";
75