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;

# 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 ));

                my $doc_draft = $from_n->_fetch_doc( id => $id, chomp_resbody => 1 );
                $to_n->shuttle_url( $to_n->{url} . '/put_doc', 'text/x-estraier-draft', $doc_draft, undef) == 200 or die "can't insert $doc_draft\n";

                $i++;
        }
        warn "$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	#!/usr/bin/perl -w
2
3	use strict;
4	use Search::Estraier;
5	use URI::Escape qw/uri_escape/;
6	use Time::HiRes;
7	use POSIX qw/strftime/;
8
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	my $max = 256;
21
22	# create and configure node
23	my $from_n = new Search::Estraier::Node(
24	url => $from,
25	croak_on_error => 1,
26	debug => $debug,
27	);
28	my $to_n = new Search::Estraier::Node(
29	url => $to,
30	croak_on_error => 1,
31	debug => $debug,
32	);
33
34	print "Copy from ",$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";
35
36	my $doc_num = $from_n->doc_num \|\| 1;
37
38	my $prev;
39	my $i = 0;
40	my $more = 1;
41
42	my $t = time();
43
44	while($more) {
45	my $res;
46	$from_n->shuttle_url( $from_n->{url} . '/list',
47	'application/x-www-form-urlencoded',
48	'max=' . $max . ( $prev ? '&prev=' . uri_escape( $prev ) : '' ),
49	\$res,
50	);
51	if (! $res \|\| $res eq '') {
52	$more = 0;
53	last;
54	}
55	foreach my $l (split(/\n/,$res)) {
56	(my $id, $prev) = split(/\t/,$l, 2);
57
58	#$to_n->put_doc( $from_n->get_doc( $id ));
59
60	my $doc_draft = $from_n->_fetch_doc( id => $id, chomp_resbody => 1 );
61	$to_n->shuttle_url( $to_n->{url} . '/put_doc', 'text/x-estraier-draft', $doc_draft, undef) == 200 or die "can't insert $doc_draft\n";
62
63	$i++;
64	}
65	warn "$prev\n";
66
67	my $rate = ( $i / (time() - $t) );
68	printf("%d records, %1.2f%% [%1.2f rec/s] estimated finish: %s\n",
69	$i,
70	($i * 100 / $doc_num),
71	$rate,
72	strftime("%Y-%m-%d %H:%M:%S", localtime( time() + int(($doc_num-$i) / $rate))),
73	);
74
75	}
76
77	print "Copy completed.\n";
78