1 |
dpavlin |
80 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
|
|
use strict; |
4 |
|
|
use Search::Estraier; |
5 |
|
|
use URI::Escape qw/uri_escape/; |
6 |
|
|
|
7 |
|
|
=head1 NAME |
8 |
|
|
|
9 |
|
|
estcp.pl - copy Hyper Estraier index from one node to another |
10 |
|
|
|
11 |
|
|
=cut |
12 |
|
|
|
13 |
|
|
my ($from,$to) = @ARGV; |
14 |
|
|
|
15 |
|
|
die "usage: $0 http://localhost:1978/node/from http://remote.example.com:1978/node/to\n" unless ($from && $to); |
16 |
|
|
|
17 |
|
|
my $debug = 0; |
18 |
|
|
|
19 |
|
|
# create and configure node |
20 |
|
|
my $from_n = new Search::Estraier::Node( |
21 |
|
|
url => $from, |
22 |
|
|
croak_on_error => 1, |
23 |
|
|
debug => $debug, |
24 |
|
|
); |
25 |
|
|
my $to_n = new Search::Estraier::Node( |
26 |
|
|
url => $to, |
27 |
|
|
croak_on_error => 1, |
28 |
|
|
debug => $debug, |
29 |
|
|
); |
30 |
|
|
|
31 |
|
|
print "Copy from ",$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"; |
32 |
|
|
|
33 |
|
|
my $tick = int($from_n->doc_num / 78); |
34 |
|
|
|
35 |
|
|
print "["; |
36 |
|
|
my $res; |
37 |
|
|
my $prev; |
38 |
|
|
my $i = 0; |
39 |
|
|
|
40 |
|
|
do { |
41 |
|
|
$from_n->shuttle_url( $from_n->{url} . '/list', |
42 |
|
|
'application/x-www-form-urlencoded', |
43 |
|
|
'max=256' . ( $prev ? '&prev=' . uri_escape( $prev ) : '' ), |
44 |
|
|
\$res, |
45 |
|
|
); |
46 |
|
|
last unless ($res); |
47 |
|
|
foreach my $l (split(/\n/,$res)) { |
48 |
|
|
(my $id, $prev) = split(/\t/,$l, 2); |
49 |
|
|
$to_n->put_doc( $from_n->get_doc( $id )); |
50 |
|
|
} |
51 |
|
|
print "$prev\n"; |
52 |
|
|
print "." if ($i++ % $tick == 0); |
53 |
|
|
} while ($res); |
54 |
|
|
|
55 |
|
|
print "]\n"; |