trunk/scripts/estcp-mt.pl

#!/usr/bin/perl -w

use strict;
use Search::Estraier;
use URI::Escape qw/uri_escape/;
use Time::HiRes;
use POSIX qw/strftime/;
use Config;
use threads; 
use Thread::Queue;

=head1 NAME

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

=cut

die "Your perl isn't compiled with support for ithreads\n" unless ($Config{useithreads});


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,
);

unless(eval{ $to_n->name }) {
        if ($to =~ m#^(http://.+)/node/(\w+)$#) {
                my ($url,$name) = ($1,$2);
                print "Creating '$name' on $url\n";
                $to_n->shuttle_url( $url . '/master?action=nodeadd',
                        'application/x-www-form-urlencoded',
                        'name=' . uri_escape($name) . '&label=' . uri_escape( $name ),
                        undef,
                );
        } else {
                die "can't extract node name from $to\n";
        }
}

# total processed elements
my $i : shared = 1;

my $q_id = Thread::Queue->new;
my $q_drafts = Thread::Queue->new;

my $get_thr = threads->new( sub {
        while (my $id = $q_id->dequeue) {
                print STDERR "get_thr, id: $id\n" if ($debug);
                my $doc_draft = $from_n->_fetch_doc( id => $id, chomp_resbody => 1 );
                $q_drafts->enqueue( $doc_draft );
        }
} );

my $t = time();
my $t_refresh = time();
my $doc_num = $from_n->doc_num || 1;

my $put_thr = threads->new( sub {
        while (my $doc_draft = $q_drafts->dequeue) {
                print STDERR "put_thr, $doc_draft\n" if ($debug);
                $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++;
                if (time() - $t_refresh > 3) {
                        my $rate = ( $i / ((time() - $t) || 1) );
                        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))),
                        );
                        $t_refresh = time();
                }

        }
} );

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 $prev;
my $more = 1;

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";

                $q_id->enqueue( $id );
        }
        warn "$prev\n";

}

$get_thr->join;
$put_thr->join;

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	86	use Config;
9			use threads;
10			use Thread::Queue;
11	dpavlin	80
12			=head1 NAME
13
14			estcp.pl - copy Hyper Estraier index from one node to another
15
16			=cut
17
18	dpavlin	86	die "Your perl isn't compiled with support for ithreads\n" unless ($Config{useithreads});
19
20
21	dpavlin	80	my ($from,$to) = @ARGV;
22
23			die "usage: $0 http://localhost:1978/node/from http://remote.example.com:1978/node/to\n" unless ($from && $to);
24
25			my $debug = 0;
26	dpavlin	83	my $max = 256;
27	dpavlin	80
28			# create and configure node
29			my $from_n = new Search::Estraier::Node(
30			url => $from,
31			croak_on_error => 1,
32			debug => $debug,
33			);
34			my $to_n = new Search::Estraier::Node(
35			url => $to,
36			croak_on_error => 1,
37			debug => $debug,
38			);
39
40	dpavlin	85	unless(eval{ $to_n->name }) {
41			if ($to =~ m#^(http://.+)/node/(\w+)$#) {
42			my ($url,$name) = ($1,$2);
43			print "Creating '$name' on $url\n";
44			$to_n->shuttle_url( $url . '/master?action=nodeadd',
45			'application/x-www-form-urlencoded',
46			'name=' . uri_escape($name) . '&label=' . uri_escape( $name ),
47			undef,
48			);
49			} else {
50			die "can't extract node name from $to\n";
51			}
52			}
53
54	dpavlin	86	# total processed elements
55			my $i : shared = 1;
56	dpavlin	80
57	dpavlin	86	my $q_id = Thread::Queue->new;
58			my $q_drafts = Thread::Queue->new;
59
60			my $get_thr = threads->new( sub {
61			while (my $id = $q_id->dequeue) {
62			print STDERR "get_thr, id: $id\n" if ($debug);
63			my $doc_draft = $from_n->_fetch_doc( id => $id, chomp_resbody => 1 );
64			$q_drafts->enqueue( $doc_draft );
65			}
66			} );
67
68			my $t = time();
69			my $t_refresh = time();
70	dpavlin	82	my $doc_num = $from_n->doc_num \|\| 1;
71	dpavlin	80
72	dpavlin	86	my $put_thr = threads->new( sub {
73			while (my $doc_draft = $q_drafts->dequeue) {
74			print STDERR "put_thr, $doc_draft\n" if ($debug);
75			$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";
76
77			$i++;
78			if (time() - $t_refresh > 3) {
79			my $rate = ( $i / ((time() - $t) \|\| 1) );
80			printf("%d records, %1.2f%% [%1.2f rec/s] estimated finish: %s\n",
81			$i,
82			($i * 100 / $doc_num),
83			$rate,
84			strftime("%Y-%m-%d %H:%M:%S", localtime( time() + int(($doc_num-$i) / $rate))),
85			);
86			$t_refresh = time();
87			}
88
89			}
90			} );
91
92			print "Copy from ",$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";
93
94	dpavlin	80	my $prev;
95	dpavlin	83	my $more = 1;
96	dpavlin	80
97	dpavlin	83	while($more) {
98			my $res;
99	dpavlin	80	$from_n->shuttle_url( $from_n->{url} . '/list',
100			'application/x-www-form-urlencoded',
101	dpavlin	83	'max=' . $max . ( $prev ? '&prev=' . uri_escape( $prev ) : '' ),
102	dpavlin	80	\$res,
103			);
104	dpavlin	83	if (! $res \|\| $res eq '') {
105			$more = 0;
106			last;
107			}
108	dpavlin	80	foreach my $l (split(/\n/,$res)) {
109			(my $id, $prev) = split(/\t/,$l, 2);
110	dpavlin	84
111			#$to_n->put_doc( $from_n->get_doc( $id ));
112
113	dpavlin	86	#my $doc_draft = $from_n->_fetch_doc( id => $id, chomp_resbody => 1 );
114			#$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";
115	dpavlin	84
116	dpavlin	86	$q_id->enqueue( $id );
117	dpavlin	80	}
118	dpavlin	82	warn "$prev\n";
119
120	dpavlin	83	}
121	dpavlin	80
122	dpavlin	86	$get_thr->join;
123			$put_thr->join;
124
125	dpavlin	83	print "Copy completed.\n";
126