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{ $to_n->name }) { |
41 |
dpavlin |
87 |
if ($to =~ m#^(http://.+)/node/([^/]+)$#) { |
42 |
dpavlin |
85 |
my ($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 |
dpavlin |
87 |
#warn "get ", $id || 'undef',"\n"; |
63 |
|
|
if ($id < 0) { |
64 |
|
|
$q_drafts->enqueue( '' ); # abort put thread |
65 |
|
|
last; |
66 |
|
|
}; |
67 |
dpavlin |
86 |
print STDERR "get_thr, id: $id\n" if ($debug); |
68 |
|
|
my $doc_draft = $from_n->_fetch_doc( id => $id, chomp_resbody => 1 ); |
69 |
|
|
$q_drafts->enqueue( $doc_draft ); |
70 |
|
|
} |
71 |
|
|
} ); |
72 |
|
|
|
73 |
|
|
my $t = time(); |
74 |
|
|
my $t_refresh = time(); |
75 |
dpavlin |
82 |
my $doc_num = $from_n->doc_num || 1; |
76 |
dpavlin |
80 |
|
77 |
dpavlin |
86 |
my $put_thr = threads->new( sub { |
78 |
|
|
while (my $doc_draft = $q_drafts->dequeue) { |
79 |
dpavlin |
87 |
last unless ($doc_draft); |
80 |
dpavlin |
86 |
print STDERR "put_thr, $doc_draft\n" if ($debug); |
81 |
|
|
$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"; |
82 |
|
|
|
83 |
|
|
$i++; |
84 |
|
|
if (time() - $t_refresh > 3) { |
85 |
|
|
my $rate = ( $i / ((time() - $t) || 1) ); |
86 |
|
|
printf("%d records, %1.2f%% [%1.2f rec/s] estimated finish: %s\n", |
87 |
|
|
$i, |
88 |
|
|
($i * 100 / $doc_num), |
89 |
|
|
$rate, |
90 |
|
|
strftime("%Y-%m-%d %H:%M:%S", localtime( time() + int(($doc_num-$i) / $rate))), |
91 |
|
|
); |
92 |
|
|
$t_refresh = time(); |
93 |
|
|
} |
94 |
|
|
|
95 |
|
|
} |
96 |
|
|
} ); |
97 |
|
|
|
98 |
|
|
print "Copy from ",$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"; |
99 |
|
|
|
100 |
dpavlin |
80 |
my $prev; |
101 |
dpavlin |
83 |
my $more = 1; |
102 |
dpavlin |
80 |
|
103 |
dpavlin |
83 |
while($more) { |
104 |
|
|
my $res; |
105 |
dpavlin |
80 |
$from_n->shuttle_url( $from_n->{url} . '/list', |
106 |
|
|
'application/x-www-form-urlencoded', |
107 |
dpavlin |
83 |
'max=' . $max . ( $prev ? '&prev=' . uri_escape( $prev ) : '' ), |
108 |
dpavlin |
80 |
\$res, |
109 |
|
|
); |
110 |
dpavlin |
83 |
if (! $res || $res eq '') { |
111 |
|
|
$more = 0; |
112 |
|
|
last; |
113 |
|
|
} |
114 |
dpavlin |
80 |
foreach my $l (split(/\n/,$res)) { |
115 |
|
|
(my $id, $prev) = split(/\t/,$l, 2); |
116 |
dpavlin |
84 |
|
117 |
|
|
#$to_n->put_doc( $from_n->get_doc( $id )); |
118 |
|
|
|
119 |
dpavlin |
86 |
#my $doc_draft = $from_n->_fetch_doc( id => $id, chomp_resbody => 1 ); |
120 |
|
|
#$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"; |
121 |
dpavlin |
84 |
|
122 |
dpavlin |
86 |
$q_id->enqueue( $id ); |
123 |
dpavlin |
80 |
} |
124 |
dpavlin |
87 |
warn "$prev\n" if ($debug); |
125 |
dpavlin |
82 |
|
126 |
dpavlin |
83 |
} |
127 |
dpavlin |
87 |
$q_id->enqueue( -1 ); # last one |
128 |
dpavlin |
80 |
|
129 |
dpavlin |
86 |
$get_thr->join; |
130 |
|
|
$put_thr->join; |
131 |
|
|
|
132 |
dpavlin |
87 |
printf "Copy of %d records completed [%1.2f rec/s]\n", $i, |
133 |
|
|
( $i / ((time() - $t) || 1) ); |
134 |
dpavlin |
83 |
|