5 |
use URI::Escape qw/uri_escape/; |
use URI::Escape qw/uri_escape/; |
6 |
use Time::HiRes; |
use Time::HiRes; |
7 |
use POSIX qw/strftime/; |
use POSIX qw/strftime/; |
8 |
|
use Config; |
9 |
|
use threads; |
10 |
|
use Thread::Queue; |
11 |
|
|
12 |
=head1 NAME |
=head1 NAME |
13 |
|
|
15 |
|
|
16 |
=cut |
=cut |
17 |
|
|
18 |
|
die "Your perl isn't compiled with support for ithreads\n" unless ($Config{useithreads}); |
19 |
|
|
20 |
|
|
21 |
my ($from,$to) = @ARGV; |
my ($from,$to) = @ARGV; |
22 |
|
|
23 |
die "usage: $0 http://localhost:1978/node/from http://remote.example.com:1978/node/to\n" unless ($from && $to); |
die "usage: $0 http://localhost:1978/node/from http://remote.example.com:1978/node/to\n" unless ($from && $to); |
24 |
|
|
25 |
my $debug = 0; |
my $debug = 0; |
26 |
|
my $max = 256; |
27 |
|
|
28 |
# create and configure node |
# create and configure node |
29 |
my $from_n = new Search::Estraier::Node( |
my $from_n = new Search::Estraier::Node( |
37 |
debug => $debug, |
debug => $debug, |
38 |
); |
); |
39 |
|
|
40 |
print "Copy from ",$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"; |
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 |
my $doc_num = $from_n->doc_num || 1; |
# total processed elements |
55 |
|
my $i : shared = 1; |
56 |
|
|
57 |
my $res; |
my $q_id = Thread::Queue->new; |
58 |
my $prev; |
my $q_drafts = Thread::Queue->new; |
59 |
my $i = 1; |
|
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(); |
my $t = time(); |
69 |
|
my $t_refresh = time(); |
70 |
|
my $doc_num = $from_n->doc_num || 1; |
71 |
|
|
72 |
|
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 |
do { |
$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 |
|
my $prev; |
95 |
|
my $more = 1; |
96 |
|
|
97 |
|
while($more) { |
98 |
|
my $res; |
99 |
$from_n->shuttle_url( $from_n->{url} . '/list', |
$from_n->shuttle_url( $from_n->{url} . '/list', |
100 |
'application/x-www-form-urlencoded', |
'application/x-www-form-urlencoded', |
101 |
'max=256' . ( $prev ? '&prev=' . uri_escape( $prev ) : '' ), |
'max=' . $max . ( $prev ? '&prev=' . uri_escape( $prev ) : '' ), |
102 |
\$res, |
\$res, |
103 |
); |
); |
104 |
last unless ($res); |
if (! $res || $res eq '') { |
105 |
|
$more = 0; |
106 |
|
last; |
107 |
|
} |
108 |
foreach my $l (split(/\n/,$res)) { |
foreach my $l (split(/\n/,$res)) { |
109 |
(my $id, $prev) = split(/\t/,$l, 2); |
(my $id, $prev) = split(/\t/,$l, 2); |
110 |
$to_n->put_doc( $from_n->get_doc( $id )); |
|
111 |
$i++; |
#$to_n->put_doc( $from_n->get_doc( $id )); |
112 |
|
|
113 |
|
#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 |
|
|
116 |
|
$q_id->enqueue( $id ); |
117 |
} |
} |
118 |
warn "$prev\n"; |
warn "$prev\n"; |
119 |
|
|
120 |
my $rate = ( $i / (time() - $t) ); |
} |
121 |
printf("%d records, %1.2f%% [%1.2f rec/s] estimated finish: %s\n", |
|
122 |
$i, |
$get_thr->join; |
123 |
($i * 100 / $doc_num), |
$put_thr->join; |
|
$rate, |
|
|
strftime("%Y-%m-%d %H:%M:%S", localtime( time() + int(($doc_num-$i) / $rate))), |
|
|
); |
|
124 |
|
|
125 |
} while ($res); |
print "Copy completed.\n"; |
126 |
|
|