1 |
dpavlin |
1 |
#!/usr/bin/perl |
2 |
|
|
|
3 |
|
|
use warnings; |
4 |
|
|
use strict; |
5 |
|
|
|
6 |
dpavlin |
21 |
our $VERSION = '0.01'; |
7 |
|
|
|
8 |
dpavlin |
1 |
use Time::HiRes qw(time); |
9 |
|
|
use Data::Dump qw(dump); |
10 |
|
|
use File::Slurp; |
11 |
dpavlin |
4 |
use Getopt::Long; |
12 |
dpavlin |
11 |
use IO::Socket::INET; |
13 |
dpavlin |
13 |
use Storable qw/freeze thaw/; |
14 |
dpavlin |
28 |
use Digest::MD5 qw/md5/; |
15 |
dpavlin |
1 |
|
16 |
dpavlin |
4 |
|
17 |
dpavlin |
19 |
my $debug = 0; |
18 |
dpavlin |
4 |
my $path = '/data/isi/full.txt'; |
19 |
dpavlin |
11 |
my $limit = 5000; |
20 |
dpavlin |
4 |
my $offset = 0; |
21 |
dpavlin |
6 |
my @views; |
22 |
dpavlin |
21 |
my $listen = 0; # off |
23 |
dpavlin |
11 |
my @nodes; |
24 |
dpavlin |
4 |
|
25 |
|
|
|
26 |
|
|
GetOptions( |
27 |
|
|
'path=s' => \$path, |
28 |
|
|
'offset=i' => \$offset, |
29 |
|
|
'limit=i' => \$limit, |
30 |
dpavlin |
6 |
'view=s' => \@views, |
31 |
dpavlin |
11 |
'listen|port=i' => \$listen, |
32 |
|
|
'connect=s' => \@nodes, |
33 |
dpavlin |
19 |
'debug!' => \$debug, |
34 |
dpavlin |
5 |
) or die $!; |
35 |
dpavlin |
4 |
|
36 |
dpavlin |
1 |
my $t = time; |
37 |
|
|
|
38 |
dpavlin |
12 |
|
39 |
dpavlin |
21 |
sub send_nodes; |
40 |
|
|
|
41 |
dpavlin |
12 |
our $prefix; |
42 |
dpavlin |
21 |
sub BEGIN { |
43 |
dpavlin |
12 |
$prefix = $0; |
44 |
|
|
if ( $prefix =~ s{^./}{} ) { |
45 |
|
|
chomp( my $pwd = `pwd` ); |
46 |
|
|
$prefix = "$pwd/$prefix"; |
47 |
|
|
} |
48 |
dpavlin |
14 |
$prefix =~ s{^(.*)/srv/Sack/bin.+$}{$1}; |
49 |
dpavlin |
12 |
warn "# prefix $prefix"; |
50 |
dpavlin |
21 |
|
51 |
|
|
$SIG{INT} = sub { |
52 |
|
|
my $signame = shift; |
53 |
|
|
send_nodes 'exit'; |
54 |
|
|
die "SIG$signame"; |
55 |
|
|
}; |
56 |
dpavlin |
12 |
} |
57 |
|
|
|
58 |
|
|
|
59 |
|
|
use lib "$prefix/srv/webpac2/lib/"; |
60 |
dpavlin |
1 |
use WebPAC::Input::ISI; |
61 |
dpavlin |
24 |
|
62 |
|
|
$WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields |
63 |
|
|
|
64 |
dpavlin |
1 |
my $input = WebPAC::Input::ISI->new( |
65 |
dpavlin |
12 |
path => "$prefix/$path", |
66 |
dpavlin |
4 |
offset => $offset, |
67 |
|
|
limit => $limit, |
68 |
dpavlin |
1 |
); |
69 |
|
|
|
70 |
dpavlin |
25 |
our $num_records = $input->size; |
71 |
dpavlin |
1 |
|
72 |
|
|
sub report { |
73 |
|
|
my $description = shift; |
74 |
|
|
my $dt = time - $t; |
75 |
|
|
printf "%s in %1.4fs %.2f/s\n", $description, $dt, $input->size / $dt; |
76 |
dpavlin |
10 |
$t = time; |
77 |
dpavlin |
1 |
} |
78 |
|
|
|
79 |
|
|
|
80 |
|
|
report $input->size . ' records loaded'; |
81 |
|
|
|
82 |
|
|
mkdir 'out' unless -e 'out'; |
83 |
|
|
|
84 |
dpavlin |
5 |
our $out; |
85 |
|
|
|
86 |
dpavlin |
8 |
our $cache; |
87 |
|
|
|
88 |
dpavlin |
13 |
our $connected; |
89 |
|
|
|
90 |
dpavlin |
11 |
sub send_nodes { |
91 |
dpavlin |
17 |
my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument! |
92 |
dpavlin |
21 |
my $header = defined $content ? length($content) : 0; |
93 |
dpavlin |
11 |
$header .= ' ' . join(' ', @_) if @_; |
94 |
|
|
|
95 |
|
|
foreach my $node ( @nodes ) { |
96 |
|
|
|
97 |
|
|
my $sock = IO::Socket::INET->new( |
98 |
|
|
PeerAddr => $node, |
99 |
|
|
Proto => 'tcp', |
100 |
dpavlin |
16 |
); |
101 |
dpavlin |
11 |
|
102 |
dpavlin |
16 |
if ( ! $sock ) { |
103 |
|
|
warn "can't connect to $node - $!"; # FIXME die? |
104 |
|
|
next; |
105 |
|
|
} |
106 |
|
|
|
107 |
dpavlin |
21 |
print ">>>> $listen $node $header\n"; |
108 |
dpavlin |
11 |
print $sock "$header\n$content" || warn "can't send $header to $node: $!"; |
109 |
|
|
|
110 |
dpavlin |
13 |
$connected->{$node} = $sock; |
111 |
dpavlin |
11 |
} |
112 |
|
|
} |
113 |
|
|
|
114 |
dpavlin |
16 |
sub get_node { |
115 |
|
|
my $node = shift; |
116 |
|
|
|
117 |
|
|
my $sock = $connected->{$node}; |
118 |
|
|
if ( ! $sock ) { |
119 |
|
|
warn "ERROR: lost connection to $node"; |
120 |
|
|
delete $connected->{$node}; |
121 |
|
|
return; |
122 |
|
|
} |
123 |
|
|
chomp( my $size = <$sock> ); |
124 |
dpavlin |
21 |
warn "<<<< $listen $node $size bytes\n"; |
125 |
dpavlin |
16 |
my $data; |
126 |
|
|
read $sock, $data, $size; |
127 |
|
|
return $data; |
128 |
|
|
} |
129 |
|
|
|
130 |
|
|
sub send_sock { |
131 |
|
|
my ( $sock, $data ) = @_; |
132 |
|
|
my $size = length $data; |
133 |
dpavlin |
26 |
warn ">>>> $listen ", $sock->peerhost, " $size bytes\n"; |
134 |
dpavlin |
16 |
print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost; |
135 |
|
|
} |
136 |
|
|
|
137 |
dpavlin |
13 |
sub merge_out { |
138 |
|
|
my $new = shift; |
139 |
|
|
|
140 |
|
|
foreach my $k1 ( keys %$new ) { |
141 |
|
|
|
142 |
|
|
foreach my $k2 ( keys %{ $new->{$k1} } ) { |
143 |
|
|
|
144 |
|
|
my $n = $new->{$k1}->{$k2}; |
145 |
|
|
my $ref = ref $out->{$k1}->{$k2}; |
146 |
|
|
|
147 |
|
|
if ( ! defined $out->{$k1}->{$k2} ) { |
148 |
|
|
$out->{$k1}->{$k2} = $n; |
149 |
|
|
} elsif ( $k1 =~ m{\+} ) { |
150 |
dpavlin |
19 |
# warn "## agregate $k1 $k2"; |
151 |
dpavlin |
13 |
$out->{$k1}->{$k2} += $n; |
152 |
dpavlin |
21 |
} elsif ( $ref eq 'ARRAY' ) { |
153 |
|
|
if ( ref $n eq 'ARRAY' ) { |
154 |
|
|
push @{ $out->{$k1}->{$k2} }, $_ foreach @$n; |
155 |
|
|
} else { |
156 |
|
|
push @{ $out->{$k1}->{$k2} }, $n; |
157 |
|
|
} |
158 |
dpavlin |
13 |
} elsif ( $ref eq '' ) { |
159 |
|
|
$out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ]; |
160 |
|
|
} else { |
161 |
|
|
die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2}); |
162 |
|
|
} |
163 |
|
|
} |
164 |
|
|
} |
165 |
|
|
|
166 |
dpavlin |
19 |
warn "## merge out ", dump $out if $debug; |
167 |
dpavlin |
13 |
} |
168 |
|
|
|
169 |
dpavlin |
11 |
sub run_code { |
170 |
|
|
my ( $view, $code ) = @_; |
171 |
|
|
|
172 |
dpavlin |
21 |
warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug; |
173 |
dpavlin |
11 |
|
174 |
|
|
send_nodes view => $view => $code; |
175 |
|
|
|
176 |
|
|
undef $out; |
177 |
|
|
|
178 |
|
|
my $affected = 0; |
179 |
|
|
$t = time; |
180 |
|
|
|
181 |
|
|
foreach my $pos ( $offset + 1 .. $offset + $input->size ) { |
182 |
|
|
my $rec = $cache->{$pos} ||= $input->fetch_rec( $pos ); |
183 |
|
|
if ( ! $rec ) { |
184 |
|
|
warn "END at $pos"; |
185 |
|
|
last; |
186 |
|
|
} |
187 |
|
|
|
188 |
|
|
eval "$code"; |
189 |
|
|
if ( $@ ) { |
190 |
dpavlin |
20 |
warn "ABORT [$pos] $@\n"; |
191 |
|
|
last; |
192 |
dpavlin |
11 |
} else { |
193 |
|
|
$affected++; |
194 |
|
|
} |
195 |
|
|
}; |
196 |
|
|
|
197 |
|
|
report "$affected affected records $view"; |
198 |
|
|
|
199 |
|
|
warn "WARN no \$out defined!" unless defined $out; |
200 |
dpavlin |
13 |
|
201 |
|
|
if ( $connected ) { |
202 |
dpavlin |
24 |
foreach my $node ( keys %$connected ) { |
203 |
|
|
warn "# $listen get_node $node\n"; |
204 |
|
|
my $o = get_node $node; |
205 |
|
|
my $s = length $o; |
206 |
|
|
$o = thaw $o; |
207 |
|
|
warn "# $listen merge $s bytes\n"; |
208 |
|
|
merge_out $o; |
209 |
|
|
} |
210 |
dpavlin |
13 |
} |
211 |
dpavlin |
11 |
} |
212 |
|
|
|
213 |
dpavlin |
1 |
sub run_views { |
214 |
dpavlin |
6 |
@views = sort glob 'views/*.pl' unless @views; |
215 |
dpavlin |
1 |
warn "# views ", dump @views; |
216 |
|
|
|
217 |
|
|
foreach my $view ( @views ) { |
218 |
|
|
|
219 |
|
|
next if system("perl -c $view") != 0; |
220 |
|
|
|
221 |
|
|
my $code = read_file $view; |
222 |
|
|
|
223 |
dpavlin |
11 |
run_code $view => $code; |
224 |
dpavlin |
1 |
|
225 |
dpavlin |
11 |
if ( defined $out ) { |
226 |
|
|
my $dump = dump $out; |
227 |
|
|
my $len = length $dump; |
228 |
dpavlin |
1 |
|
229 |
dpavlin |
11 |
my $path = $view; |
230 |
|
|
$path =~ s{views?/}{out/} || die "no view in $view"; |
231 |
|
|
$path =~ s{\.pl}{}; |
232 |
dpavlin |
1 |
|
233 |
dpavlin |
11 |
print "OUT $view $offset/$limit $len bytes $path" |
234 |
dpavlin |
5 |
, ( $len < 10000 ? " \$out = $dump" : ' SAVED ONLY' ) |
235 |
dpavlin |
1 |
, "\n" |
236 |
|
|
; |
237 |
dpavlin |
5 |
|
238 |
dpavlin |
9 |
unlink "$path.last" if -e "$path.last"; |
239 |
|
|
rename $path, "$path.last"; |
240 |
dpavlin |
1 |
write_file $path, $dump; |
241 |
dpavlin |
11 |
report "SAVE $path"; |
242 |
|
|
} |
243 |
|
|
|
244 |
|
|
} |
245 |
|
|
|
246 |
|
|
} |
247 |
|
|
|
248 |
|
|
if ( $listen ) { |
249 |
|
|
my $sock = IO::Socket::INET->new( |
250 |
|
|
Listen => SOMAXCONN, |
251 |
dpavlin |
21 |
LocalAddr => '127.0.0.1', |
252 |
dpavlin |
11 |
LocalPort => $listen, |
253 |
|
|
Proto => 'tcp', |
254 |
|
|
Reuse => 1, |
255 |
|
|
) or die $!; |
256 |
|
|
|
257 |
|
|
while (1) { |
258 |
|
|
|
259 |
dpavlin |
25 |
warn "NODE $listen ready - path: $path offset: $offset limit: $limit #recs: $num_records\n"; |
260 |
dpavlin |
11 |
|
261 |
|
|
my $client = $sock->accept(); |
262 |
|
|
|
263 |
dpavlin |
16 |
warn "<<<< $listen connect from ", $client->peerhost, $/; |
264 |
dpavlin |
11 |
|
265 |
|
|
my @header = split(/\s/, <$client>); |
266 |
dpavlin |
22 |
warn "<<<< $listen header ",dump(@header),$/; |
267 |
dpavlin |
11 |
|
268 |
|
|
my $size = shift @header; |
269 |
|
|
|
270 |
|
|
my $content; |
271 |
|
|
read $client, $content, $size; |
272 |
|
|
|
273 |
|
|
if ( $header[0] eq 'view' ) { |
274 |
|
|
run_code $header[1] => $content; |
275 |
dpavlin |
16 |
send_sock $client => freeze $out; |
276 |
|
|
} elsif ( $header[0] eq 'info' ) { |
277 |
dpavlin |
25 |
my $info = "$listen\t$offset\t$limit\t$num_records\t$path"; |
278 |
dpavlin |
21 |
$info .= "\t" . eval $header[1] if $header[1]; |
279 |
dpavlin |
16 |
warn "info $info\n"; |
280 |
|
|
send_sock $client => $info; |
281 |
|
|
} elsif ( $header[0] eq 'exit' ) { |
282 |
|
|
warn "exit $listen"; |
283 |
|
|
exit; |
284 |
dpavlin |
5 |
} else { |
285 |
dpavlin |
16 |
warn "WARN $listen unknown"; |
286 |
dpavlin |
1 |
} |
287 |
|
|
|
288 |
|
|
} |
289 |
|
|
} |
290 |
|
|
|
291 |
dpavlin |
3 |
run_views; |
292 |
|
|
|
293 |
dpavlin |
1 |
while ( 1 ) { |
294 |
|
|
|
295 |
|
|
print "sack> "; |
296 |
dpavlin |
21 |
chomp( my $cmd = <STDIN> ); |
297 |
dpavlin |
1 |
|
298 |
dpavlin |
21 |
if ( $cmd =~ m{^(h|\?)} ) { |
299 |
|
|
print << "__HELP__" |
300 |
|
|
Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit |
301 |
|
|
|
302 |
|
|
View Run run views |
303 |
|
|
VI \\e Output show output of last run |
304 |
|
|
Info [\$VERSION] instrospect |
305 |
|
|
Quit EXit shutdown |
306 |
|
|
|
307 |
|
|
__HELP__ |
308 |
|
|
} elsif ( $cmd =~ m{^(vi|\\e|o)}i ) { |
309 |
dpavlin |
3 |
system "vi out/*"; |
310 |
dpavlin |
21 |
} elsif ( $cmd =~ m{^i(?:nfo)?\s?(.+)?$}i ) { |
311 |
dpavlin |
16 |
print "# nodes: ", join(' ',@nodes), $/; |
312 |
|
|
|
313 |
dpavlin |
21 |
send_nodes 'info' => $2; |
314 |
|
|
|
315 |
dpavlin |
16 |
my @info = ( |
316 |
dpavlin |
25 |
"node\toffset\tlimit\t#recs\tpath", |
317 |
|
|
"----\t------\t-----\t-----\t----", |
318 |
|
|
"0\t$offset\t$limit\t$num_records\t$path", |
319 |
dpavlin |
16 |
); |
320 |
|
|
|
321 |
|
|
push @info, get_node $_ foreach @nodes; |
322 |
|
|
|
323 |
|
|
print "$_\n" foreach @info; |
324 |
|
|
|
325 |
dpavlin |
21 |
} elsif ( $cmd =~ m{^(q|e|x)}i ) { |
326 |
dpavlin |
16 |
warn "# exit"; |
327 |
|
|
send_nodes 'exit'; |
328 |
|
|
exit; |
329 |
dpavlin |
21 |
} elsif ( $cmd =~ m{^(v|r)}i ) { |
330 |
dpavlin |
3 |
run_views; |
331 |
dpavlin |
21 |
} elsif ( $cmd ) { |
332 |
|
|
warn "UNKNOWN ", dump $cmd; |
333 |
dpavlin |
3 |
} |
334 |
|
|
|
335 |
dpavlin |
1 |
} |
336 |
|
|
|