1 |
dpavlin |
1 |
#!/usr/bin/perl |
2 |
|
|
|
3 |
|
|
use warnings; |
4 |
|
|
use strict; |
5 |
|
|
|
6 |
dpavlin |
50 |
our $VERSION = '0.04'; |
7 |
dpavlin |
21 |
|
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 |
30 |
use Storable qw/freeze thaw store/; |
14 |
dpavlin |
1 |
|
15 |
dpavlin |
4 |
|
16 |
dpavlin |
19 |
my $debug = 0; |
17 |
dpavlin |
4 |
my $path = '/data/isi/full.txt'; |
18 |
dpavlin |
11 |
my $limit = 5000; |
19 |
dpavlin |
4 |
my $offset = 0; |
20 |
dpavlin |
6 |
my @views; |
21 |
dpavlin |
30 |
my $port = 0; # interactive |
22 |
dpavlin |
11 |
my @nodes; |
23 |
dpavlin |
4 |
|
24 |
|
|
|
25 |
|
|
GetOptions( |
26 |
|
|
'path=s' => \$path, |
27 |
|
|
'offset=i' => \$offset, |
28 |
|
|
'limit=i' => \$limit, |
29 |
dpavlin |
6 |
'view=s' => \@views, |
30 |
dpavlin |
30 |
'listen|port=i' => \$port, |
31 |
dpavlin |
11 |
'connect=s' => \@nodes, |
32 |
dpavlin |
19 |
'debug!' => \$debug, |
33 |
dpavlin |
5 |
) or die $!; |
34 |
dpavlin |
4 |
|
35 |
dpavlin |
1 |
my $t = time; |
36 |
|
|
|
37 |
dpavlin |
12 |
|
38 |
dpavlin |
21 |
sub send_nodes; |
39 |
|
|
|
40 |
dpavlin |
12 |
our $prefix; |
41 |
dpavlin |
21 |
sub BEGIN { |
42 |
dpavlin |
12 |
$prefix = $0; |
43 |
dpavlin |
30 |
if ( $prefix !~ m{^/} ) { |
44 |
dpavlin |
12 |
chomp( my $pwd = `pwd` ); |
45 |
|
|
$prefix = "$pwd/$prefix"; |
46 |
|
|
} |
47 |
dpavlin |
51 |
$prefix =~ s{^(.*)/srv/Sack/.+$}{$1}; |
48 |
dpavlin |
12 |
warn "# prefix $prefix"; |
49 |
dpavlin |
21 |
|
50 |
|
|
$SIG{INT} = sub { |
51 |
|
|
my $signame = shift; |
52 |
|
|
send_nodes 'exit'; |
53 |
dpavlin |
33 |
#clean if $clean; # FIXME |
54 |
dpavlin |
21 |
die "SIG$signame"; |
55 |
|
|
}; |
56 |
dpavlin |
12 |
} |
57 |
|
|
|
58 |
dpavlin |
33 |
use lib "$prefix/srv/Sack/lib/"; |
59 |
|
|
use Sack::Digest; |
60 |
dpavlin |
45 |
our $digest = Sack::Digest->new( port => $port, clean => 1 ); |
61 |
|
|
sub digest { $digest->to_int($_[0]) } |
62 |
dpavlin |
12 |
|
63 |
|
|
use lib "$prefix/srv/webpac2/lib/"; |
64 |
dpavlin |
1 |
use WebPAC::Input::ISI; |
65 |
dpavlin |
24 |
|
66 |
|
|
$WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields |
67 |
|
|
|
68 |
dpavlin |
1 |
my $input = WebPAC::Input::ISI->new( |
69 |
dpavlin |
12 |
path => "$prefix/$path", |
70 |
dpavlin |
4 |
offset => $offset, |
71 |
|
|
limit => $limit, |
72 |
dpavlin |
1 |
); |
73 |
|
|
|
74 |
dpavlin |
25 |
our $num_records = $input->size; |
75 |
dpavlin |
1 |
|
76 |
|
|
sub report { |
77 |
dpavlin |
30 |
my $description = join(' ',@_); |
78 |
dpavlin |
1 |
my $dt = time - $t; |
79 |
|
|
printf "%s in %1.4fs %.2f/s\n", $description, $dt, $input->size / $dt; |
80 |
dpavlin |
10 |
$t = time; |
81 |
dpavlin |
1 |
} |
82 |
|
|
|
83 |
|
|
|
84 |
|
|
report $input->size . ' records loaded'; |
85 |
|
|
|
86 |
|
|
mkdir 'out' unless -e 'out'; |
87 |
|
|
|
88 |
dpavlin |
5 |
our $out; |
89 |
|
|
|
90 |
dpavlin |
8 |
our $cache; |
91 |
|
|
|
92 |
dpavlin |
13 |
our $connected; |
93 |
|
|
|
94 |
dpavlin |
11 |
sub send_nodes { |
95 |
dpavlin |
17 |
my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument! |
96 |
dpavlin |
21 |
my $header = defined $content ? length($content) : 0; |
97 |
dpavlin |
11 |
$header .= ' ' . join(' ', @_) if @_; |
98 |
|
|
|
99 |
|
|
foreach my $node ( @nodes ) { |
100 |
|
|
|
101 |
|
|
my $sock = IO::Socket::INET->new( |
102 |
|
|
PeerAddr => $node, |
103 |
|
|
Proto => 'tcp', |
104 |
dpavlin |
16 |
); |
105 |
dpavlin |
11 |
|
106 |
dpavlin |
16 |
if ( ! $sock ) { |
107 |
|
|
warn "can't connect to $node - $!"; # FIXME die? |
108 |
|
|
next; |
109 |
|
|
} |
110 |
|
|
|
111 |
dpavlin |
30 |
warn "[$port] >>>> $node $header\n"; |
112 |
dpavlin |
11 |
print $sock "$header\n$content" || warn "can't send $header to $node: $!"; |
113 |
|
|
|
114 |
dpavlin |
13 |
$connected->{$node} = $sock; |
115 |
dpavlin |
11 |
} |
116 |
|
|
} |
117 |
|
|
|
118 |
dpavlin |
16 |
sub get_node { |
119 |
|
|
my $node = shift; |
120 |
|
|
|
121 |
|
|
my $sock = $connected->{$node}; |
122 |
|
|
if ( ! $sock ) { |
123 |
dpavlin |
30 |
warn "[$port] ERROR lost connection to $node"; |
124 |
dpavlin |
16 |
delete $connected->{$node}; |
125 |
|
|
return; |
126 |
|
|
} |
127 |
|
|
chomp( my $size = <$sock> ); |
128 |
dpavlin |
30 |
warn "[$port] <<<< $node $size bytes\n"; |
129 |
dpavlin |
16 |
my $data; |
130 |
|
|
read $sock, $data, $size; |
131 |
|
|
return $data; |
132 |
|
|
} |
133 |
|
|
|
134 |
|
|
sub send_sock { |
135 |
|
|
my ( $sock, $data ) = @_; |
136 |
|
|
my $size = length $data; |
137 |
dpavlin |
30 |
warn "[$port] >>>> ", $sock->peerhost, " $size bytes\n"; |
138 |
dpavlin |
16 |
print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost; |
139 |
|
|
} |
140 |
|
|
|
141 |
dpavlin |
13 |
sub merge_out { |
142 |
dpavlin |
50 |
my ( $from_node, $new ) = @_; |
143 |
dpavlin |
13 |
|
144 |
dpavlin |
50 |
warn "### merge $from_node"; |
145 |
|
|
|
146 |
|
|
my $from_port = $from_node; |
147 |
|
|
$from_port =~ s{.+:(\d+)$}{$1}; |
148 |
|
|
|
149 |
|
|
my $remote_digest = Sack::Digest->new( port => $from_port ); |
150 |
|
|
my ( $local, $remote ) = ( 0, 0 ); |
151 |
|
|
|
152 |
dpavlin |
13 |
foreach my $k1 ( keys %$new ) { |
153 |
|
|
|
154 |
|
|
foreach my $k2 ( keys %{ $new->{$k1} } ) { |
155 |
|
|
|
156 |
dpavlin |
30 |
my $n = delete $new->{$k1}->{$k2}; |
157 |
dpavlin |
50 |
|
158 |
|
|
if ( $k1 =~ m{#} ) { |
159 |
|
|
die "ASSERT $k1 $k2" unless $k2 =~ m{^\d+$}; |
160 |
|
|
#warn "XXX $k1 $k2"; |
161 |
|
|
my $md5 = $remote_digest->{nr_md5}->[$k2] || warn "[$port] no2md5 $n not found in $from_port\n"; |
162 |
|
|
if ( my $local_k2 = $digest->{md5_nr}->{$md5} ) { |
163 |
|
|
$k2 = $local_k2; |
164 |
|
|
$local++; |
165 |
|
|
} else { |
166 |
|
|
$k2 = $digest->to_int( $remote_digest->{md5}->{$md5} ); |
167 |
|
|
$remote++; |
168 |
|
|
} |
169 |
|
|
} |
170 |
|
|
|
171 |
dpavlin |
30 |
my $ref = ref $out->{$k1}->{$k2}; |
172 |
dpavlin |
50 |
#warn "XXXX $k1 $k2 $ref"; |
173 |
dpavlin |
13 |
if ( ! defined $out->{$k1}->{$k2} ) { |
174 |
|
|
$out->{$k1}->{$k2} = $n; |
175 |
|
|
} elsif ( $k1 =~ m{\+} ) { |
176 |
dpavlin |
19 |
# warn "## agregate $k1 $k2"; |
177 |
dpavlin |
13 |
$out->{$k1}->{$k2} += $n; |
178 |
dpavlin |
21 |
} elsif ( $ref eq 'ARRAY' ) { |
179 |
|
|
if ( ref $n eq 'ARRAY' ) { |
180 |
|
|
push @{ $out->{$k1}->{$k2} }, $_ foreach @$n; |
181 |
|
|
} else { |
182 |
|
|
push @{ $out->{$k1}->{$k2} }, $n; |
183 |
|
|
} |
184 |
dpavlin |
13 |
} elsif ( $ref eq '' ) { |
185 |
|
|
$out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ]; |
186 |
|
|
} else { |
187 |
|
|
die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2}); |
188 |
|
|
} |
189 |
|
|
} |
190 |
|
|
} |
191 |
|
|
|
192 |
dpavlin |
50 |
warn "[$port] merge local $local remote $remote from $from_port\n"; |
193 |
dpavlin |
19 |
warn "## merge out ", dump $out if $debug; |
194 |
dpavlin |
13 |
} |
195 |
|
|
|
196 |
dpavlin |
11 |
sub run_code { |
197 |
|
|
my ( $view, $code ) = @_; |
198 |
|
|
|
199 |
dpavlin |
21 |
warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug; |
200 |
dpavlin |
11 |
|
201 |
|
|
send_nodes view => $view => $code; |
202 |
|
|
|
203 |
|
|
undef $out; |
204 |
|
|
|
205 |
|
|
my $affected = 0; |
206 |
|
|
$t = time; |
207 |
|
|
|
208 |
|
|
foreach my $pos ( $offset + 1 .. $offset + $input->size ) { |
209 |
|
|
my $rec = $cache->{$pos} ||= $input->fetch_rec( $pos ); |
210 |
|
|
if ( ! $rec ) { |
211 |
dpavlin |
41 |
print STDERR "END @ $pos"; |
212 |
dpavlin |
11 |
last; |
213 |
|
|
} |
214 |
|
|
|
215 |
|
|
eval "$code"; |
216 |
|
|
if ( $@ ) { |
217 |
dpavlin |
41 |
warn "ABORT $pos $@\n"; |
218 |
dpavlin |
20 |
last; |
219 |
dpavlin |
11 |
} else { |
220 |
|
|
$affected++; |
221 |
|
|
} |
222 |
dpavlin |
30 |
|
223 |
dpavlin |
40 |
$pos % 10000 == 0 ? print STDERR $pos : |
224 |
|
|
$pos % 1000 == 0 ? print STDERR "." : 0 ; |
225 |
dpavlin |
11 |
}; |
226 |
|
|
|
227 |
dpavlin |
30 |
report "\n[$port] RECS $affected $view"; |
228 |
dpavlin |
11 |
|
229 |
|
|
warn "WARN no \$out defined!" unless defined $out; |
230 |
dpavlin |
13 |
|
231 |
dpavlin |
46 |
$digest->sync; |
232 |
|
|
|
233 |
dpavlin |
13 |
if ( $connected ) { |
234 |
dpavlin |
24 |
foreach my $node ( keys %$connected ) { |
235 |
dpavlin |
30 |
warn "[$port] get_node $node\n"; |
236 |
dpavlin |
24 |
my $o = get_node $node; |
237 |
|
|
my $s = length $o; |
238 |
|
|
$o = thaw $o; |
239 |
dpavlin |
30 |
warn "[$port] merge $node $s bytes\n"; |
240 |
dpavlin |
50 |
merge_out $node => $o; |
241 |
dpavlin |
24 |
} |
242 |
dpavlin |
13 |
} |
243 |
dpavlin |
11 |
} |
244 |
|
|
|
245 |
dpavlin |
1 |
sub run_views { |
246 |
dpavlin |
6 |
@views = sort glob 'views/*.pl' unless @views; |
247 |
dpavlin |
1 |
warn "# views ", dump @views; |
248 |
|
|
|
249 |
|
|
foreach my $view ( @views ) { |
250 |
|
|
|
251 |
|
|
next if system("perl -c $view") != 0; |
252 |
|
|
|
253 |
|
|
my $code = read_file $view; |
254 |
|
|
|
255 |
dpavlin |
11 |
run_code $view => $code; |
256 |
dpavlin |
1 |
|
257 |
dpavlin |
11 |
if ( defined $out ) { |
258 |
dpavlin |
1 |
|
259 |
dpavlin |
11 |
my $path = $view; |
260 |
|
|
$path =~ s{views?/}{out/} || die "no view in $view"; |
261 |
dpavlin |
30 |
$path =~ s{\.pl}{.storable}; |
262 |
dpavlin |
1 |
|
263 |
dpavlin |
9 |
unlink "$path.last" if -e "$path.last"; |
264 |
|
|
rename $path, "$path.last"; |
265 |
dpavlin |
30 |
|
266 |
|
|
store $out => $path; |
267 |
dpavlin |
31 |
report "[$port] SAVE $path $offset-$limit", -s $path, "bytes"; |
268 |
dpavlin |
35 |
|
269 |
|
|
if ( -s $path < 4096 ) { |
270 |
dpavlin |
45 |
print '$out = ', dump $digest->undigest_out($out); |
271 |
dpavlin |
35 |
} |
272 |
dpavlin |
11 |
} |
273 |
|
|
|
274 |
|
|
} |
275 |
|
|
|
276 |
|
|
} |
277 |
|
|
|
278 |
dpavlin |
35 |
|
279 |
|
|
sub info_tabs { |
280 |
|
|
"$port\t$offset\t$limit\t$num_records\t$path\t" |
281 |
dpavlin |
36 |
. join("\t", map { |
282 |
dpavlin |
35 |
my $b = $_; |
283 |
dpavlin |
46 |
$b =~ s{^.+\.$port\.([^/]+)$}{$1}; |
284 |
dpavlin |
35 |
"$b " . -s $_ |
285 |
dpavlin |
46 |
} glob "/dev/shm/sack.$port.*" ); |
286 |
dpavlin |
35 |
} |
287 |
|
|
|
288 |
|
|
|
289 |
dpavlin |
30 |
if ( $port ) { |
290 |
dpavlin |
11 |
my $sock = IO::Socket::INET->new( |
291 |
|
|
Listen => SOMAXCONN, |
292 |
dpavlin |
21 |
LocalAddr => '127.0.0.1', |
293 |
dpavlin |
30 |
LocalPort => $port, |
294 |
dpavlin |
11 |
Proto => 'tcp', |
295 |
|
|
Reuse => 1, |
296 |
|
|
) or die $!; |
297 |
|
|
|
298 |
|
|
while (1) { |
299 |
|
|
|
300 |
dpavlin |
30 |
warn "[$port] READY path: $path offset: $offset limit: $limit #recs: $num_records\n"; |
301 |
dpavlin |
11 |
|
302 |
|
|
my $client = $sock->accept(); |
303 |
|
|
|
304 |
dpavlin |
30 |
warn "[$port] <<<< connect from ", $client->peerhost, $/; |
305 |
dpavlin |
11 |
|
306 |
|
|
my @header = split(/\s/, <$client>); |
307 |
dpavlin |
30 |
warn "[$port] <<<< header ",dump(@header),$/; |
308 |
dpavlin |
11 |
|
309 |
|
|
my $size = shift @header; |
310 |
|
|
|
311 |
|
|
my $content; |
312 |
|
|
read $client, $content, $size; |
313 |
|
|
|
314 |
|
|
if ( $header[0] eq 'view' ) { |
315 |
|
|
run_code $header[1] => $content; |
316 |
dpavlin |
16 |
send_sock $client => freeze $out; |
317 |
|
|
} elsif ( $header[0] eq 'info' ) { |
318 |
dpavlin |
35 |
my $info = info_tabs; |
319 |
dpavlin |
30 |
warn "[$port] info $info\n"; |
320 |
dpavlin |
16 |
send_sock $client => $info; |
321 |
|
|
} elsif ( $header[0] eq 'exit' ) { |
322 |
dpavlin |
30 |
warn "[$port] exit"; |
323 |
dpavlin |
16 |
exit; |
324 |
dpavlin |
5 |
} else { |
325 |
dpavlin |
30 |
warn "[$port] UNKNOWN $header[0]"; |
326 |
dpavlin |
1 |
} |
327 |
|
|
|
328 |
|
|
} |
329 |
|
|
} |
330 |
|
|
|
331 |
dpavlin |
30 |
sub info { |
332 |
|
|
send_nodes 'info' => $2; |
333 |
|
|
|
334 |
|
|
my @info = ( |
335 |
dpavlin |
36 |
"port\toffset\tlimit\t#recs\tpath", |
336 |
|
|
"----\t------\t-----\t-----\t----", |
337 |
dpavlin |
35 |
info_tabs, |
338 |
dpavlin |
30 |
); |
339 |
|
|
|
340 |
|
|
push @info, get_node $_ foreach @nodes; |
341 |
|
|
|
342 |
|
|
print "[$port] INFO\n" |
343 |
|
|
, join("\n", @info) |
344 |
|
|
, "\n\n" ; |
345 |
|
|
|
346 |
|
|
return @info; |
347 |
|
|
} |
348 |
|
|
|
349 |
|
|
info; |
350 |
dpavlin |
3 |
run_views; |
351 |
|
|
|
352 |
dpavlin |
1 |
while ( 1 ) { |
353 |
|
|
|
354 |
|
|
print "sack> "; |
355 |
dpavlin |
21 |
chomp( my $cmd = <STDIN> ); |
356 |
dpavlin |
1 |
|
357 |
dpavlin |
21 |
if ( $cmd =~ m{^(h|\?)} ) { |
358 |
|
|
print << "__HELP__" |
359 |
|
|
Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit |
360 |
|
|
|
361 |
|
|
View Run run views |
362 |
|
|
VI \\e Output show output of last run |
363 |
|
|
Info [\$VERSION] instrospect |
364 |
|
|
Quit EXit shutdown |
365 |
|
|
|
366 |
|
|
__HELP__ |
367 |
|
|
} elsif ( $cmd =~ m{^(vi|\\e|o)}i ) { |
368 |
dpavlin |
33 |
#system "vi out/*"; |
369 |
|
|
system "bin/storableedit.pl", (glob('out/*.storable'))[0]; |
370 |
dpavlin |
21 |
} elsif ( $cmd =~ m{^i(?:nfo)?\s?(.+)?$}i ) { |
371 |
dpavlin |
30 |
info; |
372 |
dpavlin |
21 |
} elsif ( $cmd =~ m{^(q|e|x)}i ) { |
373 |
dpavlin |
16 |
warn "# exit"; |
374 |
|
|
send_nodes 'exit'; |
375 |
|
|
exit; |
376 |
dpavlin |
21 |
} elsif ( $cmd =~ m{^(v|r)}i ) { |
377 |
dpavlin |
3 |
run_views; |
378 |
dpavlin |
21 |
} elsif ( $cmd ) { |
379 |
|
|
warn "UNKNOWN ", dump $cmd; |
380 |
dpavlin |
3 |
} |
381 |
|
|
|
382 |
dpavlin |
1 |
} |
383 |
|
|
|