/[Sack]/trunk/bin/sack.pl
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Contents of /trunk/bin/sack.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 80 - (show annotations)
Sat Sep 26 10:14:03 2009 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 10527 byte(s)
huge speedup using $merge_digest_mapping to keep information in-memory
between view runs

1 #!/usr/bin/perl
2
3 use warnings;
4 use strict;
5
6 our $VERSION = '0.05';
7
8 use Time::HiRes qw(time);
9 use Data::Dump qw(dump);
10 use File::Slurp;
11 use Getopt::Long;
12 use IO::Socket::INET;
13 use Storable qw/freeze thaw store/;
14
15
16 my $debug = 0;
17 my $path = '/data/isi/full.txt';
18 my $limit = 5000;
19 my $offset = 0;
20 my @views;
21 my $port = 0; # interactive
22 my @nodes;
23
24
25 GetOptions(
26 'path=s' => \$path,
27 'offset=i' => \$offset,
28 'limit=i' => \$limit,
29 'view=s' => \@views,
30 'port|listen=i' => \$port,
31 'node|connect=i' => \@nodes,
32 'debug!' => \$debug,
33 ) or die $!;
34
35 my $t = time;
36
37
38 sub send_nodes;
39
40 our $prefix;
41 sub BEGIN {
42 $prefix = $0;
43 if ( $prefix !~ m{^/} ) {
44 chomp( my $pwd = `pwd` );
45 $prefix = "$pwd/$prefix";
46 }
47 $prefix =~ s{^(.*)/srv/Sack/.+$}{$1};
48 warn "# prefix $prefix";
49
50 $SIG{INT} = sub {
51 my $signame = shift;
52 send_nodes 'exit';
53 #clean if $clean; # FIXME
54 die "SIG$signame";
55 };
56 }
57
58 use lib "$prefix/srv/Sack/lib/";
59 use Sack::Digest;
60 our $digest = Sack::Digest->new( port => $port, clean => 1 );
61 sub digest { $digest->to_int($_[0]) }
62
63 use lib "$prefix/srv/webpac2/lib/";
64 use WebPAC::Input::ISI;
65
66 $WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields
67
68 my $input = WebPAC::Input::ISI->new(
69 path => "$prefix/$path",
70 offset => $offset,
71 limit => $limit,
72 );
73
74 our $num_records = $input->size;
75 our @reports;
76
77 sub report {
78 my $description = join(' ',@_);
79 my $dt = time - $t;
80 my $report = [ $description, $dt, $input->size / $dt ];
81 printf "[$port] %s in %1.4fs %.2f/s\n", @$report;
82 push @reports, $report;
83 $t = time;
84 }
85
86 sub show_report {
87 "\n" . join( "\n", map { sprintf "%8.4fs %10.2f/s %s", $_->[1], $_->[2], $_->[0] } @reports ) . "\n";
88 }
89
90 report $input->size , 'records loaded';
91
92 mkdir 'out' unless -e 'out';
93
94 our $out;
95
96 our $cache;
97
98 our $connected;
99
100 sub node_sock {
101 my $node = shift;
102 my $sock = IO::Socket::INET->new(
103 PeerAddr => '127.0.0.1',
104 PeerPort => $node,
105 Proto => 'tcp',
106 );
107
108 return $sock if $sock && $sock->connected;
109
110 warn "[$port] can't connect to $node - $!\n"; # FIXME die?
111 return;
112 }
113
114 sub send_nodes {
115 my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument!
116 my $header = defined $content ? length($content) : 0;
117 $header .= ' ' . join(' ', @_) if @_;
118
119 warn "# send_nodes ", dump(@_), " to ", dump @nodes;
120
121 foreach my $node ( @nodes ) {
122
123 my $sock = node_sock($node) || next;
124
125 warn "[$port] >>>> [$node] $header\n";
126 print $sock "$header\n$content" || warn "can't send $header to $node: $!";
127
128 $connected->{$node} = $sock;
129 }
130 }
131
132 sub get_node {
133 my $node = shift;
134
135 my $sock = $connected->{$node};
136 if ( ! $sock || ! $sock->connected ) {
137 warn "[$port] no connection to $node";
138 delete $connected->{$node};
139 return;
140 }
141 chomp( my $size = <$sock> );
142 warn "[$port] <<<< [$node] $size bytes\n" if $debug || $size > 1024;
143 my $data;
144 read $sock, $data, $size;
145 return $data;
146 }
147
148 sub send_sock {
149 my ( $sock, $data ) = @_;
150 my $size = length $data;
151 warn "[$port] >>>> $size bytes\n" if $debug || $size > 1024;
152 print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost;
153 }
154
155 sub pull_node_file {
156 my ( $node, $file ) = @_;
157
158 my $path = "/dev/shm/sack.$node.$file";
159 return if -e $path; # FIXME
160
161 my $sock = node_sock $node || die "not connected to $node";
162
163 print $sock "0 file $file\n";
164
165 my $size = <$sock>;
166 chomp($size);
167 warn "[$port] pull_node_file $node $file $size bytes\n";
168
169 my $block = 4096;
170 my $buff = ' ' x $block;
171
172 open(my $fh, '>', $path) || die "can't open $path";
173 while ( read $sock, $buff, $block ) {
174 print $fh $buff;
175 }
176 close($fh);
177 }
178
179 my $merge_digest_mapping;
180
181 sub merge_out {
182 my ( $from_node, $new ) = @_;
183
184 my $t_merge = time();
185
186 pull_node_file $from_node => 'nr_md5';
187 pull_node_file $from_node => 'md5';
188
189 my $remote_digest = Sack::Digest->new( port => $from_node );
190 my ( $local, $remote ) = ( 0, 0 );
191
192 my $tick = 0;
193 print STDERR "[$port] merge [$from_node]";
194
195 my $missing;
196
197 foreach my $k1 ( keys %$new ) {
198
199 foreach my $k2 ( keys %{ $new->{$k1} } ) {
200
201 my $n = delete $new->{$k1}->{$k2};
202
203 if ( $k1 =~ m{#} ) {
204 die "ASSERT $k1 $k2" unless $k2 =~ m{^\d+$};
205 #warn "XXX $k1 $k2";
206
207 if ( defined $merge_digest_mapping->{$from_node}->[ $k2 ] ) {
208 $k2 = $merge_digest_mapping->{$from_node}->[ $k2 ];
209 } else {
210
211 my $md5 = $remote_digest->{nr_md5}->[$k2];
212
213 if ( ! $md5 ) {
214 $missing->{nr_md5}->{$from_node}++; # FIXME die?
215 next;
216 }
217
218 my $local_k2;
219
220 if ( $local_k2 = $digest->{md5_nr}->{$md5} ) {
221 $local++;
222 } elsif ( my $full = $remote_digest->{md5}->{$md5} ) {
223 $local_k2 = $digest->to_int( $remote_digest->{md5}->{$md5} );
224 $remote++;
225 } else {
226 $missing->{md5}->{$from_node}++;
227 }
228
229 $k2 = $merge_digest_mapping->{$from_node}->[ $k2 ] = $local_k2;
230
231 }
232 }
233
234 my $ref = ref $out->{$k1}->{$k2};
235 #warn "XXXX $k1 $k2 $ref";
236 if ( ! defined $out->{$k1}->{$k2} ) {
237 $out->{$k1}->{$k2} = $n;
238 } elsif ( $k1 =~ m{\+} ) {
239 # warn "## agregate $k1 $k2";
240 $out->{$k1}->{$k2} += $n;
241 } elsif ( $ref eq 'ARRAY' ) {
242 if ( ref $n eq 'ARRAY' ) {
243 push @{ $out->{$k1}->{$k2} }, $_ foreach @$n;
244 } else {
245 push @{ $out->{$k1}->{$k2} }, $n;
246 }
247 } elsif ( $ref eq '' ) {
248 $out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ];
249 } else {
250 die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2});
251 }
252
253 if ( $tick++ % 1000 == 0 ) {
254 print STDERR ".";
255 } elsif ( $tick % 10000 == 0 ) {
256 print STDERR $tick;
257 }
258 }
259 }
260
261 $t_merge = time - $t_merge;
262 my $digests = $local + $remote;
263 printf STDERR "\n[$port] merge %d in %.4fs %.2f/s digests local: %.1f%% %d/%d\n", $tick, $t_merge, $digests / $t_merge, $local * 100 / ( $digests || 1 ), $local, $remote;
264 push @reports, [ "$tick merged $from_node", $t_merge, $tick / $t_merge ];
265
266 warn "[$port] missing ", dump $missing if $missing;
267
268 warn "## merge out ", dump $out if $debug;
269 }
270
271 sub run_code {
272 my ( $view, $code ) = @_;
273
274 warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug;
275
276 send_nodes view => $view => $code;
277
278 undef $out;
279
280 my $affected = 0;
281 $t = time;
282
283 foreach my $pos ( $offset + 1 .. $offset + $input->size ) {
284 my $rec = $cache->{$pos} ||= $input->fetch_rec( $pos );
285 if ( ! $rec ) {
286 print STDERR "END @ $pos";
287 last;
288 }
289
290 eval "$code";
291 if ( $@ ) {
292 warn "ABORT $pos $@\n";
293 last;
294 } else {
295 $affected++;
296 }
297
298 $pos % 10000 == 0 ? print STDERR $pos :
299 $pos % 1000 == 0 ? print STDERR "." : 0 ;
300 };
301
302 report "$affected affected $view";
303
304 warn "WARN no \$out defined!" unless defined $out;
305
306 $digest->sync;
307
308 if ( $connected ) {
309 foreach my $node ( keys %$connected ) {
310 warn "[$port] get_node [$node]\n";
311 my $o = get_node $node;
312 next unless $o;
313 my $s = length $o;
314 $o = thaw $o;
315 warn "[$port] got $s bytes from [$node]\n";
316 merge_out $node => $o;
317 }
318 }
319 }
320
321 sub run_views {
322 @views = sort glob 'views/*.pl' unless @views;
323 warn "# views ", dump @views;
324
325 foreach my $view ( @views ) {
326
327 next if system("perl -c $view") != 0;
328
329 my $code = read_file $view;
330
331 run_code $view => $code;
332
333 if ( defined $out ) {
334
335 my $path = $view;
336 $path =~ s{views?/}{out/} || die "no view in $view";
337 $path =~ s{\.pl}{.storable};
338
339 unlink "$path.last" if -e "$path.last";
340 rename $path, "$path.last";
341
342 store $out => $path;
343 report "save $path", -s $path, "bytes";
344
345 if ( -s $path < 4096 ) {
346 print '$out = ', dump $digest->undigest_out($out);
347 }
348 }
349
350 }
351
352 }
353
354
355 sub info_tabs {
356 "$port\t$offset\t$limit\t$num_records\t$path\t"
357 . join("\t", map {
358 my $b = $_;
359 $b =~ s{^.+\.$port\.([^/]+)$}{$1};
360 "$b " . -s $_
361 } glob "/dev/shm/sack.$port.*" );
362 }
363
364
365 if ( $port ) {
366
367 my $pid_path = "/tmp/sack.$port.pid";
368 if ( -e $pid_path ) {
369 my $pid = read_file $pid_path;
370 kill 9, $pid && warn "[$port] kill old $pid\n";
371 }
372 write_file $pid_path, $$;
373
374
375 my $sock = IO::Socket::INET->new(
376 Listen => SOMAXCONN,
377 LocalAddr => '127.0.0.1',
378 LocalPort => $port,
379 Proto => 'tcp',
380 Reuse => 1,
381 ) or die $!;
382
383 while (1) {
384
385 warn "[$port] accept path: $path offset: $offset limit: $limit #recs: $num_records\n";
386
387 my $client = $sock->accept();
388
389 warn "[$port] <<<< connect from ", $client->peerhost, $/;
390
391 my @header = split(/\s/, <$client>);
392 warn "[$port] <<<< header ",dump(@header),$/;
393
394 my $size = shift @header;
395
396 my $content;
397 read $client, $content, $size;
398
399 if ( $header[0] eq 'view' ) {
400 run_code $header[1] => $content;
401 send_sock $client => freeze $out;
402 } elsif ( $header[0] eq 'info' ) {
403 my $info = info_tabs;
404 warn "[$port] info $info\n";
405 $info .= "\n" . show_report if $content =~ m{r}i;
406 send_sock $client => $info;
407 } elsif ( $header[0] eq 'exit' ) {
408 warn "[$port] exit";
409 exit;
410 } elsif ( $header[0] eq 'file' ) {
411 $digest->close;
412 my $path = "/dev/shm/sack.$port.$header[1]";
413 my $size = -s $path;
414 warn "[$port] >>>> file $path $size bytes\n";
415 print $client "$size\n";
416 my $block = 4096;
417 my $buff = ' ' x $block;
418 open(my $fh, '<', $path) || die "can't open $path";
419 while ( read $fh, $buff, $block ) {
420 print $client $buff;
421 }
422 $digest->open;
423 } else {
424 warn "[$port] UNKNOWN $header[0]";
425 }
426
427 $client->close;
428
429 }
430 }
431
432 sub info {
433 my $detail = shift || '';
434
435 send_nodes 'info' => $detail;
436
437 my @info = (
438 "port\toffset\tlimit\t#recs\tpath",
439 "----\t------\t-----\t-----\t----",
440 info_tabs,
441 );
442
443 push @info, show_report if $detail =~ m{r}i;
444
445 push @info, get_node $_ foreach @nodes;
446
447 print "[$port] INFO", $detail ? " $detail" : '', " \n"
448 , join("\n", @info)
449 , "\n" ;
450
451 return @info;
452 }
453
454 info;
455 while ( keys %$connected != @nodes ) {
456 warn "[$port] wait for ", join(' ', grep { ! defined $connected->{$_} } @nodes );
457 sleep 1;
458 }
459 run_views;
460
461 while ( 1 ) {
462
463 print "sack> ";
464 chomp( my $cmd = <STDIN> );
465
466 if ( $cmd =~ m{^(h|\?)} ) {
467 print << "__HELP__"
468 Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit
469
470 View Run run views
471 VI \\e Output show output of last run
472 Info [\$VERSION] instrospect
473 Quit EXit shutdown
474
475 __HELP__
476 } elsif ( $cmd =~ m{^(vi|\\e|o)}i ) {
477 #system "vi out/*";
478 $digest->sync;
479 system "bin/storableedit.pl", (glob('out/*.storable'))[0];
480 } elsif ( $cmd =~ m{^i(?:nfo)?\s?(\S+)?$}i ) {
481 info $1;
482 } elsif ( $cmd =~ m{^(q|e|x)}i ) {
483 warn "# exit";
484 send_nodes 'exit';
485 exit;
486 } elsif ( $cmd =~ m{^(v|r)}i ) {
487 run_views;
488 } elsif ( $cmd =~ m{^n(ode)?\s*(\d+)}i ) {
489 push @nodes, $2;
490 info;
491 } elsif ( $cmd ) {
492 warn "UNKNOWN ", dump $cmd;
493 }
494
495 }
496

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26