/[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 84 - (show annotations)
Tue Sep 29 16:54:05 2009 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 11135 byte(s)
speedup 10% by evaling one once into coderef

1 #!/usr/bin/perl
2
3 use warnings;
4 use strict;
5
6 our $VERSION = '0.06';
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 sub port2color {
58 my $port = shift;
59 return "\e[1m0\e[0m" if $port == 0;
60
61 my $c = ( $port % 6 ) + 31;
62 return "\e[${c}m$port\e[0m";
63 }
64
65 $SIG{__WARN__} = sub {
66 return unless @_;
67 my $msg = join('', @_);
68 if ( $msg !~ m{[\n\r]$} ) {
69 my @loc = caller(1);
70 $msg .= " in $loc[1] +$loc[2]\n" if @loc;
71 }
72 $msg =~ s{\[(0|\d\d\d\d)\]}{ '[' . port2color($1) . ']' }eg;
73 print STDERR $msg;
74 return 1;
75 };
76
77 }
78
79 use lib "$prefix/srv/Sack/lib/";
80 use Sack::Digest;
81 our $digest = Sack::Digest->new( port => $port, clean => 1 );
82 sub digest { $digest->to_int($_[0]) }
83
84 use lib "$prefix/srv/webpac2/lib/";
85 use WebPAC::Input::ISI;
86
87 $WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields
88
89 my $input = WebPAC::Input::ISI->new(
90 path => "$prefix/$path",
91 offset => $offset,
92 limit => $limit,
93 );
94
95 our $num_records = $input->size;
96 our @reports;
97
98 sub report {
99 my $description = join(' ',@_);
100 my $dt = time - $t;
101 my $report = [ $description, $dt, $input->size / $dt ];
102 warn sprintf "[$port] %s in %1.4fs %.2f/s\n", @$report;
103 push @reports, $report;
104 $t = time;
105 }
106
107 sub show_report {
108 "\n" . join( "\n", map { sprintf "%8.4fs %10.2f/s %s", $_->[1], $_->[2], $_->[0] } @reports ) . "\n";
109 }
110
111 report $input->size , 'records loaded';
112
113 mkdir 'out' unless -e 'out';
114
115 our $out;
116
117 our $cache;
118
119 our $connected;
120
121 sub node_sock {
122 my $node = shift;
123 my $sock = IO::Socket::INET->new(
124 PeerAddr => '127.0.0.1',
125 PeerPort => $node,
126 Proto => 'tcp',
127 );
128
129 return $sock if $sock && $sock->connected;
130
131 warn "[$port] can't connect to $node - $!\n"; # FIXME die?
132 return;
133 }
134
135 sub send_nodes {
136 my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument!
137 my $header = defined $content ? length($content) : 0;
138 $header .= ' ' . join(' ', @_) if @_;
139
140 warn "# send_nodes ", dump(@_), " to ", dump @nodes;
141
142 foreach my $node ( @nodes ) {
143
144 my $sock = node_sock($node) || next;
145
146 warn "[$port] >>>> [$node] $header\n";
147 print $sock "$header\n$content" || warn "can't send $header to $node: $!";
148
149 $connected->{$node} = $sock;
150 }
151 }
152
153 sub get_node {
154 my $node = shift;
155
156 my $sock = $connected->{$node};
157 if ( ! $sock || ! $sock->connected ) {
158 warn "[$port] no connection to $node";
159 delete $connected->{$node};
160 return;
161 }
162 chomp( my $size = <$sock> );
163 warn "[$port] <<<< [$node] $size bytes\n" if $debug || $size > 1024;
164 my $data;
165 read $sock, $data, $size;
166 return $data;
167 }
168
169 sub send_sock {
170 my ( $sock, $data ) = @_;
171 my $size = length $data;
172 warn "[$port] >>>> $size bytes\n" if $debug || $size > 1024;
173 print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost;
174 }
175
176 sub pull_node_file {
177 my ( $node, $file ) = @_;
178
179 my $path = "/dev/shm/sack.$node.$file";
180 return if -e $path; # FIXME
181
182 my $sock = node_sock $node || die "not connected to $node";
183
184 print $sock "0 file $file\n";
185
186 my $size = <$sock>;
187 chomp($size);
188 warn "[$port] pull_node_file [$node] $file $size bytes\n";
189
190 my $block = 4096;
191 my $buff = ' ' x $block;
192
193 open(my $fh, '>', $path) || die "can't open $path";
194 while ( read $sock, $buff, $block ) {
195 print $fh $buff;
196 }
197 close($fh);
198 }
199
200 my $merge_digest_mapping;
201
202 sub merge_out {
203 my ( $from_node, $new ) = @_;
204
205 my $t_merge = time();
206
207 pull_node_file $from_node => 'nr_md5';
208 pull_node_file $from_node => 'md5';
209
210 my $remote_digest = Sack::Digest->new( port => $from_node );
211 my ( $local, $remote ) = ( 0, 0 );
212
213 my $tick = 0;
214 warn "[$port] merge [$from_node]";
215
216 my $missing;
217
218 foreach my $k1 ( keys %$new ) {
219
220 foreach my $k2 ( keys %{ $new->{$k1} } ) {
221
222 my $n = delete $new->{$k1}->{$k2};
223
224 if ( $k1 =~ m{#} ) {
225 die "ASSERT $k1 $k2" unless $k2 =~ m{^\d+$};
226 #warn "XXX $k1 $k2";
227
228 if ( defined $merge_digest_mapping->{$from_node}->[ $k2 ] ) {
229 $k2 = $merge_digest_mapping->{$from_node}->[ $k2 ];
230 } else {
231
232 my $md5 = $remote_digest->{nr_md5}->[$k2];
233
234 if ( ! $md5 ) {
235 $missing->{nr_md5}->{$from_node}++; # FIXME die?
236 next;
237 }
238
239 my $local_k2;
240
241 if ( $local_k2 = $digest->{md5_nr}->{$md5} ) {
242 $local++;
243 } elsif ( my $full = $remote_digest->{md5}->{$md5} ) {
244 $local_k2 = $digest->to_int( $remote_digest->{md5}->{$md5} );
245 $remote++;
246 } else {
247 $missing->{md5}->{$from_node}++;
248 }
249
250 $k2 = $merge_digest_mapping->{$from_node}->[ $k2 ] = $local_k2;
251
252 }
253 }
254
255 my $ref = ref $out->{$k1}->{$k2};
256 #warn "XXXX $k1 $k2 $ref";
257 if ( ! defined $out->{$k1}->{$k2} ) {
258 $out->{$k1}->{$k2} = $n;
259 } elsif ( $k1 =~ m{\+} ) {
260 # warn "## agregate $k1 $k2";
261 $out->{$k1}->{$k2} += $n;
262 } elsif ( $ref eq 'ARRAY' ) {
263 if ( ref $n eq 'ARRAY' ) {
264 push @{ $out->{$k1}->{$k2} }, $_ foreach @$n;
265 } else {
266 push @{ $out->{$k1}->{$k2} }, $n;
267 }
268 } elsif ( $ref eq '' ) {
269 $out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ];
270 } else {
271 die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2});
272 }
273
274 if ( $tick++ % 1000 == 0 ) {
275 print STDERR ".";
276 } elsif ( $tick % 10000 == 0 ) {
277 print STDERR $tick;
278 }
279 }
280 }
281
282 $t_merge = time - $t_merge;
283 my $digests = $local + $remote;
284 warn sprintf "\n[$port] merge %d in %.4fs %.2f/s digests local: %.1f%% %d/%d\n", $tick, $t_merge, $digests / $t_merge, $local * 100 / ( $digests || 1 ), $local, $remote;
285 push @reports, [ "$tick merged [$from_node]", $t_merge, $digests / $t_merge ];
286
287 warn "[$port] missing ", dump $missing if $missing;
288
289 warn "## merge out ", dump $out if $debug;
290 }
291
292 our $rec;
293
294 sub run_code {
295 my ( $view, $code ) = @_;
296
297 warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug;
298
299 send_nodes view => $view => $code;
300
301 undef $out;
302
303 my $affected = 0;
304 $t = time;
305
306 my $coderef = eval "sub { $code }";
307 if ( $@ ) {
308 warn "ABORT code: $@";
309 return;
310 }
311
312 foreach my $pos ( $offset + 1 .. $offset + $input->size ) {
313 $rec = $cache->{$pos} ||= $input->fetch_rec( $pos );
314 if ( ! $rec ) {
315 print STDERR "END @ $pos";
316 last;
317 }
318
319 eval { $coderef->() };
320 if ( $@ ) {
321 warn "ABORT $pos $@\n";
322 last;
323 } else {
324 $affected++;
325 }
326
327 $pos % 10000 == 0 ? print STDERR $pos :
328 $pos % 1000 == 0 ? print STDERR "." : 0 ;
329 };
330
331 report "$affected affected $view";
332
333 warn "WARN no \$out defined!" unless defined $out;
334
335 $digest->sync;
336
337 if ( $connected ) {
338 foreach my $node ( keys %$connected ) {
339 warn "[$port] get_node [$node]\n";
340 my $o = get_node $node;
341 next unless $o;
342 my $s = length $o;
343 $o = thaw $o;
344 warn "[$port] got $s bytes from [$node]\n";
345 merge_out $node => $o;
346 }
347 }
348 }
349
350 sub run_views {
351 @views = sort glob 'views/*.pl' unless @views;
352 warn "# views ", dump @views;
353
354 foreach my $view ( @views ) {
355
356 next if system("perl -c $view") != 0;
357
358 my $code = read_file $view;
359
360 run_code $view => $code;
361
362 if ( defined $out ) {
363
364 my $path = $view;
365 $path =~ s{views?/}{out/} || die "no view in $view";
366 $path =~ s{\.pl}{.storable};
367
368 unlink "$path.last" if -e "$path.last";
369 rename $path, "$path.last";
370
371 store $out => $path;
372 report "save $path", -s $path, "bytes";
373
374 if ( -s $path < 4096 ) {
375 print '$out = ', dump $digest->undigest_out($out);
376 }
377 }
378
379 }
380
381 }
382
383
384 sub info_tabs {
385 my $port_col = port2color($port);
386 "$port_col\t$offset\t$limit\t$num_records\t$path\t"
387 . join("\t", map {
388 my $b = $_;
389 $b =~ s{^.+\.$port\.([^/]+)$}{$1};
390 "$b " . -s $_
391 } glob "/dev/shm/sack.$port.*" );
392 }
393
394
395 if ( $port ) {
396
397 my $pid_path = "/tmp/sack.$port.pid";
398 if ( -e $pid_path ) {
399 my $pid = read_file $pid_path;
400 kill 9, $pid && warn "[$port] kill old $pid\n";
401 }
402 write_file $pid_path, $$;
403
404
405 my $sock = IO::Socket::INET->new(
406 Listen => SOMAXCONN,
407 LocalAddr => '127.0.0.1',
408 LocalPort => $port,
409 Proto => 'tcp',
410 Reuse => 1,
411 ) or die "[$port] die $!";
412
413 while (1) {
414
415 warn "[$port] accept path: $path offset: $offset limit: $limit #recs: $num_records\n";
416
417 my $client = $sock->accept();
418
419 warn "[$port] <<<< connect from ", $client->peerhost, $/;
420
421 my @header = split(/\s/, <$client>);
422 warn "[$port] <<<< header ",dump(@header),$/;
423
424 my $size = shift @header;
425
426 my $content;
427 read $client, $content, $size;
428
429 if ( $header[0] eq 'view' ) {
430 run_code $header[1] => $content;
431 send_sock $client => freeze $out;
432 } elsif ( $header[0] eq 'info' ) {
433 my $info = info_tabs;
434 warn "[$port] info $info\n";
435 $info .= "\n" . show_report if $content =~ m{r}i;
436 send_sock $client => $info;
437 } elsif ( $header[0] eq 'exit' ) {
438 warn "[$port] exit";
439 exit;
440 } elsif ( $header[0] eq 'file' ) {
441 $digest->close;
442 my $path = "/dev/shm/sack.$port.$header[1]";
443 my $size = -s $path;
444 warn "[$port] >>>> file $path $size bytes\n";
445 print $client "$size\n";
446 my $block = 4096;
447 my $buff = ' ' x $block;
448 open(my $fh, '<', $path) || die "can't open $path";
449 while ( read $fh, $buff, $block ) {
450 print $client $buff;
451 }
452 $digest->open;
453 } else {
454 warn "[$port] UNKNOWN $header[0]";
455 }
456
457 $client->close;
458
459 }
460 }
461
462 sub info {
463 my $detail = shift || '';
464
465 send_nodes 'info' => $detail;
466
467 my @info = (
468 "port\toffset\tlimit\t#recs\tpath",
469 "----\t------\t-----\t-----\t----",
470 info_tabs,
471 );
472
473 push @info, show_report if $detail =~ m{r}i;
474
475 push @info, get_node $_ foreach @nodes;
476
477 print "[$port] INFO", $detail ? " $detail" : '', " \n"
478 , join("\n", @info)
479 , "\n" ;
480
481 return @info;
482 }
483
484 info;
485 while ( keys %$connected != @nodes ) {
486 warn "[$port] wait for [", join('] [', grep { ! defined $connected->{$_} } @nodes ), "]\n";
487 sleep 1;
488 info;
489 }
490 run_views;
491
492 while ( 1 ) {
493
494 print "sack> ";
495 chomp( my $cmd = <STDIN> );
496
497 if ( $cmd =~ m{^(h|\?)} ) {
498 print << "__HELP__"
499 Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit
500
501 View Run run views
502 VI \\e Output show output of last run
503 Info [Report] info with optional report
504 Quit EXit shutdown
505
506 __HELP__
507 } elsif ( $cmd =~ m{^(vi|\\e|o)}i ) {
508 #system "vi out/*";
509 $digest->sync;
510 system "bin/storableedit.pl", (glob('out/*.storable'))[0];
511 } elsif ( $cmd =~ m{^i(?:nfo)?\s?(\S+)?$}i ) {
512 info $1;
513 } elsif ( $cmd =~ m{^(q|e|x)}i ) {
514 warn "# exit";
515 send_nodes 'exit';
516 exit;
517 } elsif ( $cmd =~ m{^(v|r)}i ) {
518 run_views;
519 } elsif ( $cmd =~ m{^n(ode)?\s*(\d+)}i ) {
520 push @nodes, $2;
521 info;
522 } elsif ( $cmd ) {
523 warn "UNKNOWN ", dump $cmd;
524 }
525
526 }
527

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26