/[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 81 - (show annotations)
Sat Sep 26 10:40:05 2009 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 10821 byte(s)
use ansi colors for port numbers in warn messages
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 sub port2color {
58 my $port = shift;
59 my $c = $port & 0x07;
60 $c ||= 1; # no black
61 $c |= 30;
62 return "[\e[${c}m$port\e[0m]";
63 }
64
65 $SIG{__WARN__} = sub {
66 my @loc = caller(1);
67 my $msg = join('', @_);
68 $msg =~ s{\[(0|\d\d\d\d)\]}{port2color($1)}e;
69 print STDOUT $msg;
70 return 1;
71 };
72
73 }
74
75 use lib "$prefix/srv/Sack/lib/";
76 use Sack::Digest;
77 our $digest = Sack::Digest->new( port => $port, clean => 1 );
78 sub digest { $digest->to_int($_[0]) }
79
80 use lib "$prefix/srv/webpac2/lib/";
81 use WebPAC::Input::ISI;
82
83 $WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields
84
85 my $input = WebPAC::Input::ISI->new(
86 path => "$prefix/$path",
87 offset => $offset,
88 limit => $limit,
89 );
90
91 our $num_records = $input->size;
92 our @reports;
93
94 sub report {
95 my $description = join(' ',@_);
96 my $dt = time - $t;
97 my $report = [ $description, $dt, $input->size / $dt ];
98 printf "[$port] %s in %1.4fs %.2f/s\n", @$report;
99 push @reports, $report;
100 $t = time;
101 }
102
103 sub show_report {
104 "\n" . join( "\n", map { sprintf "%8.4fs %10.2f/s %s", $_->[1], $_->[2], $_->[0] } @reports ) . "\n";
105 }
106
107 report $input->size , 'records loaded';
108
109 mkdir 'out' unless -e 'out';
110
111 our $out;
112
113 our $cache;
114
115 our $connected;
116
117 sub node_sock {
118 my $node = shift;
119 my $sock = IO::Socket::INET->new(
120 PeerAddr => '127.0.0.1',
121 PeerPort => $node,
122 Proto => 'tcp',
123 );
124
125 return $sock if $sock && $sock->connected;
126
127 warn "[$port] can't connect to $node - $!\n"; # FIXME die?
128 return;
129 }
130
131 sub send_nodes {
132 my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument!
133 my $header = defined $content ? length($content) : 0;
134 $header .= ' ' . join(' ', @_) if @_;
135
136 warn "# send_nodes ", dump(@_), " to ", dump @nodes;
137
138 foreach my $node ( @nodes ) {
139
140 my $sock = node_sock($node) || next;
141
142 warn "[$port] >>>> [$node] $header\n";
143 print $sock "$header\n$content" || warn "can't send $header to $node: $!";
144
145 $connected->{$node} = $sock;
146 }
147 }
148
149 sub get_node {
150 my $node = shift;
151
152 my $sock = $connected->{$node};
153 if ( ! $sock || ! $sock->connected ) {
154 warn "[$port] no connection to $node";
155 delete $connected->{$node};
156 return;
157 }
158 chomp( my $size = <$sock> );
159 warn "[$port] <<<< [$node] $size bytes\n" if $debug || $size > 1024;
160 my $data;
161 read $sock, $data, $size;
162 return $data;
163 }
164
165 sub send_sock {
166 my ( $sock, $data ) = @_;
167 my $size = length $data;
168 warn "[$port] >>>> $size bytes\n" if $debug || $size > 1024;
169 print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost;
170 }
171
172 sub pull_node_file {
173 my ( $node, $file ) = @_;
174
175 my $path = "/dev/shm/sack.$node.$file";
176 return if -e $path; # FIXME
177
178 my $sock = node_sock $node || die "not connected to $node";
179
180 print $sock "0 file $file\n";
181
182 my $size = <$sock>;
183 chomp($size);
184 warn "[$port] pull_node_file $node $file $size bytes\n";
185
186 my $block = 4096;
187 my $buff = ' ' x $block;
188
189 open(my $fh, '>', $path) || die "can't open $path";
190 while ( read $sock, $buff, $block ) {
191 print $fh $buff;
192 }
193 close($fh);
194 }
195
196 my $merge_digest_mapping;
197
198 sub merge_out {
199 my ( $from_node, $new ) = @_;
200
201 my $t_merge = time();
202
203 pull_node_file $from_node => 'nr_md5';
204 pull_node_file $from_node => 'md5';
205
206 my $remote_digest = Sack::Digest->new( port => $from_node );
207 my ( $local, $remote ) = ( 0, 0 );
208
209 my $tick = 0;
210 print STDERR "[$port] merge [$from_node]";
211
212 my $missing;
213
214 foreach my $k1 ( keys %$new ) {
215
216 foreach my $k2 ( keys %{ $new->{$k1} } ) {
217
218 my $n = delete $new->{$k1}->{$k2};
219
220 if ( $k1 =~ m{#} ) {
221 die "ASSERT $k1 $k2" unless $k2 =~ m{^\d+$};
222 #warn "XXX $k1 $k2";
223
224 if ( defined $merge_digest_mapping->{$from_node}->[ $k2 ] ) {
225 $k2 = $merge_digest_mapping->{$from_node}->[ $k2 ];
226 } else {
227
228 my $md5 = $remote_digest->{nr_md5}->[$k2];
229
230 if ( ! $md5 ) {
231 $missing->{nr_md5}->{$from_node}++; # FIXME die?
232 next;
233 }
234
235 my $local_k2;
236
237 if ( $local_k2 = $digest->{md5_nr}->{$md5} ) {
238 $local++;
239 } elsif ( my $full = $remote_digest->{md5}->{$md5} ) {
240 $local_k2 = $digest->to_int( $remote_digest->{md5}->{$md5} );
241 $remote++;
242 } else {
243 $missing->{md5}->{$from_node}++;
244 }
245
246 $k2 = $merge_digest_mapping->{$from_node}->[ $k2 ] = $local_k2;
247
248 }
249 }
250
251 my $ref = ref $out->{$k1}->{$k2};
252 #warn "XXXX $k1 $k2 $ref";
253 if ( ! defined $out->{$k1}->{$k2} ) {
254 $out->{$k1}->{$k2} = $n;
255 } elsif ( $k1 =~ m{\+} ) {
256 # warn "## agregate $k1 $k2";
257 $out->{$k1}->{$k2} += $n;
258 } elsif ( $ref eq 'ARRAY' ) {
259 if ( ref $n eq 'ARRAY' ) {
260 push @{ $out->{$k1}->{$k2} }, $_ foreach @$n;
261 } else {
262 push @{ $out->{$k1}->{$k2} }, $n;
263 }
264 } elsif ( $ref eq '' ) {
265 $out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ];
266 } else {
267 die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2});
268 }
269
270 if ( $tick++ % 1000 == 0 ) {
271 print STDERR ".";
272 } elsif ( $tick % 10000 == 0 ) {
273 print STDERR $tick;
274 }
275 }
276 }
277
278 $t_merge = time - $t_merge;
279 my $digests = $local + $remote;
280 printf STDERR "\n[$port] merge %d in %.4fs %.2f/s digests local: %.1f%% %d/%d\n", $tick, $t_merge, $digests / $t_merge, $local * 100 / ( $digests || 1 ), $local, $remote;
281 push @reports, [ "$tick merged $from_node", $t_merge, $tick / $t_merge ];
282
283 warn "[$port] missing ", dump $missing if $missing;
284
285 warn "## merge out ", dump $out if $debug;
286 }
287
288 sub run_code {
289 my ( $view, $code ) = @_;
290
291 warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug;
292
293 send_nodes view => $view => $code;
294
295 undef $out;
296
297 my $affected = 0;
298 $t = time;
299
300 foreach my $pos ( $offset + 1 .. $offset + $input->size ) {
301 my $rec = $cache->{$pos} ||= $input->fetch_rec( $pos );
302 if ( ! $rec ) {
303 print STDERR "END @ $pos";
304 last;
305 }
306
307 eval "$code";
308 if ( $@ ) {
309 warn "ABORT $pos $@\n";
310 last;
311 } else {
312 $affected++;
313 }
314
315 $pos % 10000 == 0 ? print STDERR $pos :
316 $pos % 1000 == 0 ? print STDERR "." : 0 ;
317 };
318
319 report "$affected affected $view";
320
321 warn "WARN no \$out defined!" unless defined $out;
322
323 $digest->sync;
324
325 if ( $connected ) {
326 foreach my $node ( keys %$connected ) {
327 warn "[$port] get_node [$node]\n";
328 my $o = get_node $node;
329 next unless $o;
330 my $s = length $o;
331 $o = thaw $o;
332 warn "[$port] got $s bytes from [$node]\n";
333 merge_out $node => $o;
334 }
335 }
336 }
337
338 sub run_views {
339 @views = sort glob 'views/*.pl' unless @views;
340 warn "# views ", dump @views;
341
342 foreach my $view ( @views ) {
343
344 next if system("perl -c $view") != 0;
345
346 my $code = read_file $view;
347
348 run_code $view => $code;
349
350 if ( defined $out ) {
351
352 my $path = $view;
353 $path =~ s{views?/}{out/} || die "no view in $view";
354 $path =~ s{\.pl}{.storable};
355
356 unlink "$path.last" if -e "$path.last";
357 rename $path, "$path.last";
358
359 store $out => $path;
360 report "save $path", -s $path, "bytes";
361
362 if ( -s $path < 4096 ) {
363 print '$out = ', dump $digest->undigest_out($out);
364 }
365 }
366
367 }
368
369 }
370
371
372 sub info_tabs {
373 "$port\t$offset\t$limit\t$num_records\t$path\t"
374 . join("\t", map {
375 my $b = $_;
376 $b =~ s{^.+\.$port\.([^/]+)$}{$1};
377 "$b " . -s $_
378 } glob "/dev/shm/sack.$port.*" );
379 }
380
381
382 if ( $port ) {
383
384 my $pid_path = "/tmp/sack.$port.pid";
385 if ( -e $pid_path ) {
386 my $pid = read_file $pid_path;
387 kill 9, $pid && warn "[$port] kill old $pid\n";
388 }
389 write_file $pid_path, $$;
390
391
392 my $sock = IO::Socket::INET->new(
393 Listen => SOMAXCONN,
394 LocalAddr => '127.0.0.1',
395 LocalPort => $port,
396 Proto => 'tcp',
397 Reuse => 1,
398 ) or die $!;
399
400 while (1) {
401
402 warn "[$port] accept path: $path offset: $offset limit: $limit #recs: $num_records\n";
403
404 my $client = $sock->accept();
405
406 warn "[$port] <<<< connect from ", $client->peerhost, $/;
407
408 my @header = split(/\s/, <$client>);
409 warn "[$port] <<<< header ",dump(@header),$/;
410
411 my $size = shift @header;
412
413 my $content;
414 read $client, $content, $size;
415
416 if ( $header[0] eq 'view' ) {
417 run_code $header[1] => $content;
418 send_sock $client => freeze $out;
419 } elsif ( $header[0] eq 'info' ) {
420 my $info = info_tabs;
421 warn "[$port] info $info\n";
422 $info .= "\n" . show_report if $content =~ m{r}i;
423 send_sock $client => $info;
424 } elsif ( $header[0] eq 'exit' ) {
425 warn "[$port] exit";
426 exit;
427 } elsif ( $header[0] eq 'file' ) {
428 $digest->close;
429 my $path = "/dev/shm/sack.$port.$header[1]";
430 my $size = -s $path;
431 warn "[$port] >>>> file $path $size bytes\n";
432 print $client "$size\n";
433 my $block = 4096;
434 my $buff = ' ' x $block;
435 open(my $fh, '<', $path) || die "can't open $path";
436 while ( read $fh, $buff, $block ) {
437 print $client $buff;
438 }
439 $digest->open;
440 } else {
441 warn "[$port] UNKNOWN $header[0]";
442 }
443
444 $client->close;
445
446 }
447 }
448
449 sub info {
450 my $detail = shift || '';
451
452 send_nodes 'info' => $detail;
453
454 my @info = (
455 "port\toffset\tlimit\t#recs\tpath",
456 "----\t------\t-----\t-----\t----",
457 info_tabs,
458 );
459
460 push @info, show_report if $detail =~ m{r}i;
461
462 push @info, get_node $_ foreach @nodes;
463
464 print "[$port] INFO", $detail ? " $detail" : '', " \n"
465 , join("\n", @info)
466 , "\n" ;
467
468 return @info;
469 }
470
471 info;
472 while ( keys %$connected != @nodes ) {
473 warn "[$port] wait for ", join(' ', grep { ! defined $connected->{$_} } @nodes );
474 sleep 1;
475 }
476 run_views;
477
478 while ( 1 ) {
479
480 print "sack> ";
481 chomp( my $cmd = <STDIN> );
482
483 if ( $cmd =~ m{^(h|\?)} ) {
484 print << "__HELP__"
485 Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit
486
487 View Run run views
488 VI \\e Output show output of last run
489 Info [\$VERSION] instrospect
490 Quit EXit shutdown
491
492 __HELP__
493 } elsif ( $cmd =~ m{^(vi|\\e|o)}i ) {
494 #system "vi out/*";
495 $digest->sync;
496 system "bin/storableedit.pl", (glob('out/*.storable'))[0];
497 } elsif ( $cmd =~ m{^i(?:nfo)?\s?(\S+)?$}i ) {
498 info $1;
499 } elsif ( $cmd =~ m{^(q|e|x)}i ) {
500 warn "# exit";
501 send_nodes 'exit';
502 exit;
503 } elsif ( $cmd =~ m{^(v|r)}i ) {
504 run_views;
505 } elsif ( $cmd =~ m{^n(ode)?\s*(\d+)}i ) {
506 push @nodes, $2;
507 info;
508 } elsif ( $cmd ) {
509 warn "UNKNOWN ", dump $cmd;
510 }
511
512 }
513

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26