/[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

Annotation of /trunk/bin/sack.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 91 - (hide annotations)
Sat Oct 3 21:08:36 2009 UTC (14 years, 7 months ago) by dpavlin
File MIME type: text/plain
File size: 11374 byte(s)
use $digest_module for remote digests
1 dpavlin 1 #!/usr/bin/perl
2    
3     use warnings;
4     use strict;
5    
6 dpavlin 82 our $VERSION = '0.06';
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 64 'port|listen=i' => \$port,
31     'node|connect=i' => \@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 81
57     sub port2color {
58     my $port = shift;
59 dpavlin 82 return "\e[1m0\e[0m" if $port == 0;
60    
61     my $c = ( $port % 6 ) + 31;
62     return "\e[${c}m$port\e[0m";
63 dpavlin 81 }
64    
65     $SIG{__WARN__} = sub {
66 dpavlin 82 return unless @_;
67 dpavlin 81 my $msg = join('', @_);
68 dpavlin 82 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 dpavlin 81 return 1;
75     };
76    
77 dpavlin 12 }
78    
79 dpavlin 87 our $digest_module = 'Sack::Digest::' . ( $ENV{SACK_DIGEST} || 'BerkeleyDB' );
80 dpavlin 33 use lib "$prefix/srv/Sack/lib/";
81 dpavlin 87 my $digest_path = $digest_module;
82     $digest_path =~ s{::}{/}g;
83     require "lib/$digest_path.pm";
84    
85     our $digest = $digest_module->new( port => $port, clean => 1 );
86 dpavlin 45 sub digest { $digest->to_int($_[0]) }
87 dpavlin 12
88 dpavlin 87 warn "[$port] using $digest_module\n";
89    
90 dpavlin 12 use lib "$prefix/srv/webpac2/lib/";
91 dpavlin 1 use WebPAC::Input::ISI;
92 dpavlin 24
93     $WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields
94    
95 dpavlin 1 my $input = WebPAC::Input::ISI->new(
96 dpavlin 12 path => "$prefix/$path",
97 dpavlin 4 offset => $offset,
98     limit => $limit,
99 dpavlin 1 );
100    
101 dpavlin 25 our $num_records = $input->size;
102 dpavlin 74 our @reports;
103 dpavlin 1
104     sub report {
105 dpavlin 30 my $description = join(' ',@_);
106 dpavlin 1 my $dt = time - $t;
107 dpavlin 74 my $report = [ $description, $dt, $input->size / $dt ];
108 dpavlin 82 warn sprintf "[$port] %s in %1.4fs %.2f/s\n", @$report;
109 dpavlin 74 push @reports, $report;
110 dpavlin 10 $t = time;
111 dpavlin 1 }
112    
113 dpavlin 74 sub show_report {
114     "\n" . join( "\n", map { sprintf "%8.4fs %10.2f/s %s", $_->[1], $_->[2], $_->[0] } @reports ) . "\n";
115     }
116 dpavlin 1
117 dpavlin 74 report $input->size , 'records loaded';
118 dpavlin 1
119     mkdir 'out' unless -e 'out';
120    
121 dpavlin 5 our $out;
122    
123 dpavlin 8 our $cache;
124    
125 dpavlin 13 our $connected;
126    
127 dpavlin 64 sub node_sock {
128     my $node = shift;
129     my $sock = IO::Socket::INET->new(
130     PeerAddr => '127.0.0.1',
131     PeerPort => $node,
132     Proto => 'tcp',
133     );
134    
135     return $sock if $sock && $sock->connected;
136    
137     warn "[$port] can't connect to $node - $!\n"; # FIXME die?
138     return;
139     }
140    
141 dpavlin 11 sub send_nodes {
142 dpavlin 17 my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument!
143 dpavlin 21 my $header = defined $content ? length($content) : 0;
144 dpavlin 11 $header .= ' ' . join(' ', @_) if @_;
145    
146 dpavlin 74 warn "# send_nodes ", dump(@_), " to ", dump @nodes;
147 dpavlin 60
148 dpavlin 11 foreach my $node ( @nodes ) {
149    
150 dpavlin 64 my $sock = node_sock($node) || next;
151 dpavlin 11
152 dpavlin 67 warn "[$port] >>>> [$node] $header\n";
153 dpavlin 11 print $sock "$header\n$content" || warn "can't send $header to $node: $!";
154    
155 dpavlin 13 $connected->{$node} = $sock;
156 dpavlin 11 }
157     }
158    
159 dpavlin 16 sub get_node {
160     my $node = shift;
161    
162     my $sock = $connected->{$node};
163 dpavlin 64 if ( ! $sock || ! $sock->connected ) {
164     warn "[$port] no connection to $node";
165 dpavlin 16 delete $connected->{$node};
166     return;
167     }
168     chomp( my $size = <$sock> );
169 dpavlin 67 warn "[$port] <<<< [$node] $size bytes\n" if $debug || $size > 1024;
170 dpavlin 16 my $data;
171     read $sock, $data, $size;
172     return $data;
173     }
174    
175     sub send_sock {
176     my ( $sock, $data ) = @_;
177     my $size = length $data;
178 dpavlin 58 warn "[$port] >>>> $size bytes\n" if $debug || $size > 1024;
179 dpavlin 16 print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost;
180     }
181    
182 dpavlin 64 sub pull_node_file {
183     my ( $node, $file ) = @_;
184    
185     my $path = "/dev/shm/sack.$node.$file";
186 dpavlin 72 return if -e $path; # FIXME
187 dpavlin 64
188     my $sock = node_sock $node || die "not connected to $node";
189    
190     print $sock "0 file $file\n";
191    
192     my $size = <$sock>;
193     chomp($size);
194 dpavlin 82 warn "[$port] pull_node_file [$node] $file $size bytes\n";
195 dpavlin 64
196     my $block = 4096;
197     my $buff = ' ' x $block;
198    
199     open(my $fh, '>', $path) || die "can't open $path";
200     while ( read $sock, $buff, $block ) {
201     print $fh $buff;
202     }
203     close($fh);
204     }
205    
206 dpavlin 80 my $merge_digest_mapping;
207    
208 dpavlin 13 sub merge_out {
209 dpavlin 50 my ( $from_node, $new ) = @_;
210 dpavlin 13
211 dpavlin 67 my $t_merge = time();
212    
213 dpavlin 64 pull_node_file $from_node => 'nr_md5';
214     pull_node_file $from_node => 'md5';
215    
216 dpavlin 91 my $remote_digest = $digest_module->new( port => $from_node );
217 dpavlin 59 my ( $local, $remote ) = ( 0, 0 );
218 dpavlin 50
219 dpavlin 59 my $tick = 0;
220 dpavlin 82 warn "[$port] merge [$from_node]";
221 dpavlin 50
222 dpavlin 64 my $missing;
223    
224 dpavlin 13 foreach my $k1 ( keys %$new ) {
225    
226     foreach my $k2 ( keys %{ $new->{$k1} } ) {
227    
228 dpavlin 30 my $n = delete $new->{$k1}->{$k2};
229 dpavlin 50
230     if ( $k1 =~ m{#} ) {
231     die "ASSERT $k1 $k2" unless $k2 =~ m{^\d+$};
232     #warn "XXX $k1 $k2";
233 dpavlin 64
234 dpavlin 80 if ( defined $merge_digest_mapping->{$from_node}->[ $k2 ] ) {
235     $k2 = $merge_digest_mapping->{$from_node}->[ $k2 ];
236     } else {
237 dpavlin 64
238 dpavlin 80 my $md5 = $remote_digest->{nr_md5}->[$k2];
239    
240     if ( ! $md5 ) {
241     $missing->{nr_md5}->{$from_node}++; # FIXME die?
242     next;
243     }
244    
245     my $local_k2;
246    
247     if ( $local_k2 = $digest->{md5_nr}->{$md5} ) {
248     $local++;
249     } elsif ( my $full = $remote_digest->{md5}->{$md5} ) {
250     $local_k2 = $digest->to_int( $remote_digest->{md5}->{$md5} );
251     $remote++;
252     } else {
253     $missing->{md5}->{$from_node}++;
254     }
255    
256     $k2 = $merge_digest_mapping->{$from_node}->[ $k2 ] = $local_k2;
257    
258 dpavlin 50 }
259     }
260    
261 dpavlin 30 my $ref = ref $out->{$k1}->{$k2};
262 dpavlin 50 #warn "XXXX $k1 $k2 $ref";
263 dpavlin 13 if ( ! defined $out->{$k1}->{$k2} ) {
264     $out->{$k1}->{$k2} = $n;
265     } elsif ( $k1 =~ m{\+} ) {
266 dpavlin 19 # warn "## agregate $k1 $k2";
267 dpavlin 13 $out->{$k1}->{$k2} += $n;
268 dpavlin 21 } elsif ( $ref eq 'ARRAY' ) {
269     if ( ref $n eq 'ARRAY' ) {
270     push @{ $out->{$k1}->{$k2} }, $_ foreach @$n;
271     } else {
272     push @{ $out->{$k1}->{$k2} }, $n;
273     }
274 dpavlin 13 } elsif ( $ref eq '' ) {
275     $out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ];
276     } else {
277     die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2});
278     }
279 dpavlin 59
280     if ( $tick++ % 1000 == 0 ) {
281     print STDERR ".";
282     } elsif ( $tick % 10000 == 0 ) {
283     print STDERR $tick;
284     }
285 dpavlin 13 }
286     }
287    
288 dpavlin 67 $t_merge = time - $t_merge;
289 dpavlin 79 my $digests = $local + $remote;
290 dpavlin 82 warn sprintf "\n[$port] merge %d in %.4fs %.2f/s digests local: %.1f%% %d/%d\n", $tick, $t_merge, $digests / $t_merge, $local * 100 / ( $digests || 1 ), $local, $remote;
291     push @reports, [ "$tick merged [$from_node]", $t_merge, $digests / $t_merge ];
292 dpavlin 59
293 dpavlin 64 warn "[$port] missing ", dump $missing if $missing;
294    
295 dpavlin 19 warn "## merge out ", dump $out if $debug;
296 dpavlin 13 }
297    
298 dpavlin 84 our $rec;
299    
300 dpavlin 11 sub run_code {
301     my ( $view, $code ) = @_;
302    
303 dpavlin 21 warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug;
304 dpavlin 11
305     send_nodes view => $view => $code;
306    
307     undef $out;
308    
309     my $affected = 0;
310     $t = time;
311    
312 dpavlin 84 my $coderef = eval "sub { $code }";
313     if ( $@ ) {
314     warn "ABORT code: $@";
315     return;
316     }
317    
318 dpavlin 11 foreach my $pos ( $offset + 1 .. $offset + $input->size ) {
319 dpavlin 84 $rec = $cache->{$pos} ||= $input->fetch_rec( $pos );
320 dpavlin 11 if ( ! $rec ) {
321 dpavlin 41 print STDERR "END @ $pos";
322 dpavlin 11 last;
323     }
324    
325 dpavlin 84 eval { $coderef->() };
326 dpavlin 11 if ( $@ ) {
327 dpavlin 41 warn "ABORT $pos $@\n";
328 dpavlin 20 last;
329 dpavlin 11 } else {
330     $affected++;
331     }
332 dpavlin 30
333 dpavlin 40 $pos % 10000 == 0 ? print STDERR $pos :
334     $pos % 1000 == 0 ? print STDERR "." : 0 ;
335 dpavlin 11 };
336    
337 dpavlin 74 report "$affected affected $view";
338 dpavlin 11
339     warn "WARN no \$out defined!" unless defined $out;
340 dpavlin 13
341 dpavlin 46 $digest->sync;
342    
343 dpavlin 13 if ( $connected ) {
344 dpavlin 24 foreach my $node ( keys %$connected ) {
345 dpavlin 74 warn "[$port] get_node [$node]\n";
346 dpavlin 24 my $o = get_node $node;
347 dpavlin 64 next unless $o;
348 dpavlin 24 my $s = length $o;
349     $o = thaw $o;
350 dpavlin 74 warn "[$port] got $s bytes from [$node]\n";
351 dpavlin 50 merge_out $node => $o;
352 dpavlin 24 }
353 dpavlin 13 }
354 dpavlin 11 }
355    
356 dpavlin 1 sub run_views {
357 dpavlin 6 @views = sort glob 'views/*.pl' unless @views;
358 dpavlin 1 warn "# views ", dump @views;
359    
360     foreach my $view ( @views ) {
361    
362     next if system("perl -c $view") != 0;
363    
364     my $code = read_file $view;
365    
366 dpavlin 11 run_code $view => $code;
367 dpavlin 1
368 dpavlin 11 if ( defined $out ) {
369 dpavlin 1
370 dpavlin 11 my $path = $view;
371     $path =~ s{views?/}{out/} || die "no view in $view";
372 dpavlin 30 $path =~ s{\.pl}{.storable};
373 dpavlin 1
374 dpavlin 9 unlink "$path.last" if -e "$path.last";
375     rename $path, "$path.last";
376 dpavlin 30
377     store $out => $path;
378 dpavlin 74 report "save $path", -s $path, "bytes";
379 dpavlin 35
380     if ( -s $path < 4096 ) {
381 dpavlin 45 print '$out = ', dump $digest->undigest_out($out);
382 dpavlin 35 }
383 dpavlin 11 }
384    
385     }
386    
387     }
388    
389 dpavlin 35
390     sub info_tabs {
391 dpavlin 82 my $port_col = port2color($port);
392     "$port_col\t$offset\t$limit\t$num_records\t$path\t"
393 dpavlin 36 . join("\t", map {
394 dpavlin 35 my $b = $_;
395 dpavlin 46 $b =~ s{^.+\.$port\.([^/]+)$}{$1};
396 dpavlin 35 "$b " . -s $_
397 dpavlin 46 } glob "/dev/shm/sack.$port.*" );
398 dpavlin 35 }
399    
400    
401 dpavlin 30 if ( $port ) {
402 dpavlin 75
403     my $pid_path = "/tmp/sack.$port.pid";
404     if ( -e $pid_path ) {
405     my $pid = read_file $pid_path;
406 dpavlin 79 kill 9, $pid && warn "[$port] kill old $pid\n";
407 dpavlin 75 }
408     write_file $pid_path, $$;
409    
410    
411 dpavlin 11 my $sock = IO::Socket::INET->new(
412     Listen => SOMAXCONN,
413 dpavlin 21 LocalAddr => '127.0.0.1',
414 dpavlin 30 LocalPort => $port,
415 dpavlin 11 Proto => 'tcp',
416     Reuse => 1,
417 dpavlin 82 ) or die "[$port] die $!";
418 dpavlin 11
419     while (1) {
420    
421 dpavlin 74 warn "[$port] accept path: $path offset: $offset limit: $limit #recs: $num_records\n";
422 dpavlin 11
423     my $client = $sock->accept();
424    
425 dpavlin 30 warn "[$port] <<<< connect from ", $client->peerhost, $/;
426 dpavlin 11
427     my @header = split(/\s/, <$client>);
428 dpavlin 30 warn "[$port] <<<< header ",dump(@header),$/;
429 dpavlin 11
430     my $size = shift @header;
431    
432     my $content;
433     read $client, $content, $size;
434    
435     if ( $header[0] eq 'view' ) {
436     run_code $header[1] => $content;
437 dpavlin 16 send_sock $client => freeze $out;
438     } elsif ( $header[0] eq 'info' ) {
439 dpavlin 35 my $info = info_tabs;
440 dpavlin 30 warn "[$port] info $info\n";
441 dpavlin 74 $info .= "\n" . show_report if $content =~ m{r}i;
442 dpavlin 16 send_sock $client => $info;
443     } elsif ( $header[0] eq 'exit' ) {
444 dpavlin 30 warn "[$port] exit";
445 dpavlin 16 exit;
446 dpavlin 64 } elsif ( $header[0] eq 'file' ) {
447     $digest->close;
448     my $path = "/dev/shm/sack.$port.$header[1]";
449     my $size = -s $path;
450     warn "[$port] >>>> file $path $size bytes\n";
451     print $client "$size\n";
452     my $block = 4096;
453     my $buff = ' ' x $block;
454     open(my $fh, '<', $path) || die "can't open $path";
455     while ( read $fh, $buff, $block ) {
456     print $client $buff;
457     }
458 dpavlin 67 $digest->open;
459 dpavlin 5 } else {
460 dpavlin 30 warn "[$port] UNKNOWN $header[0]";
461 dpavlin 1 }
462    
463 dpavlin 75 $client->close;
464    
465 dpavlin 1 }
466     }
467    
468 dpavlin 30 sub info {
469 dpavlin 74 my $detail = shift || '';
470 dpavlin 30
471 dpavlin 74 send_nodes 'info' => $detail;
472    
473 dpavlin 30 my @info = (
474 dpavlin 36 "port\toffset\tlimit\t#recs\tpath",
475     "----\t------\t-----\t-----\t----",
476 dpavlin 35 info_tabs,
477 dpavlin 30 );
478    
479 dpavlin 74 push @info, show_report if $detail =~ m{r}i;
480    
481 dpavlin 30 push @info, get_node $_ foreach @nodes;
482    
483 dpavlin 74 print "[$port] INFO", $detail ? " $detail" : '', " \n"
484 dpavlin 30 , join("\n", @info)
485 dpavlin 74 , "\n" ;
486 dpavlin 30
487     return @info;
488     }
489    
490     info;
491 dpavlin 76 while ( keys %$connected != @nodes ) {
492 dpavlin 82 warn "[$port] wait for [", join('] [', grep { ! defined $connected->{$_} } @nodes ), "]\n";
493 dpavlin 76 sleep 1;
494 dpavlin 82 info;
495 dpavlin 76 }
496 dpavlin 3 run_views;
497    
498 dpavlin 1 while ( 1 ) {
499    
500     print "sack> ";
501    
502 dpavlin 89 my $cmd = <STDIN>;
503     exit unless defined $cmd; # CTRL+D
504     chomp $cmd;
505    
506 dpavlin 21 if ( $cmd =~ m{^(h|\?)} ) {
507     print << "__HELP__"
508     Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit
509    
510     View Run run views
511     VI \\e Output show output of last run
512 dpavlin 82 Info [Report] info with optional report
513 dpavlin 21 Quit EXit shutdown
514    
515     __HELP__
516     } elsif ( $cmd =~ m{^(vi|\\e|o)}i ) {
517 dpavlin 33 #system "vi out/*";
518 dpavlin 53 $digest->sync;
519 dpavlin 33 system "bin/storableedit.pl", (glob('out/*.storable'))[0];
520 dpavlin 74 } elsif ( $cmd =~ m{^i(?:nfo)?\s?(\S+)?$}i ) {
521     info $1;
522 dpavlin 21 } elsif ( $cmd =~ m{^(q|e|x)}i ) {
523 dpavlin 16 warn "# exit";
524     send_nodes 'exit';
525     exit;
526 dpavlin 21 } elsif ( $cmd =~ m{^(v|r)}i ) {
527 dpavlin 3 run_views;
528 dpavlin 55 } elsif ( $cmd =~ m{^n(ode)?\s*(\d+)}i ) {
529 dpavlin 60 push @nodes, $2;
530 dpavlin 55 info;
531 dpavlin 21 } elsif ( $cmd ) {
532     warn "UNKNOWN ", dump $cmd;
533 dpavlin 3 }
534    
535 dpavlin 1 }
536    

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26