/[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 81 - (hide 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 dpavlin 1 #!/usr/bin/perl
2    
3     use warnings;
4     use strict;
5    
6 dpavlin 74 our $VERSION = '0.05';
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     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 $msg;
70     return 1;
71     };
72    
73 dpavlin 12 }
74    
75 dpavlin 33 use lib "$prefix/srv/Sack/lib/";
76     use Sack::Digest;
77 dpavlin 45 our $digest = Sack::Digest->new( port => $port, clean => 1 );
78     sub digest { $digest->to_int($_[0]) }
79 dpavlin 12
80     use lib "$prefix/srv/webpac2/lib/";
81 dpavlin 1 use WebPAC::Input::ISI;
82 dpavlin 24
83     $WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields
84    
85 dpavlin 1 my $input = WebPAC::Input::ISI->new(
86 dpavlin 12 path => "$prefix/$path",
87 dpavlin 4 offset => $offset,
88     limit => $limit,
89 dpavlin 1 );
90    
91 dpavlin 25 our $num_records = $input->size;
92 dpavlin 74 our @reports;
93 dpavlin 1
94     sub report {
95 dpavlin 30 my $description = join(' ',@_);
96 dpavlin 1 my $dt = time - $t;
97 dpavlin 74 my $report = [ $description, $dt, $input->size / $dt ];
98     printf "[$port] %s in %1.4fs %.2f/s\n", @$report;
99     push @reports, $report;
100 dpavlin 10 $t = time;
101 dpavlin 1 }
102    
103 dpavlin 74 sub show_report {
104     "\n" . join( "\n", map { sprintf "%8.4fs %10.2f/s %s", $_->[1], $_->[2], $_->[0] } @reports ) . "\n";
105     }
106 dpavlin 1
107 dpavlin 74 report $input->size , 'records loaded';
108 dpavlin 1
109     mkdir 'out' unless -e 'out';
110    
111 dpavlin 5 our $out;
112    
113 dpavlin 8 our $cache;
114    
115 dpavlin 13 our $connected;
116    
117 dpavlin 64 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 dpavlin 11 sub send_nodes {
132 dpavlin 17 my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument!
133 dpavlin 21 my $header = defined $content ? length($content) : 0;
134 dpavlin 11 $header .= ' ' . join(' ', @_) if @_;
135    
136 dpavlin 74 warn "# send_nodes ", dump(@_), " to ", dump @nodes;
137 dpavlin 60
138 dpavlin 11 foreach my $node ( @nodes ) {
139    
140 dpavlin 64 my $sock = node_sock($node) || next;
141 dpavlin 11
142 dpavlin 67 warn "[$port] >>>> [$node] $header\n";
143 dpavlin 11 print $sock "$header\n$content" || warn "can't send $header to $node: $!";
144    
145 dpavlin 13 $connected->{$node} = $sock;
146 dpavlin 11 }
147     }
148    
149 dpavlin 16 sub get_node {
150     my $node = shift;
151    
152     my $sock = $connected->{$node};
153 dpavlin 64 if ( ! $sock || ! $sock->connected ) {
154     warn "[$port] no connection to $node";
155 dpavlin 16 delete $connected->{$node};
156     return;
157     }
158     chomp( my $size = <$sock> );
159 dpavlin 67 warn "[$port] <<<< [$node] $size bytes\n" if $debug || $size > 1024;
160 dpavlin 16 my $data;
161     read $sock, $data, $size;
162     return $data;
163     }
164    
165     sub send_sock {
166     my ( $sock, $data ) = @_;
167     my $size = length $data;
168 dpavlin 58 warn "[$port] >>>> $size bytes\n" if $debug || $size > 1024;
169 dpavlin 16 print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost;
170     }
171    
172 dpavlin 64 sub pull_node_file {
173     my ( $node, $file ) = @_;
174    
175     my $path = "/dev/shm/sack.$node.$file";
176 dpavlin 72 return if -e $path; # FIXME
177 dpavlin 64
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 dpavlin 80 my $merge_digest_mapping;
197    
198 dpavlin 13 sub merge_out {
199 dpavlin 50 my ( $from_node, $new ) = @_;
200 dpavlin 13
201 dpavlin 67 my $t_merge = time();
202    
203 dpavlin 64 pull_node_file $from_node => 'nr_md5';
204     pull_node_file $from_node => 'md5';
205    
206 dpavlin 59 my $remote_digest = Sack::Digest->new( port => $from_node );
207     my ( $local, $remote ) = ( 0, 0 );
208 dpavlin 50
209 dpavlin 59 my $tick = 0;
210 dpavlin 67 print STDERR "[$port] merge [$from_node]";
211 dpavlin 50
212 dpavlin 64 my $missing;
213    
214 dpavlin 13 foreach my $k1 ( keys %$new ) {
215    
216     foreach my $k2 ( keys %{ $new->{$k1} } ) {
217    
218 dpavlin 30 my $n = delete $new->{$k1}->{$k2};
219 dpavlin 50
220     if ( $k1 =~ m{#} ) {
221     die "ASSERT $k1 $k2" unless $k2 =~ m{^\d+$};
222     #warn "XXX $k1 $k2";
223 dpavlin 64
224 dpavlin 80 if ( defined $merge_digest_mapping->{$from_node}->[ $k2 ] ) {
225     $k2 = $merge_digest_mapping->{$from_node}->[ $k2 ];
226     } else {
227 dpavlin 64
228 dpavlin 80 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 dpavlin 50 }
249     }
250    
251 dpavlin 30 my $ref = ref $out->{$k1}->{$k2};
252 dpavlin 50 #warn "XXXX $k1 $k2 $ref";
253 dpavlin 13 if ( ! defined $out->{$k1}->{$k2} ) {
254     $out->{$k1}->{$k2} = $n;
255     } elsif ( $k1 =~ m{\+} ) {
256 dpavlin 19 # warn "## agregate $k1 $k2";
257 dpavlin 13 $out->{$k1}->{$k2} += $n;
258 dpavlin 21 } 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 dpavlin 13 } 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 dpavlin 59
270     if ( $tick++ % 1000 == 0 ) {
271     print STDERR ".";
272     } elsif ( $tick % 10000 == 0 ) {
273     print STDERR $tick;
274     }
275 dpavlin 13 }
276     }
277    
278 dpavlin 67 $t_merge = time - $t_merge;
279 dpavlin 79 my $digests = $local + $remote;
280 dpavlin 80 printf STDERR "\n[$port] merge %d in %.4fs %.2f/s digests local: %.1f%% %d/%d\n", $tick, $t_merge, $digests / $t_merge, $local * 100 / ( $digests || 1 ), $local, $remote;
281 dpavlin 74 push @reports, [ "$tick merged $from_node", $t_merge, $tick / $t_merge ];
282 dpavlin 59
283 dpavlin 64 warn "[$port] missing ", dump $missing if $missing;
284    
285 dpavlin 19 warn "## merge out ", dump $out if $debug;
286 dpavlin 13 }
287    
288 dpavlin 11 sub run_code {
289     my ( $view, $code ) = @_;
290    
291 dpavlin 21 warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug;
292 dpavlin 11
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 dpavlin 41 print STDERR "END @ $pos";
304 dpavlin 11 last;
305     }
306    
307     eval "$code";
308     if ( $@ ) {
309 dpavlin 41 warn "ABORT $pos $@\n";
310 dpavlin 20 last;
311 dpavlin 11 } else {
312     $affected++;
313     }
314 dpavlin 30
315 dpavlin 40 $pos % 10000 == 0 ? print STDERR $pos :
316     $pos % 1000 == 0 ? print STDERR "." : 0 ;
317 dpavlin 11 };
318    
319 dpavlin 74 report "$affected affected $view";
320 dpavlin 11
321     warn "WARN no \$out defined!" unless defined $out;
322 dpavlin 13
323 dpavlin 46 $digest->sync;
324    
325 dpavlin 13 if ( $connected ) {
326 dpavlin 24 foreach my $node ( keys %$connected ) {
327 dpavlin 74 warn "[$port] get_node [$node]\n";
328 dpavlin 24 my $o = get_node $node;
329 dpavlin 64 next unless $o;
330 dpavlin 24 my $s = length $o;
331     $o = thaw $o;
332 dpavlin 74 warn "[$port] got $s bytes from [$node]\n";
333 dpavlin 50 merge_out $node => $o;
334 dpavlin 24 }
335 dpavlin 13 }
336 dpavlin 11 }
337    
338 dpavlin 1 sub run_views {
339 dpavlin 6 @views = sort glob 'views/*.pl' unless @views;
340 dpavlin 1 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 dpavlin 11 run_code $view => $code;
349 dpavlin 1
350 dpavlin 11 if ( defined $out ) {
351 dpavlin 1
352 dpavlin 11 my $path = $view;
353     $path =~ s{views?/}{out/} || die "no view in $view";
354 dpavlin 30 $path =~ s{\.pl}{.storable};
355 dpavlin 1
356 dpavlin 9 unlink "$path.last" if -e "$path.last";
357     rename $path, "$path.last";
358 dpavlin 30
359     store $out => $path;
360 dpavlin 74 report "save $path", -s $path, "bytes";
361 dpavlin 35
362     if ( -s $path < 4096 ) {
363 dpavlin 45 print '$out = ', dump $digest->undigest_out($out);
364 dpavlin 35 }
365 dpavlin 11 }
366    
367     }
368    
369     }
370    
371 dpavlin 35
372     sub info_tabs {
373     "$port\t$offset\t$limit\t$num_records\t$path\t"
374 dpavlin 36 . join("\t", map {
375 dpavlin 35 my $b = $_;
376 dpavlin 46 $b =~ s{^.+\.$port\.([^/]+)$}{$1};
377 dpavlin 35 "$b " . -s $_
378 dpavlin 46 } glob "/dev/shm/sack.$port.*" );
379 dpavlin 35 }
380    
381    
382 dpavlin 30 if ( $port ) {
383 dpavlin 75
384     my $pid_path = "/tmp/sack.$port.pid";
385     if ( -e $pid_path ) {
386     my $pid = read_file $pid_path;
387 dpavlin 79 kill 9, $pid && warn "[$port] kill old $pid\n";
388 dpavlin 75 }
389     write_file $pid_path, $$;
390    
391    
392 dpavlin 11 my $sock = IO::Socket::INET->new(
393     Listen => SOMAXCONN,
394 dpavlin 21 LocalAddr => '127.0.0.1',
395 dpavlin 30 LocalPort => $port,
396 dpavlin 11 Proto => 'tcp',
397     Reuse => 1,
398     ) or die $!;
399    
400     while (1) {
401    
402 dpavlin 74 warn "[$port] accept path: $path offset: $offset limit: $limit #recs: $num_records\n";
403 dpavlin 11
404     my $client = $sock->accept();
405    
406 dpavlin 30 warn "[$port] <<<< connect from ", $client->peerhost, $/;
407 dpavlin 11
408     my @header = split(/\s/, <$client>);
409 dpavlin 30 warn "[$port] <<<< header ",dump(@header),$/;
410 dpavlin 11
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 dpavlin 16 send_sock $client => freeze $out;
419     } elsif ( $header[0] eq 'info' ) {
420 dpavlin 35 my $info = info_tabs;
421 dpavlin 30 warn "[$port] info $info\n";
422 dpavlin 74 $info .= "\n" . show_report if $content =~ m{r}i;
423 dpavlin 16 send_sock $client => $info;
424     } elsif ( $header[0] eq 'exit' ) {
425 dpavlin 30 warn "[$port] exit";
426 dpavlin 16 exit;
427 dpavlin 64 } 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 dpavlin 67 $digest->open;
440 dpavlin 5 } else {
441 dpavlin 30 warn "[$port] UNKNOWN $header[0]";
442 dpavlin 1 }
443    
444 dpavlin 75 $client->close;
445    
446 dpavlin 1 }
447     }
448    
449 dpavlin 30 sub info {
450 dpavlin 74 my $detail = shift || '';
451 dpavlin 30
452 dpavlin 74 send_nodes 'info' => $detail;
453    
454 dpavlin 30 my @info = (
455 dpavlin 36 "port\toffset\tlimit\t#recs\tpath",
456     "----\t------\t-----\t-----\t----",
457 dpavlin 35 info_tabs,
458 dpavlin 30 );
459    
460 dpavlin 74 push @info, show_report if $detail =~ m{r}i;
461    
462 dpavlin 30 push @info, get_node $_ foreach @nodes;
463    
464 dpavlin 74 print "[$port] INFO", $detail ? " $detail" : '', " \n"
465 dpavlin 30 , join("\n", @info)
466 dpavlin 74 , "\n" ;
467 dpavlin 30
468     return @info;
469     }
470    
471     info;
472 dpavlin 76 while ( keys %$connected != @nodes ) {
473     warn "[$port] wait for ", join(' ', grep { ! defined $connected->{$_} } @nodes );
474     sleep 1;
475     }
476 dpavlin 3 run_views;
477    
478 dpavlin 1 while ( 1 ) {
479    
480     print "sack> ";
481 dpavlin 21 chomp( my $cmd = <STDIN> );
482 dpavlin 1
483 dpavlin 21 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 dpavlin 33 #system "vi out/*";
495 dpavlin 53 $digest->sync;
496 dpavlin 33 system "bin/storableedit.pl", (glob('out/*.storable'))[0];
497 dpavlin 74 } elsif ( $cmd =~ m{^i(?:nfo)?\s?(\S+)?$}i ) {
498     info $1;
499 dpavlin 21 } elsif ( $cmd =~ m{^(q|e|x)}i ) {
500 dpavlin 16 warn "# exit";
501     send_nodes 'exit';
502     exit;
503 dpavlin 21 } elsif ( $cmd =~ m{^(v|r)}i ) {
504 dpavlin 3 run_views;
505 dpavlin 55 } elsif ( $cmd =~ m{^n(ode)?\s*(\d+)}i ) {
506 dpavlin 60 push @nodes, $2;
507 dpavlin 55 info;
508 dpavlin 21 } elsif ( $cmd ) {
509     warn "UNKNOWN ", dump $cmd;
510 dpavlin 3 }
511    
512 dpavlin 1 }
513    

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26