/[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 79 - (hide annotations)
Fri Sep 25 23:22:32 2009 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 10262 byte(s)
fix digest output
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 12 }
57    
58 dpavlin 33 use lib "$prefix/srv/Sack/lib/";
59     use Sack::Digest;
60 dpavlin 45 our $digest = Sack::Digest->new( port => $port, clean => 1 );
61     sub digest { $digest->to_int($_[0]) }
62 dpavlin 12
63     use lib "$prefix/srv/webpac2/lib/";
64 dpavlin 1 use WebPAC::Input::ISI;
65 dpavlin 24
66     $WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields
67    
68 dpavlin 1 my $input = WebPAC::Input::ISI->new(
69 dpavlin 12 path => "$prefix/$path",
70 dpavlin 4 offset => $offset,
71     limit => $limit,
72 dpavlin 1 );
73    
74 dpavlin 25 our $num_records = $input->size;
75 dpavlin 74 our @reports;
76 dpavlin 1
77     sub report {
78 dpavlin 30 my $description = join(' ',@_);
79 dpavlin 1 my $dt = time - $t;
80 dpavlin 74 my $report = [ $description, $dt, $input->size / $dt ];
81     printf "[$port] %s in %1.4fs %.2f/s\n", @$report;
82     push @reports, $report;
83 dpavlin 10 $t = time;
84 dpavlin 1 }
85    
86 dpavlin 74 sub show_report {
87     "\n" . join( "\n", map { sprintf "%8.4fs %10.2f/s %s", $_->[1], $_->[2], $_->[0] } @reports ) . "\n";
88     }
89 dpavlin 1
90 dpavlin 74 report $input->size , 'records loaded';
91 dpavlin 1
92     mkdir 'out' unless -e 'out';
93    
94 dpavlin 5 our $out;
95    
96 dpavlin 8 our $cache;
97    
98 dpavlin 13 our $connected;
99    
100 dpavlin 64 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 dpavlin 11 sub send_nodes {
115 dpavlin 17 my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument!
116 dpavlin 21 my $header = defined $content ? length($content) : 0;
117 dpavlin 11 $header .= ' ' . join(' ', @_) if @_;
118    
119 dpavlin 74 warn "# send_nodes ", dump(@_), " to ", dump @nodes;
120 dpavlin 60
121 dpavlin 11 foreach my $node ( @nodes ) {
122    
123 dpavlin 64 my $sock = node_sock($node) || next;
124 dpavlin 11
125 dpavlin 67 warn "[$port] >>>> [$node] $header\n";
126 dpavlin 11 print $sock "$header\n$content" || warn "can't send $header to $node: $!";
127    
128 dpavlin 13 $connected->{$node} = $sock;
129 dpavlin 11 }
130     }
131    
132 dpavlin 16 sub get_node {
133     my $node = shift;
134    
135     my $sock = $connected->{$node};
136 dpavlin 64 if ( ! $sock || ! $sock->connected ) {
137     warn "[$port] no connection to $node";
138 dpavlin 16 delete $connected->{$node};
139     return;
140     }
141     chomp( my $size = <$sock> );
142 dpavlin 67 warn "[$port] <<<< [$node] $size bytes\n" if $debug || $size > 1024;
143 dpavlin 16 my $data;
144     read $sock, $data, $size;
145     return $data;
146     }
147    
148     sub send_sock {
149     my ( $sock, $data ) = @_;
150     my $size = length $data;
151 dpavlin 58 warn "[$port] >>>> $size bytes\n" if $debug || $size > 1024;
152 dpavlin 16 print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost;
153     }
154    
155 dpavlin 64 sub pull_node_file {
156     my ( $node, $file ) = @_;
157    
158     my $path = "/dev/shm/sack.$node.$file";
159 dpavlin 72 return if -e $path; # FIXME
160 dpavlin 64
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 dpavlin 13 sub merge_out {
180 dpavlin 50 my ( $from_node, $new ) = @_;
181 dpavlin 13
182 dpavlin 67 my $t_merge = time();
183    
184 dpavlin 64 pull_node_file $from_node => 'nr_md5';
185     pull_node_file $from_node => 'md5';
186    
187 dpavlin 59 my $remote_digest = Sack::Digest->new( port => $from_node );
188     my ( $local, $remote ) = ( 0, 0 );
189 dpavlin 50
190 dpavlin 59 my $tick = 0;
191 dpavlin 67 print STDERR "[$port] merge [$from_node]";
192 dpavlin 50
193 dpavlin 64 my $missing;
194    
195 dpavlin 13 foreach my $k1 ( keys %$new ) {
196    
197     foreach my $k2 ( keys %{ $new->{$k1} } ) {
198    
199 dpavlin 30 my $n = delete $new->{$k1}->{$k2};
200 dpavlin 50
201     if ( $k1 =~ m{#} ) {
202     die "ASSERT $k1 $k2" unless $k2 =~ m{^\d+$};
203     #warn "XXX $k1 $k2";
204 dpavlin 64 my $md5 = $remote_digest->{nr_md5}->[$k2];
205    
206     if ( ! $md5 ) {
207 dpavlin 67 $missing->{nr_md5}->{$from_node}++; # FIXME die?
208 dpavlin 64 next;
209     }
210    
211 dpavlin 50 if ( my $local_k2 = $digest->{md5_nr}->{$md5} ) {
212     $k2 = $local_k2;
213     $local++;
214 dpavlin 67 } elsif ( my $full = $remote_digest->{md5}->{$md5} ) {
215 dpavlin 50 $k2 = $digest->to_int( $remote_digest->{md5}->{$md5} );
216     $remote++;
217 dpavlin 67 } else {
218     $missing->{md5}->{$from_node}++;
219 dpavlin 50 }
220     }
221    
222 dpavlin 30 my $ref = ref $out->{$k1}->{$k2};
223 dpavlin 50 #warn "XXXX $k1 $k2 $ref";
224 dpavlin 13 if ( ! defined $out->{$k1}->{$k2} ) {
225     $out->{$k1}->{$k2} = $n;
226     } elsif ( $k1 =~ m{\+} ) {
227 dpavlin 19 # warn "## agregate $k1 $k2";
228 dpavlin 13 $out->{$k1}->{$k2} += $n;
229 dpavlin 21 } elsif ( $ref eq 'ARRAY' ) {
230     if ( ref $n eq 'ARRAY' ) {
231     push @{ $out->{$k1}->{$k2} }, $_ foreach @$n;
232     } else {
233     push @{ $out->{$k1}->{$k2} }, $n;
234     }
235 dpavlin 13 } elsif ( $ref eq '' ) {
236     $out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ];
237     } else {
238     die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2});
239     }
240 dpavlin 59
241     if ( $tick++ % 1000 == 0 ) {
242     print STDERR ".";
243     } elsif ( $tick % 10000 == 0 ) {
244     print STDERR $tick;
245     }
246 dpavlin 13 }
247     }
248    
249 dpavlin 67 $t_merge = time - $t_merge;
250 dpavlin 79 my $digests = $local + $remote;
251     printf STDERR "\n[$port] merge %d in %.4fs %.2f/s digests local: %.1f%% %d/%d\n", $tick, $t_merge, $digests / $t_merge, $local * 100 / $digests, $local, $remote;
252 dpavlin 74 push @reports, [ "$tick merged $from_node", $t_merge, $tick / $t_merge ];
253 dpavlin 59
254 dpavlin 64 warn "[$port] missing ", dump $missing if $missing;
255    
256 dpavlin 19 warn "## merge out ", dump $out if $debug;
257 dpavlin 13 }
258    
259 dpavlin 11 sub run_code {
260     my ( $view, $code ) = @_;
261    
262 dpavlin 21 warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug;
263 dpavlin 11
264     send_nodes view => $view => $code;
265    
266     undef $out;
267    
268     my $affected = 0;
269     $t = time;
270    
271     foreach my $pos ( $offset + 1 .. $offset + $input->size ) {
272     my $rec = $cache->{$pos} ||= $input->fetch_rec( $pos );
273     if ( ! $rec ) {
274 dpavlin 41 print STDERR "END @ $pos";
275 dpavlin 11 last;
276     }
277    
278     eval "$code";
279     if ( $@ ) {
280 dpavlin 41 warn "ABORT $pos $@\n";
281 dpavlin 20 last;
282 dpavlin 11 } else {
283     $affected++;
284     }
285 dpavlin 30
286 dpavlin 40 $pos % 10000 == 0 ? print STDERR $pos :
287     $pos % 1000 == 0 ? print STDERR "." : 0 ;
288 dpavlin 11 };
289    
290 dpavlin 74 report "$affected affected $view";
291 dpavlin 11
292     warn "WARN no \$out defined!" unless defined $out;
293 dpavlin 13
294 dpavlin 46 $digest->sync;
295    
296 dpavlin 13 if ( $connected ) {
297 dpavlin 24 foreach my $node ( keys %$connected ) {
298 dpavlin 74 warn "[$port] get_node [$node]\n";
299 dpavlin 24 my $o = get_node $node;
300 dpavlin 64 next unless $o;
301 dpavlin 24 my $s = length $o;
302     $o = thaw $o;
303 dpavlin 74 warn "[$port] got $s bytes from [$node]\n";
304 dpavlin 50 merge_out $node => $o;
305 dpavlin 24 }
306 dpavlin 13 }
307 dpavlin 11 }
308    
309 dpavlin 1 sub run_views {
310 dpavlin 6 @views = sort glob 'views/*.pl' unless @views;
311 dpavlin 1 warn "# views ", dump @views;
312    
313     foreach my $view ( @views ) {
314    
315     next if system("perl -c $view") != 0;
316    
317     my $code = read_file $view;
318    
319 dpavlin 11 run_code $view => $code;
320 dpavlin 1
321 dpavlin 11 if ( defined $out ) {
322 dpavlin 1
323 dpavlin 11 my $path = $view;
324     $path =~ s{views?/}{out/} || die "no view in $view";
325 dpavlin 30 $path =~ s{\.pl}{.storable};
326 dpavlin 1
327 dpavlin 9 unlink "$path.last" if -e "$path.last";
328     rename $path, "$path.last";
329 dpavlin 30
330     store $out => $path;
331 dpavlin 74 report "save $path", -s $path, "bytes";
332 dpavlin 35
333     if ( -s $path < 4096 ) {
334 dpavlin 45 print '$out = ', dump $digest->undigest_out($out);
335 dpavlin 35 }
336 dpavlin 11 }
337    
338     }
339    
340     }
341    
342 dpavlin 35
343     sub info_tabs {
344     "$port\t$offset\t$limit\t$num_records\t$path\t"
345 dpavlin 36 . join("\t", map {
346 dpavlin 35 my $b = $_;
347 dpavlin 46 $b =~ s{^.+\.$port\.([^/]+)$}{$1};
348 dpavlin 35 "$b " . -s $_
349 dpavlin 46 } glob "/dev/shm/sack.$port.*" );
350 dpavlin 35 }
351    
352    
353 dpavlin 30 if ( $port ) {
354 dpavlin 75
355     my $pid_path = "/tmp/sack.$port.pid";
356     if ( -e $pid_path ) {
357     my $pid = read_file $pid_path;
358 dpavlin 79 kill 9, $pid && warn "[$port] kill old $pid\n";
359 dpavlin 75 }
360     write_file $pid_path, $$;
361    
362    
363 dpavlin 11 my $sock = IO::Socket::INET->new(
364     Listen => SOMAXCONN,
365 dpavlin 21 LocalAddr => '127.0.0.1',
366 dpavlin 30 LocalPort => $port,
367 dpavlin 11 Proto => 'tcp',
368     Reuse => 1,
369     ) or die $!;
370    
371     while (1) {
372    
373 dpavlin 74 warn "[$port] accept path: $path offset: $offset limit: $limit #recs: $num_records\n";
374 dpavlin 11
375     my $client = $sock->accept();
376    
377 dpavlin 30 warn "[$port] <<<< connect from ", $client->peerhost, $/;
378 dpavlin 11
379     my @header = split(/\s/, <$client>);
380 dpavlin 30 warn "[$port] <<<< header ",dump(@header),$/;
381 dpavlin 11
382     my $size = shift @header;
383    
384     my $content;
385     read $client, $content, $size;
386    
387     if ( $header[0] eq 'view' ) {
388     run_code $header[1] => $content;
389 dpavlin 16 send_sock $client => freeze $out;
390     } elsif ( $header[0] eq 'info' ) {
391 dpavlin 35 my $info = info_tabs;
392 dpavlin 30 warn "[$port] info $info\n";
393 dpavlin 74 $info .= "\n" . show_report if $content =~ m{r}i;
394 dpavlin 16 send_sock $client => $info;
395     } elsif ( $header[0] eq 'exit' ) {
396 dpavlin 30 warn "[$port] exit";
397 dpavlin 16 exit;
398 dpavlin 64 } elsif ( $header[0] eq 'file' ) {
399     $digest->close;
400     my $path = "/dev/shm/sack.$port.$header[1]";
401     my $size = -s $path;
402     warn "[$port] >>>> file $path $size bytes\n";
403     print $client "$size\n";
404     my $block = 4096;
405     my $buff = ' ' x $block;
406     open(my $fh, '<', $path) || die "can't open $path";
407     while ( read $fh, $buff, $block ) {
408     print $client $buff;
409     }
410 dpavlin 67 $digest->open;
411 dpavlin 5 } else {
412 dpavlin 30 warn "[$port] UNKNOWN $header[0]";
413 dpavlin 1 }
414    
415 dpavlin 75 $client->close;
416    
417 dpavlin 1 }
418     }
419    
420 dpavlin 30 sub info {
421 dpavlin 74 my $detail = shift || '';
422 dpavlin 30
423 dpavlin 74 send_nodes 'info' => $detail;
424    
425 dpavlin 30 my @info = (
426 dpavlin 36 "port\toffset\tlimit\t#recs\tpath",
427     "----\t------\t-----\t-----\t----",
428 dpavlin 35 info_tabs,
429 dpavlin 30 );
430    
431 dpavlin 74 push @info, show_report if $detail =~ m{r}i;
432    
433 dpavlin 30 push @info, get_node $_ foreach @nodes;
434    
435 dpavlin 74 print "[$port] INFO", $detail ? " $detail" : '', " \n"
436 dpavlin 30 , join("\n", @info)
437 dpavlin 74 , "\n" ;
438 dpavlin 30
439     return @info;
440     }
441    
442     info;
443 dpavlin 76 while ( keys %$connected != @nodes ) {
444     warn "[$port] wait for ", join(' ', grep { ! defined $connected->{$_} } @nodes );
445     sleep 1;
446     }
447 dpavlin 3 run_views;
448    
449 dpavlin 1 while ( 1 ) {
450    
451     print "sack> ";
452 dpavlin 21 chomp( my $cmd = <STDIN> );
453 dpavlin 1
454 dpavlin 21 if ( $cmd =~ m{^(h|\?)} ) {
455     print << "__HELP__"
456     Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit
457    
458     View Run run views
459     VI \\e Output show output of last run
460     Info [\$VERSION] instrospect
461     Quit EXit shutdown
462    
463     __HELP__
464     } elsif ( $cmd =~ m{^(vi|\\e|o)}i ) {
465 dpavlin 33 #system "vi out/*";
466 dpavlin 53 $digest->sync;
467 dpavlin 33 system "bin/storableedit.pl", (glob('out/*.storable'))[0];
468 dpavlin 74 } elsif ( $cmd =~ m{^i(?:nfo)?\s?(\S+)?$}i ) {
469     info $1;
470 dpavlin 21 } elsif ( $cmd =~ m{^(q|e|x)}i ) {
471 dpavlin 16 warn "# exit";
472     send_nodes 'exit';
473     exit;
474 dpavlin 21 } elsif ( $cmd =~ m{^(v|r)}i ) {
475 dpavlin 3 run_views;
476 dpavlin 55 } elsif ( $cmd =~ m{^n(ode)?\s*(\d+)}i ) {
477 dpavlin 60 push @nodes, $2;
478 dpavlin 55 info;
479 dpavlin 21 } elsif ( $cmd ) {
480     warn "UNKNOWN ", dump $cmd;
481 dpavlin 3 }
482    
483 dpavlin 1 }
484    

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26