/[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 40 - (hide annotations)
Wed Sep 23 23:54:38 2009 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 7024 byte(s)
display real position in ticks
1 dpavlin 1 #!/usr/bin/perl
2    
3     use warnings;
4     use strict;
5    
6 dpavlin 36 our $VERSION = '0.03';
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 30 'listen|port=i' => \$port,
31 dpavlin 11 'connect=s' => \@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 30 $prefix =~ s{^(.*)/srv/Sack/[\./]+bin.+$}{$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 35 Sack::Digest->clean;
61 dpavlin 33 Sack::Digest->open( $port );
62 dpavlin 35 sub digest { Sack::Digest->to_int($_[0]) }
63 dpavlin 12
64     use lib "$prefix/srv/webpac2/lib/";
65 dpavlin 1 use WebPAC::Input::ISI;
66 dpavlin 24
67     $WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields
68    
69 dpavlin 1 my $input = WebPAC::Input::ISI->new(
70 dpavlin 12 path => "$prefix/$path",
71 dpavlin 4 offset => $offset,
72     limit => $limit,
73 dpavlin 1 );
74    
75 dpavlin 25 our $num_records = $input->size;
76 dpavlin 1
77     sub report {
78 dpavlin 30 my $description = join(' ',@_);
79 dpavlin 1 my $dt = time - $t;
80     printf "%s in %1.4fs %.2f/s\n", $description, $dt, $input->size / $dt;
81 dpavlin 10 $t = time;
82 dpavlin 1 }
83    
84    
85     report $input->size . ' records loaded';
86    
87     mkdir 'out' unless -e 'out';
88    
89 dpavlin 5 our $out;
90    
91 dpavlin 8 our $cache;
92    
93 dpavlin 13 our $connected;
94    
95 dpavlin 11 sub send_nodes {
96 dpavlin 17 my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument!
97 dpavlin 21 my $header = defined $content ? length($content) : 0;
98 dpavlin 11 $header .= ' ' . join(' ', @_) if @_;
99    
100     foreach my $node ( @nodes ) {
101    
102     my $sock = IO::Socket::INET->new(
103     PeerAddr => $node,
104     Proto => 'tcp',
105 dpavlin 16 );
106 dpavlin 11
107 dpavlin 16 if ( ! $sock ) {
108     warn "can't connect to $node - $!"; # FIXME die?
109     next;
110     }
111    
112 dpavlin 30 warn "[$port] >>>> $node $header\n";
113 dpavlin 11 print $sock "$header\n$content" || warn "can't send $header to $node: $!";
114    
115 dpavlin 13 $connected->{$node} = $sock;
116 dpavlin 11 }
117     }
118    
119 dpavlin 16 sub get_node {
120     my $node = shift;
121    
122     my $sock = $connected->{$node};
123     if ( ! $sock ) {
124 dpavlin 30 warn "[$port] ERROR lost connection to $node";
125 dpavlin 16 delete $connected->{$node};
126     return;
127     }
128     chomp( my $size = <$sock> );
129 dpavlin 30 warn "[$port] <<<< $node $size bytes\n";
130 dpavlin 16 my $data;
131     read $sock, $data, $size;
132     return $data;
133     }
134    
135     sub send_sock {
136     my ( $sock, $data ) = @_;
137     my $size = length $data;
138 dpavlin 30 warn "[$port] >>>> ", $sock->peerhost, " $size bytes\n";
139 dpavlin 16 print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost;
140     }
141    
142 dpavlin 13 sub merge_out {
143     my $new = shift;
144    
145     foreach my $k1 ( keys %$new ) {
146    
147     foreach my $k2 ( keys %{ $new->{$k1} } ) {
148    
149 dpavlin 30 my $n = delete $new->{$k1}->{$k2};
150     my $ref = ref $out->{$k1}->{$k2};
151 dpavlin 13
152     if ( ! defined $out->{$k1}->{$k2} ) {
153     $out->{$k1}->{$k2} = $n;
154     } elsif ( $k1 =~ m{\+} ) {
155 dpavlin 19 # warn "## agregate $k1 $k2";
156 dpavlin 13 $out->{$k1}->{$k2} += $n;
157 dpavlin 21 } elsif ( $ref eq 'ARRAY' ) {
158     if ( ref $n eq 'ARRAY' ) {
159     push @{ $out->{$k1}->{$k2} }, $_ foreach @$n;
160     } else {
161     push @{ $out->{$k1}->{$k2} }, $n;
162     }
163 dpavlin 13 } elsif ( $ref eq '' ) {
164     $out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ];
165     } else {
166     die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2});
167     }
168     }
169     }
170    
171 dpavlin 19 warn "## merge out ", dump $out if $debug;
172 dpavlin 13 }
173    
174 dpavlin 11 sub run_code {
175     my ( $view, $code ) = @_;
176    
177 dpavlin 21 warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug;
178 dpavlin 11
179     send_nodes view => $view => $code;
180    
181     undef $out;
182    
183     my $affected = 0;
184     $t = time;
185    
186     foreach my $pos ( $offset + 1 .. $offset + $input->size ) {
187     my $rec = $cache->{$pos} ||= $input->fetch_rec( $pos );
188     if ( ! $rec ) {
189     warn "END at $pos";
190     last;
191     }
192    
193     eval "$code";
194     if ( $@ ) {
195 dpavlin 20 warn "ABORT [$pos] $@\n";
196     last;
197 dpavlin 11 } else {
198     $affected++;
199     }
200 dpavlin 30
201 dpavlin 40 $pos % 10000 == 0 ? print STDERR $pos :
202     $pos % 1000 == 0 ? print STDERR "." : 0 ;
203 dpavlin 11 };
204    
205 dpavlin 30 report "\n[$port] RECS $affected $view";
206 dpavlin 11
207     warn "WARN no \$out defined!" unless defined $out;
208 dpavlin 13
209     if ( $connected ) {
210 dpavlin 24 foreach my $node ( keys %$connected ) {
211 dpavlin 30 warn "[$port] get_node $node\n";
212 dpavlin 24 my $o = get_node $node;
213     my $s = length $o;
214     $o = thaw $o;
215 dpavlin 30 warn "[$port] merge $node $s bytes\n";
216 dpavlin 24 merge_out $o;
217     }
218 dpavlin 13 }
219 dpavlin 11 }
220    
221 dpavlin 1 sub run_views {
222 dpavlin 6 @views = sort glob 'views/*.pl' unless @views;
223 dpavlin 1 warn "# views ", dump @views;
224    
225     foreach my $view ( @views ) {
226    
227     next if system("perl -c $view") != 0;
228    
229     my $code = read_file $view;
230    
231 dpavlin 11 run_code $view => $code;
232 dpavlin 1
233 dpavlin 11 if ( defined $out ) {
234 dpavlin 1
235 dpavlin 11 my $path = $view;
236     $path =~ s{views?/}{out/} || die "no view in $view";
237 dpavlin 30 $path =~ s{\.pl}{.storable};
238 dpavlin 1
239 dpavlin 9 unlink "$path.last" if -e "$path.last";
240     rename $path, "$path.last";
241 dpavlin 30
242     store $out => $path;
243 dpavlin 31 report "[$port] SAVE $path $offset-$limit", -s $path, "bytes";
244 dpavlin 35
245     if ( -s $path < 4096 ) {
246 dpavlin 36 print '$out = ', dump Sack::Digest->undigest_out($out);
247 dpavlin 35 }
248 dpavlin 11 }
249    
250     }
251    
252     }
253    
254 dpavlin 35
255     sub info_tabs {
256     "$port\t$offset\t$limit\t$num_records\t$path\t"
257 dpavlin 36 . join("\t", map {
258 dpavlin 35 my $b = $_;
259     $b =~ s{^.+/([^/]+)$}{$1};
260     "$b " . -s $_
261     } glob '/dev/shm/sack.*' );
262     }
263    
264    
265 dpavlin 30 if ( $port ) {
266 dpavlin 11 my $sock = IO::Socket::INET->new(
267     Listen => SOMAXCONN,
268 dpavlin 21 LocalAddr => '127.0.0.1',
269 dpavlin 30 LocalPort => $port,
270 dpavlin 11 Proto => 'tcp',
271     Reuse => 1,
272     ) or die $!;
273    
274     while (1) {
275    
276 dpavlin 30 warn "[$port] READY path: $path offset: $offset limit: $limit #recs: $num_records\n";
277 dpavlin 11
278     my $client = $sock->accept();
279    
280 dpavlin 30 warn "[$port] <<<< connect from ", $client->peerhost, $/;
281 dpavlin 11
282     my @header = split(/\s/, <$client>);
283 dpavlin 30 warn "[$port] <<<< header ",dump(@header),$/;
284 dpavlin 11
285     my $size = shift @header;
286    
287     my $content;
288     read $client, $content, $size;
289    
290     if ( $header[0] eq 'view' ) {
291     run_code $header[1] => $content;
292 dpavlin 16 send_sock $client => freeze $out;
293     } elsif ( $header[0] eq 'info' ) {
294 dpavlin 35 my $info = info_tabs;
295 dpavlin 30 warn "[$port] info $info\n";
296 dpavlin 16 send_sock $client => $info;
297     } elsif ( $header[0] eq 'exit' ) {
298 dpavlin 30 warn "[$port] exit";
299 dpavlin 16 exit;
300 dpavlin 5 } else {
301 dpavlin 30 warn "[$port] UNKNOWN $header[0]";
302 dpavlin 1 }
303    
304     }
305     }
306    
307 dpavlin 30 sub info {
308     send_nodes 'info' => $2;
309    
310     my @info = (
311 dpavlin 36 "port\toffset\tlimit\t#recs\tpath",
312     "----\t------\t-----\t-----\t----",
313 dpavlin 35 info_tabs,
314 dpavlin 30 );
315    
316     push @info, get_node $_ foreach @nodes;
317    
318     print "[$port] INFO\n"
319     , join("\n", @info)
320     , "\n\n" ;
321    
322     return @info;
323     }
324    
325     info;
326 dpavlin 3 run_views;
327    
328 dpavlin 1 while ( 1 ) {
329    
330     print "sack> ";
331 dpavlin 21 chomp( my $cmd = <STDIN> );
332 dpavlin 1
333 dpavlin 21 if ( $cmd =~ m{^(h|\?)} ) {
334     print << "__HELP__"
335     Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit
336    
337     View Run run views
338     VI \\e Output show output of last run
339     Info [\$VERSION] instrospect
340     Quit EXit shutdown
341    
342     __HELP__
343     } elsif ( $cmd =~ m{^(vi|\\e|o)}i ) {
344 dpavlin 33 #system "vi out/*";
345 dpavlin 37 Sack::Digest->sync;
346 dpavlin 33 system "bin/storableedit.pl", (glob('out/*.storable'))[0];
347 dpavlin 21 } elsif ( $cmd =~ m{^i(?:nfo)?\s?(.+)?$}i ) {
348 dpavlin 30 info;
349 dpavlin 21 } elsif ( $cmd =~ m{^(q|e|x)}i ) {
350 dpavlin 16 warn "# exit";
351     send_nodes 'exit';
352     exit;
353 dpavlin 21 } elsif ( $cmd =~ m{^(v|r)}i ) {
354 dpavlin 3 run_views;
355 dpavlin 21 } elsif ( $cmd ) {
356     warn "UNKNOWN ", dump $cmd;
357 dpavlin 3 }
358    
359 dpavlin 1 }
360    

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26