/[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 24 - (hide annotations)
Tue Sep 22 21:27:13 2009 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 6487 byte(s)
don't parse subfields in WebPAC::Input::ISI for nice memory usage
check if we got anything from client before unserializing it

1 dpavlin 1 #!/usr/bin/perl
2    
3     use warnings;
4     use strict;
5    
6 dpavlin 21 our $VERSION = '0.01';
7    
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 13 use Storable qw/freeze thaw/;
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 21 my $listen = 0; # off
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 11 'listen|port=i' => \$listen,
31     '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     if ( $prefix =~ s{^./}{} ) {
44     chomp( my $pwd = `pwd` );
45     $prefix = "$pwd/$prefix";
46     }
47 dpavlin 14 $prefix =~ s{^(.*)/srv/Sack/bin.+$}{$1};
48 dpavlin 12 warn "# prefix $prefix";
49 dpavlin 21
50     $SIG{INT} = sub {
51     my $signame = shift;
52     send_nodes 'exit';
53     die "SIG$signame";
54     };
55 dpavlin 12 }
56    
57    
58     use lib "$prefix/srv/webpac2/lib/";
59 dpavlin 1 use WebPAC::Input::ISI;
60 dpavlin 24
61     $WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields
62    
63 dpavlin 1 my $input = WebPAC::Input::ISI->new(
64 dpavlin 12 path => "$prefix/$path",
65 dpavlin 4 offset => $offset,
66     limit => $limit,
67 dpavlin 1 );
68    
69    
70     sub report {
71     my $description = shift;
72     my $dt = time - $t;
73     printf "%s in %1.4fs %.2f/s\n", $description, $dt, $input->size / $dt;
74 dpavlin 10 $t = time;
75 dpavlin 1 }
76    
77    
78     report $input->size . ' records loaded';
79    
80     mkdir 'out' unless -e 'out';
81    
82 dpavlin 5 our $out;
83    
84 dpavlin 8 our $cache;
85    
86 dpavlin 13 our $connected;
87    
88 dpavlin 11 sub send_nodes {
89 dpavlin 17 my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument!
90 dpavlin 21 my $header = defined $content ? length($content) : 0;
91 dpavlin 11 $header .= ' ' . join(' ', @_) if @_;
92    
93     foreach my $node ( @nodes ) {
94    
95     my $sock = IO::Socket::INET->new(
96     PeerAddr => $node,
97     Proto => 'tcp',
98 dpavlin 16 );
99 dpavlin 11
100 dpavlin 16 if ( ! $sock ) {
101     warn "can't connect to $node - $!"; # FIXME die?
102     next;
103     }
104    
105 dpavlin 21 print ">>>> $listen $node $header\n";
106 dpavlin 11 print $sock "$header\n$content" || warn "can't send $header to $node: $!";
107    
108 dpavlin 13 $connected->{$node} = $sock;
109 dpavlin 11 }
110     }
111    
112 dpavlin 16 sub get_node {
113     my $node = shift;
114    
115     my $sock = $connected->{$node};
116     if ( ! $sock ) {
117     warn "ERROR: lost connection to $node";
118     delete $connected->{$node};
119     return;
120     }
121     chomp( my $size = <$sock> );
122 dpavlin 21 warn "<<<< $listen $node $size bytes\n";
123 dpavlin 16 my $data;
124     read $sock, $data, $size;
125     return $data;
126     }
127    
128     sub send_sock {
129     my ( $sock, $data ) = @_;
130     my $size = length $data;
131 dpavlin 21 warn ">>>> $listen ", $sock->peerhost, " $size bytes";
132 dpavlin 16 print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost;
133     }
134    
135 dpavlin 13 sub merge_out {
136     my $new = shift;
137    
138     foreach my $k1 ( keys %$new ) {
139    
140     foreach my $k2 ( keys %{ $new->{$k1} } ) {
141    
142     my $n = $new->{$k1}->{$k2};
143     my $ref = ref $out->{$k1}->{$k2};
144    
145     if ( ! defined $out->{$k1}->{$k2} ) {
146     $out->{$k1}->{$k2} = $n;
147     } elsif ( $k1 =~ m{\+} ) {
148 dpavlin 19 # warn "## agregate $k1 $k2";
149 dpavlin 13 $out->{$k1}->{$k2} += $n;
150 dpavlin 21 } elsif ( $ref eq 'ARRAY' ) {
151     if ( ref $n eq 'ARRAY' ) {
152     push @{ $out->{$k1}->{$k2} }, $_ foreach @$n;
153     } else {
154     push @{ $out->{$k1}->{$k2} }, $n;
155     }
156 dpavlin 13 } elsif ( $ref eq '' ) {
157     $out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ];
158     } else {
159     die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2});
160     }
161     }
162     }
163    
164 dpavlin 19 warn "## merge out ", dump $out if $debug;
165 dpavlin 13 }
166    
167 dpavlin 11 sub run_code {
168     my ( $view, $code ) = @_;
169    
170 dpavlin 21 warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug;
171 dpavlin 11
172     send_nodes view => $view => $code;
173    
174     undef $out;
175    
176     my $affected = 0;
177     $t = time;
178    
179     foreach my $pos ( $offset + 1 .. $offset + $input->size ) {
180     my $rec = $cache->{$pos} ||= $input->fetch_rec( $pos );
181     if ( ! $rec ) {
182     warn "END at $pos";
183     last;
184     }
185    
186     eval "$code";
187     if ( $@ ) {
188 dpavlin 20 warn "ABORT [$pos] $@\n";
189     last;
190 dpavlin 11 } else {
191     $affected++;
192     }
193     };
194    
195     report "$affected affected records $view";
196    
197     warn "WARN no \$out defined!" unless defined $out;
198 dpavlin 13
199     if ( $connected ) {
200 dpavlin 24 foreach my $node ( keys %$connected ) {
201     warn "# $listen get_node $node\n";
202     my $o = get_node $node;
203     my $s = length $o;
204     $o = thaw $o;
205     warn "# $listen merge $s bytes\n";
206     merge_out $o;
207     }
208 dpavlin 13 }
209 dpavlin 11 }
210    
211 dpavlin 1 sub run_views {
212 dpavlin 6 @views = sort glob 'views/*.pl' unless @views;
213 dpavlin 1 warn "# views ", dump @views;
214    
215     foreach my $view ( @views ) {
216    
217     next if system("perl -c $view") != 0;
218    
219     my $code = read_file $view;
220    
221 dpavlin 11 run_code $view => $code;
222 dpavlin 1
223 dpavlin 11 if ( defined $out ) {
224     my $dump = dump $out;
225     my $len = length $dump;
226 dpavlin 1
227 dpavlin 11 my $path = $view;
228     $path =~ s{views?/}{out/} || die "no view in $view";
229     $path =~ s{\.pl}{};
230 dpavlin 1
231 dpavlin 11 print "OUT $view $offset/$limit $len bytes $path"
232 dpavlin 5 , ( $len < 10000 ? " \$out = $dump" : ' SAVED ONLY' )
233 dpavlin 1 , "\n"
234     ;
235 dpavlin 5
236 dpavlin 9 unlink "$path.last" if -e "$path.last";
237     rename $path, "$path.last";
238 dpavlin 1 write_file $path, $dump;
239 dpavlin 11 report "SAVE $path";
240     }
241    
242     }
243    
244     }
245    
246     if ( $listen ) {
247     my $sock = IO::Socket::INET->new(
248     Listen => SOMAXCONN,
249 dpavlin 21 LocalAddr => '127.0.0.1',
250 dpavlin 11 LocalPort => $listen,
251     Proto => 'tcp',
252     Reuse => 1,
253     ) or die $!;
254    
255     while (1) {
256    
257 dpavlin 22 warn "NODE $listen ready - wating for connection\n";
258 dpavlin 11
259     my $client = $sock->accept();
260    
261 dpavlin 16 warn "<<<< $listen connect from ", $client->peerhost, $/;
262 dpavlin 11
263     my @header = split(/\s/, <$client>);
264 dpavlin 22 warn "<<<< $listen header ",dump(@header),$/;
265 dpavlin 11
266     my $size = shift @header;
267    
268     my $content;
269     read $client, $content, $size;
270    
271     if ( $header[0] eq 'view' ) {
272     run_code $header[1] => $content;
273 dpavlin 16 send_sock $client => freeze $out;
274     } elsif ( $header[0] eq 'info' ) {
275     my $info = "$listen\t$offset\t$limit\t$path";
276 dpavlin 21 $info .= "\t" . eval $header[1] if $header[1];
277 dpavlin 16 warn "info $info\n";
278     send_sock $client => $info;
279     } elsif ( $header[0] eq 'exit' ) {
280     warn "exit $listen";
281     exit;
282 dpavlin 5 } else {
283 dpavlin 16 warn "WARN $listen unknown";
284 dpavlin 1 }
285    
286     }
287     }
288    
289 dpavlin 3 run_views;
290    
291 dpavlin 1 while ( 1 ) {
292    
293     print "sack> ";
294 dpavlin 21 chomp( my $cmd = <STDIN> );
295 dpavlin 1
296 dpavlin 21 if ( $cmd =~ m{^(h|\?)} ) {
297     print << "__HELP__"
298     Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit
299    
300     View Run run views
301     VI \\e Output show output of last run
302     Info [\$VERSION] instrospect
303     Quit EXit shutdown
304    
305     __HELP__
306     } elsif ( $cmd =~ m{^(vi|\\e|o)}i ) {
307 dpavlin 3 system "vi out/*";
308 dpavlin 21 } elsif ( $cmd =~ m{^i(?:nfo)?\s?(.+)?$}i ) {
309 dpavlin 16 print "# nodes: ", join(' ',@nodes), $/;
310    
311 dpavlin 21 send_nodes 'info' => $2;
312    
313 dpavlin 16 my @info = (
314     "node\toffset\tlimit\tpath",
315     "----\t------\t-----\t----",
316 dpavlin 21 "0\t$offset\t$limit\t$path",
317 dpavlin 16 );
318    
319     push @info, get_node $_ foreach @nodes;
320    
321     print "$_\n" foreach @info;
322    
323 dpavlin 21 } elsif ( $cmd =~ m{^(q|e|x)}i ) {
324 dpavlin 16 warn "# exit";
325     send_nodes 'exit';
326     exit;
327 dpavlin 21 } elsif ( $cmd =~ m{^(v|r)}i ) {
328 dpavlin 3 run_views;
329 dpavlin 21 } elsif ( $cmd ) {
330     warn "UNKNOWN ", dump $cmd;
331 dpavlin 3 }
332    
333 dpavlin 1 }
334    

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26