/[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 26 - (hide annotations)
Tue Sep 22 21:45:04 2009 UTC (14 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 6604 byte(s)
no backtrace from warn
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 dpavlin 25 our $num_records = $input->size;
70 dpavlin 1
71     sub report {
72     my $description = shift;
73     my $dt = time - $t;
74     printf "%s in %1.4fs %.2f/s\n", $description, $dt, $input->size / $dt;
75 dpavlin 10 $t = time;
76 dpavlin 1 }
77    
78    
79     report $input->size . ' records loaded';
80    
81     mkdir 'out' unless -e 'out';
82    
83 dpavlin 5 our $out;
84    
85 dpavlin 8 our $cache;
86    
87 dpavlin 13 our $connected;
88    
89 dpavlin 11 sub send_nodes {
90 dpavlin 17 my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument!
91 dpavlin 21 my $header = defined $content ? length($content) : 0;
92 dpavlin 11 $header .= ' ' . join(' ', @_) if @_;
93    
94     foreach my $node ( @nodes ) {
95    
96     my $sock = IO::Socket::INET->new(
97     PeerAddr => $node,
98     Proto => 'tcp',
99 dpavlin 16 );
100 dpavlin 11
101 dpavlin 16 if ( ! $sock ) {
102     warn "can't connect to $node - $!"; # FIXME die?
103     next;
104     }
105    
106 dpavlin 21 print ">>>> $listen $node $header\n";
107 dpavlin 11 print $sock "$header\n$content" || warn "can't send $header to $node: $!";
108    
109 dpavlin 13 $connected->{$node} = $sock;
110 dpavlin 11 }
111     }
112    
113 dpavlin 16 sub get_node {
114     my $node = shift;
115    
116     my $sock = $connected->{$node};
117     if ( ! $sock ) {
118     warn "ERROR: lost connection to $node";
119     delete $connected->{$node};
120     return;
121     }
122     chomp( my $size = <$sock> );
123 dpavlin 21 warn "<<<< $listen $node $size bytes\n";
124 dpavlin 16 my $data;
125     read $sock, $data, $size;
126     return $data;
127     }
128    
129     sub send_sock {
130     my ( $sock, $data ) = @_;
131     my $size = length $data;
132 dpavlin 26 warn ">>>> $listen ", $sock->peerhost, " $size bytes\n";
133 dpavlin 16 print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost;
134     }
135    
136 dpavlin 13 sub merge_out {
137     my $new = shift;
138    
139     foreach my $k1 ( keys %$new ) {
140    
141     foreach my $k2 ( keys %{ $new->{$k1} } ) {
142    
143     my $n = $new->{$k1}->{$k2};
144     my $ref = ref $out->{$k1}->{$k2};
145    
146     if ( ! defined $out->{$k1}->{$k2} ) {
147     $out->{$k1}->{$k2} = $n;
148     } elsif ( $k1 =~ m{\+} ) {
149 dpavlin 19 # warn "## agregate $k1 $k2";
150 dpavlin 13 $out->{$k1}->{$k2} += $n;
151 dpavlin 21 } elsif ( $ref eq 'ARRAY' ) {
152     if ( ref $n eq 'ARRAY' ) {
153     push @{ $out->{$k1}->{$k2} }, $_ foreach @$n;
154     } else {
155     push @{ $out->{$k1}->{$k2} }, $n;
156     }
157 dpavlin 13 } elsif ( $ref eq '' ) {
158     $out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ];
159     } else {
160     die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2});
161     }
162     }
163     }
164    
165 dpavlin 19 warn "## merge out ", dump $out if $debug;
166 dpavlin 13 }
167    
168 dpavlin 11 sub run_code {
169     my ( $view, $code ) = @_;
170    
171 dpavlin 21 warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug;
172 dpavlin 11
173     send_nodes view => $view => $code;
174    
175     undef $out;
176    
177     my $affected = 0;
178     $t = time;
179    
180     foreach my $pos ( $offset + 1 .. $offset + $input->size ) {
181     my $rec = $cache->{$pos} ||= $input->fetch_rec( $pos );
182     if ( ! $rec ) {
183     warn "END at $pos";
184     last;
185     }
186    
187     eval "$code";
188     if ( $@ ) {
189 dpavlin 20 warn "ABORT [$pos] $@\n";
190     last;
191 dpavlin 11 } else {
192     $affected++;
193     }
194     };
195    
196     report "$affected affected records $view";
197    
198     warn "WARN no \$out defined!" unless defined $out;
199 dpavlin 13
200     if ( $connected ) {
201 dpavlin 24 foreach my $node ( keys %$connected ) {
202     warn "# $listen get_node $node\n";
203     my $o = get_node $node;
204     my $s = length $o;
205     $o = thaw $o;
206     warn "# $listen merge $s bytes\n";
207     merge_out $o;
208     }
209 dpavlin 13 }
210 dpavlin 11 }
211    
212 dpavlin 1 sub run_views {
213 dpavlin 6 @views = sort glob 'views/*.pl' unless @views;
214 dpavlin 1 warn "# views ", dump @views;
215    
216     foreach my $view ( @views ) {
217    
218     next if system("perl -c $view") != 0;
219    
220     my $code = read_file $view;
221    
222 dpavlin 11 run_code $view => $code;
223 dpavlin 1
224 dpavlin 11 if ( defined $out ) {
225     my $dump = dump $out;
226     my $len = length $dump;
227 dpavlin 1
228 dpavlin 11 my $path = $view;
229     $path =~ s{views?/}{out/} || die "no view in $view";
230     $path =~ s{\.pl}{};
231 dpavlin 1
232 dpavlin 11 print "OUT $view $offset/$limit $len bytes $path"
233 dpavlin 5 , ( $len < 10000 ? " \$out = $dump" : ' SAVED ONLY' )
234 dpavlin 1 , "\n"
235     ;
236 dpavlin 5
237 dpavlin 9 unlink "$path.last" if -e "$path.last";
238     rename $path, "$path.last";
239 dpavlin 1 write_file $path, $dump;
240 dpavlin 11 report "SAVE $path";
241     }
242    
243     }
244    
245     }
246    
247     if ( $listen ) {
248     my $sock = IO::Socket::INET->new(
249     Listen => SOMAXCONN,
250 dpavlin 21 LocalAddr => '127.0.0.1',
251 dpavlin 11 LocalPort => $listen,
252     Proto => 'tcp',
253     Reuse => 1,
254     ) or die $!;
255    
256     while (1) {
257    
258 dpavlin 25 warn "NODE $listen ready - path: $path offset: $offset limit: $limit #recs: $num_records\n";
259 dpavlin 11
260     my $client = $sock->accept();
261    
262 dpavlin 16 warn "<<<< $listen connect from ", $client->peerhost, $/;
263 dpavlin 11
264     my @header = split(/\s/, <$client>);
265 dpavlin 22 warn "<<<< $listen header ",dump(@header),$/;
266 dpavlin 11
267     my $size = shift @header;
268    
269     my $content;
270     read $client, $content, $size;
271    
272     if ( $header[0] eq 'view' ) {
273     run_code $header[1] => $content;
274 dpavlin 16 send_sock $client => freeze $out;
275     } elsif ( $header[0] eq 'info' ) {
276 dpavlin 25 my $info = "$listen\t$offset\t$limit\t$num_records\t$path";
277 dpavlin 21 $info .= "\t" . eval $header[1] if $header[1];
278 dpavlin 16 warn "info $info\n";
279     send_sock $client => $info;
280     } elsif ( $header[0] eq 'exit' ) {
281     warn "exit $listen";
282     exit;
283 dpavlin 5 } else {
284 dpavlin 16 warn "WARN $listen unknown";
285 dpavlin 1 }
286    
287     }
288     }
289    
290 dpavlin 3 run_views;
291    
292 dpavlin 1 while ( 1 ) {
293    
294     print "sack> ";
295 dpavlin 21 chomp( my $cmd = <STDIN> );
296 dpavlin 1
297 dpavlin 21 if ( $cmd =~ m{^(h|\?)} ) {
298     print << "__HELP__"
299     Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit
300    
301     View Run run views
302     VI \\e Output show output of last run
303     Info [\$VERSION] instrospect
304     Quit EXit shutdown
305    
306     __HELP__
307     } elsif ( $cmd =~ m{^(vi|\\e|o)}i ) {
308 dpavlin 3 system "vi out/*";
309 dpavlin 21 } elsif ( $cmd =~ m{^i(?:nfo)?\s?(.+)?$}i ) {
310 dpavlin 16 print "# nodes: ", join(' ',@nodes), $/;
311    
312 dpavlin 21 send_nodes 'info' => $2;
313    
314 dpavlin 16 my @info = (
315 dpavlin 25 "node\toffset\tlimit\t#recs\tpath",
316     "----\t------\t-----\t-----\t----",
317     "0\t$offset\t$limit\t$num_records\t$path",
318 dpavlin 16 );
319    
320     push @info, get_node $_ foreach @nodes;
321    
322     print "$_\n" foreach @info;
323    
324 dpavlin 21 } elsif ( $cmd =~ m{^(q|e|x)}i ) {
325 dpavlin 16 warn "# exit";
326     send_nodes 'exit';
327     exit;
328 dpavlin 21 } elsif ( $cmd =~ m{^(v|r)}i ) {
329 dpavlin 3 run_views;
330 dpavlin 21 } elsif ( $cmd ) {
331     warn "UNKNOWN ", dump $cmd;
332 dpavlin 3 }
333    
334 dpavlin 1 }
335    

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26