| 1 |
#!/usr/bin/perl |
| 2 |
|
| 3 |
use warnings; |
| 4 |
use strict; |
| 5 |
|
| 6 |
use Time::HiRes qw(time); |
| 7 |
use Data::Dump qw(dump); |
| 8 |
use File::Slurp; |
| 9 |
use Getopt::Long; |
| 10 |
use IO::Socket::INET; |
| 11 |
use Storable qw/freeze thaw/; |
| 12 |
|
| 13 |
|
| 14 |
my $path = '/data/isi/full.txt'; |
| 15 |
my $limit = 5000; |
| 16 |
my $offset = 0; |
| 17 |
my @views; |
| 18 |
my $listen; |
| 19 |
my @nodes; |
| 20 |
|
| 21 |
|
| 22 |
GetOptions( |
| 23 |
'path=s' => \$path, |
| 24 |
'offset=i' => \$offset, |
| 25 |
'limit=i' => \$limit, |
| 26 |
'view=s' => \@views, |
| 27 |
'listen|port=i' => \$listen, |
| 28 |
'connect=s' => \@nodes, |
| 29 |
) or die $!; |
| 30 |
|
| 31 |
my $t = time; |
| 32 |
|
| 33 |
|
| 34 |
our $prefix; |
| 35 |
BEGIN { |
| 36 |
$prefix = $0; |
| 37 |
if ( $prefix =~ s{^./}{} ) { |
| 38 |
chomp( my $pwd = `pwd` ); |
| 39 |
$prefix = "$pwd/$prefix"; |
| 40 |
} |
| 41 |
$prefix =~ s{^(.*)/srv/Sack/bin.+$}{$1}; |
| 42 |
warn "# prefix $prefix"; |
| 43 |
} |
| 44 |
|
| 45 |
|
| 46 |
use lib "$prefix/srv/webpac2/lib/"; |
| 47 |
use WebPAC::Input::ISI; |
| 48 |
my $input = WebPAC::Input::ISI->new( |
| 49 |
path => "$prefix/$path", |
| 50 |
offset => $offset, |
| 51 |
limit => $limit, |
| 52 |
); |
| 53 |
|
| 54 |
|
| 55 |
sub report { |
| 56 |
my $description = shift; |
| 57 |
my $dt = time - $t; |
| 58 |
printf "%s in %1.4fs %.2f/s\n", $description, $dt, $input->size / $dt; |
| 59 |
$t = time; |
| 60 |
} |
| 61 |
|
| 62 |
|
| 63 |
report $input->size . ' records loaded'; |
| 64 |
|
| 65 |
mkdir 'out' unless -e 'out'; |
| 66 |
|
| 67 |
our $out; |
| 68 |
|
| 69 |
our $cache; |
| 70 |
|
| 71 |
our $connected; |
| 72 |
|
| 73 |
sub send_nodes { |
| 74 |
my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument! |
| 75 |
my $header = length($content); |
| 76 |
$header .= ' ' . join(' ', @_) if @_; |
| 77 |
|
| 78 |
foreach my $node ( @nodes ) { |
| 79 |
|
| 80 |
my $sock = IO::Socket::INET->new( |
| 81 |
PeerAddr => $node, |
| 82 |
Proto => 'tcp', |
| 83 |
); |
| 84 |
|
| 85 |
if ( ! $sock ) { |
| 86 |
warn "can't connect to $node - $!"; # FIXME die? |
| 87 |
next; |
| 88 |
} |
| 89 |
|
| 90 |
print ">>>> $node $header\n"; |
| 91 |
print $sock "$header\n$content" || warn "can't send $header to $node: $!"; |
| 92 |
|
| 93 |
$connected->{$node} = $sock; |
| 94 |
} |
| 95 |
} |
| 96 |
|
| 97 |
sub get_node { |
| 98 |
my $node = shift; |
| 99 |
|
| 100 |
my $sock = $connected->{$node}; |
| 101 |
if ( ! $sock ) { |
| 102 |
warn "ERROR: lost connection to $node"; |
| 103 |
delete $connected->{$node}; |
| 104 |
return; |
| 105 |
} |
| 106 |
chomp( my $size = <$sock> ); |
| 107 |
warn "<<<< $node $size bytes\n"; |
| 108 |
my $data; |
| 109 |
read $sock, $data, $size; |
| 110 |
return $data; |
| 111 |
} |
| 112 |
|
| 113 |
sub send_sock { |
| 114 |
my ( $sock, $data ) = @_; |
| 115 |
my $size = length $data; |
| 116 |
warn ">>>> ", $sock->peerhost, " $size bytes"; |
| 117 |
print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost; |
| 118 |
} |
| 119 |
|
| 120 |
sub merge_out { |
| 121 |
my $new = shift; |
| 122 |
|
| 123 |
warn "## merge $new\n"; |
| 124 |
|
| 125 |
foreach my $k1 ( keys %$new ) { |
| 126 |
|
| 127 |
foreach my $k2 ( keys %{ $new->{$k1} } ) { |
| 128 |
|
| 129 |
my $n = $new->{$k1}->{$k2}; |
| 130 |
my $ref = ref $out->{$k1}->{$k2}; |
| 131 |
|
| 132 |
if ( ! defined $out->{$k1}->{$k2} ) { |
| 133 |
$out->{$k1}->{$k2} = $n; |
| 134 |
} elsif ( $k1 =~ m{\+} ) { |
| 135 |
warn "# agregate $k1 $k2"; |
| 136 |
$out->{$k1}->{$k2} += $n; |
| 137 |
} elsif ( $ref eq 'ARRAY' ) { |
| 138 |
push @{ $out->{$k1}->{$k2} }, $n; |
| 139 |
} elsif ( $ref eq '' ) { |
| 140 |
$out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ]; |
| 141 |
} else { |
| 142 |
die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2}); |
| 143 |
} |
| 144 |
} |
| 145 |
} |
| 146 |
|
| 147 |
warn "## merge out ", dump $out; |
| 148 |
} |
| 149 |
|
| 150 |
sub run_code { |
| 151 |
my ( $view, $code ) = @_; |
| 152 |
|
| 153 |
warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n"; |
| 154 |
|
| 155 |
send_nodes view => $view => $code; |
| 156 |
|
| 157 |
undef $out; |
| 158 |
|
| 159 |
my $affected = 0; |
| 160 |
$t = time; |
| 161 |
|
| 162 |
foreach my $pos ( $offset + 1 .. $offset + $input->size ) { |
| 163 |
my $rec = $cache->{$pos} ||= $input->fetch_rec( $pos ); |
| 164 |
if ( ! $rec ) { |
| 165 |
warn "END at $pos"; |
| 166 |
last; |
| 167 |
} |
| 168 |
|
| 169 |
eval "$code"; |
| 170 |
if ( $@ ) { |
| 171 |
warn "ERROR [$pos] $@\n"; |
| 172 |
} else { |
| 173 |
$affected++; |
| 174 |
} |
| 175 |
}; |
| 176 |
|
| 177 |
report "$affected affected records $view"; |
| 178 |
|
| 179 |
warn "WARN no \$out defined!" unless defined $out; |
| 180 |
|
| 181 |
if ( $connected ) { |
| 182 |
warn "# get results from ", join(' ', keys %$connected ); |
| 183 |
merge_out( thaw( get_node( $_ ) ) ) foreach keys %$connected; |
| 184 |
} |
| 185 |
} |
| 186 |
|
| 187 |
sub run_views { |
| 188 |
@views = sort glob 'views/*.pl' unless @views; |
| 189 |
warn "# views ", dump @views; |
| 190 |
|
| 191 |
foreach my $view ( @views ) { |
| 192 |
|
| 193 |
next if system("perl -c $view") != 0; |
| 194 |
|
| 195 |
my $code = read_file $view; |
| 196 |
|
| 197 |
run_code $view => $code; |
| 198 |
|
| 199 |
if ( defined $out ) { |
| 200 |
my $dump = dump $out; |
| 201 |
my $len = length $dump; |
| 202 |
|
| 203 |
my $path = $view; |
| 204 |
$path =~ s{views?/}{out/} || die "no view in $view"; |
| 205 |
$path =~ s{\.pl}{}; |
| 206 |
|
| 207 |
print "OUT $view $offset/$limit $len bytes $path" |
| 208 |
, ( $len < 10000 ? " \$out = $dump" : ' SAVED ONLY' ) |
| 209 |
, "\n" |
| 210 |
; |
| 211 |
|
| 212 |
unlink "$path.last" if -e "$path.last"; |
| 213 |
rename $path, "$path.last"; |
| 214 |
write_file $path, $dump; |
| 215 |
report "SAVE $path"; |
| 216 |
} |
| 217 |
|
| 218 |
} |
| 219 |
|
| 220 |
} |
| 221 |
|
| 222 |
if ( $listen ) { |
| 223 |
my $sock = IO::Socket::INET->new( |
| 224 |
Listen => SOMAXCONN, |
| 225 |
# LocalAddr => '0.0.0.0', |
| 226 |
LocalPort => $listen, |
| 227 |
Proto => 'tcp', |
| 228 |
Reuse => 1, |
| 229 |
) or die $!; |
| 230 |
|
| 231 |
while (1) { |
| 232 |
|
| 233 |
warn "NODE listen on $listen\n"; |
| 234 |
|
| 235 |
my $client = $sock->accept(); |
| 236 |
|
| 237 |
warn "<<<< $listen connect from ", $client->peerhost, $/; |
| 238 |
|
| 239 |
my @header = split(/\s/, <$client>); |
| 240 |
warn "# header ",dump @header; |
| 241 |
|
| 242 |
my $size = shift @header; |
| 243 |
|
| 244 |
my $content; |
| 245 |
read $client, $content, $size; |
| 246 |
|
| 247 |
if ( $header[0] eq 'view' ) { |
| 248 |
run_code $header[1] => $content; |
| 249 |
send_sock $client => freeze $out; |
| 250 |
} elsif ( $header[0] eq 'info' ) { |
| 251 |
my $info = "$listen\t$offset\t$limit\t$path"; |
| 252 |
warn "info $info\n"; |
| 253 |
send_sock $client => $info; |
| 254 |
} elsif ( $header[0] eq 'exit' ) { |
| 255 |
warn "exit $listen"; |
| 256 |
exit; |
| 257 |
} else { |
| 258 |
warn "WARN $listen unknown"; |
| 259 |
} |
| 260 |
|
| 261 |
} |
| 262 |
} |
| 263 |
|
| 264 |
run_views; |
| 265 |
|
| 266 |
while ( 1 ) { |
| 267 |
|
| 268 |
print "sack> "; |
| 269 |
my $cmd = <STDIN>; |
| 270 |
|
| 271 |
if ( $cmd =~ m{^(vi?|\\e|o(?:ut)?)}i ) { |
| 272 |
system "vi out/*"; |
| 273 |
} elsif ( $cmd =~ m{^i(nfo)?}i ) { |
| 274 |
print "# nodes: ", join(' ',@nodes), $/; |
| 275 |
|
| 276 |
my @info = ( |
| 277 |
"node\toffset\tlimit\tpath", |
| 278 |
"----\t------\t-----\t----", |
| 279 |
"here\t$offset\t$limit\t$path", |
| 280 |
); |
| 281 |
|
| 282 |
send_nodes 'info'; |
| 283 |
push @info, get_node $_ foreach @nodes; |
| 284 |
|
| 285 |
print "$_\n" foreach @info; |
| 286 |
|
| 287 |
} elsif ( $cmd =~ m{^(q(uit)|e(xit))}i ) { |
| 288 |
warn "# exit"; |
| 289 |
send_nodes 'exit'; |
| 290 |
exit; |
| 291 |
} else { |
| 292 |
run_views; |
| 293 |
} |
| 294 |
|
| 295 |
} |
| 296 |
|
Parent Directory
| 
Revision Log