| 1 |
#!/usr/bin/perl |
| 2 |
|
| 3 |
use warnings; |
| 4 |
use strict; |
| 5 |
|
| 6 |
our $VERSION = '0.05'; |
| 7 |
|
| 8 |
use Time::HiRes qw(time); |
| 9 |
use Data::Dump qw(dump); |
| 10 |
use File::Slurp; |
| 11 |
use Getopt::Long; |
| 12 |
use IO::Socket::INET; |
| 13 |
use Storable qw/freeze thaw store/; |
| 14 |
|
| 15 |
|
| 16 |
my $debug = 0; |
| 17 |
my $path = '/data/isi/full.txt'; |
| 18 |
my $limit = 5000; |
| 19 |
my $offset = 0; |
| 20 |
my @views; |
| 21 |
my $port = 0; # interactive |
| 22 |
my @nodes; |
| 23 |
|
| 24 |
|
| 25 |
GetOptions( |
| 26 |
'path=s' => \$path, |
| 27 |
'offset=i' => \$offset, |
| 28 |
'limit=i' => \$limit, |
| 29 |
'view=s' => \@views, |
| 30 |
'port|listen=i' => \$port, |
| 31 |
'node|connect=i' => \@nodes, |
| 32 |
'debug!' => \$debug, |
| 33 |
) or die $!; |
| 34 |
|
| 35 |
my $t = time; |
| 36 |
|
| 37 |
|
| 38 |
sub send_nodes; |
| 39 |
|
| 40 |
our $prefix; |
| 41 |
sub BEGIN { |
| 42 |
$prefix = $0; |
| 43 |
if ( $prefix !~ m{^/} ) { |
| 44 |
chomp( my $pwd = `pwd` ); |
| 45 |
$prefix = "$pwd/$prefix"; |
| 46 |
} |
| 47 |
$prefix =~ s{^(.*)/srv/Sack/.+$}{$1}; |
| 48 |
warn "# prefix $prefix"; |
| 49 |
|
| 50 |
$SIG{INT} = sub { |
| 51 |
my $signame = shift; |
| 52 |
send_nodes 'exit'; |
| 53 |
#clean if $clean; # FIXME |
| 54 |
die "SIG$signame"; |
| 55 |
}; |
| 56 |
} |
| 57 |
|
| 58 |
use lib "$prefix/srv/Sack/lib/"; |
| 59 |
use Sack::Digest; |
| 60 |
our $digest = Sack::Digest->new( port => $port, clean => 1 ); |
| 61 |
sub digest { $digest->to_int($_[0]) } |
| 62 |
|
| 63 |
use lib "$prefix/srv/webpac2/lib/"; |
| 64 |
use WebPAC::Input::ISI; |
| 65 |
|
| 66 |
$WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields |
| 67 |
|
| 68 |
my $input = WebPAC::Input::ISI->new( |
| 69 |
path => "$prefix/$path", |
| 70 |
offset => $offset, |
| 71 |
limit => $limit, |
| 72 |
); |
| 73 |
|
| 74 |
our $num_records = $input->size; |
| 75 |
our @reports; |
| 76 |
|
| 77 |
sub report { |
| 78 |
my $description = join(' ',@_); |
| 79 |
my $dt = time - $t; |
| 80 |
my $report = [ $description, $dt, $input->size / $dt ]; |
| 81 |
printf "[$port] %s in %1.4fs %.2f/s\n", @$report; |
| 82 |
push @reports, $report; |
| 83 |
$t = time; |
| 84 |
} |
| 85 |
|
| 86 |
sub show_report { |
| 87 |
"\n" . join( "\n", map { sprintf "%8.4fs %10.2f/s %s", $_->[1], $_->[2], $_->[0] } @reports ) . "\n"; |
| 88 |
} |
| 89 |
|
| 90 |
report $input->size , 'records loaded'; |
| 91 |
|
| 92 |
mkdir 'out' unless -e 'out'; |
| 93 |
|
| 94 |
our $out; |
| 95 |
|
| 96 |
our $cache; |
| 97 |
|
| 98 |
our $connected; |
| 99 |
|
| 100 |
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 |
sub send_nodes { |
| 115 |
my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument! |
| 116 |
my $header = defined $content ? length($content) : 0; |
| 117 |
$header .= ' ' . join(' ', @_) if @_; |
| 118 |
|
| 119 |
warn "# send_nodes ", dump(@_), " to ", dump @nodes; |
| 120 |
|
| 121 |
foreach my $node ( @nodes ) { |
| 122 |
|
| 123 |
my $sock = node_sock($node) || next; |
| 124 |
|
| 125 |
warn "[$port] >>>> [$node] $header\n"; |
| 126 |
print $sock "$header\n$content" || warn "can't send $header to $node: $!"; |
| 127 |
|
| 128 |
$connected->{$node} = $sock; |
| 129 |
} |
| 130 |
} |
| 131 |
|
| 132 |
sub get_node { |
| 133 |
my $node = shift; |
| 134 |
|
| 135 |
my $sock = $connected->{$node}; |
| 136 |
if ( ! $sock || ! $sock->connected ) { |
| 137 |
warn "[$port] no connection to $node"; |
| 138 |
delete $connected->{$node}; |
| 139 |
return; |
| 140 |
} |
| 141 |
chomp( my $size = <$sock> ); |
| 142 |
warn "[$port] <<<< [$node] $size bytes\n" if $debug || $size > 1024; |
| 143 |
my $data; |
| 144 |
read $sock, $data, $size; |
| 145 |
return $data; |
| 146 |
} |
| 147 |
|
| 148 |
sub send_sock { |
| 149 |
my ( $sock, $data ) = @_; |
| 150 |
my $size = length $data; |
| 151 |
warn "[$port] >>>> $size bytes\n" if $debug || $size > 1024; |
| 152 |
print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost; |
| 153 |
} |
| 154 |
|
| 155 |
sub pull_node_file { |
| 156 |
my ( $node, $file ) = @_; |
| 157 |
|
| 158 |
my $path = "/dev/shm/sack.$node.$file"; |
| 159 |
return if -e $path; # FIXME |
| 160 |
|
| 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 |
sub merge_out { |
| 180 |
my ( $from_node, $new ) = @_; |
| 181 |
|
| 182 |
my $t_merge = time(); |
| 183 |
|
| 184 |
pull_node_file $from_node => 'nr_md5'; |
| 185 |
pull_node_file $from_node => 'md5'; |
| 186 |
|
| 187 |
my $remote_digest = Sack::Digest->new( port => $from_node ); |
| 188 |
my ( $local, $remote ) = ( 0, 0 ); |
| 189 |
|
| 190 |
my $tick = 0; |
| 191 |
print STDERR "[$port] merge [$from_node]"; |
| 192 |
|
| 193 |
my $missing; |
| 194 |
|
| 195 |
foreach my $k1 ( keys %$new ) { |
| 196 |
|
| 197 |
foreach my $k2 ( keys %{ $new->{$k1} } ) { |
| 198 |
|
| 199 |
my $n = delete $new->{$k1}->{$k2}; |
| 200 |
|
| 201 |
if ( $k1 =~ m{#} ) { |
| 202 |
die "ASSERT $k1 $k2" unless $k2 =~ m{^\d+$}; |
| 203 |
#warn "XXX $k1 $k2"; |
| 204 |
my $md5 = $remote_digest->{nr_md5}->[$k2]; |
| 205 |
|
| 206 |
if ( ! $md5 ) { |
| 207 |
$missing->{nr_md5}->{$from_node}++; # FIXME die? |
| 208 |
next; |
| 209 |
} |
| 210 |
|
| 211 |
if ( my $local_k2 = $digest->{md5_nr}->{$md5} ) { |
| 212 |
$k2 = $local_k2; |
| 213 |
$local++; |
| 214 |
} elsif ( my $full = $remote_digest->{md5}->{$md5} ) { |
| 215 |
$k2 = $digest->to_int( $remote_digest->{md5}->{$md5} ); |
| 216 |
$remote++; |
| 217 |
} else { |
| 218 |
$missing->{md5}->{$from_node}++; |
| 219 |
} |
| 220 |
} |
| 221 |
|
| 222 |
my $ref = ref $out->{$k1}->{$k2}; |
| 223 |
#warn "XXXX $k1 $k2 $ref"; |
| 224 |
if ( ! defined $out->{$k1}->{$k2} ) { |
| 225 |
$out->{$k1}->{$k2} = $n; |
| 226 |
} elsif ( $k1 =~ m{\+} ) { |
| 227 |
# warn "## agregate $k1 $k2"; |
| 228 |
$out->{$k1}->{$k2} += $n; |
| 229 |
} 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 |
} 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 |
|
| 241 |
if ( $tick++ % 1000 == 0 ) { |
| 242 |
print STDERR "."; |
| 243 |
} elsif ( $tick % 10000 == 0 ) { |
| 244 |
print STDERR $tick; |
| 245 |
} |
| 246 |
} |
| 247 |
} |
| 248 |
|
| 249 |
$t_merge = time - $t_merge; |
| 250 |
printf STDERR "%d in %.4fs %.2f/s local %.1f%% %d/%d\n", $tick, $t_merge, $tick / $t_merge, $local * 100 / $tick, $local, $remote; |
| 251 |
push @reports, [ "$tick merged $from_node", $t_merge, $tick / $t_merge ]; |
| 252 |
|
| 253 |
warn "[$port] missing ", dump $missing if $missing; |
| 254 |
|
| 255 |
warn "## merge out ", dump $out if $debug; |
| 256 |
} |
| 257 |
|
| 258 |
sub run_code { |
| 259 |
my ( $view, $code ) = @_; |
| 260 |
|
| 261 |
warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug; |
| 262 |
|
| 263 |
send_nodes view => $view => $code; |
| 264 |
|
| 265 |
undef $out; |
| 266 |
|
| 267 |
my $affected = 0; |
| 268 |
$t = time; |
| 269 |
|
| 270 |
foreach my $pos ( $offset + 1 .. $offset + $input->size ) { |
| 271 |
my $rec = $cache->{$pos} ||= $input->fetch_rec( $pos ); |
| 272 |
if ( ! $rec ) { |
| 273 |
print STDERR "END @ $pos"; |
| 274 |
last; |
| 275 |
} |
| 276 |
|
| 277 |
eval "$code"; |
| 278 |
if ( $@ ) { |
| 279 |
warn "ABORT $pos $@\n"; |
| 280 |
last; |
| 281 |
} else { |
| 282 |
$affected++; |
| 283 |
} |
| 284 |
|
| 285 |
$pos % 10000 == 0 ? print STDERR $pos : |
| 286 |
$pos % 1000 == 0 ? print STDERR "." : 0 ; |
| 287 |
}; |
| 288 |
|
| 289 |
report "$affected affected $view"; |
| 290 |
|
| 291 |
warn "WARN no \$out defined!" unless defined $out; |
| 292 |
|
| 293 |
$digest->sync; |
| 294 |
|
| 295 |
if ( $connected ) { |
| 296 |
foreach my $node ( keys %$connected ) { |
| 297 |
warn "[$port] get_node [$node]\n"; |
| 298 |
my $o = get_node $node; |
| 299 |
next unless $o; |
| 300 |
my $s = length $o; |
| 301 |
$o = thaw $o; |
| 302 |
warn "[$port] got $s bytes from [$node]\n"; |
| 303 |
merge_out $node => $o; |
| 304 |
} |
| 305 |
} |
| 306 |
} |
| 307 |
|
| 308 |
sub run_views { |
| 309 |
@views = sort glob 'views/*.pl' unless @views; |
| 310 |
warn "# views ", dump @views; |
| 311 |
|
| 312 |
foreach my $view ( @views ) { |
| 313 |
|
| 314 |
next if system("perl -c $view") != 0; |
| 315 |
|
| 316 |
my $code = read_file $view; |
| 317 |
|
| 318 |
run_code $view => $code; |
| 319 |
|
| 320 |
if ( defined $out ) { |
| 321 |
|
| 322 |
my $path = $view; |
| 323 |
$path =~ s{views?/}{out/} || die "no view in $view"; |
| 324 |
$path =~ s{\.pl}{.storable}; |
| 325 |
|
| 326 |
unlink "$path.last" if -e "$path.last"; |
| 327 |
rename $path, "$path.last"; |
| 328 |
|
| 329 |
store $out => $path; |
| 330 |
report "save $path", -s $path, "bytes"; |
| 331 |
|
| 332 |
if ( -s $path < 4096 ) { |
| 333 |
print '$out = ', dump $digest->undigest_out($out); |
| 334 |
} |
| 335 |
} |
| 336 |
|
| 337 |
} |
| 338 |
|
| 339 |
} |
| 340 |
|
| 341 |
|
| 342 |
sub info_tabs { |
| 343 |
"$port\t$offset\t$limit\t$num_records\t$path\t" |
| 344 |
. join("\t", map { |
| 345 |
my $b = $_; |
| 346 |
$b =~ s{^.+\.$port\.([^/]+)$}{$1}; |
| 347 |
"$b " . -s $_ |
| 348 |
} glob "/dev/shm/sack.$port.*" ); |
| 349 |
} |
| 350 |
|
| 351 |
|
| 352 |
if ( $port ) { |
| 353 |
|
| 354 |
my $pid_path = "/tmp/sack.$port.pid"; |
| 355 |
if ( -e $pid_path ) { |
| 356 |
my $pid = read_file $pid_path; |
| 357 |
kill 9, $pid && warn "[$port] kill old $pid"; |
| 358 |
} |
| 359 |
write_file $pid_path, $$; |
| 360 |
|
| 361 |
|
| 362 |
my $sock = IO::Socket::INET->new( |
| 363 |
Listen => SOMAXCONN, |
| 364 |
LocalAddr => '127.0.0.1', |
| 365 |
LocalPort => $port, |
| 366 |
Proto => 'tcp', |
| 367 |
Reuse => 1, |
| 368 |
) or die $!; |
| 369 |
|
| 370 |
while (1) { |
| 371 |
|
| 372 |
warn "[$port] accept path: $path offset: $offset limit: $limit #recs: $num_records\n"; |
| 373 |
|
| 374 |
my $client = $sock->accept(); |
| 375 |
|
| 376 |
warn "[$port] <<<< connect from ", $client->peerhost, $/; |
| 377 |
|
| 378 |
my @header = split(/\s/, <$client>); |
| 379 |
warn "[$port] <<<< header ",dump(@header),$/; |
| 380 |
|
| 381 |
my $size = shift @header; |
| 382 |
|
| 383 |
my $content; |
| 384 |
read $client, $content, $size; |
| 385 |
|
| 386 |
if ( $header[0] eq 'view' ) { |
| 387 |
run_code $header[1] => $content; |
| 388 |
send_sock $client => freeze $out; |
| 389 |
} elsif ( $header[0] eq 'info' ) { |
| 390 |
my $info = info_tabs; |
| 391 |
warn "[$port] info $info\n"; |
| 392 |
$info .= "\n" . show_report if $content =~ m{r}i; |
| 393 |
send_sock $client => $info; |
| 394 |
} elsif ( $header[0] eq 'exit' ) { |
| 395 |
warn "[$port] exit"; |
| 396 |
exit; |
| 397 |
} elsif ( $header[0] eq 'file' ) { |
| 398 |
$digest->close; |
| 399 |
my $path = "/dev/shm/sack.$port.$header[1]"; |
| 400 |
my $size = -s $path; |
| 401 |
warn "[$port] >>>> file $path $size bytes\n"; |
| 402 |
print $client "$size\n"; |
| 403 |
my $block = 4096; |
| 404 |
my $buff = ' ' x $block; |
| 405 |
open(my $fh, '<', $path) || die "can't open $path"; |
| 406 |
while ( read $fh, $buff, $block ) { |
| 407 |
print $client $buff; |
| 408 |
} |
| 409 |
$digest->open; |
| 410 |
} else { |
| 411 |
warn "[$port] UNKNOWN $header[0]"; |
| 412 |
} |
| 413 |
|
| 414 |
$client->close; |
| 415 |
|
| 416 |
} |
| 417 |
} |
| 418 |
|
| 419 |
sub info { |
| 420 |
my $detail = shift || ''; |
| 421 |
|
| 422 |
send_nodes 'info' => $detail; |
| 423 |
|
| 424 |
my @info = ( |
| 425 |
"port\toffset\tlimit\t#recs\tpath", |
| 426 |
"----\t------\t-----\t-----\t----", |
| 427 |
info_tabs, |
| 428 |
); |
| 429 |
|
| 430 |
push @info, show_report if $detail =~ m{r}i; |
| 431 |
|
| 432 |
push @info, get_node $_ foreach @nodes; |
| 433 |
|
| 434 |
print "[$port] INFO", $detail ? " $detail" : '', " \n" |
| 435 |
, join("\n", @info) |
| 436 |
, "\n" ; |
| 437 |
|
| 438 |
return @info; |
| 439 |
} |
| 440 |
|
| 441 |
info; |
| 442 |
run_views; |
| 443 |
|
| 444 |
while ( 1 ) { |
| 445 |
|
| 446 |
print "sack> "; |
| 447 |
chomp( my $cmd = <STDIN> ); |
| 448 |
|
| 449 |
if ( $cmd =~ m{^(h|\?)} ) { |
| 450 |
print << "__HELP__" |
| 451 |
Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit |
| 452 |
|
| 453 |
View Run run views |
| 454 |
VI \\e Output show output of last run |
| 455 |
Info [\$VERSION] instrospect |
| 456 |
Quit EXit shutdown |
| 457 |
|
| 458 |
__HELP__ |
| 459 |
} elsif ( $cmd =~ m{^(vi|\\e|o)}i ) { |
| 460 |
#system "vi out/*"; |
| 461 |
$digest->sync; |
| 462 |
system "bin/storableedit.pl", (glob('out/*.storable'))[0]; |
| 463 |
} elsif ( $cmd =~ m{^i(?:nfo)?\s?(\S+)?$}i ) { |
| 464 |
info $1; |
| 465 |
} elsif ( $cmd =~ m{^(q|e|x)}i ) { |
| 466 |
warn "# exit"; |
| 467 |
send_nodes 'exit'; |
| 468 |
exit; |
| 469 |
} elsif ( $cmd =~ m{^(v|r)}i ) { |
| 470 |
run_views; |
| 471 |
} elsif ( $cmd =~ m{^n(ode)?\s*(\d+)}i ) { |
| 472 |
push @nodes, $2; |
| 473 |
info; |
| 474 |
} elsif ( $cmd ) { |
| 475 |
warn "UNKNOWN ", dump $cmd; |
| 476 |
} |
| 477 |
|
| 478 |
} |
| 479 |
|
Parent Directory
| 
Revision Log