trunk/bin/sack.pl

#!/usr/bin/perl

use warnings;
use strict;

our $VERSION = '0.04';

use Time::HiRes qw(time);
use Data::Dump qw(dump);
use File::Slurp;
use Getopt::Long;
use IO::Socket::INET;
use Storable qw/freeze thaw store/;


my $debug  = 0;
my $path   = '/data/isi/full.txt';
my $limit  = 5000;
my $offset = 0; 
my @views;
my $port   = 0; # interactive
my @nodes;


GetOptions(
        'path=s'   => \$path,
        'offset=i' => \$offset,
        'limit=i'  => \$limit,
        'view=s'   => \@views,
        'listen|port=i' => \$port,
        'connect=s'   => \@nodes,
        'debug!'   => \$debug,
) or die $!;

my $t = time;


sub send_nodes;

our $prefix;
sub BEGIN {
        $prefix = $0;
        if ( $prefix !~ m{^/} ) {
                chomp( my $pwd = `pwd` );
                $prefix = "$pwd/$prefix";
        }
        $prefix =~ s{^(.*)/srv/Sack/.+$}{$1};
        warn "# prefix $prefix";

        $SIG{INT} = sub {
                my $signame = shift;
                send_nodes 'exit';
                #clean if $clean;       # FIXME
                die "SIG$signame";
        };
}

use lib "$prefix/srv/Sack/lib/";
use Sack::Digest;
our $digest = Sack::Digest->new( port => $port, clean => 1 );
sub digest { $digest->to_int($_[0]) }

use lib "$prefix/srv/webpac2/lib/";
use WebPAC::Input::ISI;

$WebPAC::Input::ISI::subfields = undef; # disable parsing of subfields

my $input = WebPAC::Input::ISI->new(
        path   => "$prefix/$path",
        offset => $offset,
        limit  => $limit,
);

our $num_records = $input->size;

sub report {
        my $description = join(' ',@_);
        my $dt = time - $t;
        printf "%s in %1.4fs %.2f/s\n", $description, $dt, $input->size / $dt;
        $t = time;
}


report $input->size . ' records loaded';

mkdir 'out' unless -e 'out';

our $out;

our $cache;

our $connected;

sub send_nodes {
        my $content = $#_ > 0 ? pop @_ : '';    # no content with just one argument!
        my $header = defined $content ? length($content) : 0;
        $header .= ' ' . join(' ', @_) if @_;

        foreach my $node ( @nodes ) {

                my $sock = IO::Socket::INET->new(
                        PeerAddr => '127.0.0.1',
                        PeerPort => $node,
                        Proto    => 'tcp',
                );

                if ( ! $sock ) {
                        warn "can't connect to $node - $!"; # FIXME die?
                        next;
                }

                warn "[$port] >>>> $node $header\n";
                print $sock "$header\n$content" || warn "can't send $header to $node: $!";

                $connected->{$node} = $sock;
        }
}

sub get_node {
        my $node = shift;

        my $sock = $connected->{$node};
        if ( ! $sock ) {
                warn "[$port] ERROR lost connection to $node";
                delete $connected->{$node};
                return;
        }
        chomp( my $size = <$sock> );
        warn "[$port] <<<< $node $size bytes\n" if $debug || $size > 1024;
        my $data;
        read $sock, $data, $size;
        return $data;
}

sub send_sock {
        my ( $sock, $data ) = @_;
        my $size   = length $data;
        warn "[$port] >>>> $size bytes\n" if $debug || $size > 1024;
        print $sock "$size\n$data" || warn "can't send $size bytes to ", $sock->peerhost;
}

sub merge_out {
        my ( $from_node, $new ) = @_;

        my $remote_digest = Sack::Digest->new( port => $from_node );
        my ( $local, $remote ) = ( 0, 0 );

        my $tick = 0;
        print STDERR "[$port] merge $from_node";

        foreach my $k1 ( keys %$new ) {

                foreach my $k2 ( keys %{ $new->{$k1} } ) {

                        my $n   = delete $new->{$k1}->{$k2};

                        if ( $k1 =~ m{#} ) {
                                die "ASSERT $k1 $k2" unless $k2 =~ m{^\d+$};
#warn "XXX $k1 $k2";
                                my $md5 = $remote_digest->{nr_md5}->[$k2] || warn "[$port] no2md5 $n not found in $from_node\n";
                                if ( my $local_k2 = $digest->{md5_nr}->{$md5} ) {
                                        $k2 = $local_k2;
                                        $local++;
                                } else {
                                        $k2 = $digest->to_int( $remote_digest->{md5}->{$md5} );
                                        $remote++;
                                }
                        }

                        my $ref = ref    $out->{$k1}->{$k2};
#warn "XXXX $k1 $k2 $ref";
                        if ( ! defined $out->{$k1}->{$k2} ) {
                                $out->{$k1}->{$k2} = $n;
                        } elsif ( $k1 =~ m{\+} ) {
#                               warn "## agregate $k1 $k2";
                                $out->{$k1}->{$k2} += $n;
                        } elsif ( $ref  eq 'ARRAY' ) {
                                if ( ref $n eq 'ARRAY' ) {
                                        push @{ $out->{$k1}->{$k2} }, $_ foreach @$n;
                                } else {
                                        push @{ $out->{$k1}->{$k2} }, $n;
                                }
                        } elsif ( $ref eq '' ) {
                                $out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ];
                        } else {
                                die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2});
                        }

                        if ( $tick++ % 1000 == 0 ) {
                                print STDERR ".";
                        } elsif ( $tick % 10000 == 0 ) {
                                print STDERR $tick;
                        }
                }
        }

        print STDERR "$tick\n";

        warn "[$port] merge local $local remote $remote from $from_node\n";
        warn "## merge out ", dump $out if $debug;
}

sub run_code {
        my ( $view, $code ) = @_;

        warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug;

        send_nodes view => $view => $code;

        undef $out;

        my $affected = 0;
        $t = time;

        foreach my $pos ( $offset + 1 .. $offset + $input->size ) {
                my $rec = $cache->{$pos} ||= $input->fetch_rec( $pos );
                if ( ! $rec ) {
                        print STDERR "END @ $pos";
                        last;
                }

                eval "$code";
                if ( $@ ) {
                        warn "ABORT $pos $@\n";
                        last;
                } else {
                        $affected++;
                }

                $pos % 10000 == 0 ? print STDERR $pos :
                $pos % 1000  == 0 ? print STDERR "."  : 0 ;
        };

        report "\n[$port] RECS $affected $view";

        warn "WARN no \$out defined!" unless defined $out;

        $digest->sync;

        if ( $connected ) {
                foreach my $node ( keys %$connected ) {
                        warn "[$port] get_node $node\n";
                        my $o = get_node $node;
                        my $s = length $o;
                        $o = thaw $o;
                        warn "[$port] merge $node $s bytes\n";
                        merge_out $node => $o;
                }
        }
}

sub run_views {
        @views = sort glob 'views/*.pl' unless @views;
        warn "# views ", dump @views;

        foreach my $view ( @views ) {

                next if system("perl -c $view") != 0;

                my $code = read_file $view;

                run_code $view => $code;

                if ( defined $out ) {

                        my $path = $view;
                        $path =~ s{views?/}{out/} || die "no view in $view";
                        $path =~ s{\.pl}{.storable};

                        unlink "$path.last" if -e "$path.last";
                        rename $path, "$path.last";

                        store $out => $path;
                        report "[$port] SAVE $path $offset-$limit", -s $path, "bytes";

                        if ( -s $path < 4096 ) {
                                print '$out = ', dump $digest->undigest_out($out);
                        }
                }

        }

}


sub info_tabs {
        "$port\t$offset\t$limit\t$num_records\t$path\t"
        . join("\t", map {
                my $b = $_;
                $b =~ s{^.+\.$port\.([^/]+)$}{$1};
                "$b " . -s $_
        } glob "/dev/shm/sack.$port.*" );
}


if ( $port ) {
        my $sock = IO::Socket::INET->new(
                Listen    => SOMAXCONN,
                LocalAddr => '127.0.0.1',
                LocalPort => $port,
                Proto     => 'tcp',
                Reuse     => 1,
        ) or die $!;

        while (1) {

                warn "[$port] READY path: $path offset: $offset limit: $limit #recs: $num_records\n";

                my $client = $sock->accept();

                warn "[$port] <<<< connect from ", $client->peerhost, $/;

                my @header = split(/\s/, <$client>);
                warn "[$port] <<<< header ",dump(@header),$/;

                my $size = shift @header;

                my $content;
                read $client, $content, $size;

                if ( $header[0] eq 'view' ) {
                        run_code $header[1] => $content;
                        send_sock $client => freeze $out;
                } elsif ( $header[0] eq 'info' ) {
                        my $info = info_tabs;
                        warn "[$port] info $info\n";
                        send_sock $client => $info;
                } elsif ( $header[0] eq 'exit' ) {
                        warn "[$port] exit";
                        exit;
                } else {
                        warn "[$port] UNKNOWN $header[0]";
                }

        }
}

sub info {
        send_nodes 'info' => $2;

        my @info = (
                "port\toffset\tlimit\t#recs\tpath",
                "----\t------\t-----\t-----\t----",
                info_tabs,
        );

        push @info, get_node $_ foreach @nodes;

        print "[$port] INFO\n"
                , join("\n", @info)
                , "\n\n" ;

        return @info;
}

info;
run_views;

while ( 1 ) {

        print "sack> ";
        chomp( my $cmd = <STDIN> );

        if ( $cmd =~ m{^(h|\?)} ) {
                print << "__HELP__"
Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit

        View Run        run views
        VI \\e Output   show output of last run
        Info [\$VERSION]        instrospect
        Quit EXit       shutdown

__HELP__
        } elsif ( $cmd =~ m{^(vi|\\e|o)}i ) {
                #system "vi out/*";
                $digest->sync;
                system "bin/storableedit.pl", (glob('out/*.storable'))[0];
        } elsif ( $cmd =~ m{^i(?:nfo)?\s?(.+)?$}i ) {
                info;
        } elsif ( $cmd =~ m{^(q|e|x)}i ) {
                warn "# exit";
                send_nodes 'exit';
                exit;
        } elsif ( $cmd =~ m{^(v|r)}i ) {
                run_views;
        } elsif ( $cmd =~ m{^n(ode)?\s*(\d+)}i ) {
                push @nodes, $1;
                info;
        } elsif ( $cmd ) {
                warn "UNKNOWN ", dump $cmd;
        }

}

1	#!/usr/bin/perl
2
3	use warnings;
4	use strict;
5
6	our $VERSION = '0.04';
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	'listen\|port=i' => \$port,
31	'connect=s' => \@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
76	sub report {
77	my $description = join(' ',@_);
78	my $dt = time - $t;
79	printf "%s in %1.4fs %.2f/s\n", $description, $dt, $input->size / $dt;
80	$t = time;
81	}
82
83
84	report $input->size . ' records loaded';
85
86	mkdir 'out' unless -e 'out';
87
88	our $out;
89
90	our $cache;
91
92	our $connected;
93
94	sub send_nodes {
95	my $content = $#_ > 0 ? pop @_ : ''; # no content with just one argument!
96	my $header = defined $content ? length($content) : 0;
97	$header .= ' ' . join(' ', @_) if @_;
98
99	foreach my $node ( @nodes ) {
100
101	my $sock = IO::Socket::INET->new(
102	PeerAddr => '127.0.0.1',
103	PeerPort => $node,
104	Proto => 'tcp',
105	);
106
107	if ( ! $sock ) {
108	warn "can't connect to $node - $!"; # FIXME die?
109	next;
110	}
111
112	warn "[$port] >>>> $node $header\n";
113	print $sock "$header\n$content" \|\| warn "can't send $header to $node: $!";
114
115	$connected->{$node} = $sock;
116	}
117	}
118
119	sub get_node {
120	my $node = shift;
121
122	my $sock = $connected->{$node};
123	if ( ! $sock ) {
124	warn "[$port] ERROR lost connection to $node";
125	delete $connected->{$node};
126	return;
127	}
128	chomp( my $size = <$sock> );
129	warn "[$port] <<<< $node $size bytes\n" if $debug \|\| $size > 1024;
130	my $data;
131	read $sock, $data, $size;
132	return $data;
133	}
134
135	sub send_sock {
136	my ( $sock, $data ) = @_;
137	my $size = length $data;
138	warn "[$port] >>>> $size bytes\n" if $debug \|\| $size > 1024;
139	print $sock "$size\n$data" \|\| warn "can't send $size bytes to ", $sock->peerhost;
140	}
141
142	sub merge_out {
143	my ( $from_node, $new ) = @_;
144
145	my $remote_digest = Sack::Digest->new( port => $from_node );
146	my ( $local, $remote ) = ( 0, 0 );
147
148	my $tick = 0;
149	print STDERR "[$port] merge $from_node";
150
151	foreach my $k1 ( keys %$new ) {
152
153	foreach my $k2 ( keys %{ $new->{$k1} } ) {
154
155	my $n = delete $new->{$k1}->{$k2};
156
157	if ( $k1 =~ m{#} ) {
158	die "ASSERT $k1 $k2" unless $k2 =~ m{^\d+$};
159	#warn "XXX $k1 $k2";
160	my $md5 = $remote_digest->{nr_md5}->[$k2] \|\| warn "[$port] no2md5 $n not found in $from_node\n";
161	if ( my $local_k2 = $digest->{md5_nr}->{$md5} ) {
162	$k2 = $local_k2;
163	$local++;
164	} else {
165	$k2 = $digest->to_int( $remote_digest->{md5}->{$md5} );
166	$remote++;
167	}
168	}
169
170	my $ref = ref $out->{$k1}->{$k2};
171	#warn "XXXX $k1 $k2 $ref";
172	if ( ! defined $out->{$k1}->{$k2} ) {
173	$out->{$k1}->{$k2} = $n;
174	} elsif ( $k1 =~ m{\+} ) {
175	# warn "## agregate $k1 $k2";
176	$out->{$k1}->{$k2} += $n;
177	} elsif ( $ref eq 'ARRAY' ) {
178	if ( ref $n eq 'ARRAY' ) {
179	push @{ $out->{$k1}->{$k2} }, $_ foreach @$n;
180	} else {
181	push @{ $out->{$k1}->{$k2} }, $n;
182	}
183	} elsif ( $ref eq '' ) {
184	$out->{$k1}->{$k2} = [ $out->{$k1}->{$k2}, $n ];
185	} else {
186	die "can't merge $k2 [$ref] from ",dump($n), " into ", dump($out->{$k1}->{$k2});
187	}
188
189	if ( $tick++ % 1000 == 0 ) {
190	print STDERR ".";
191	} elsif ( $tick % 10000 == 0 ) {
192	print STDERR $tick;
193	}
194	}
195	}
196
197	print STDERR "$tick\n";
198
199	warn "[$port] merge local $local remote $remote from $from_node\n";
200	warn "## merge out ", dump $out if $debug;
201	}
202
203	sub run_code {
204	my ( $view, $code ) = @_;
205
206	warn "\n#### CODE $view START ####\n$code\n#### CODE $view END ####\n" if $debug;
207
208	send_nodes view => $view => $code;
209
210	undef $out;
211
212	my $affected = 0;
213	$t = time;
214
215	foreach my $pos ( $offset + 1 .. $offset + $input->size ) {
216	my $rec = $cache->{$pos} \|\|= $input->fetch_rec( $pos );
217	if ( ! $rec ) {
218	print STDERR "END @ $pos";
219	last;
220	}
221
222	eval "$code";
223	if ( $@ ) {
224	warn "ABORT $pos $@\n";
225	last;
226	} else {
227	$affected++;
228	}
229
230	$pos % 10000 == 0 ? print STDERR $pos :
231	$pos % 1000 == 0 ? print STDERR "." : 0 ;
232	};
233
234	report "\n[$port] RECS $affected $view";
235
236	warn "WARN no \$out defined!" unless defined $out;
237
238	$digest->sync;
239
240	if ( $connected ) {
241	foreach my $node ( keys %$connected ) {
242	warn "[$port] get_node $node\n";
243	my $o = get_node $node;
244	my $s = length $o;
245	$o = thaw $o;
246	warn "[$port] merge $node $s bytes\n";
247	merge_out $node => $o;
248	}
249	}
250	}
251
252	sub run_views {
253	@views = sort glob 'views/*.pl' unless @views;
254	warn "# views ", dump @views;
255
256	foreach my $view ( @views ) {
257
258	next if system("perl -c $view") != 0;
259
260	my $code = read_file $view;
261
262	run_code $view => $code;
263
264	if ( defined $out ) {
265
266	my $path = $view;
267	$path =~ s{views?/}{out/} \|\| die "no view in $view";
268	$path =~ s{\.pl}{.storable};
269
270	unlink "$path.last" if -e "$path.last";
271	rename $path, "$path.last";
272
273	store $out => $path;
274	report "[$port] SAVE $path $offset-$limit", -s $path, "bytes";
275
276	if ( -s $path < 4096 ) {
277	print '$out = ', dump $digest->undigest_out($out);
278	}
279	}
280
281	}
282
283	}
284
285
286	sub info_tabs {
287	"$port\t$offset\t$limit\t$num_records\t$path\t"
288	. join("\t", map {
289	my $b = $_;
290	$b =~ s{^.+\.$port\.([^/]+)$}{$1};
291	"$b " . -s $_
292	} glob "/dev/shm/sack.$port.*" );
293	}
294
295
296	if ( $port ) {
297	my $sock = IO::Socket::INET->new(
298	Listen => SOMAXCONN,
299	LocalAddr => '127.0.0.1',
300	LocalPort => $port,
301	Proto => 'tcp',
302	Reuse => 1,
303	) or die $!;
304
305	while (1) {
306
307	warn "[$port] READY path: $path offset: $offset limit: $limit #recs: $num_records\n";
308
309	my $client = $sock->accept();
310
311	warn "[$port] <<<< connect from ", $client->peerhost, $/;
312
313	my @header = split(/\s/, <$client>);
314	warn "[$port] <<<< header ",dump(@header),$/;
315
316	my $size = shift @header;
317
318	my $content;
319	read $client, $content, $size;
320
321	if ( $header[0] eq 'view' ) {
322	run_code $header[1] => $content;
323	send_sock $client => freeze $out;
324	} elsif ( $header[0] eq 'info' ) {
325	my $info = info_tabs;
326	warn "[$port] info $info\n";
327	send_sock $client => $info;
328	} elsif ( $header[0] eq 'exit' ) {
329	warn "[$port] exit";
330	exit;
331	} else {
332	warn "[$port] UNKNOWN $header[0]";
333	}
334
335	}
336	}
337
338	sub info {
339	send_nodes 'info' => $2;
340
341	my @info = (
342	"port\toffset\tlimit\t#recs\tpath",
343	"----\t------\t-----\t-----\t----",
344	info_tabs,
345	);
346
347	push @info, get_node $_ foreach @nodes;
348
349	print "[$port] INFO\n"
350	, join("\n", @info)
351	, "\n\n" ;
352
353	return @info;
354	}
355
356	info;
357	run_views;
358
359	while ( 1 ) {
360
361	print "sack> ";
362	chomp( my $cmd = <STDIN> );
363
364	if ( $cmd =~ m{^(h\|\?)} ) {
365	print << "__HELP__"
366	Sacks Lorry v$VERSION - path: $path offset: $offset limit: $limit
367
368	View Run run views
369	VI \\e Output show output of last run
370	Info [\$VERSION] instrospect
371	Quit EXit shutdown
372
373	__HELP__
374	} elsif ( $cmd =~ m{^(vi\|\\e\|o)}i ) {
375	#system "vi out/*";
376	$digest->sync;
377	system "bin/storableedit.pl", (glob('out/*.storable'))[0];
378	} elsif ( $cmd =~ m{^i(?:nfo)?\s?(.+)?$}i ) {
379	info;
380	} elsif ( $cmd =~ m{^(q\|e\|x)}i ) {
381	warn "# exit";
382	send_nodes 'exit';
383	exit;
384	} elsif ( $cmd =~ m{^(v\|r)}i ) {
385	run_views;
386	} elsif ( $cmd =~ m{^n(ode)?\s*(\d+)}i ) {
387	push @nodes, $1;
388	info;
389	} elsif ( $cmd ) {
390	warn "UNKNOWN ", dump $cmd;
391	}
392
393	}
394