trunk/bin/sack.pl

#!/usr/bin/perl

use warnings;
use strict;

our $VERSION = '0.01';

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/;


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


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

my $t = time;


sub send_nodes;

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

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


use lib "$prefix/srv/webpac2/lib/";
use WebPAC::Input::ISI;
my $input = WebPAC::Input::ISI->new(
        path   => "$prefix/$path",
        offset => $offset,
        limit  => $limit,
);


sub report {
        my $description = shift;
        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 => $node,
                        Proto    => 'tcp',
                );

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

                print ">>>> $listen $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 "ERROR: lost connection to $node";
                delete $connected->{$node};
                return;
        }
        chomp( my $size = <$sock> );
        warn "<<<< $listen $node $size bytes\n";
        my $data;
        read $sock, $data, $size;
        return $data;
}

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

sub merge_out {
        my $new = shift;

        foreach my $k1 ( keys %$new ) {

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

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

                        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});
                        }
                }
        }

        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 ) {
                        warn "END at $pos";
                        last;
                }

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

        report "$affected affected records $view";

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

        if ( $connected ) {
                warn "# get results from ", join(' ', keys %$connected );
                merge_out( thaw( get_node( $_ ) ) ) foreach keys %$connected;
        }
}

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 $dump = dump $out;
                        my $len  = length $dump;

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

                        print "OUT $view $offset/$limit $len bytes $path"
                                , ( $len < 10000 ?  " \$out = $dump" : ' SAVED ONLY' )
                                , "\n"
                                ;

                        unlink "$path.last" if -e "$path.last";
                        rename $path, "$path.last";
                        write_file $path, $dump;
                        report "SAVE $path";
                }

        }

}

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

        while (1) {

                warn "NODE listen on $listen\n";

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

                warn "<<<< $listen connect from ", $client->peerhost, $/;

                my @header = split(/\s/, <$client>);
                warn "# $listen 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 $info = "$listen\t$offset\t$limit\t$path";
                        $info .= "\t" . eval $header[1] if $header[1];
                        warn "info $info\n";
                        send_sock $client => $info;
                } elsif ( $header[0] eq 'exit' ) {
                        warn "exit $listen";
                        exit;
                } else {
                        warn "WARN $listen unknown";
                }

        }
}

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/*";
        } elsif ( $cmd =~ m{^i(?:nfo)?\s?(.+)?$}i ) {
                print "# nodes: ", join(' ',@nodes), $/;

                send_nodes 'info' => $2;

                my @info = (
                        "node\toffset\tlimit\tpath",
                        "----\t------\t-----\t----",
                        "0\t$offset\t$limit\t$path",
                );

                push @info, get_node $_ foreach @nodes;

                print "$_\n" foreach @info;

        } elsif ( $cmd =~ m{^(q|e|x)}i ) {
                warn "# exit";
                send_nodes 'exit';
                exit;
        } elsif ( $cmd =~ m{^(v|r)}i ) {
                run_views;
        } elsif ( $cmd ) {
                warn "UNKNOWN ", dump $cmd;
        }

}

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