trunk/bin/sack.pl

#!/usr/bin/perl

use warnings;
use strict;

use Time::HiRes qw(time);
use Data::Dump qw(dump);
use File::Slurp;
use Getopt::Long;


my $path   = '/data/isi/full.txt';
my $limit  = 10000;
my $offset = 0; 
my @views;


GetOptions(
        'path=s'   => \$path,
        'offset=i' => \$offset,
        'limit=i'  => \$limit,
        'view=s'   => \@views,
) or die $!;

my $t = time;

use lib '/srv/webpac2/lib/';
use WebPAC::Input::ISI;
my $input = WebPAC::Input::ISI->new(
        path   => $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;
}


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

mkdir 'out' unless -e 'out';

our $out;

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

        foreach my $view ( @views ) {

                my ( $nr, $package ) = ( $1, $2 )
                        if $view =~ m{/(\d+)\.([^/]+(\.pl)?$)};

                undef $out;

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

                my $code = read_file $view;
                warn "## CODE\n$code\n## CODE\n";

                my $affected = 0;
                $t = time;

                foreach my $pos ( $offset + 1 .. $input->size ) {
                        my $rec = $input->fetch_rec($pos);
                        if ( ! $rec ) {
                                warn "END at $pos";
                                last;
                        }

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

                report "$affected affected records $view";

                if ( defined $out ) {
                        my $dump = dump $out;
                        my $len  = length $dump;
                        my $path = "out/$nr.$package";
                        print "out $view $offset/$limit $len bytes $path"
                                , ( $len < 10000 ?  " \$out = $dump" : ' SAVED ONLY' )
                                , "\n"
                                ;

                        write_file $path, $dump;
                        report "save $path";
                } else {
                        warn "W: no \$out defined!";
                }

        }
}

run_views;

while ( 1 ) {

        print "sack> ";
        my $cmd = <STDIN>;

        if ( $cmd =~ m{(vi|\\e|out)}i ) {
                system "vi out/*";
        } else {
                run_views;
        }

}

1	dpavlin	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	dpavlin	4	use Getopt::Long;
10	dpavlin	1
11	dpavlin	4
12			my $path = '/data/isi/full.txt';
13			my $limit = 10000;
14			my $offset = 0;
15	dpavlin	6	my @views;
16	dpavlin	4
17
18			GetOptions(
19			'path=s' => \$path,
20			'offset=i' => \$offset,
21			'limit=i' => \$limit,
22	dpavlin	6	'view=s' => \@views,
23	dpavlin	5	) or die $!;
24	dpavlin	4
25	dpavlin	1	my $t = time;
26
27			use lib '/srv/webpac2/lib/';
28			use WebPAC::Input::ISI;
29			my $input = WebPAC::Input::ISI->new(
30	dpavlin	4	path => $path,
31			offset => $offset,
32			limit => $limit,
33	dpavlin	1	);
34
35
36			sub report {
37			my $description = shift;
38			my $dt = time - $t;
39			printf "%s in %1.4fs %.2f/s\n", $description, $dt, $input->size / $dt;
40			}
41
42
43			report $input->size . ' records loaded';
44
45			mkdir 'out' unless -e 'out';
46
47	dpavlin	5	our $out;
48
49	dpavlin	1	sub run_views {
50	dpavlin	6	@views = sort glob 'views/*.pl' unless @views;
51	dpavlin	1	warn "# views ", dump @views;
52
53			foreach my $view ( @views ) {
54
55			my ( $nr, $package ) = ( $1, $2 )
56			if $view =~ m{/(\d+)\.([^/]+(\.pl)?$)};
57
58	dpavlin	5	undef $out;
59	dpavlin	1
60			next if system("perl -c $view") != 0;
61
62			my $code = read_file $view;
63	dpavlin	5	warn "## CODE\n$code\n## CODE\n";
64	dpavlin	1
65	dpavlin	5	my $affected = 0;
66	dpavlin	1	$t = time;
67
68	dpavlin	5	foreach my $pos ( $offset + 1 .. $input->size ) {
69			my $rec = $input->fetch_rec($pos);
70			if ( ! $rec ) {
71			warn "END at $pos";
72			last;
73			}
74	dpavlin	1
75	dpavlin	5	eval "$code";
76			if ( $@ ) {
77			warn "ERROR [$pos] $@\n";
78			} else {
79			$affected++;
80			}
81	dpavlin	1	};
82
83	dpavlin	5	report "$affected affected records $view";
84	dpavlin	1
85			if ( defined $out ) {
86			my $dump = dump $out;
87			my $len = length $dump;
88			my $path = "out/$nr.$package";
89	dpavlin	5	print "out $view $offset/$limit $len bytes $path"
90			, ( $len < 10000 ? " \$out = $dump" : ' SAVED ONLY' )
91	dpavlin	1	, "\n"
92			;
93	dpavlin	5
94	dpavlin	1	write_file $path, $dump;
95			report "save $path";
96	dpavlin	5	} else {
97			warn "W: no \$out defined!";
98	dpavlin	1	}
99
100			}
101			}
102
103	dpavlin	3	run_views;
104
105	dpavlin	1	while ( 1 ) {
106
107			print "sack> ";
108			my $cmd = <STDIN>;
109
110	dpavlin	3	if ( $cmd =~ m{(vi\|\\e\|out)}i ) {
111			system "vi out/*";
112			} else {
113			run_views;
114			}
115
116	dpavlin	1	}
117