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; 


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