# Long running program with interaction

# - Reports progress to STDERR
# - Traps ^C to avoid it killing the program
# - accepts a signal to cause it to give an intermediate result

use strict;
use warnings;
use English;
use Hit;

$SIG{"INT"} = \&nodie;
$SIG{"TERM"} = \&nodie;

$SIG{"USR1"} = \&tell;

our $lasttime = 0;
our $pinged = 0;
our $verbose = 0;

my @met1 = times();

sub nodie {
        $pinged++;
        }

sub tell {
        $verbose++;
        }

sub react {
        my $thistime = time();
        if ($thistime - $lasttime < 6) {
                print "\n";
                exit();
        }
        print "\nPlease do that again quickly\n";
        $lasttime = $thistime;
        $pinged-- ;
        }

my $fname = "huge.txt";

open FH,$fname or die;
my $syze = -s $fname;
my $bytes = 0;

my @interest = ();
our $lines = 0;
our $found = 0;
my $start = time();

my $from = localtime($start);

print "My program is $0\n";
print "My program is $PROGRAM_NAME $PID $]\n";

open (FHX,">/tmp/huge.pid");
print FHX "$PID\n";
close FHX;

while (<FH>) { # Scalar context, line by line. NOT foreach!
        $bytes += length;
        my $access = new Hit($_);
        $lines++;
        if ($access->isaSearch()) {
                $found++;
                }
        push @interest,$access;

        select STDERR;
        $| = 1;
        select STDOUT;

        my $now = time();
        my $runfor = $now - $start;
        my $percent = $bytes * 100 / $syze;

        printf STDERR ("Processed %12d lines in %6d secs. %6.2f%% done\r",
                $lines, $runfor, $percent)
        if ($lines % 80000 == 0);

        if ($percent > 0) {
                my $estimate = $runfor * 100 / $percent;
                my $to = localtime($start + $estimate);

                printf STDERR ("From %20s to % 20s\r",
                        $from,$to)
                                if ($lines % 80000 == 40000);
        }

        react() if ($pinged>0);
        talk() if ($verbose>0);

        }
print STDERR "\n";
print "Found $found out of $lines\n";

sub talk {
        $verbose = 0;
        print "\n";
        print "So far matched $found in $lines\n";
        }

sub byterm {
        our %timesaver;

        if (! $timesaver{$a} or 1 == 1) {
                my $x1 = $a -> getsearch();
                $x1 =~ s/^\s+//;
                $x1 =~ s/\s+$//;
                my @x1f = reverse(split(/\s+/,$x1));
                my $jx1 = join(" ",@x1f);
                $timesaver{$a} = $jx1;
                }

        if (! $timesaver{$b} or 1 == 1) {
                my $x2 = $b -> getsearch();
                $x2 =~ s/^\s+//;
                $x2 =~ s/\s+$//;
                my @x2f = reverse(split(/\s+/,$x2));
                my $jx2 = join(" ",@x2f);
                $timesaver{$b} = $jx2;
                }

        return $timesaver{$a} cmp $timesaver{$b};
        }

my $elapsed = time() - $start;
my @met2 = times();
print "Elapsed time $elapsed\n";
print "@met1\n@met2\n";

my @sinter = sort byterm (@interest) ;
my $k = 0;
foreach my $sample (@sinter) {
        if ($sample->isaSearch() ) {
                print $sample->getsearch(),"\n";
                $k++;
                }
        last if ($k>198) ;
        }
my @met3 = times();
print "@met1\n@met2\n@met3\n";

__END__

** Renning that - WITH caching

munchkin:j11pl grahamellis$ perl h1
Copyright me, this year
My program is h1
My program is h1 151 5.008009
Processed 800000 lines in 36 secs. 97.40% done
Found 41621 out of 822244
Elapsed time 37
0.01 0 0 0
34.85 1.52 0 0
ksh !ne
invalid command name " "
set prompt "(%|\\\$|#) "
expect interact can not find channel named " return "
perl substitute "
[etc]
perl $_
perl $_
0.01 0 0 0
34.85 1.52 0 0
62.2 1.79 0 0

WITHOUT caching - takes 20 seconds longer!

perl $_
0.01 0 0 0
34.88 1.56 0 0
82.19 1.82 0 0