in reply to Regex question

TIMTOWTDI (and inelegant and verbose, but I think, 'linear and explicit'). Think in terms of removing 6 lines, rather than 5 and 1.

#!C:/perl/bin
use strict;
use warnings;  # /me is too bleary at this hour to eliminate the "Usel
+ess use of private variable in void context at ..." (lines 56 & 61)

if (!$ARGV[0]) {
    print "Useage: perl 738855.pl String_to_search_for\n";
    exit;
}

my $regex_string = $ARGV[0];
my $re = qr/$regex_string/;

my @data_array;
while (my $line = <DATA>) {    
    chomp($line);
    push @data_array, $line;
    }

my $flag = '';
my $test_line;
my @cache;
for $test_line(@data_array) {
    $test_line =~ /($re)/;
    if ($1) {
        $flag = "Found";
        push @cache, "$flag";
        next;
    } else {
        push @cache, $test_line
    }
}

# print Dumper @cache;

my $i = 0;
my $cache;
for $cache(@cache) {
    if ( $cache =~ /Found/ ) {
        last;
    } else {
        $i++;
    }
}

# print "\$i: $i\n";            # $i = 6 when using the __DATA__ below
my $j = ( $i - 5 );            # "-5":  5 preceding lines (and current
+ line) to delete

for ($j; $j <= $i; ++$j ) {
    $cache[$j] = '';
}

$j = $i;
for ( $j; $j <= ($i+3); ++$j ) {    # "$i+3" 3 following lines to disc
+ard
    $cache[$j] = '';
}

for $cache(@cache) {
    if ($cache) {
        print "$cache \n";
    } else {
    next;
    }
}


__DATA__
Line 1
Line 2 should be removed
line 3 should be removed
line 4 should be removed
line 5 should be removed
line 6 should be removed
line 7 should be removed this has more than 5 preceding lines and more
+ than 3 following lines. FOO
line 8 should be removed
line 9 should be removed
line 10 should be removed
line 11

[download]

Except for the warnings noted above, output is as desired:

Line 1
line 11

[download]
hth

Replies are listed 'Best First'.
Re^2: Regex question
by jamsda (Novice) on Jan 26, 2009 at 05:07 UTC

    Thanks for responses guys!

    Here's my first whack at it

    #!/usr/bin/perl 

use Getopt::Long;

GetOptions(
    'd=s'     => \$delete,
    'i=s'     => \$patternmatch,
    'c=s'     => \$confile,
)

open (FILE, $confile);
while (<FILE>) {    
    $line[$.] = $_;    
    if (/$patternmatch/) {    
       delete $line[$_] for ($.-5 .. $.);     
    }    
}
close(FILE);
open (OUTPUT, ">$confile");
print (OUTPUT @line);
close (OUTPUT);

    [download]

    perl ./dellines.pl -i <patternmatch> -c ./someconfile -d

    I'm also trying to get the (-d) to be a stand-alone flag. What would be the best approach to this?

    Thanks, Jim

[reply]
[d/l]

      That can delete too many lines. 

      01 keep 
02 keep        <- gets discarded by your code.
03 keep        <- gets discarded by your code.
04 keep        <- gets discarded by your code.
05 keep        <- gets discarded by your code.
06 keep        <- gets discarded by your code.
07 discard B5
08 discard B4
09 discard B3
10 discard B2
11 discard B1
12 somepat
13 somepat

      [download]

      That's in addition to the problems with $.-5 going negative.

      By the way, splice would be more efficient than delete since it wouldn't shrink the allocated buffer.

      I'm also trying to get the (-d) to be a stand-alone flag. What would be the best approach to this?

          'd!'      => \$delete,

      [download]

      -c ./someconfile

      That's weird. Why not just print to STDOUT and redirect output? It's way more flexible at no cost. 

      perl dellines.pl -i somepat -d > someconfile

      [download]

      The name of the option doesn't even make sense. It's an output file, not a config file as far as the tool is concerned.

[reply]
[d/l]
[select]

In Section Seekers of Perl Wisdom