remove elements from file using Perl - perl

Question

Input.txt
CASE
REPEAT 1 TIMES
ENDREPEAT
ENDCASE
REPEAT
ENDREPEAT
CASE
REPEAT 2 TIMES
ENDREPEAT
ENDCASE
code.pl
open (FH, "input.txt");
my #arr = <FH>;
foreach (#arr) {
if ($_ =~ s/ENDCASE.*?CASE//gsi) {
$_ = s/ENDCASE.*?CASE//gsi;
}
}
print #arr;
Output : perl code.pl
It prints the Array without modifying........
CASE
REPEAT 1 TIMES
ENDREPEAT
ENDCASE
REPEAT ===> To be Removed
ENDREPEAT ===> To be Removed
CASE
REPEAT 2 TIMES
ENDREPEAT
ENDCASE
Output Needed is, ***||||||||||||****
CASE
REPEAT 1 TIMES
ENDREPEAT
ENDCASE
************Content Removed*****************
CASE
REPEAT 2 TIMES
ENDREPEAT
ENDCASE
Please Guide me to get this output.
Thanks in advance.........

Answer 1

This can be done through the command line as well à la flip-flop operator.
To just output the result to screen
$ perl -ne 'print if /^CASE/ .. /^ENDCASE/' Input.txt
To direct the output to another file
$ perl -ne 'print if /^CASE/ .. /^ENDCASE/' Input.txt > output.txt
To modify the file in-place
$ perl -ni.bak -e 'print if /^CASE/ .. /^ENDCASE/' Input.txt
Replace ' (single-quotes) with "(double-quotes) if on Windows.

Answer 2

You've got a couple of suggestions of ways to address your problem, but you might be interested to hear why your solution didn't work. There are a couple of reasons.
Firstly, When you read your file into #arr you get one line of the file in each element of the array. And when you process the array an element at at time, no element contains both ENDCASE and CASE so your regex never matches and nothing is changed.
For your approach to work, you need to rewrite the program to process the whole file in one go. (I've also cleaned up your code a little.)
#!/usr/bin/perl
use strict;
use warnings;
open (my $fh, '<', 'input.txt') or die $!;
my $file = do { local $/; <$fh> };
$file =~ s/ENDCASE.*?CASE//gsi;
print $file;
But this doesn't fix the problem. It gives the output:
CASE
REPEAT 1 TIMES
ENDREPEAT
REPEAT 2 TIMES
ENDREPEAT
ENDCASE
That's because the ENDCASE and CASE are included in your regex so they get removed. You'll need to look at lookahead and lookbehind assertions in perlre to fix this issue. I'll leave that as an exercise for the reader.

Answer 3

Tie your file using Tie::File:
tie #array, 'Tie::File', filename or die ...;
Manipulate the lines, in any way you see fit, and then untie the array:
untie #array;
Thus, your modifications will be reflected in the original file.

Answer 4

Here's a weird idea that just might work.
use English qw<$INPLACE_EDIT $RS>;
$INPLACE_EDIT = '.bak';
local $RS = "CASE\n";
while ( <$input> ) {
print(( !/^(END)?CASE\n\z/ms or $1 ) ? $_ : $RS );
}
The idea is that you break up your records not by newlines, but by CASE + \n and thus you get to treat all the lines between an ENDCASE and a CASE as one record that you can simply replace with "CASE\n".
Note that we simply print the record unless we see a line start before 'ENDCASE' or 'CASE' followed by a newline. So even though we make a pretty brittle assumption when breaking up the records, we check our assumption before modifying the record. Also if it matches "ENDCASE\n" then $1 is 'END' and we print that record unmodified.
This can break, though. If for some reason you were capable of having a comment here:
ENDCASE
REPEAT ===> This prints because it ends with CASE
ENDREPEAT
CASE
Then the first line would be printed. So we could do this:
my $match = 0;
my $old_1;
while ( <$input> ) {
if ( m/^(END)?CASE\n\z/ms and not $1 ) {
print $RS;
}
else {
next if $old_1;
print;
}
$old_1 = $1;
}