Extract word

[etheral]

####################################################################
##                               MAIN
####################################################################

my $input_file = get_file ('no_sense_hogwash.txt'); #define an input f
+ile

my %ignore = (

    'and' => 1,

  );

while($input_file =~ /(\w[\w'-]*)/g) {
  
  my $word = lc $1;
  
  if (defined $ignore{$word}) {
    
    next;
    
  }
  
}

print "$input_file\n";

####################################################################
##                               SUBS
####################################################################

sub get_file { #to open a file (currently restriction.txt)
               #and store all the data within $sequence

  my ($input_file) = @_;
  
  open (IN, $input_file) or die "Cannot open $input_file for reading: 
+$OS_ERROR\n"; #open a filehandle or die  
  
  my $sequence = ''; 
  
  foreach my $line (<IN>) { #for each line in the filehandle IN

    if ($line =~ /^\s*$/) { # discard blank line
  
      next; #skip the rest of the statement block and continue with th
+e next iteration of the loop
    
    }

    $sequence .= $line # add (concatenate) to a string sequence 
      
  }
  
  return $sequence; #return the string sequence
  
  close IN;

}

I'd like to extract from a file all the text apart from the word 'and', but this code doesn't do that (but I tell it to do that, don't I?). Please help me, where is the bug?

[ikegami]

The first problem is that you do the same thing regardless of the truth of defined $ignore{$word}.

You could use the following:

perl -pe's/\band\b//g;' in > out

Or maybe you want to skip the entire line:

perl -ne'print if !/\band\b/;' in > out

[suaveant]

You ARE skipping and... but you aren't actually DOING anything when you skip it. You load the WHOLE file into $input_file and match words and then skip, but you aren't removing the word from the file. You'd need a substitute s/// to REMOVE text, something like s/(\w[\w'-]*)/$ignore{lc $1} ? '' : $1/g though you'd probably be better off generating a regular expression that matches any words you want, rather than matching all words and comparing against a hash. There are modules out there to do that, I see Regexp::List, though I think there was another one...

That's assuming you just want to remove certain words not the whole line.

Edit Regexp::Assemble is the one I was trying to think of

[Lotus1]

return $sequence; #return the string sequence
close IN;

The 'return' statement should come after the close. Otherwise the close statement will never be executed.