How to remove non numeric charcter from numeric string? - perl

Question

I have following data. I would like to print last column without non numeric character from a string. Kindly help me
N THR K 149A
CA THR K 149A
C THR K 149A
O THR K 149A
CB THR K 149A
OG1 THR K 149A
CG2 THR K 149A
N SER K 149B
CA SER K 149B
C SER K 149B
O SER K 149B
CB SER K 149B
for solving the above problem I have tried by following program.
#!/usr/bin/perl -w
open(F1, "$ARGV[0]") or die;
chomp(#arr=<F1>);
close F1;
for($i=0;$i<=$#arr;$i++)
{
#pdb=split(/\h/,$arr[$i]);
if($pdb[3] =~ /[A-Z]/*$);{
$pdb[3] =~ s/\D//g;
print "$pdb[1] $pdb[2] $pdb[3]\n";
}
}

Answer 1

Ok, unless this is a typo, it is the thing wrong with your code.
if($pdb[3] =~ /[A-Z]/*$);{
In this code, you have placed the slash / in the middle of your regex, and also placed a semi-colon there which does not belong anywhere on that line. Also, you are using * as the quantifier, which will not work as intended, because it will allow a match on the empty string (zero matches), which will match all strings. The correct line is:
if($pdb[3] =~ /[A-Z]+$/) {
However, this entire line is incorrect, when taken in context:
if($pdb[3] =~ /[A-Z]*$/) {
$pdb[3] =~ s/\D//g;
Here you only remove non-digits if upper case letters are found. Besides the fact that you are checking for two different things, you do not need to check before substituting, because a substitution will not do anything if it does not match. So... something like this:
if ($foo =~ /A/) {
$foo =~ s/A//g;
is completely redundant, because s/A//g will not do anything unless there is already an A in the string.
Also, a few more things you should know:
Always use
use strict;
use warnings;
As it will help you prevent a lot of simple mistakes.
Use three argument open, with lexical file handle, and check the return value including the error:
open my $fh, "<", $file or die "Cannot open $file: $!";
You do not need to quote variables, such as with "$ARGV[0]". You leave out the quotes: $ARGV[0].
You are using a C-style for loop. Using a Perl-style loop is preferred, in my opinion:
for my $i (0 .. $#arr)
But you should not be using array indexes unless you need the indexes themselves, so the better loop is:
for my $line (#arr)
But again, as a general rule, it is better to read a file line-by-line than slurping it into an array. For this purpose you would use a while loop, which iterates over the file handle instead of exhausting it all at once:
while (<$fh>) {
# process line $_
}
Using /\h/ as the field delimiter for split is wrong, unless you intended that consecutive whitespace indicates empty fields. The default split is ' ', which splits on multiple whitespace /\s+/, and also strips leading whitespace. With CSV data, it is possibly correct to split on single delimiters, but in that case you should use the specific delimiter, and not a character class like \h.
Like I said before, using the * quantifier in a regex match is horribly wrong. You might notice that a regex such as /[A-Z]*/ matches anything if you try it out: perl -lnwe 'print /[A-Z]*/ ? "match!" : "no match";' That is because it is allowed to match the empty string, and all strings match the empty string.
And like I also said, you do not need to check before you substitute. At least not for the same thing. So, when simplified, your code becomes:
open my $fh, "<", $ARGV[0] or die "Cannot open $ARGV[0]: $!";
while (<$fh>) { # short for while ($_ = <$fh>)
chomp; # short for chomp($_)
my #fields = split; # short for split(' ', $_)
$fields[3] =~ s/\D//g;
print "#fields[1,2,3]\n"; # quoting an array inserts spaces between elements
}
Note that I used an array slice, where we only use the elements with the indicated elements. You can also write this such as:
print join(" ", $fields[1], $fields[2], $fields[3]), "\n";
You might note also that this is possible to achieve using a one-liner:
perl -anlwe '$F[3] =~ s/\D//g; print "#F[1,2,3]"'
The -a switch autosplits the line on whitespace, storing the fields in #F. The -l switch chomps the line and adds newline to print. And the -n switch reads input from STDIN or argument files, whichever is supplied.

Answer 2

Try this
perl -ne 'print "$1\n" if m/(\d+)\D$/' datafile