my $count = () = $currentline =~ /\b$pattern\b/g;
[download]

Although, none of this will match your string of abxyxyp, because it doesn't have any spaces at all, so I suspect you mistyped that.

I will try this, thank you for such quick responses :) Yes the second line was a typo sorry, it should have spaces. I need to keep the spaces as some categories are more than 1 char long eg. qwe a x o p .Without the spaces "qwe" could be interrpted as "q w e" which I do not want. Thanks again ! :) P.

You might still look at \b. It's a zero width assertion that there's a word boundary there. Which means it's pretty much exactly like the other solution I gave except for three aspects:
1. It's based on the difference between \w and \W, not between \s and \S (that is, it will detect a boundary after "qwe" in "qwe-rty")
2. If your $pattern begins or ends with a non-word character, a \b will assert that it's next to a word character (while the solution based on \s will still assert that there is a space next to it
3. Finally, and possibly most importantly, it will also match at the begining and end of the string. If you want the other to work there, you'll have to add spaces to the begining and end.
Probably none of that is all that imporant, but it's often very useful in cases like this.

Show us what is contained in $pattern. If it is just "xy", what you have posted wont find it anyway if xy is embedded between non-space characters. Your existing regex doesnt seem to reflect the behavior you're describing.

Hi, A requirement is that it is preceeded and followed by a space. I have included a script that demo's what I am saying if that helps. Thanks :)

#!/usr/bin/perl

$countsAllExample="a b x y i x y p";
$doesntCountAll="a b x y x y p";
$pattern = "x y";
$count=0;

while($countsAllExample =~ /\s$pattern\s/g){
$count +=1;
}

print "found $pattern $count times in $countsAllExample\n";

#now error occurs as only 1 "x y" reported
###########################################
$count=0;

while($doesntCountAll =~ /\s$pattern\s/g){
$count +=1;
}

print "found $pattern $count times in $doesntCountAll\n";

Ok, take your second test case, the one that doesn't work as you expect, and let's walk through it.

1. Match "_x_y_" (I used _ to indicate a space character). One is found, and the pattern match position pointer is advanced to the 2nd 'x'.
2. Look for another "_x_y_", but there isn't one, because you're already at position 'x', not position '_' (space).

What you may want is something like this:

m/\s$pattern(?=\s)/

This won't gobble the second space char.




Dave

You are searching for spaces around your pattern so yes it should be expected to only match once in the second case. Knowing your exact pattern would be the biggest help in solving this quandry. Also I would reccomend this construct

foreach(@lines)
{
    while(m{$pattern}g)
    { 
       $count++;
    }
}
[download]

Reap this node please (is there a special way of requesting NodeReaper to come along?) Reason: alcohol clouded reasoning.

• Considered by bart: Delete as per the author's request
• 2004-11-17 Unconsidered by Arunbear: 2+ keep votes prevents reaping; Keep/Edit/Delete: 9/1/25

while ($doesntCountAll =~ /\s$pattern\s/g) {
    $count++; pos($doesntCountAll) -= 1;
}
[download]