4.2 Match Delimiters

4.2  Match Delimiters

   The regular expression match operator is =~. On the left hand side, we have the target of the match. On the right hand side we have the regular expression along with any modifiers, if any. An example is

$string =~ m/theoretical/

Here, we are looking for the pattern theoretical in $string.

If on the left hand side, the variable over which we are doing the pattern matching is the special variable $_, we need not write the variable as well as the operator =~. Therefore,

$_ =~ m/theoretical/

and

m/theoretical/

mean exactly the same. In addition, if / is used as the delimiter, the use of m is not necessary. Therefore, we can write the previous expression as

/theoretical/

also.

Actually, Perl allows us to use any non-alphanumeric, non-whitespace character as the delimiter if we use m to indicate the match operation. In the specific example, we have shown earlier, any of the following will work instead of m/$pattern/.

m@$pattern@

m#$pattern#

m=$pattern=

m!$pattern!

m~$pattern~

m%$pattern%

m^$pattern^

m*$pattern*

m-$pattern-

m+$pattern+

m:$pattern:

m;$pattern;

m"$pattern"

m|$pattern|

m\$pattern\

m,$pattern,

However, there are some exceptions. For a complete discussion, one should look at the book Mastering Regular Expressions [Fri97]. Here are some of the exceptions. Although we said any non-alphanumeric character will work as the delimiter, the four “natural” delimiters need to be used in complementary pairs. In other words, the following do not work:

m{$pattern{

m[$pattern[

m($pattern(

m<$pattern<

but they work in natural pairs as in the following.

m{$pattern}

m[$pattern]

m($pattern)

m<$pattern>

If we use ?  as the delimiter, the pattern matches only once. For example, if the previous example were changed to

Program 4.2

#!/usr/bin/perl 
$pattern = shift(@ARGV);

while (<>){
    if (m?$pattern?){
        print "File $ARGV:";
        print "\t$_";     #prints the line
    }
}

the result will be

File nature.tex: theoretically satisfying, will be

That is, only the first match will take place.

Finally, if we use the single quote as the delimiter, the pattern behaves like a string delimited with a single quote. That is, variables inside the string are not be interpolated. Specifically, in this example, if we change the pattern matching to

m'$pattern'

Perl looks for the literal presence of the string $pattern in the files.