Slashes and hashes in Perl and metacharacters

Question

Thanks for the previous assistance everyone!. I have a query regarding RegExp in Perl

My issue is..

I know, when matching you can write m// or // or ## (mus

Answer 1

When you build a regexp, you define a character as a delimiter for your regexp i.e. doing // or ##.
If you need to use that character inside your regexp, you will need to escape it so that the regexp engine does not see it as the end of the regexp.

If you build your regexp between forward slashes /, you will need to escape the forward slashes contained in your regexp, hence the escaping in your second example.

Of course, the same rule apply with any character you use as a regexp delimiter, not just forward slashes.

Answer 2

The question itself has been properly answered in several answers. But everything you always wanted to know about Perl regular expressions, but may or may not have been afraid to ask, can be found in perldoc perlre, perldoc perlrequick and perldoc perlretut. I recommend you read through them.

Answer 3

The regex match-operator allows to define i a custom non-whitespace-character as seperator.

In your first example the '#' is used as seperator. So in this regex you don't need to escape the '/' because it hase no special meaning. In the second regex, the seperator char isn't changed. So the default '/' is used. Now you have to escape all '/' in your pattern. Otherwise the parser is confused. :)

Answer 4

The forward slashes are not meta characters in themselves - only the use of them in the second example as expression separators makes them "special".

The format of a substitute expression is:

s<expression separator char><expression to look for><expression separator char><expression to replace with><expression separator char>

In the first example, using a hash as the first character after the =~ s, makes that character the expression separator, so forward slash is not special and does not require any escaping. in the second example, the expression separator is indeed the forward slash, so it must be escaped within the expressions themselves.

Answer 5

If you are not use slashes, the recommend practice is to use the curly braces and the /x modifier.

$date=~ s{ (\d+) \/ (\d+) \/ (\d+) }{$1/$2/$3}x;

Escaping the non-alphanumerics is also a standard even if they are not meta-characters. See perldoc -f quotemeta.

Answer 6

There is another depth to this question about escaping forward slashes with the s modifier. With my example the capturing becomes the problem.

$image_name =~ s/((http:\/\/.+\/)\/)/$2/g;

For this to work the typo with the addition of a second forward slash, had to be captured. Also, trying to work with just the two slashes did not work. The first slash has to be led by more than one character.

Changing "http://world.com/Photos//space_shots/out_of_this_world.jpg"
To: "http://world.com/Photos/space_shots/out_of_this_world.jpg"