Perl - Problem with “]” in a regular expression

Question

I have a string :

my $string = "name_of_my_function(arg1,arg2,[arg3,arg4])";

and I want to extract the name of the function "name_of_my_function" and the parameters :

$arg1 = "arg1"
$arg2 = "arg2"
@arg_list = ("arg3", "arg4")

the code I use to extract the function is :

$row =~ m/^([^\(]*)\(([^\)]*)\)/;
$function = $1;

However, it works when the string doesn't have any "]" , for example :

my $string = "name_of_my_function(arg1,arg2,arg3)";

but it doesn't return anything when there is a "]"

Any idea?

Thanks,

SLP

Answer 1

The regex you show captures the function name, and all other arguments in a string, which is a very reasonble first step. Then parse the arguments out of that second string. I expand your $string so to have multiple bracketed lists of arguments, interleaved with non-bracketed ones

perl -wE'
    $s = "name_of_my_function(arg1,arg2,[arg3,arg4],arg5,[arg6,arg7])"; 
    @m = $s =~ /^([^\(]*)\(([^\)]*)\)/; 
    @p = grep { $_ } split /\s*,\s*|\[(.*?)\]/, $m[1];
    for (@p) { 
        if (/,/) { push @arg_list, $_ }
        else     { push @args, $_ }
    }
    say $m[0];
    say for @args; 
    say for @arg_list
'

This prints

name_of_my_function
arg1
arg2
arg5
arg3,arg4
arg6,arg7

The split is where individual arguments are extracted, as well as bracketed argument list(s), each as a string. That may return empty elements thus grep { $_ } to filter them out.

Then you can proceed to extract individual arguments from lists that were in brackets, by splitting each string in @arg_list by , again.

---

The main part of the above can , as the problem stands, go in one statement

@p = grep { $_ } split /\( | \) | \[(.*?)\] |,/x, $s;

where I added /x modifier so to be able to space it out for readability. This delivers to @p the function name, individual arguments, and a string with (comma separated) argument list from each [] .

However, I think that it is far more sensible to break this up into several steps.

Answer 2

Well, if the number of arguments is variable, that is not that simple to do it with rgex only (arguments will be matched with + quantifier, so they won't be stored in capturing group, which would be easy to extract). Having in mind the above, you could use this pattern (\\w+)\\(((\\w+|\\[(\\w+,?)+\\]),?)+\\)

Explanation:

(\\w+) - match one or more word characters (name of a function) and store it in first capturing group,

(\\w+|\\[(\\w+,?)+\\]) - alternation: match \\w+ (same as above) or \\[(\\w+,?)+\\] : \\[ - match [ literally, (\\w+,?)+ - match on or more times \\w+, pattern which is one or more word characters followed by one or zero commas ( ,? ), \\] - match ] literally,

((\\w+|\\[(\\w+,?)+\\]),?)+ - match whole above pattern, optionally followed by comma ( ,? ) one or more times. This would match argument list.

\\( , \\) 0 match ( , ) literally

Further processing - extract whats between brackets () in order to extract arguments list programatically - it would be easier that doing it with complex regular expression

Demo

UPDATE :

Try pattern: https://regex101.com/r/wBcJZ0/3

I omitted explanation, as it is very similair to previous pattern.

Updted demo