perl -c -Mre=debug -e 'm/123*/'
Freeing REx: `","'
Compiling REx `123*'
size 6 Got 52 bytes for offset annotations.
first at 1
rarest char 2 at 1
   1: EXACT <12>(3)
   3: STAR(6)
   4:   EXACT <3>(0)
   6: END(0)
...
[download]

The indented part is the interesting one. If two regexes show the same output in the indented part, they will match the same pattern (the reversion isn't always true, because (a|b) and (b|a) behave the same in some regexes, and behave differently in others (not quite sure, but you can construct cases with alternation where the order somtimes matter, and somtimes not)).

I don't know if you can access this information from inside perl, perhaps with one of the B:: modules, if not you can still use backticks or qx to execute a command line like the one I showed.

My Regexp::Compare does that - not for all regexps (that would solve the halting problem :-) ), but "1233*" is recognized as more specific than "123*". On the other hand, the implementation is quite dependent on Perl internals and currently broken for perl 5.10 - the public demand didn't justify an upgrade yet...

not for all regexps (that would solve the halting problem :-) ),

To digress a bit on theory: for regular expressions that isn't hard at all.

Once you have a DFA for your regex, you can build the minimal DFA (that can be done in O(n²) or O(n³), not sure...). You can do that for both regexes you compare, and if the DFAs are isomorphic, both regexes are equivalent.

Note that checking if two DFAs are isomorphic isn't hard either, because you know the start state.

But of course Perl's regexps are not regular (in the CS sense), so you can forget everything I said if you're only after a practical solution.

That doesn't work for two reasons.

1. Regexp::Optimizer is a no-op if no alternation is used.

2. Even if Regexp::Optimizer normalized regexps, the normalized regexps still wouldn't be equal since the two regexps are not equivalent.

use Regexp::Optimizer qw( );
my $o = Regexp::Optimizer->new();
print $o->optimize(qr/123*/),  "\n";  # (?-xism:123*)
print $o->optimize(qr/1233*/), "\n";  # (?-xism:1233*)
print $o->optimize(qr/123+/),  "\n";  # (?-xism:123+)
[download]

The OP didn't ask for a method to check if two regexp are equivalent, but for a method to check if a regexp will match at least everything a second regexp matches.

Update: Added code. Changed formatting.

use re 'eval';  # Open Pandora's box
[download]

* (at least, not in the same section of code, and not without turning on taint.)

Actually your regular expressions are:

$exp1 = "123*";
$exp2 = "123+";
[download]

* is "any number of" the preceding character, including zero.
+ is "one or more of" the preceding character.

Hence: /x+/ == /xx*/

Apart from the pedantry, I'm afraid don't have an answer for your question.

no flames if this changed for 5.10, because that's the Perl I live in.

^S.*$[\r\n]
^SYS.*$[\r\n]
[download]

[root@host]# perl -Mre=debug -e '/12/';
Compiling REx "12"
Final program:
   1: EXACT <12> (3)
   3: END (0)
anchored "12" at 0 (checking anchored isall) minlen 2 
Freeing REx: "12"
[root@host]# perl -Mre=debug -e '/1+/';
Compiling REx "1+"
Final program:
   1: PLUS (4)
   2:   EXACT <1> (0)
   4: END (0)
anchored "1" at 0 (checking anchored) plus minlen 1 
Freeing REx: "1+"
[download]