Micz has asked for the wisdom of the Perl Monks concerning the following question:

I have a bunch of regexs which operate on the same string, doing s/ and tr/. can I put this all into one regex? it needs to be a perl5 or posix compliant regex as I want to use it in perl, php, java etc. which is why I would like one regex, to make it easier to transport. the code currently looks like: 
$print =~ s/([b-z]|[A-Z]):/\1/g;
$print =~ s/a |V /A /g;
$print =~ s/a: /A: /g;
$print =~ s/2:/2 %/g;
$print =~ s/9|3:/&/g;
$print =~ s/([a,e,O])I/\1j/g;
$print =~ tr/QRC@6{/Orx**E/;
...

[download]
you get the point. please excuse my coding. thanks for your help!

Replies are listed 'Best First'.
Re: Concatenating regexs?
by japhy (Canon) on May 04, 2001 at 17:48 UTC
    It's hard to just "group" together regexes into one regex, unless you make a "dispatch table". 
    s/([b-zA-Z]):/\1/g;
s/[aV] /A /g;
s/a: /A: /g;
s/2:/2 %/g;
s/9|3:/&/g;
s/([aeO])I/\1j/g;
tr/QRC@6{/Orx**E/;

my %trans;
$trans{"$_:"} = $_ for 'b' .. 'z', 'A' .. 'Z';
$trans{"$_ "} = "A " for 'a', 'V';
$trans{"a: "} = "A: ";
$trans{"2:"} = "2 %";
$trans{$_} = "&" for 9, '3:';
$trans{"${_}I"} = "${_}j" for 'a', 'e', 'O';

my $regex = join '|', map quotemeta($_), keys %trans;

### WARNING -- this does not allow for nested
### translations (like (A => 'BC', B => 'D') on
### the string "A" will yield "BC", not "DC")
$string =~ s/($regex)/$trans{$1}/g;
$string =~ tr/QRC@6{/Orx**E/;

    [download]


    japhy -- Perl and Regex Hacker
[reply]
[d/l]
Re: Concatenating regexs?
by suaveant (Parson) on May 04, 2001 at 17:45 UTC
    not easily, since they all do quite different things... much easier as you have it, but if you are diong them often maybe you should look into using qr{} to precompile the regexps...

    also, you realize that $print =~ s/([a,e,O])I/\1j/g; matches , as well as a e and O? if you don't want to match comma use [aeO] [] is a character class, so it looks at characters, so a separator is not necessary.

    also I believe that using $1 in the second half of the substitution is preferred to \1... perl -w would tell you that...

    you can do $print =~ s/([b-zA-Z]):/\1/g; which should be much faster than $print =~ s/([b-z]|[A-Z]):/\1/g; character classes can have as many ranges as you want in them.... but the (|) will slow you down compared to a straight character class...

    Update Oops... 9|3: does match 9 or 3:... my bad, ignore the next line
    not sure, do you want 9|3: to match '9|3:' or do you want it to match '9' or '3:'... it will match the former...
                    - Ant

[reply]
[d/l]
[select]
      thanks for your help, I have a lot to learn. I have taken the first points and used them. How would I find either '9' or '3:' ? using 9\|3: ? thanks again, jan
[reply]
        Ack! oops, sorry... my bad /9|3:/ will do the or...
                        - Ant
[reply]

Back to Seekers of Perl Wisdom