Concatenating sequences of RegExp search and replaces ... Possible?

[SparceMatrix]

I would like to be able to take this series of simple search-and-replaces and replace it with a single or fewer regular expressions. Is it possible? Or, is there a more economical way to handle a series of search and replaces?

#!/usr/bin/perl

$SampleString = "P1rs2*this*for*thrEE*diFF2r2nt*p1TT2rns";

$SampleString =~ s/\*/ /g;

$SampleString =~ s/1/a/g;

$SampleString =~ s/2/e/g;

$SampleString =~ s/((\w)\2)/\L$1\E/g;

print $SampleString;

 

[rharsh]

All the single character substitutions can be taken care of using tr instead:

$SampleString =~ tr/*12/ ae/;

 

[SparceMatrix]

Great! But is that as far as I can take it? Could I make it just one RegExp? Anyone?

 

[KevinADC]

this is contrived for your example but does what you want:

$SampleString = "P1rs2*this*for*thrEE*diFF2r2nt*p1TT2rns";
$SampleString =~ tr/*12A-Z/ aea-z/;
$SampleString = ucfirst($SampleString);
print $SampleString;

one regexp and then uses ucfirst() to capatalize the first character in the string.

 

[SparceMatrix]

I'm convinced of the power of tr///, but generally, is it possible to combine two or more RegExp's into one RegExp? I'm wondering if there is a general technique. Maybe there is a general way of handling that in PERL. In VBS, I have been moving the series of RegExp's into an array and looping through them.

 

[KevinADC]

I'm convinced of the power of tr///, but generally, is it possible to combine two or more RegExp's into one RegExp? I'm wondering if there is a general technique. Maybe there is a general way of handling that in PERL. In VBS, I have been moving the series of RegExp's into an array and looping through them.

In general is it possible? It really depends on what you are doing is the only answer I can think of. It is not always beneficial to do so however. Sometimes two or more regexps is more efficient than one regexp even though they do the same thing in the end.

The best thing is to understand all the various options and the differences between the three basic regexps: matching, substitution, and transliteration.

Moving a series of regexp's into an array and looping through them is fine, but it's really the same as using a series of them, I am not aware of any gain by doing that. My thinking is it's probably less efficient than using a series of regexp's although probably not by much. If you find you need to use a series of regexp's over and over, you probably need to write a dedicated sub routine just for that purpose.

The tr/// regexp is an effiecent but limited regexp. When you need to only do a straight individual for individual character replacement (like your example) it's great, but it has no concept of the various options available for the more powerful s/// form of regexp. No wild card matching, no pattern matching, no string anchors, can't use the powerful 'e' (eval) option, no variable interpolation, etc etc.