Perl regex substitute from hash

Question

Is there an efficient way to substitute a bunch a strings using values from a Perl hash?

For example,

$regex{foo} = "bar";
$regex{hello} = "world";
$regex{python} = "perl";

open(F, "myfile.txt");
while (<F>) {
      foreach $key (keys %regex) {
            s/$key/$regex{$key}/g;
      }
}
close(F);

Is there a way to accomplish the above in Perl?

Answer 1

First question: are you sure that what you have is inefficient?

Second, the most obvious next step would be to pull everything into a single regex:

my $check = join '|', keys %regex;

And then you can do the substitution as:

s/($check)/$regex{$1}/g;

This can still be "slow" with sufficient overlap of the keys where the regex engine has to recheck the same letters constantly. You can possibly use something like Regexp::Optimizer to eliminate the overlap. But the cost of optimising may be more than the cost of just doing everything, depending on how many changes (key/values in your hash) and how many lines you're modifying. Premature optimisation-- !

Note that, of course, your example code isn't doing anything with the text after the substitution. It won't modify the file in-place, so I'm assuming you're handling that separately.

Answer 2

Define a regexp that matches any of the keys.

$regex = join("|", map {quotemeta} keys %regex);

Replace any match of $regex by $regex{$1}.

s/($regex)/$regex{$1}/go;

Omit the o modifier if $regex changes during the execution of the program.

Note that if there are keys that are a prefix of another key (e.g. f and foo), whichever comes first in the joined regexp will be seen as a match (e.g. f|foo matches f but foo|f matches foo in foobar). If that can happen, you may need to sort keys %regex according to which match you want to win. (Thanks to ysth for pointing this out.)