I agree that a solution where you start by just concatenating the strings, and then deal with duplicate // characters later is probably a fairly simple approach. Here's a solution similar to yours but using a negative lookbehind assertion to preserve the // if it follows a word character and a colon. That should prevent "http://" from becoming "http:/" (just as your solution also prevents this problem).

One issue with both of our solutions is that "file:///...." would be changed to "file://".

Here it is...

use strict;
use warnings;

while ( <DATA> ) {
    chomp;
    next unless $_;
    print "$_\t=>\n";
    s{(?<!\w:)//}[/]g;
    print "$_\n\n";
}

__DATA__
http://www.perlmonks.org/index.html
http://www.perlmonks.org//index.html
http:///davido.perlmonk.org/index.html
http://www.yahoo.com:8080/index.html
[download]