cat file | perl -lpe '
         }
         {
         *
         _
         =
         *.
         }
         {
         '
[download]

Very nice. Actually, we spoke briefly about this technique at last years YAPC::EU. ;-)

I asked you how you came up with the idea. You also signed a Perl book that I had with me and wrote a variation of the above on it. :-)

--
John.

Okay, this hurts my brain.

Any chance of a few pointers as to how this works in order to relieve this brain pain?

Horizontally, Abigail-II's code looks like this:     perl -lpe '}{*_=*.}{' file

Which is really quite beautiful.

Anyway, it (ab)uses the way that the -p command line option works. Consider the following output from deparse:

    $ perl -MO=Deparse  -lpe '' file

    LINE: while (defined($_ = <ARGV>)) {
        chomp $_;
    }
    continue {
        print $_;
    }
[download]

So for a simple bare -p you get all of that extra and useful code.

Now if we deparse Abigail-II's code:

    $ perl -MO=Deparse  -lpe '}{*_=*.}{' file

    LINE: while (defined($_ = <ARGV>)) {
        chomp $_;
    }
    {
        *_ = *.;
    }
    {
        ();
    }
    continue {
        die "-p destination: $!\n" unless print $_;
    }
[download]

The addition of the extra braces has created a while loop that loops through the file(s), a block with an assignment and a block with an empty statement and a continue. In effect it has disassociated the continue from the while.

The typeglob assignment *_ = *. has the effect, among other things, of setting $_ = $.. Since the while has already looped through the file(s) $. is now the number of lines in the file(s).

The last action of the program will be to enter the continue and print $_ so that the number of lines is output. The -l command line option helpfully appends a newline.

Update: Abigail-II's own explanation is here


John.

You could always use the deparser....

Abigail

perl -ne END{print$.} file
[download]

% ruby -e 'puts open(ARGV.shift).read.count("\n")' file
[download]

-Blake