2016 twenty-four merry days of Perl Feed

Making Perl Functional

List::Util - 2016-12-15

"So you see, I simply have too many nice children to sort from the naughty ones. There's only a few days until its time to take presents to only the nice ones," Santa said sadly. "To make things worse, our database provider got DDoS'd by the naughty children's botnets, and all I have is this CSV backup!"

"Santa! Don't worry!" said Isabella the elf cheerily. "I think I have a solution to your problem!"

Santa looked at Isabella with keen interest. "Well, what's that then?" he asked curiously.

Using functional techniques on Perl lists

Perl is not often thought of as a functional programming language, but it has the key list evaluation primitives of filter known as grep in Perl, map, and reduce (sometimes called fold in other languages.)

Perl also supports functions as first-class variables, so it has all of the necessary ingredients to be a fruitful functional programming environment. (If this idea intrigues you, I very highly recommend Mark Jason Dominus' excellent book Higher Order Perl which you may read and download for free from his website.)

grep takes a list and applies a "predicate" - a code block that returns a true or false value. If the predicate returns true, the list item is passed into a new list. If the predicate returns false, the list item is discarded.

Here's part of Santa's CSV file:

    id, name, nice, siblings, present
    1, Hermione, 1, 0, Wand
    2, Delores, 0, 1, Coal
    3, Draco, 0, 0, Dirty sock
    4, Ronald, 1, 6, Scarf

Using grep

Our task here is to filter the naughty names from the nice ones. We're going to use Text::CSV to parse the data, because friends don't let friends parse CSV using split.

In this particular case, our predicate is very simple. We can check to see if the nice field contains a 1 or a 0. If it has a 1, we add the child to our list of nice children.

use 5.014;
use Text::CSV; # could use Text::CSV_XS instead
use IO::String;

my $data = <<'DATA';
id, name, nice, siblings, present
1, Hermione, 1, 0, Wand
2, Delores, 0, 1, Coal
3, Draco, 0, 0, Dirty sock
4, Ronald, 1, 6, Scarf
DATA

my $fh = IO::String->new($data);
my $csv = Text::CSV->new( { allow_whitespace => 1 } );

$csv->column_names( $csv->getline($fh) );
my @nice = grep {; $_->{nice} } @{ $csv->getline_hr_all($fh) };
close $fh;

Why do I use ; as the first character in my code block?

I want to make it abundantly clear to Perl that I am using a code block. There are various ways to ensure that curly braces are interpreted as a code block but I never remember them. Using a semi-colon (;) as the very first token in my block ensures that Perl knows that whatever follows should be considered a code block (and not, for example a hash operation.)

Using map

Santa has a database with latitude and longitude for each child's home. We need to add this information to each nice child entry.

To do this we can use a map function. A map takes each element of a list and applies a function to it, and returns the transformed item in a new list.

my %locations = (
    'Wrigley Field' => [41.94757, -87.6562],
    'The Getty'     => [34.07905, -118.4744],
    'Austin'        => [30.26759, -97.74299],
    'Honolulu'      => [21.30485, -157.8578],
);

We also need to do some calculations of how much distance Santa expects to cover when he's delivering the gifts, so instead of just using the latitude and longitude as given in the database, we're going to represent these points using Geo::Calc::XS

use Geo::Calc::XS;
my @locations = map {;
    my $loc = $locations{$_};
    Geo::Calc::XS->new(
        lat  => $loc->[0],
        lon  => $loc->[1],
        units => 'mi'
    )
} @nice;

Using reduce

Now we can compute the total distance traveled between each location using a reduction. This is a list operation that performs a function on each list element and adds its results into an accumulator variable. When the list is exhausted, the function returns the accumulator's value.

In Perl, reduce is found in List::Util and has been in core since 5.7.3, so it should (almost) always be available, the same as map and grep.

use List::Util qw(reduce);

my $miles_traveled = reduce {;
    $a->distance_to( $b )
} @deliveries;

say $miles_traveled;

I hope Santa gets plenty of cocoa and cookies for that long trip.

Pipelines

A really powerful technique is to build a pipeline of successive map and grep operations.

my $miles_traveled = reduce {;
    $a->distance_to( $b )
} map {;
    my $loc = $locations{$_};
    Geo::Calc::XS->new(
        lat  => $loc->[0],
        lon  => $loc->[1],
       units => 'mi'
    )
}  grep {;
    $_->{nice}
} @{ $csv->getline_hr_all($fh) };

This is a compact and expressive way to signal your intention of how to modify data which comes in a list form. Some areas where this is especially useful Perl programmers are likely to encounter in everyday use are DBI results and JSON input and/or transformation.

I hope you enjoyed this exploration of grep, map, and reduce. I find them tremendously useful and I use them frequently instead of foreach or while or other such loops.

SEE ALSO

Gravatar Image This article contributed by: Mark Allen <mrallen1@cpan.org>