Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:61021
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.217.170 as permitted sender)
MIME-Version: 1.0
Date: Thu, 28 Jun 2012 12:48:56 +0200
Message-ID: <CAF+90c8M9nxhGTXXh5dkKYaM75iYaPuKiSxfr7pC3KDkLyEwrA@mail.gmail.com>
To: PHP internals <internals@lists.php.net>
Content-Type: text/plain; charset=ISO-8859-1
Subject: List comprehensions and generator expressions for PHP
From: nikita.ppv@gmail.com (Nikita Popov)

Hi internals!

Python and several other languages include support for list
comprehensions and generator expressions and I'd like to see something
similar in PHP too.

I created a hacky proof of concept implementation here:
https://github.com/nikic/php-src/tree/addListComprehensions. It's
really dirty, but it implements both features in about ~150 lines of
code.

Currently I'm using the following syntax:

    $firstNames = [foreach ($users as $user) yield $user->firstName];

This code is roughly equivalent to writing:

    $firstNames = [];
    foreach ($users as $user) {
        $firstNames[] = $user->firstName;
    }

You may notice that this particular list comprehension provides the
same functionality as array_column(), just in a little more
generalized way. E.g. you could use all of the following without
having special functions for them all:

    $firstNames = [foreach ($users as $user) yield $user->firstName];

    $firstNames = [foreach ($users as $user) yield $user->getFirstName()];

    $firstNames = [foreach ($users as $user) yield $user['firstName']];

It's also possible to explicitly specify a key:

    $firstNames = [foreach ($users as $user) yield $user->id =>
$user->firstName];

It is also possible to filter elements using list comprehensions:

    $underageUsers = [foreach ($users as $user) if ($user->age < 18)
yield $user];
    // or just the names
    $underageUserNames = [foreach ($users as $user) if ($user->age <
18) yield $user->firstName];

It is also possible to nest multiple foreach loops:

    $aList = ['A', 'B'];
    $bList = [1, 2];
    $combinations = [foreach ($aList as $a) foreach ($bList as $b)
yield [$a, $b]];
    // gives: [ ['A', 1], ['A', 2], ['B', 1], ['B', 2] ]

All the above are list comprehensions (or in PHP rather array
comprehensions), i.e. they create an array as the result.

If this is not needed it is also possible to compute the values lazily
using generator expressions, which use () instead of [].

    $firstNames = (foreach ($users as $user) yield $user->firstName);

In this case $firstNames will no longer be an array of first names,
but instead will be a generator producing first names.

This is handy if you only need to iterate the resulting "list" only
once as it saves you holding the whole list in memory.

Also it allows you to work with infinite lists easily:

    function *naturalNumbers() {
        for ($i = 0; ; ++$i) {
            yield $i;
        }
    }

    // all natural numbers
    $numbers = naturalNumbers();
    // only the odd ones
    $oddNumbers = (foreach ($numbers as $n) if ($n % 2) yield $n);
    // ...

(At this point I wonder whether one should include support for
for-loops in list comprehensions. So the naturalNumbers() function
could be replaced with (for ($i = 0;; ++$i) yield $i), etc)

So, what do you think? Do we want something like this in PHP?

Nikita