Apply array_map on column of the array - php

Question

I was trying to get a solution to speed up the process of PHP.
We are running into execution time issues on a specific page.
We have an array with around 10.000 rows and we need to apply several callback functions on some of the columns of that array.
What is the best way and the fastest execution possible to go over an array and only apply a callback on several columns.
<?php
$records = [
['id' => 2135, 'first_name' => 'John', 'price' => 1000, 'unit' => 5, 'discount' => 30],
['id' => 3245, 'first_name' => 'Sally', 'price' => 2000, 'unit' => 8, 'discount' => 80],
['id' => 5342, 'first_name' => 'Jane', 'price' => 4000, 'unit' => 5, 'discount' => 34],
['id' => 5623, 'first_name' => 'Peter', 'price' => 1500, 'unit' => 4, 'discount' => 25]
];
function simpleMultiply($value)
{
return $value * 2;
}
$applyToColumn = ['price','unit','discount'];
// $expectedRecords = array_map('simpleMultiply', array_column($records, 'id'));
$expectedRecords = [
['id' => 2135, 'first_name' => 'John', 'price' => 2000, 'unit' => 10, 'discount' => 60],
['id' => 3245, 'first_name' => 'Sally', 'price' => 4000, 'unit' => 16, 'discount' => 160],
['id' => 5342, 'first_name' => 'Jane', 'price' => 8000, 'unit' => 10, 'discount' => 68],
['id' => 5623, 'first_name' => 'Peter', 'price' => 3000, 'unit' => 8, 'discount' => 50]
];
?>

Answer 1

array_column is an expensive operation if the array is large, because it has to create an entirely new array. Calling it repeatedly for different columns will multiply that expense. I suggest you just use a simple foreach loop.
$expectedRecords = array();
foreach ($records as $r) {
foreach ($applyToColumn as $col) {
$r[$col] = simpleMultiply($r[$col]);
}
$expectedRecords[] = $r;
}
If you can modify the original $records instead of creating a new $expectedRecords, you can use a reference in the first foreach.
foreach ($records as &$r) {
foreach ($applyToColumn as $col) {
$r[$col] = simpleMultiply($r[$col]);
}
}

Answer 2

You can wrap the result set in an iterator (or a generator, which looks rather similar to an iterator) that modifies the current row before returning it as the iteration result, e.g.
<?php
class Foo extends IteratorIterator {
public function current() {
$rv = parent::current();
if ( $rv ) {
// ... and more checks here ....
// this example just assumes the fields/elements exist
$rv['x'] *= 17;
return $rv;
}
}
}
$pdo = new PDO('mysql:host=localhost;dbname=test;charset=utf8', 'localonly', 'localonly', array(
PDO::ATTR_EMULATE_PREPARES=>false,
PDO::MYSQL_ATTR_DIRECT_QUERY=>false,
PDO::ATTR_ERRMODE=>PDO::ERRMODE_EXCEPTION
));
setup($pdo);
// the point is not to have the complete result set in memory
// see: e.g. php.net/mysqlinfo.concepts.buffering (or whatever database you use ;-) )
$pdo->setAttribute(PDO::MYSQL_ATTR_USE_BUFFERED_QUERY, false);
$result = new Foo( $pdo->query('SELECT x,y FROM sofoo') );
foreach( $result as $row ) {
echo join(', ', $row), PHP_EOL;
}
function setup($pdo) {
$pdo->exec('
CREATE TEMPORARY TABLE sofoo(
id int auto_increment, # I just throw this in casually....
x int,
y int,
primary key(id)
)
');
$stmt = $pdo->prepare('INSERT INTO sofoo (x,y) VALUES (?,?)');
foreach( range(1,5) as $x ) {
$stmt->execute( array($x, $x*13) );
}
}
or (without most of the boilerplate, using a generator - and invoking the anonymous function immediately)
$pdo->setAttribute(PDO::MYSQL_ATTR_USE_BUFFERED_QUERY, false);
$result = (function($iter) {
foreach( $iter as $row ) {
$row['x'] *= 17;
yield $row;
}
})($pdo->query('SELECT x,y FROM sofoo') );
foreach( $result as $row ) {
echo join(', ', $row), PHP_EOL;
}
You can build this "system" arbitrarily complex. Maybe passing the field names and functions to modify the elements to the generator like
$pdo->setAttribute(PDO::MYSQL_ATTR_USE_BUFFERED_QUERY, false);
$result = my_generator(
$pdo->query('SELECT x,y FROM sofoo'),
array('x'=>function($e) { return $e*17; })
);
foreach( $result as $row ) {
echo join(', ', $row), PHP_EOL;
}
function my_generator($iter, $modifiers) {
foreach( $iter as $row ) {
foreach( $modifiers as $field=>$func) {
$row[$field] = $func($row[$field]);
}
yield $row;
}
}
or add some chaining mechanism so that multiple functions can modify the same field. Or passing the complete row to the functions so they can change the structure as well. Or ... or ...