I have an array
$users = array();
$users[0]['id']=1;
$users[0]['name']="user1";
$users[1]['id']=2;
$users[1]['name']="user2";
$users[2]['id']=3;
$users[2]['name']="user3";
And i need to extract id's into another array $ids; such that
$ids[0]=1;
$ids[1]=2;
$ids[2]=3;
I know one way to do it...
$ids = array();
foreach($users as $user){
$ids[] = $user['id'];
}
But
1.is it is the best way to doit? 2.is it possible to do it without loops. 3.is it is the fastest way....?
All the comments above address the question very well, but since noone has actually posted any actual answer, here's some additional info (to justify my answering it):
Probably, but it is certainly the cleaner and most readable way to do it
Yes, but as people have said, it's only a trick, as loops will be used in the background.
Now this calls for some investigation. I have recreated a similar array as yours, using 100000 entries:
for ($i=0;$i<100000;$i++) {
$users[] = array('id' => rand(),
'name' => 'default');
}
And ran a few tests using different cases:
1.Plain old for loop (the one you have used yourself):
$ids=array();
$t = microtime(true);
foreach ($users as $key => $value) {
$ids[] = $value['id'];
}
echo microtime(true)-$t;
This required 0.085'' on average
2.Using array_walk():
$t = microtime(true);
array_walk($users, create_function('$v,$k,&$ids', '$ids[0][] = $v["id"];'), array(&$ids));
echo microtime(true)-$t;
This required on average 0.22'' (the same when using $GLOBALS['ids'] instead of this "reference" hack)
3.Using splFixedArray: This iterator is supposed to be faster than plain arrays. Indeed, the code above requires 0.075'' on average:
$users = new SplFixedArray(100000);
for ($i=0;$i<100000;$i++) {
$users[$i] = array('id' => rand(),
'name' => 'default');
}
$ids=array();
$t = microtime(true);
foreach ($users as $key => $value) {
$ids[$key] = $value['id'];
}
echo microtime(true)-$t;
While the code below, where we use splFixedArray for both arrays, performed even faster, around 0.062'':
$users = new SplFixedArray(100000);
for ($i=0;$i<100000;$i++) {
$users[$i] = array('id' => rand(),
'name' => 'default');
}
$ids=new SplFixedArray(100000);
$t = microtime(true);
foreach ($users as $key => $value) {
$ids[$key] = $value['id'];
}
echo microtime(true)-$t;
So, the bottom line is that no, it's not the fastest way, but it's the best if you take into account all 3 parameters you posed in your initial question
array_map is probably more appropriate than array_walk here. Shouldn't alter the performance numbers much though.
$users = array();
$users[0]['id'] = 100;
$users[1]['id'] = 101;
$users[2]['id'] = 102;
$ids = array();
array_walk_recursive($users, function($val, $key) use(&$ids){
if ($key == 'id') $ids[] = $val;
});
print_r($ids);
Array
(
[0] => 100
[1] => 101
[2] => 102
)
array_walk()/array_map()etc - but this is just a loop in a different coat and in this case I suspect it would be less efficient than your simple, effective and easy to readforeach. So in short1.is it is the best way to doit?- probably, yes.2.is it possible to do it without loops.- definitely not, no. Even things that don't look like loops are just loops underneath (in the C++ source).3.is it is the fastest way....?- again, probably yes. I'd be surprised if it wasn't. But with something that will take <10usec, speed is not really a concern.1. No- care to expand? Or did you meanYes?