Checking for empty arrays: count vs empty

问题:

This question on 'How to tell if a PHP array is empty' had me thinking of this question

Is there a reason that count should be used instead of empty when determining if an array is empty or not?

My personal thought would be if the 2 are equivalent for the case of empty arrays you should use empty because it gives a boolean answer to a boolean question. From the question linked above, it seems that count($var) == 0 is the popular method. To me, while technically correct, makes no sense. E.g. Q: $var, are you empty? A: 7. Hmmm...

Is there a reason I should use count == 0 instead or just a matter of personal taste?

As pointed out by others in comments for a now deleted answer, count will have performance impacts for large arrays because it will have to count all elements, whereas empty can stop as soon as it knows it isn't empty. So, if they give the same results in this case, but count is potentially inefficient, why would we ever use count($var) == 0?

回答1:

I generally use empty. Im not sure why people would use count really - If the array is large then count takes longer/has more overhead. If you simply need to know whether or not the array is empty then use empty.

回答2:

I was curious to see which one was actually faster so I made a simple script to benchmark those functions.

<?php

function benchmark($name, $iterations, $action){
    $time=microtime(true);
    for($i=0;$i<=$iterations;++$i){
        $action();
    }
    echo $name . ' ' . round(microtime(true)-$time, 6) . "\n";
}

$iterations = 1000000;
$x = array();
$y = range(0, 10000000);
$actions = array(
    "Empty empty()" => function() use($x){
        empty($x);
    },
    "Empty count()" => function() use($x){
        count($x);
    },
    "Full empty()" => function() use($y){
        empty($y);
    },
    "Full count()" => function() use($y){
        count($y);
    },
    ############
    "IF empty empty()" => function() use($x){
        if(empty($x)){ $t=1; }
    },
    "IF empty count()" => function() use($x){
        if(count($x)){ $t=1; }
    },
    "IF full empty()" => function() use($y){
        if(empty($y)){ $t=1; }
    },
    "IF full count()" => function() use($y){
        if(count($y)){ $t=1; }
    },
    ############
    "OR empty empty()" => function() use($x){
        empty($x) OR $t=1;
    },
    "OR empty count()" => function() use($x){
        count($x) OR $t=1;
    },
    "OR full empty()" => function() use($y){
        empty($y) OR $t=1;
    },
    "OR full count()" => function() use($y){
        count($y) OR $t=1;
    },
    ############
    "IF/ELSE empty empty()" => function() use($x){
        if(empty($x)){ $t=1; } else { $t=2; }
    },
    "IF/ELSE empty count()" => function() use($x){
        if(count($x)){ $t=1; } else { $t=2; }
    },
    "IF/ELSE full empty()" => function() use($y){
        if(empty($y)){ $t=1; } else { $t=2; }
    },
    "IF/ELSE full count()" => function() use($y){
        if(count($y)){ $t=1; } else { $t=2; }
    },
    ############
    "( ? : ) empty empty()" => function() use($x){
        $t = (empty($x) ? 1 : 2);
    },
    "( ? : ) empty count()" => function() use($x){
        $t = (count($x) ? 1 : 2);
    },
    "( ? : ) full empty()" => function() use($y){
        $t = (empty($y) ? 1 : 2);
    },
    "( ? : ) full count()" => function() use($y){
        $t = (count($y) ? 1 : 2);
    }
);

foreach($actions as $name => $action){
    benchmark($name, $iterations, $action);
}
//END

Since I was doing it I also tried to check the performance doing operations that would normally be associated with count()/empty()

Using PHP 5.4.39:

Empty empty() 0.118691
Empty count() 0.218974
Full empty() 0.133747
Full count() 0.216424
IF empty empty() 0.166474
IF empty count() 0.235922
IF full empty() 0.120642
IF full count() 0.248273
OR empty empty() 0.123875
OR empty count() 0.258665
OR full empty() 0.157839
OR full count() 0.224869
IF/ELSE empty empty() 0.167004
IF/ELSE empty count() 0.263351
IF/ELSE full empty() 0.145794
IF/ELSE full count() 0.248425
( ? : ) empty empty() 0.169487
( ? : ) empty count() 0.265701
( ? : ) full empty() 0.149847
( ? : ) full count() 0.252891

Using HipHop VM 3.6.1 (dbg)

Empty empty() 0.210652
Empty count() 0.212123
Full empty() 0.206016
Full count() 0.204722
IF empty empty() 0.227852
IF empty count() 0.219821
IF full empty() 0.220823
IF full count() 0.221397
OR empty empty() 0.218813
OR empty count() 0.220105
OR full empty() 0.229118
OR full count() 0.221787
IF/ELSE empty empty() 0.221499
IF/ELSE empty count() 0.221274
IF/ELSE full empty() 0.221879
IF/ELSE full count() 0.228737
( ? : ) empty empty() 0.224143
( ? : ) empty count() 0.222459
( ? : ) full empty() 0.221606
( ? : ) full count() 0.231288

Conclusions if you're using PHP:

1. empty() is much much faster than count() in both scenarios, with an empty and populated array

2. count() performs the same with a full or empty array.

3. Doing a simple IF or just a Boolean operation is the same.

4. IF/ELSE is very slightly more efficient than ( ? : ). Unless you're doing billions of iterations with expressions in the middle it is completely insignificant.

Conclusions if you're using HHVM:

1. empty() is a teeny-weeny bit faster than count() but insignificantly so.

   [ The rest is the same as in PHP ]

In conclusion of the conclusion, if you just need to know if the array is empty always use empty();

This was just a curious test simply done without taking many things into account. It is just a proof of concept and might not reflect operations in production.

回答3:

I think it's only personal preference. Some people might say empty is faster (e.g. http://jamessocol.com/projects/count_vs_empty.php) while others might say count is better since it was originally made for arrays. empty is more general and can be applied to other types.

php.net gives the following warning for count though :

count() may return 0 for a variable that isn't set, but it may also return 0 for a variable that has been initialized with an empty array. Use isset() to test if a variable is set.

In other words, if the variable is not set, you will get a notice from PHP saying it's undefined. Therefore, before using count, it would be preferable to check the variable with isset. This is not necessary with empty.

回答4:

Is there a reason that count should be used instead of empty when determining if an array is empty or not?

There is, when you need to do something on non-empty array knowing it's size:

if( 0 < ( $cnt = count($array) ) )
{
 echo "Your array size is: $cnt";
}
else
 echo "Too bad, your array is empty :(";

But I wouldn't recommend using count, unless you are 100% sure, that what you are counting is an array. Lately I have been debugging code, where on error function was returning FALSE instead of empty array, and what I discovered was:

var_dump(count(FALSE));

output:

int 1

So since then I am using empty or if(array() === $array) to be sure that I have array that is empty.

回答5:

Alternatively, you can cast the variable as a boolean (implicitly or explicitly):

if( $value )
{
  // array is not empty
}

if( (bool) $value )
{
  // array is still not empty
}

This method does generate an E_NOTICE if the variable is not defined, similarly to count().

For more information, see the PHP Manual page on type comparisons.

回答6:

count() seems to work better with array-like interfaces that implement ArrayAccess/Countable. empty() returns true for these kinds of objects even if they have no elements. Typically these classes will implement the Countable interface, so if the question is "Does this collection contain elements?" without making an assumption about the implementation, then count() is a better option.

回答7:

My personal preference is more for coding elegance (in relation to my specific use-case). I agree with Dan McG inasmuch that count() isn't responding with the correct datatype (in this case boolean) for the test in question forcing the developer to write more code to fill an 'if' statement.

Whether this has any significant impact on performance is only debatable for extremely large arrays (which you probably won't have enough memory allocation for anyway in most setups).

Particularly when it comes to PHP's $_POST array, it seems much more "logical" in my opinion to write/see:

if ( !empty ( $_POST ) ) {
    // deal with postdata
}

回答8:

Hope this might help someone even though it has already been answered (and debated some what). In my own scenario, I know all my arrays all have 7 elements (checks were made earlier in my code) and I am performing an array_diff which of course returns an array of zero when equal.

I had 34 sec for count and 17 sec for empty. Both give me the same calculations so my code is still fine.

However you can also try the == or === as in PHP - Check if two arrays are equal. The best I would say is try count vs empty vs == empty array, then see which gives your own best perfs. In my case count was the slowest so I am using empty now... will be checking serialize next

回答9:

There is no strong reason to prefer count($myArray) == 0 over empty($myArray). They have identical semantics. Some might find one more readable than the other. One might perform marginally better than the other but it's not likely to be a significant factor in the vast majority of php applications. For all practical purposes, the choice is a matter of taste.

回答10:

Sometimes using empty is a must. For example this code:

$myarray = array();

echo "myarray:"; var_dump($myarray); echo "<br>";
echo "case1 count: ".count($myarray)."<br>";
echo "case1 empty: ".empty($myarray)."<br>";

$glob = glob('sdfsdfdsf.txt');

echo "glob:"; var_dump($glob); echo "<br>";
echo "case2 count: ".count($glob)."<br>";
echo "case2 empty: ".empty($glob);

If you run this code like this: http://phpfiddle.org/main/code/g9x-uwi

You get this output:

myarray:array(0) { } 
case1 count: 0
case1 empty: 1

glob:bool(false) 
case2 count: 1
case2 empty: 1

So if you count the empty glob output you get wrong output. You should check for emptiness.

From glob documentation:

Returns an array containing the matched files/directories, an empty array if no file matched or FALSE on error.
Note: On some systems it is impossible to distinguish between empty match and an error.

Also check this question: Why count(false) return 1?

回答11:

I remade my mind guys, thanks.

Ok, there is no difference between the usage of empty and count. Technically, count should be used for arrays, and empty could be used for arrays as well as strings. So in most cases, they are interchangeable and if you see the php docs, you will see the suggestion list of count if you are at empty and vice versa.

回答12:

Since a variable parsed as negative would return int(1) with count()

I prefer ($array === [] || !$array) to test for an empty array.

Yes, we should expect an empty array, but we shouldn't expect a good implementation on functions without enforced return types.

Examples with count()

var_dump(count(0));
> int(1)
var_dump(count(false));
> int(1)