How to check if PHP array is associative or sequen

Question

PHP treats all arrays as associative, so there aren't any built in functions. Can anyone recommend a fairly efficient way to check if an array contains only numeric keys?

Basically, I want to be able to differentiate between this:

$sequentialArray = array('apple', 'orange', 'tomato', 'carrot');

and this:

$assocArray = array('fruit1' => 'apple', 
                    'fruit2' => 'orange', 
                    'veg1' => 'tomato', 
                    'veg2' => 'carrot');

Answer 1

Here's the method I use:

function is_associative ( $a )
{
    return in_array(false, array_map('is_numeric', array_keys($a)));
}

assert( true === is_associative(array(1, 2, 3, 4)) );

assert( false === is_associative(array('foo' => 'bar', 'bar' => 'baz')) );

assert( false === is_associative(array(1, 2, 3, 'foo' => 'bar')) );

Note that this doesn't account for special cases like:

$a = array( 1, 2, 3, 4 );

unset($a[1]);

assert( true === is_associative($a) );

Sorry, can't help you with that. It's also somewhat performant for decently sized arrays, as it doesn't make needless copies. It is these little things that makes Python and Ruby so much nicer to write in... :P

Answer 2

As stated by the OP:

PHP treats all arrays as associative

it is not quite sensible (IMHO) to write a function that checks if an array is associative. So first thing first: what is a key in a PHP array?:

The key can either be an integer or a string.

That means there are 3 possible cases:

• Case 1. all keys are numeric / integers.
• Case 2. all keys are strings.
• Case 3. some keys are strings, some keys are numeric / integers.

We can check each case with the following functions.

Case 1: all keys are numeric / integers.

Note: This function returns true for empty arrays too.

//! Check whether the input is an array whose keys are all integers.
/*!
    \param[in] $InputArray          (array) Input array.
    \return                         (bool) \b true iff the input is an array whose keys are all integers.
*/
function IsArrayAllKeyInt($InputArray)
{
    if(!is_array($InputArray))
    {
        return false;
    }

    if(count($InputArray) <= 0)
    {
        return true;
    }

    return array_unique(array_map("is_int", array_keys($InputArray))) === array(true);
}

Case 2: all keys are strings.

Note: This function returns true for empty arrays too.

//! Check whether the input is an array whose keys are all strings.
/*!
    \param[in] $InputArray          (array) Input array.
    \return                         (bool) \b true iff the input is an array whose keys are all strings.
*/
function IsArrayAllKeyString($InputArray)
{
    if(!is_array($InputArray))
    {
        return false;
    }

    if(count($InputArray) <= 0)
    {
        return true;
    }

    return array_unique(array_map("is_string", array_keys($InputArray))) === array(true);
}

Case 3. some keys are strings, some keys are numeric / integers.

Note: This function returns true for empty arrays too.

//! Check whether the input is an array with at least one key being an integer and at least one key being a string.
/*!
    \param[in] $InputArray          (array) Input array.
    \return                         (bool) \b true iff the input is an array with at least one key being an integer and at least one key being a string.
*/
function IsArraySomeKeyIntAndSomeKeyString($InputArray)
{
    if(!is_array($InputArray))
    {
        return false;
    }

    if(count($InputArray) <= 0)
    {
        return true;
    }

    return count(array_unique(array_map("is_string", array_keys($InputArray)))) >= 2;
}

It follows that:

• If the value is not an array, all 3 functions return false.
• If the value is an empty array, all 3 functions return true
  (which is by definition, as in "the empty set is a subset of any set A because all its elements belong to A").
• If the value is a non-empty array, exactly 1 function returns true.

---

Now, for an array to be a "genuine" array that we are all accustomed to, meaning:

• Its keys are all numeric / integers.
• Its keys are sequential (i.e. increasing by step 1).
• Its keys start from zero.

We can check with the following function.

Case 3a. keys are numeric / integers, sequential, and zero-based.

Note: This function returns true for empty arrays too.

//! Check whether the input is an array whose keys are numeric, sequential, and zero-based.
/*!
    \param[in] $InputArray          (array) Input array.
    \return                         (bool) \b true iff the input is an array whose keys are numeric, sequential, and zero-based.
*/
function IsArrayKeyNumericSequentialZeroBased($InputArray)
{
    if(!is_array($InputArray))
    {
        return false;
    }

    if(count($InputArray) <= 0)
    {
        return true;
    }

    return array_keys($InputArray) === range(0, count($InputArray) - 1);
}

Caveats / Pitfalls (or, even more peculiar facts about array keys in PHP)

Integer keys

The keys for these arrays are integers:

array(0 => "b");
array(13 => "b");
array(-13 => "b");          // Negative integers are also integers.
array(0x1A => "b");         // Hexadecimal notation.

String keys

The keys for these arrays are strings:

array("fish and chips" => "b");
array("" => "b");                                   // An empty string is also a string.
array("[e-mail]" => "b");    // Strings may contain non-alphanumeric characters.
array("stack\t\"over\"\r\nflow's cool" => "b");     // Strings may contain special characters.
array('$tα€k↔øv∈rflöw⛄' => "b");                    // Strings may contain all kinds of symbols.
array("functіon" => "b");                           // You think this looks fine? Think again! (see https://stackoverflow.com/q/9246051/1402846)
array("ま말轉转ДŁ" => "b");                         // How about Japanese/Korean/Chinese/Russian/Polish?
array("fi\x0sh" => "b");                            // Strings may contain null characters.
array(file_get_contents("https://www.google.com/images/nav_logo114.png") => "b");   // Strings may even be binary!

Integer keys that look like strings

If you think the key in array("13" => "b") is a string, you are wrong. From the doc here:

Strings containing valid integers will be cast to the integer type. E.g. the key "8" will actually be stored under 8. On the other hand "08" will not be cast, as it isn't a valid decimal integer.

For example, the key for these arrays are integers:

array("13" => "b");
array("-13" => "b");                        // Negative, ok.

But the key for these arrays are strings:

array("13." => "b");
array("+13" => "b");                        // Positive, not ok.
array("-013" => "b");
array("0x1A" => "b");                       // Not converted to integers even though it's a valid hexadecimal number.
array("013" => "b");                        // Not converted to integers even though it's a valid octal number.
array("18446744073709551616" => "b");       // Not converted to integers as it can't fit into a 64-bit integer.

What's more, according to the doc,

The size of an integer is platform-dependent, although a maximum value of about two billion is the usual value (that's 32 bits signed). 64-bit platforms usually have a maximum value of about 9E18, except for Windows, which is always 32 bit. PHP does not support unsigned integers.

So the key for this array may or may not be an integer - it depends on your platform.

array("60000000000" => "b");                // Array key could be integer or string, it can fit into a 64-bit (but not 32-bit) integer.

Even worse, PHP tends to be buggy if the integer is near the 2³¹ = 2,147,483,648 boundary (see bug 51430, bug 52899). For example, on my local environment (PHP 5.3.8 on XAMPP 1.7.7 on Windows 7), var_dump(array("2147483647" => "b")) gives

array(1) {
    [2147483647]=>
    string(1) "b"
}   

but on this live demo on codepad (PHP 5.2.5), the same expression gives

array(1) {
    ["2147483647"]=>
    string(1) "b"
}

So the key is an integer in one environment but a string in another, even though 2147483647 is a valid signed 32-bit integer.

Answer 3

I've used both array_keys($obj) !== range(0, count($obj) - 1) and array_values($arr) !== $arr (which are duals of each other, although the second is cheaper than the first) but both fail for very large arrays.

This is because array_keys and array_values are both very costly operations (since they build a whole new array of size roughly that of the original).

The following function is more robust than the methods provided above:

function array_type( $obj ){
    $last_key = -1;
    $type = 'index';
    foreach( $obj as $key => $val ){
        if( !is_int( $key ) || $key < 0 ){
            return 'assoc';
        }
        if( $key !== $last_key + 1 ){
            $type = 'sparse';
        }
        $last_key = $key;
    }
    return $type;
}

Also note that if you don't care to differentiate sparse arrays from associative arrays you can simply return 'assoc' from both if blocks.

Finally, while this might seem much less "elegant" than a lot of "solutions" on this page, in practice it is vastly more efficient. Almost any associative array will be detected instantly. Only indexed arrays will get checked exhaustively, and the methods outlined above not only check indexed arrays exhaustively, they duplicate them.

Answer 4

I know it's a bit pointless adding an answer to this huge queue, but here's a readable O(n) solution that doesn't require duplicating any values:

function isNumericArray($array) {
    $count = count($array);
    for ($i = 0; $i < $count; $i++) {
        if (!isset($array[$i])) {
            return FALSE;
        }
    }
    return TRUE;
}

Rather than check the keys to see if they are all numeric, you iterate over the keys that would be there for a numeric array and make sure they exist.

Answer 5

One more fast from source. Fit encoding of json_encode (and bson_encode). So has javascript Array compliance.

function isSequential($value){
    if(is_array($value) || ($value instanceof \Countable && $value instanceof \ArrayAccess)){
        for ($i = count($value) - 1; $i >= 0; $i--) {
            if (!isset($value[$i]) && !array_key_exists($i, $value)) {
                return false;
            }
        }
        return true;
    } else {
        throw new \InvalidArgumentException(
            sprintf('Data type "%s" is not supported by method %s', gettype($value), __METHOD__)
        );
    }
}

Answer 6

After some local benchmarking, debugging, compiler probing, profiling, and abusing 3v4l.org to benchmark across more versions (yes, I got a warning to stop) and comparing against every variation I could find...

I give you an organically derived best-average-worst-case scenario associative array test function that is at worst roughly as good as or better than all other average-case scenarios.

/**
 * Tests if an array is an associative array.
 *
 * @param array $array An array to test.
 * @return boolean True if the array is associative, otherwise false.
 */
function is_assoc(array &$arr) {
    // don't try to check non-arrays or empty arrays
    if (FALSE === is_array($arr) || 0 === ($l = count($arr))) {
        return false;
    }

    // shortcut by guessing at the beginning
    reset($arr);
    if (key($arr) !== 0) {
        return true;
    }

    // shortcut by guessing at the end
    end($arr);
    if (key($arr) !== $l-1) {
        return true;
    }

    // rely on php to optimize test by reference or fast compare
    return array_values($arr) !== $arr;
}

From https://3v4l.org/rkieX:

<?php

// array_values
function method_1(Array &$arr) {
    return $arr === array_values($arr);
}

// method_2 was DQ; did not actually work

// array_keys
function method_3(Array &$arr) {
    return array_keys($arr) === range(0, count($arr) - 1);
}

// foreach
function method_4(Array &$arr) {
    $idx = 0;
    foreach( $arr as $key => $val ){
        if( $key !== $idx )
            return FALSE;
        ++$idx;
    }
    return TRUE;
}

// guessing
function method_5(Array &$arr) {
    global $METHOD_5_KEY;
    $i = 0;
    $l = count($arr)-1;

    end($arr);
    if ( key($arr) !== $l )
        return FALSE;

    reset($arr);
    do {
        if ( $i !== key($arr) )
            return FALSE;
        ++$i;
        next($arr);
    } while ($i < $l);
    return TRUE;
}

// naieve
function method_6(Array &$arr) {
    $i = 0;
    $l = count($arr);
    do {
        if ( NULL === @$arr[$i] )
            return FALSE;
        ++$i;
    } while ($i < $l);
    return TRUE;
}

// deep reference reliance
function method_7(Array &$arr) {
    return array_keys(array_values($arr)) === array_keys($arr);
}


// organic (guessing + array_values)
function method_8(Array &$arr) {
    reset($arr);
    if ( key($arr) !== 0 )
        return FALSE;

    end($arr);
    if ( key($arr) !== count($arr)-1 )
        return FALSE;

    return array_values($arr) === $arr;
}

function benchmark(Array &$methods, Array &$target, $expected){    
    foreach($methods as $method){
        $start = microtime(true);
        for ($i = 0; $i < 2000; ++$i) {
            //$dummy = call_user_func($method, $target);
            if ( $method($target) !== $expected ) {
                echo "Method $method is disqualified for returning an incorrect result.\n";
                unset($methods[array_search($method,$methods,true)]);
                $i = 0;
                break;
            }
        }
        if ( $i != 0 ) {
            $end = microtime(true);
            echo "Time taken with $method = ".round(($end-$start)*1000.0,3)."ms\n";
        }
    }
}



$true_targets = [
    'Giant array' => range(0, 500),
    'Tiny array' => range(0, 20),
];


$g = range(0,10);
unset($g[0]);

$false_targets = [
    'Large array 1' => range(0, 100) + ['a'=>'a'] + range(101, 200),
    'Large array 2' => ['a'=>'a'] + range(0, 200),
    'Tiny array' => range(0, 10) + ['a'=>'a'] + range(11, 20),
    'Gotcha array' => $g,
];

$methods = [
    'method_1',
    'method_3',
    'method_4',
    'method_5',
    'method_6',
    'method_7',
    'method_8'
];


foreach($false_targets as $targetName => $target){
    echo "==== Benchmark using $targetName expecing FALSE ====\n";
    benchmark($methods, $target, false);
    echo "\n";
}
foreach($true_targets as $targetName => $target){
    echo "==== Benchmark using $targetName expecting TRUE ====\n";
    benchmark($methods, $target, true);
    echo "\n";
}