Test if value is larger or smaller then PHP_INT_MAX?

Question

How to check if double value is larger or smaller then PHP_INT_MAX:

<?php

$result = (double)PHP_INT_MAX;
$result += 100.0;

if ($result > PHP_INT_MAX) {
    echo "result is > then PHP_INT_MAX";
} else {
    echo "result is <= then PHP_INT_MAX";
}

And the output is:

result is <= then PHP_INT_MAX

which is incorrect.

Then I tried with:

<?php

$result = (double)PHP_INT_MAX;
$result += 100.0;
//$result -= 100.0;


if ((int) $result < 0) {
    echo "Solution 1: result is > then PHP_INT_MAX\n";
} else {
    echo "Solution 1: result is <= then PHP_INT_MAX\n";
}

if (!filter_var(
    $result,
    FILTER_VALIDATE_INT,
    array(
        'options' => array(
            'min_range' => 0,
            'max_range' => PHP_INT_MAX,
        )
    )
)) {
    echo "Solution 2: result is > then PHP_INT_MAX\n";
} else {
    echo "Solution 2: result is <= then PHP_INT_MAX\n";
}

And it seemed to work:

Solution 1: result is > then PHP_INT_MAX
Solution 2: result is > then PHP_INT_MAX

But then I did this (decrease $result for 100):

<?php

$result = (double)PHP_INT_MAX;
//$result += 100.0;
$result -= 100.0;


if ((int) $result < 0) {
    echo "Solution 1: result is > then PHP_INT_MAX\n";
} else {
    echo "Solution 1: result is <= then PHP_INT_MAX\n";
}

if (!filter_var(
    $result,
    FILTER_VALIDATE_INT,
    array(
        'options' => array(
            'min_range' => 0,
            'max_range' => PHP_INT_MAX,
        )
    )
)) {
    echo "Solution 2: result is > then PHP_INT_MAX\n";
} else {
    echo "Solution 2: result is <= then PHP_INT_MAX\n";
}

And output is wrong:

Solution 1: result is > then PHP_INT_MAX
Solution 2: result is > then PHP_INT_MAX

Does anybody knows what is going on, and how to reliably check the value against PHP_INT_MAX?

Tested on Linux, 64bit system, PHP 7.3.18.

Thanks!

UPDATE:

1. After reading all the comments, answers and after lots of experimentation:

In order to roughly detect when double value will go over PHP_INT_MAX you can use:

!filter_var(number_format($result, 0, '.', ''), FILTER_VALIDATE_INT)

$result >= PHP_INT_MAX

(int) $result < 0

(This one assumes that it is expected that $result will be >=0 to start with).

Checkout this example, where I am testing 20,000 numbers around PHP_INT_MAX:

<?php

function from_scientific_notation($value)
{
    $decimalNotation = rtrim(sprintf('%f', $value), '0');
    return rtrim($decimalNotation, '.');
}


for ($i = -10000; $i <= 10000; $i += 1) {
    $result = (double) PHP_INT_MAX + (double) $i;

    if (!filter_var(number_format($result, 0, '.', ''), FILTER_VALIDATE_INT)) {
        echo "S1:";
    }

    if ($result >= PHP_INT_MAX) {
        echo "S2:";
    }

    if ((int) $result < 0) {
        echo "S3:";
    }

    echo "\nPHP_INT_MAX + $i = " .from_scientific_notation($result) . "\n";
}

Output:

...

PHP_INT_MAX + -515 = 9223372036854774784

PHP_INT_MAX + -514 = 9223372036854774784

PHP_INT_MAX + -513 = 9223372036854774784
S1:S2:S3:
PHP_INT_MAX + -512 = 9223372036854775808
S1:S2:S3:
PHP_INT_MAX + -511 = 9223372036854775808
....

Please note loss of precision, the number PHP_INT_MAX - 512 is detected to be too larger then PHP_INT_MAX. Depending on the situation this inaccuracy might be or might not be acceptable.

2. But really, the only correct way of working with such large numbers would be to use libraries like bcmath or gmp (as mentioned in the answer below). But again - that might require a major refactor which might be out of the question.

Answer 1

Use the BC Math Functions. See comments for explanation.

<?php

$int = PHP_INT_MAX;

var_dump($int); // int(9223372036854775807)

$larger = bcadd($int, '100');

var_dump($larger); // string(19) "9223372036854775907"

// bccomp() will return 1 if first argument is larger than second.
var_dump(bccomp($larger, $int)); // int(1)

UPDATE

Updated demo, showing how bcadd() shows the intended output of your initial question. This outputs:

result is > then PHP_INT_MAX
result is <= then PHP_INT_MAX

Code:

<?php

$result = (double)PHP_INT_MAX;

// Add 100 to PHP_INT_MAX.
// Use sprintf() to prevent type juggling from formatting the float
// using scientific syntax, which is unsupported by bcadd().
$result = bcadd(sprintf('%f', $result), '100.0');

if (bccomp($result, sprintf("%f", (double)PHP_INT_MAX)) === 1) {
    echo "result is > then PHP_INT_MAX\n";
} else {
    echo "result is <= then PHP_INT_MAX\n";
}


// Subtract 200 from first result, making it less than PHP_INT_MAX.
$result = bcsub(sprintf('%f', $result), '200.0');

if (bccomp($result, sprintf("%f", (double)PHP_INT_MAX)) === 1) {
    echo "result is > then PHP_INT_MAX\n";
} else {
    echo "result is <= then PHP_INT_MAX\n";
}

Answer 2

Well the "easy" thing to do is:

filter_var(number_format($result, 0,'.',''), FILTER_VALIDATE_INT);

Where $result is a floating point number.

However the value will need to be significantly different than PHP_INT_MAX for this to work due to the way floating point numbers lose accuracy. Values close to PHP_INT_MAX are rather unpredictable.

There is a big red warning here you need to read carefully before working with floating point numbers but the point is the accuracy is not unlimited and the larger the number the smaller the accuracy gets.

The IEE 754 wikipedia page is also a good read, but the focus here is that a double precision floating point number has about 16 decimal digits so anything that has more than 16 significant digits will not be accurate. PHP_INT_MAX on 64 bit machines is about 19 digits so will lose about 3 significant digits of accuracy.