Fastest way to check if a string is JSON in PHP?

Question

I need a really, really fast method of checking if a string is JSON or not. I feel like this is not the best way:

function isJson($string) {
    return ((is_string($string) &&
            (is_object(json_decode($string)) ||
            is_array(json_decode($string))))) ? true : false;
}

Any performance enthusiasts out there want to improve this method?

Answer 1

The custom function

function custom_json_decode(&$contents=NULL, $normalize_contents=true, $force_array=true){

    //---------------decode contents---------------------

    $decoded_contents=NULL;

    if(is_string($contents)){

        $decoded_contents=json_decode($contents,$force_array);

    }

    //---------------normalize contents---------------------

    if($normalize_contents===true){

        if(is_string($decoded_contents)){

            if($decoded_contents==='NULL'||$decoded_contents==='null'){

                $contents=NULL;
            }
            elseif($decoded_contents==='FALSE'||$decoded_contents==='false'){

                $contents=false;
            }
        }
        elseif(!is_null($decoded_contents)){

            $contents=$decoded_contents;
        }
    }
    else{

        //---------------validation contents---------------------

        $contents=$decoded_contents;
    }

    return $contents;
}

Cases

$none_json_str='hello';

//------------decoding a none json str---------------

$contents=custom_json_decode($none_json_str); // returns 'hello'

//------------checking a none json str---------------

custom_json_decode($none_json_str,false);

$valid_json=false;

if(!is_null($none_json_str)){

    $valid_json=true;

}

Resources

https://gist.github.com/rafasashi/93d06bae83cc1a1f440b

Answer 2

I don't know about performance or elegance of my solution, but it's what I'm using:

if (preg_match('/^[\[\{]\"/', $string)) {
    $aJson = json_decode($string, true);
    if (!is_null($aJson)) {
       ... do stuff here ...
    }
}

Since all my JSON encoded strings start with {" it suffices to test for this with a RegEx. I'm not at all fluent with RegEx, so there might be a better way to do this. Also: strpos() might be quicker.

Just trying to give in my tuppence worth.

P.S. Just updated the RegEx string to /^[\[\{]\"/ to also find JSON array strings. So it now looks for either [" or {" at the beginning of the string.

Answer 3

Another simple way

function is_json($str)
{
    return is_array(json_decode($str,true));
}

Answer 4

Hi here's a little snippet from my library, in this first condition I'm just checking if the data is json then return it if correctly decoded, please note the substr usage for performance ( I haven't seen yet any json file not begining neither by { or [

$input=trim($input);
if ((substr($input, 0, 1) == '{' && substr($input, -1) == '}') or (substr($input, 0, 1) == '[' && substr($input, -1) == ']')) {
    $output = json_decode($input, 1);
    if (in_array(gettype($output),['object','array'])) {
        #then it's definitely JSON
    }
}

Answer 5

Freshly-made function for PHP 5.2 compatibility, if you need the decoded data on success:

function try_json_decode( $json, & $success = null ){
  // non-strings may cause warnings
  if( !is_string( $json )){
    $success = false;
    return $json;
  }

  $data = json_decode( $json );

  // output arg
  $success =

    // non-null data: success!
    $data !==  null  ||

    // null data from 'null' json: success!
    $json === 'null' ||

    // null data from '  null  ' json padded with whitespaces: success!
    preg_match('/^\s*null\s*$/', $json );

  // return decoded or original data
  return $success ? $data : $json;
}

Usage:

$json_or_not = ...;

$data = try_json_decode( $json_or_not, $success );

if( $success )
     process_data( $data );
else what_the_hell_is_it( $data );

Some tests:

var_dump( try_json_decode( array(), $success ), $success );
// ret = array(0){}, $success == bool(false)

var_dump( try_json_decode( 123, $success ), $success );
// ret = int(123), $success == bool(false)

var_dump( try_json_decode('      ', $success ), $success );
// ret = string(6) "      ", $success == bool(false)

var_dump( try_json_decode( null, $success ), $success );
// ret = NULL, $success == bool(false)

var_dump( try_json_decode('null', $success ), $success );
// ret = NULL, $success == bool(true)

var_dump( try_json_decode('  null  ', $success ), $success );
// ret = NULL, $success == bool(true)

var_dump( try_json_decode('  true  ', $success ), $success );
// ret = bool(true), $success == bool(true)

var_dump( try_json_decode('  "hello"  ', $success ), $success );
// ret = string(5) "hello", $success == bool(true)

var_dump( try_json_decode('  {"a":123}  ', $success ), $success );
// ret = object(stdClass)#2 (1) { ["a"]=> int(123) }, $success == bool(true)

Answer 6

Using json_decode to "probe" it might not actually be the fastest way. If it's a deeply nested structure, then instantiating a lot of objects of arrays to just throw them away is a waste of memory and time.

So it might be faster to use preg_match and the RFC4627 regex to also ensure validity:

  // in JS:
  var my_JSON_object = !(/[^,:{}\[\]0-9.\-+Eaeflnr-u \n\r\t]/.test(
         text.replace(/"(\\.|[^"\\])*"/g, '')));

The same in PHP:

  return !preg_match('/[^,:{}\\[\\]0-9.\\-+Eaeflnr-u \\n\\r\\t]/',
       preg_replace('/"(\\.|[^"\\\\])*"/', '', $json_string));

Not enough of a performance enthusiast to bother with benchmarks here however.