json_encode() returns false when I pass it non-ascii characters

Question

When I run the following code json_encode() returns false (error). Does anyone know why?

$r = json_encode(chr(128));
var_dump($r);

The result is:

bool(false)

Answer 1

var_dump(chr(128));

The chr(128) is a malformed UTF-8 character. json_encode() can't encode malformed UTF-8 characters.

JSON standards

https://datatracker.ietf.org/doc/html/rfc8259#section-1

JSON can represent four primitive types (strings, numbers, booleans, and null) and two structured types (objects and arrays).

A string is a sequence of zero or more Unicode characters

json_encode() with flags.

If you are running PHP 7.3 or newer, you can try this code to throw the error message.

var_dump(json_encode(chr(128), JSON_THROW_ON_ERROR));

Result:

Fatal error: Uncaught JsonException: Malformed UTF-8 characters, possibly incorrectly encoded.

To prevent the errors, use JSON_INVALID_UTF8_IGNORE flag but this still can not encode invalid characters.

Example:

var_dump(json_encode(chr(128), JSON_INVALID_UTF8_IGNORE));

Result:

string '""' (length=2)

Cleanup string.

You may clean up the string to remove any invalid/malformed characters with this code.

$string = chr(128);
$string = mb_convert_encoding($string, 'UTF-8', 'UTF-8');
// then encode
var_dump(json_encode($string));// result is string '"?"' (length=3)