PHP decoding and encoding json with unicode characters

StackOverflow https://stackoverflow.com/questions/7381900

Question

I have some json I need to decode, alter and then encode without messing up any characters.

If I have a unicode character in a json string it will not decode. I'm not sure why since json.org says a string can contain: any-Unicode-character- except-"-or-\-or- control-character. But it doesn't work in python either.

{"Tag":"Odómetro"}

I can use utf8_encode which will allow the string to be decoded with json_decode, however the character gets mangled into something else. This is the result from a print_r of the result array. Two characters.

[Tag] => Odómetro

When I encode the array again I the character escaped to ascii, which is correct according to the json spec:

"Tag"=>"Od\u00f3metro"

Is there some way I can un-escape this? json_encode gives no such option, utf8_encode does not seem to work either.

Edit I see there is an unescaped_unicode option for json_encode. However it's not working as expected. Oh damn, it's only on php 5.4. I will have to use some regex as I only have 5.3.

$json = json_encode($array, JSON_UNESCAPED_UNICODE);
Warning: json_encode() expects parameter 2 to be long, string ...
Was it helpful?

Solution

Judging from everything you've said, it seems like the original Odómetro string you're dealing with is encoded with ISO 8859-1, not UTF-8.

Here's why I think so:

  • json_encode produced parseable output after you ran the input string through utf8_encode, which converts from ISO 8859-1 to UTF-8.
  • You did say that you got "mangled" output when using print_r after doing utf8_encode, but the mangled output you got is actually exactly what would happen by trying to parse UTF-8 text as ISO 8859-1 (ó is \x63\xb3 in UTF-8, but that sequence is ó in ISO 8859-1.
  • Your htmlentities hackaround solution worked. htmlentities needs to know what the encoding of the input string to work correctly. If you don't specify one, it assumes ISO 8859-1. (html_entity_decode, confusingly, defaults to UTF-8, so your method had the effect of converting from ISO 8859-1 to UTF-8.)
  • You said you had the same problem in Python, which would seem to exclude PHP from being the issue.

PHP will use the \uXXXX escaping, but as you noted, this is valid JSON.

So, it seems like you need to configure your connection to Postgres so that it will give you UTF-8 strings. The PHP manual indicates you'd do this by appending options='--client_encoding=UTF8' to the connection string. There's also the possibility that the data currently stored in the database is in the wrong encoding. (You could simply use utf8_encode, but this will only support characters that are part of ISO 8859-1).

Finally, as another answer noted, you do need to make sure that you're declaring the proper charset, with an HTTP header or otherwise (of course, this particular issue might have just been an artifact of the environment where you did your print_r testing).

OTHER TIPS

I have found following way to fix this issue... I hope this can help you.

json_encode($data,JSON_UNESCAPED_UNICODE|JSON_UNESCAPED_SLASHES);

JSON_UNESCAPED_UNICODE was added in PHP 5.4 so it looks like you need upgrade your version of PHP to take advantage of it. 5.4 is not released yet though! :(

There is a 5.4 alpha release candidate on QA though if you want to play on your development machine.

A hacky way of doing JSON_UNESCAPED_UNICODE in PHP 5.3. Really disappointed by PHP json support. Maybe this will help someone else.

$array = some_json();
// Encode all string children in the array to html entities.
array_walk_recursive($array, function(&$item, $key) {
    if(is_string($item)) {
        $item = htmlentities($item);
    }
});
$json = json_encode($array);

// Decode the html entities and end up with unicode again.
$json = html_entity_decode($rson);

try setting the utf-8 encoding in your page:

header('content-type:text/html;charset=utf-8');

this works for me:

$arr = array('tag' => 'Odómetro');
$encoded = json_encode($arr);
$decoded = json_decode($encoded);
echo $decoded->{'tag'};

$json = array('tag' => 'Odómetro'); // Original array
$json = json_encode($json); // {"Tag":"Od\u00f3metro"}
$json = json_decode($json); // Od\u00f3metro becomes  Odómetro
echo $json->{'tag'}; // Odómetro
echo utf8_decode($json->{'tag'}); // Odómetro

You were close, just use utf8_decode.

Try Using:

utf8_decode() and utf8_encode

To encode an array that contains special characters, ISO 8859-1 to UTF8. (If utf8_encode & utf8_decode is not what is working for you, this might be an option)

Everything that is in ISO-8859-1 should be converted to UTF8:

$utf8 = utf8_encode('이 감사의 마음을 전합니다!'); //contains UTF8 & ISO 8859-1 characters;    
$iso88591 = mb_convert_encoding($utf8, 'ISO-8859-1', 'UTF-8');
$data = $iso88591;

Encode should work after this: 

$encoded_data = json_encode($data);

Convert UTF-8 to & from ISO 8859-1

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top