Convert HTML numbered entities in php to unicode for use on iPhone

I'm creating a web service to transfer json to an iPhone app. I'm using json-framework to receive the json, and that works great because it automatically decodes things like "\u2018". The problem I'm running into is there doesn't seem to be a comprehensive way to get all the characters in one fell swoop.

For example html_entity_decode() gets most things, but it leaves behind stuff like ‘ (‘). In order to catch these entities and convert them to something json-framework can use (e.g., \u2018), I'm using this code to convert the &# to \u, convert the numbers to hex, and then strip the ending semicolon.

function func($matches) {
  return "\u" . dechex($matc开发者_如何学Pythonhes[1]);
}
$json = preg_replace_callback("/&#(\d{4});/", "func", $json);

This is working for me at the moment, but it just doesn't feel right. It seems like I'm surely missing some characters that are going to come back to haunt me later.

Does anyone see flaws in this approach? Can anyone think of characters this approach will miss?

Any help would be most appreciated!

From where are you getting this HTML-encoded input? If you're scraping a web page you should be using an HTML parser, which will decode both entity and character references for you. If you are getting them in form input data, you've got a problem with encodings (make sure to serve the page containing the form as UTF-8 to avoid this).

If you must convert an HTML-encoded stretch of literal text to JSON, you should do it by HTML-decoding first then JSON-encoding, rather than attempting to go straight to JSON format (which will fail for a bunch of other characters that need escaping). Use the built-in decoder and encoder functions rather than trying to create JSON-encoded characters like \u.... yourself (as there are traps there).

$html= 'abc " def Ӓ ghi ሴ jkl \n mno';
$raw= html_entity_decode($html, ENT_COMPAT, 'utf-8');
$json= json_encode($raw);

"abc \" def \u04d2 ghi \u1234 jkl \\n mno"

‘ is a decimal numbered entity, while I believe \u2018 is a hexadecimal representation. HTML also supports hexadecimal numbered entities (e.g., ‘), but once you've found # as the entity prefix you're looking at either decimal or hex. There are also named entities (e.g., &) but it doesn't sound like you need to cover those cases in your code.

$html_escape = ""Love sex magic rise" & 尹真希 ‘";
$utf8 = mb_convert_encoding($html_escape, 'UTF-8', 'HTML-ENTITIES');
echo json_encode(array(
    "title" => $utf8
));

// {"title":"\"Love sex magic rise\" & \u5c39\u771f\u5e0c \u2018"}

This work well for me