Whenever I use the \\u2028 character literal in my javascript source with the content type set to \"text/html; charset=utf-8\" I get a javascript parse errors.
Examp
Alright,to answer my own question.
Normally a JSON parser strips out these problem characters, because I was retrieving JSONP I wasn't using a JSON parser, in stead the browser tried to parse the JSON itself as soon as the callback was called.
The only way to fix it was to make sure the server never returns these characters when requesting a JSONP resource.
p.s. My question was about u2028, according to Douglas Crockford's json2 library all of the following characters can cause these problems:
'\u0000\u00ad\u0600-\u0604\u070f\u17b4\u17b5\u200c-\u200f\u2028-\u202f\u2060-\u206f\ufeff\ufff0-\uffff'
Could you just use \u2028
, instead of real character?, because U+2028 is unicode line seperator, browsers would think that as real line break character like \n
.
We cannot do like
x = "
"
Right? but we do x = "\n"
, so might be same concept.
Well, that makes sense, since you are telling the browser that the HTML and script are both using UTF-8, but then you specify a character that is not UTF-8 encoded. When you specify "charset=UTF-8", you are respoonsible for making sure the bytes transmitted to the browser are actually UTF-8. The web server and and browser will not do it for you in this situation.
Yes, it's a feature of the JavaScript language, documented in the ECMAScript standard (3rd edition section 7.3), that the U+2028 and U+2029 characters count as line endings. Consequently a JavaScript parser will treat any unencoded U+2028/9 character in the same way as a newline. Since you can't put a newline inside a string literal, you get a syntax error.
This is an unfortunate oversight in the design of JSON: it is not actually a proper subset of JavaScript. Raw U+2028/9 characters are valid in string literals in JSON, and will be accepted by JSON.parse
, but not so in JavaScript itself.
Hence it is only safe to generate JavaScript code using a JSON parser if you're sure it explicitly \u
-escapes those characters. Some do, some don't; many \u
-escape all non-ASCII characters, which avoids the problem.