Re: Request: ?json_string, similar to ?js_string

Tuesday, December 20, 2011, 11:45:50 AM, Eirik Lygre wrote:

> Hi,
>
> FreeMarker today comes with the built-in "?js_string", which
> escapes strings according to JavaScript rules. From the FreeMarker docs:
>
> Escapes the string with the escaping rules of JavaScript language
> string literals, so it is safe to insert the value into a string
> literal. Both quotation mark (") and apostrophe-quoate (') are
> escaped. Starting from FreeMarker 2.3.1, it also escapes > as \> (to
> avoid </script>). Furthermore, all characters under UCS code point
> 0x20, that has no dedicated escape sequence in JavaScript language,
> will be replaced with hexadecimal escape (\xXX). (Of course,
> according the JavaScript language string literal syntax, backslash
> (\) will be escaped too, line-feed will be escaped as \n, ...etc.)
>
> However, this is not valid json.

Because each languages has to have its own string escaping rules...

> In particular, the native JSON javascript object in newer browsers
> do not support all those escape characters. An example of problems
> can be found on stackoverflow:
> http://stackoverflow.com/questions/2275359/jquery-single-quote-in-json-response
>
> In case a single-quote pops up, ?js_script will escape it which
> again leads to invalid JSON. The hotfix for it is:
> ${variable?js_string?replace("\'", "\'")} and if you really want to
> be picky: ${variable?js_string?replace("\'", "\'")?replace("\>",">")}
>
> The formal grammar for legal json escapes are shown in the spec, at
> http://es5.github.com/#x15.12.1. The excerpt below shows legal
> escape characters (in additon, any character can be escaped using its numeric unicode sequence):
>
> JSONEscapeCharacter :: one of
> " / \ b f n r t
>
> The questions, then, would be something like this:
>
> a) Should FreeMarker support ?json_string, similar to js_string?

Sure. (Actually, someone earlier wanted to dump whole graphs of
FreeMarker variables into JSON... a not so trivial task.)

> b) Is there interest in a patch, or is it better that those who
> know the code inside-out create it in the first place?

The only problem with contributing this is that it needs
disproportional amount of bureaucracy on your side (signing +
physically posting CLA). Unless, you plan to do more contributions in
the future, of course. So I assume I have to add this to my TODO for
the next release.

-- 
Best regards,
 Daniel Dekany


------------------------------------------------------------------------------
Write once. Port to many.
Get the SDK and tools to simplify cross-platform app development. Create 
new or port existing apps to sell to consumers worldwide. Explore the 
Intel AppUpSM program developer opportunity. appdeveloper.intel.com/join
http://p.sf.net/sfu/intel-appdev