SRP_JSON converts text differently

wa_laan · October 2018

I have some code which uses SRP_JSON to convert text to JSON
In the past it converts the text correctly, now it seems something has changed since the last update of SRP utilities

I have an example of my code here:

If SRP_JSON(RootHandle, "NEW") then
	If SRP_JSON(ArrayHandle, "NEW", "ARRAY") then
		If SRP_JSON(ObjectId, "NEW") then
			SRP_JSON(ObjectId, 'setValue', 'Name', 'André', 'STRING')
			SRP_JSON(ArrayHandle, 'ADD', ObjectId)
			SRP_JSON(ObjectId, 'RELEASE')
		end
		
		SRP_JSON(RootHandle, 'SET', "Login", ArrayHandle)
		SRP_JSON(ArrayHandle, 'RELEASE')
	end
	
	// Now get the actual JSON
	rv = SRP_JSON(RootHandle, "STRINGIFY", "Fast")
	
	// All done with the root object
	SRP_JSON(RootHandle, "RELEASE")
end

this results in :
{"Login":[{"Name":"Andr\uffffffe9"}]}

while it used to result in :
{"Login":[{"Name":"André"}]}
It seems this has something to do with some encoding issue

Is it possible to make it optionable which encoding is used?

r.od.akker · October 2018

Even if UTF-16 was intended by SRP_JSON, the output should have been
{"Login":[{"Name":"Andr\u00e9"}]}

So converting é to \uffffffe9 is invalid anyways, imho.

KevinFournier · October 2018

SRP Utilities 2.0.3 RC3 fixes this. It was never meant to output \u sequences for anything except control characters, so that was the first problem. The second problem, as @r.od.akker point out, was the 32-bit length sequence instead of the 16-bit sequence. Both have been corrected.

In short, it should now output as you preferred without having to ask it to.

wa_laan · October 2018

Thanks Kevin! We installed the new RC and the problem seems to be fixed

SRP_JSON converts text differently

Comments