How to: what is the last update on Json/array conversion?

[trevix]

Hello.
As usual, when some problem emerges, I realise that I may have not chosen the best solution, or that OS/LC version updates have surpassed me, leaving me with wrong solutions.

In this case, my problem is how to do "json to array" and "array to json" conversion between my standalone and the data sent and returned by my FileMaker Server and by Microsoft Azure.
The data exchanged are multidimensional array, that may contain numbers, text, image data and so on.

As of now, in the inclusion of the standalone, I have the "JSON library". But I also use a substack as library with the " FASTJSON v1.1.1" by Bob Hall.
I have to use the JSON library (jsonImport) for converting Json to array

Code: Select all

put textDecode(pJSON,"UTF8") into tJSON
put JsonImport(tJSON) into tArray --JSON library in inclusion

while using the FASTJSON lib for converting Array to json.

Code: Select all

 put ArrayToJSON(pArray) into tArrayJson --convert to json in fastjson library
put textEncode(tArrayJson, "utf-8") into tArrayJson

It wasn't easy but, with some fix here and there, it worked, until I started dealing with image content.
"jsonImport" got VERY slow dealing with a 150k image data in the content of the array, and sometime things go wrong.

And then there is mergJSON, with I never dealt with, but may be I should?

So, the point is: is there a unique solution for dealing with JSON in LC, fast and reliable for any situation, or this is a wishful thinking and I need to always patch things here and there?

Sorry if my post demonstrates naivety.
Thanks

[prometheus]

Hi Trevix

Have you tried this one?
https://github.com/Ferruslogic/PhotonJSON

I've got success with this one when others has failed

[jacque]

I think I know what the problem is but I'm not sure how to solve it. Images are binary so using textDecode on them will mangle them. You'd need a way to identify images and leave their binary content alone while still decoding the text portions.

[trevix]

mmmmh...
I could compress and decompress the binary data?

[FourthWorld]

Is the data in the JSON truly binary, or Base64?

If compressing the JSON is an option it'll help for transport, but not decoding.

If you have control over how the FMP server packs the data, could externalizing the images be a solution? Then the JSON would include only a URL, which could then be used to obtain the image in truly binary form. Image-format-specific compression is hard to beat, and it keeps the JSON light and nimble.

[jacque]

I have no idea if this would work, but maybe encoding the binary to UTF8 would decode it correctly later? Otherwise I think I'd do what Richard suggests and use base64, which would require decoding the base64 before display but wouldn't interfere with UTF8...I think.

Using URLs instead of the actual binary is probably the best way.

[FourthWorld]

To clarify, I'm not suggesting Base64, just asking if that's what's being used in FMP's JSON. JSON is a plain text format, so the images are unlikely represented in binary within it.

My suggestion, more a general preference without knowing the specifics of this DB, is to consider not putting binary data like images in the DB at all, instead leaving them in a separate HTTP-accesible folder and putting only the file path in the DB.

[trevix]

Saving the binary data in a folder is something that eventually will have to do, but now it would be too complex to change.
As of now, I came up with the below mess, that works on oSX, iOS and (with some trouble) Android.
This is what I do:

Code: Select all

export snapshot from rect (the rect of graphic "PhotoCircle") of this card to tFileVar as PNG

the snapshot tFileVar is the user logo or photo in the standalone.
I then encode the tFileVar and put it in an multidimensional array, together with other prefs info.

Code: Select all

put "Bill" into tPref["UserName"]
put base64Encode(textEncode(tFileVar,"UTF-8")) into tPref["UserPhoto"]

The all package is sent to FMS DB using its API:

Code: Select all

put ArrayToJSON(tPref) into tPrefJson --convert to json in fastjson library
put textEncode(tPrefJson, "utf-8") into tPrefJson

In the DB, the UserPhoto data is saved in a text column (doesn't need to be seen here).
The reverse is done when the user needs to login in a different device, with the same account.

The scary part is always as to have not intercepted some strange char that will disrupt the all thing, linke in the password or other prefs data.
Did I made some error?

[trevix]

I specified "...with some trouble in Android" because an image that is good in iOS/OSX devices, does not show on Android. And viceversa.
I guess something is wrong in the way I encode things.
May be the texEncode should use utf-16? Or else?
The fact that the FMS is running on a OSX may have something to do with it?