Different behaviour of execute JS between Android and IOS

[Ninerian]

While working with the executeJS function of the tabs module I found an issue.

Sample Code

function y (result) { console.log('result',result)}
function x () {  return [[{"ss":"zz"}]]; }

// a native tab, bound to a global variable
window.testTab.executeJS('(' + x.toString() + ')();',y)

Expected Result

A log entry with: [FORGE] '[[{"ss":"zz"}]]'

Actual Result

Android log entry: [FORGE] '[{"ss":"zz"}]'
IOS log entry: [FORGE] '[[{"ss":"zz"}]]'

On android the first and last character is removed from the output. This makes it impossible to send a valid JSON string as return value and parse it on the application side. Also if nothing is return, the callback parameter is ul, as null gets the same handling.

It think this line causes the behaviour:

trigger.io-tabs/inspector/an-inspector/ForgeModule/flavors/crosswalk/src/io/trigger/forge/android/modules/tabs/WebViewProxy.java

Line 190 in 20a21d5

task.success(value.substring(1, value.length()-1));

I also couldn't get into the value==null path.

How is the recommended way to return a value from an injected JS function?

[antoinevg]

Wow, looks like that line has been in there since the code was originally written 6 years ago!

I have no idea what the reason was for that originally but it a) looks like a rather ugly hack, b) doesn't look safe as there are no bounds checks on the variable and c) frankly doesn't make any sense.

That said, I've modified it in dec3258 and the module tests are still passing.

Can you please give it a shot on your side?

Published as:

"tabs": {
    "version": "3.6.1",
    "config": { }
},

Needs:

"platform_version": "v2.8.5"

[Ninerian]

Thank you @antoinevg. I'll try it out.