Ability to provide arguments to getScriptEngine

[sinaa]

Would it be possible to extend the library, such that it'd be possible to pass arguments to getScriptEngine such as "-strict", "--no-java", "--no-syntax-extensions" ?

It'd be a great help since this library already provides sandboxing features such as CPU limits, etc.

[mxro]

Hi Sina,

Thank you for submitting this issue. I think it is always great to find ways to make the API more useful. Please see in the linked commit the change which will allow to create a new Sandbox as follows:

NashornSandbox sandbox = NashornSandboxes.create(args);

Could you point me to documentation what arguments are allowed? Could you think of a test case we could create to test this feature?

Thank you!

[sinaa]

@mxro , thanks a lot for this!! really appreciate that you implemented the feature in the library.

Regarding documentation of the arguments, there's this official documentation ( http://hg.openjdk.java.net/jdk8/jdk8/nashorn/file/tip/docs/DEVELOPER_README ), but I don't know if there's a more accessible version of it.

Here's also some source reference to these properties: https://github.com/JetBrains/jdk8u_nashorn/blob/master/src/jdk/nashorn/internal/runtime/resources/Options.properties

There are quite a few options such as strict, no-java, no-syntax-extensions, log (log level), lazy-compilation, etc.

Re. test cases, I'm particularly keen on --no-java. Enabling it should make Java.XXX not work.

[mxro]

Thank you for the information. I have added this to the documentation, creating a test case and released a new version to Maven central with the additional method in NashornSandboxes added.

I hope this solves this issue for you!

[sinaa]

Thanks a lot. This is really great work. I'll update here if I find an issue in production.

[sinaa]

Hello again.

Perhaps it'd make sense if I create a separate issue, but since it's relevant to this change, I comment here.

It seems that enabling strict mode has a conflict with delight. In particular, assertScriptEngine() in NashornSandboxImpl evaluates this expression:

quit=function(){};exit=function(){};print=function(){};echo = function(){};readFully=function(){};readLine=function(){};load=function(){};loadWithNewGlobal=function(){};$ARG=null;$ENV=null;$EXEC=null;$OPTIONS=null;$OUT=null;$ERR=null;$EXIT=null;

Would appending var to the beginning of each global var above conflict with the library in any way? (is it intentional? would adding var cause a problem?)

[mxro]

Thank you for reopening the issue and reporting this. I have changed the declaration of the variables. Can you test if this would work for you?

[Frontrider]

For testing, I could use scenarios in the following format:
-parameter
-script(s) what should no longer run.