How do I access sci.core/eval-string* inside of babushka?

[kkinnear]

In my library zprint, I use sci (specifically sci.core/eval-string*) to turn maps (which sometimes contain functions) read in from files into Clojure maps. sci works great, and produces functions which work fine but are also essentially "sandboxed", and so won't do terrible things to someone's environment.

When I try to run the zprint library inside of babushka, I can't seem to figure out how to access sci.core/eval-string* in order to do the same thing. If I just use read-string, it will read a string but not produce a working function when one appears in the string. I'm using the reader-conditional :bb to do all of this, for what that's worth.

I know that babushka is implemented with sci, so it isn't like sci isn't "in there" somewhere, but despite plenty of trial and error, I can't seem to figure out any way to get the equivalent functionality to my calls to sci.core/eval-string* when actually running in babushka as opposed to Clojure or Clojurescipt. To be clear, since I'm using reader-conditionals, I can do whatever is necessary in babushka without impacting the normal usage in the Clojure and Clojurescript environments. So any equivalent to sci.core/eval-string* would be fine. The only other sci thing I even use is sci.core/init to make some of my functions available to sci processed input.

Any help would be greatly appreciated!

[borkdude]

load-string in babashka will do the same thing as sci.core/eval-string*

[kkinnear]

Thank you so much! load-string work great -- except for one detail. When I use sci.core/eval-string*, I'm giving it a ctx that I created with sci.core/init. Is there some way to make load-string aware of my ctx? Or the information that I would like to put in a ctx, more specifically.

I've looked at the code in sci for load-string, and while the impl version seems to take a ctx, it looks like the load-string that is exposed has one given to it. Not that I can call init to create a ctx anyway in babashka. Perhaps there is some way to get the information that I would like to have in a ctx known by babashka or sci prior to the call to load-string?

Near as I can tell, this is the only thing missing in order to have a fully functional zprint that will work inside of babashka. Which would be a neat thing in general, and particularly useful for the testing environment that I'm working for for zprint itself (which is the source of my current questions).

Thanks!

[borkdude]

load-string in babashka has an implicit context, the global context which is babashka's context. but maybe I could just expose SCI's interface (or parts of it) to babashka. Which functions are you using? init, eval-string*, more?

[kkinnear]

Wow, that would be great. I use only init and eval-string*. It would be really nice to be able to access these functions in babashka!

Thank you so much for sci! sci has been a wonderful way to handle processing configuration files for zprint. It just works (which is no small thing), and protects people from bad things that might show up in the files. It was trivial to integrate into zprint, and is the only library I use other than rewrite-clj. It is an amazing piece of work in it own right, to say nothing of enabling babashka as well. Thank you for all of your excellent work on both of these Clojure capabilities that are becoming fundamental!

[borkdude]

I've exposed sci.core on master. You can test this with:

bash <(curl https://raw.githubusercontent.com/babashka/babashka/master/install) --dev-build --dir /tmp

as soon as the master build finishes.

[kkinnear]

Thank you very very much!

That's a very impressive string to download the snapshot version to test. The download worked flawlessly!

√ projects/zprint % /tmp/bb --version
babashka v1.3.183-SNAPSHOT
√ projects/zprint %

And indeed, the zprint library also now works as well! I can now use the same code for babashka as for Clojure (no reader-conditionals necessary), and when I read in a string with sci.core/eval-string* it finds functions that I have added to a context using sci.core/init. That's great, and I will be letting people know that zprint now works completely when used within babashka.

One final question -- is there some way to add to the internal sci context within babashka itself? I can certainly put everything in the ns statement at the top of the babashka file, but if I have 15 functions or so, and they are all set up to be fed to sci.core/init, is there something I can call at the start of my file to add them to the internal babashka sci context? This isn't critical, because I can certainly make babashka aware of them using the ns statement, but if there is some way to add them to the internal babashka sci context, that would be convenient.

Thank you again for making all of this possible. It is a huge help to me in creating a system test for zprint using babashka, and I think it will be a useful thing for others that might want to use zprint inside of babashka. To say nothing of other folks or libraries that use sci that can now have them operate within babashka unchanged!

[borkdude]

The internal SCI context of babashka isn't exposed so I recommend doing it via the ns form.

[kkinnear]

Great thanks. Thanks again for exposing sci.core!