Yea I don’t use real atoms anywhere in the code. What I ended up doing is in the parsing stage whenever I need to “create” and atom I actually create a string with a placeholder for later use in the eval step.
You can see an example of it being used here
And the actual atom function that sets the placeholder:.
Then in the eval step all we do is generate the code with the Macro.to_string helper and regex out the placeholder atom to a “real” but not really a real atom.
Thank you for your tool! It now works as expected… well kind of.
I noticed it duplicates/attempts to redefine recursive data structures instead of normalizing and defining them once.
Sure! You can see that happening on my previous JSON. Some entities are being defined multiple times with varying properties, instead of being merged as a single entity composed of all properties inferred by each instance.