The fake data that I've bothered to model are weighted age ranges. Fortunately, as of Python3.6, you can access it from random.choices  in the stdlib
PS. that's a really cool python tip!
joke2k/faker is python and the data is stored in code (all or most of the random values are in .py files around the codebase), perhaps leading to its slowness.
stympy/faker is ruby and its random values are in yaml files, with some fields defined as ruby functions (those are not supported by plait.py).
can use 'plait -l' and 'plait -ll name' and 'plait -ll name.name' (more info in the README) to get a list of fake fields available.
Probably also available in other languages.
“hypothesis” package in Python (+pytest plugin)
“rapidcheck” in C++
“quickcheck” in Rust
Granted, I come from Ruby, and writing DSLs is pretty typical. Maybe not so popular in Python.
I am asking this because I become suspicious of config languages that read like code. Is not a bonafide programming language the better choice in this scenario? i.e. all overly-configurable formats (e.g. Terraform .tf files, JSON schemas...) converge on just being a new scripting language?
yaml was a format I chose because it is easy to write (close to human), but can not express full programming concepts (but yes to some metaprogramming). i did not want the templates to be full powered as they are meant to be able to express relationships between variables, but not much more (especially not side effects). they also support lazy evaluation - statements do not need to be in order. this is closer to a "mathematical language" for me.
the choice for yaml was also based on the premise that if performance becomes an issue, can hopefully move to another language but retain templates (will have to re-implement python's "random" compat, though)
if that function uses an import, you might also need to add an "imports" field, like in this example: https://github.com/plaitpy/plaitpy/blob/master/templates/web...
otherwise, that's a feature that can be added here:
https://github.com/plaitpy/plaitpy/blob/master/src/fields.py..., if it works for you (and is added as a flag), i'd be happy to take patches.