Kata

JS translation added. No leading minuses ( correctly this time ).

not 7kyu at all

I am still unhappy about the possible leading minus sign: it's a unary minus, where all the others are binary ones.

Especially now that this is approved at 7kyu, I would like to suggest to remove this special case. For a 7kyu, it's already complicated enough ( can you tell I voted 6? ).

When solving with eval, this is not a problem, but when solving with a parser, it definitely complicates things.

the input parameter string overwrites the Python module of the same name

N.B. that str should not be used as a replacement, as that would overwrite the Python function by that name

perhaps consider a variable name based on the concept of what's presented, which is a "calculation" or possibly an "expression"...

( Python )

random tests generate a lot of cases that expect 0.

generating smaller numbers would help, removing 0 from available digits would also help.

note that the current random generator can generate numbers that start with leading zeroes; that is a choice - solvers might not expect that and might or might not handle it correctly ( most code will handle leading zeroes gracefully ofc ).

Haskell translation

this translation modifies the description. check you're OK with that.
this translation generates slightly different random inputs than the Python version. if you can read Haskell, check you're OK with that. if you can't, just take my word for it. :P

in general, accepting translations before approval is .. a choice. but I don't expect the specs to change, and the translation is probably solid ( famous last words .. ).