Is there a reason why the two parsers are using different algorithms? Additionally, implement the max and min functions.

[light-weaver]

math_json_parser/math_json_parser/parser_numexpr.py

Line 63 in 4f40955

def parser(self, math_json):

In the polars parser you are using recursion. In here, you are using stack operations. This can make things unnecessarily difficult for future maintainers.

Also, if I am not wrong... The algorithm here only works for functions/operators with known arity (one or two in the cases you have implemented). We may eventually have cases with an unknown number of inputs. Infact, I would like you to implement two such functions: max and min.

You can find two valid JSON structures for these functions in the MathJSON documentation: https://cortexjs.io/compute-engine/reference/arithmetic/

Choose whichever one is easier to handle in your parsers and document the choice.