Introducing the Builtin Class

Most of the time you’ll probably need to pass information into the Function you want to add. For this, use the Builtin class. In WL these things functions and variables defined this way are tagged as a “Built-in Symbol”.

The method that you should define in the Builtin class that should get invoked needs to start off with the name eval.

As before, this method has an evaluation parameter at the end.

Other parameters that are appropriate for the function can be added. However those parameters must also be listed suffixed with an underscore (_) in the Python method’s docstring in a special way.

The docstring is used by Builtin’s default evaluate() method when trying to resolve what Python method to call. The docstring looks pretty much the same as it would look if you were defining this in WL.

For example, let’s add a string parameter. In Mathics3 the function might look like this:

Hello[s_String] := Print["Hello, " <> s <> "!"]

The evaluation method, looks like this:

def eval(self, person: String, evaluation: Evaluation) -> String:
  "Hello[person_String]"
      return String(f"Hello, {person.value}!")

Here is the complete code:

from mathics.builtin.base import Builtin
from mathics.core.atom import String
from mathics.core.evaluation import Evaluation

class Hello(Builtin):
  def eval(self, person: String, evaluation: Evaluation) -> String:
    "Hello[person_String]"
        return String(f"Hello, {person.value}!")

The parameter person has type Mathics3 String. Therefore, to use that for use in Python, we need to convert it to a Python value. This is done with person.value. And the the return value also needs to be a Mathics3 String so we need to convert the Python string in Python back to a Mathics3 String.

Previously, since we weren’t modifying the Mathics3 String, no conversions to Python and from Python were needed.

See Patterns for more information about how to specify expressions with patterns in them that you might use in an eval() method docstring.

Next: