Question about missing keys behavior in rule evaluation

[anietopltx]

Hi,

I’ve noticed something inconsistent when evaluating rules with missing keys in the input dictionary.

For example:

person = {'name': 'Lebron'}

rule = rule_engine.Rule(
'surname == "James"'
)

If I apply this rule to person, the result is simply False.

However, if the rule is numeric:

rule = rule_engine.Rule(
'age > 10'
)

then I get an error:
errors.EvaluationError('data type mismatch')

Why is this happening? Is it the expected behavior or could it be a bug?

Extra info: I’m using a custom context with context = rule_engine.Context(default_value=None).

Thanks in advance!

[zeroSteiner]

The extra info is super relevant here. So since you defined the default_value as None, that will effectively treat missing values as None, allowing you to avoid undefined symbol errors being raised, thus making rule evaluation a bit more permissive. As for why the one rule works and the other doesn't, it's the same behavior you'd see in Python. It's safe to compare None to a string via the equality operator but you can't compare None to an integer with the greater-than operator.

>>> surname = None
>>> age = None
>>> surname == 'James'
False
>>> age > 10
Traceback (most recent call last):
  File "<python-input-3>", line 1, in <module>
    age > 10
TypeError: '>' not supported between instances of 'NoneType' and 'int'
>>> 

Rule engine raises the data type mismatch instead of the type error for the same reason as Python, you can't compare None and int using that operator.

If you don't set the default_value on the context, it should start to raise exceptions for symbols that don't exist instead of treating them as None.

[anietopltx]

Thanks a lot for the quick and clear answer! That explanation really helps.

It also makes sense now why I was confused — I’m more used to reasoning in JavaScript, where something like null > 10 doesn’t raise an error but just evaluates to false (because null is coerced to 0). So I expected Python to behave similarly.