python-jsonlogic 0.1.0

An extensible and sane implementation of JsonLogic

python-jsonlogic is an extensible and sane implementation of JsonLogic, making use of the JSON Schema specification.

Motivation

While the JsonLogic format can be great to serialize logic, it lacks a formal specification and some aspects are unclear/unspecified:

• operators arguments can take any value. For instance, comparison operators are said to work with “numeric” values, however the JavaScript playground doesn’t validate inputs. It is also convenient to allow such comparison operators for date and datetime objects as well.

• Operators accessing data use a dot-like notation, which is ambiguous when dealing with keys such as my.key.

• Operators such as map provides their own data scope, making it impossible to access higher-level data inside the operator expression.

For these reasons, python-jsonlogic provides a way to typecheck your JSON Logic expressions at “compile” time, before applying input data to them.

Installation

From PyPI:

pip install python-jsonlogic

The library can be imported from the jsonlogic module.

Usage

# 1. Create or use an already existing operator registry:
from jsonlogic.operators import operator_registry

# 2. Parse the JSON Logic expression:
from jsonlogic import JSONLogicExpression

expr = JSONLogicExpression.from_json({"map": [
    [1, 2],
    {"*": [{"var": ""}, {"var": "/my_int@1"}]},
]})

# 3. Create an operator tree:
root_op = expr.as_operator_tree(operator_registry)

# 4. Typecheck the expression:
from jsonlogic.typechecking import typecheck

typ, diagnostics = typecheck(
    root_op,
    data_schema={
      "type": "object",
      "properties": {
            "my_int": {"type": "integer"}
        },
    }
)
print(typ)
#> ArrayType(IntegerType())

# 5. Evaluate with data:
from jsonlogic.evaluation import evaluate
value = evaluate(
    root_op,
    data={"my_int": 2},
    data_schema=None,
)
print(value)
#> [2, 4]