A simple lazy Python Calculation Engine.

The module is still in development. You can install it by cloning this repository and using the poetry install command.

poetry add git+https://github.com/bsdz/calcengine.git

Alternatively, you can use pip install -r requirements.txt. This was generated using poetry export without hashes option.

The core module for the calculation engine only uses core python standard library.

The demo spreadsheet application uses pyqt5, pandas, matplotlib and pillow.

First instantiate a CalcEngine and use watch decorator to register functions as nodes. Note that a function along with any arguments and keyword arguments make a unique node.

from calcengine import CalcEngine ce = CalcEngine() @ce.watch() def a(): print("..in a") return 100 @ce.watch() def b(): print("..in b") return a() @ce.watch() def c(x, y): print(f"..in c with x={x} and y={y}") return 2 * a() + x * y @ce.watch() def d(x, y=0): print(f"..in d with x={x} and y={y}") return 3 * b() + x - y @ce.watch() def e(): print("..in e") _x = d(5, y=-3) return c(2, 3) - 5 + _x @ce.watch() def f(): print("..in f") return d(0) + e()

Calling a function will cache all values and path during first run.

>>> f() ..in f ..in d with x=0 and y=0 ..in b ..in a ..in e ..in d with x=5 and y=-3 ..in c with x=2 and y=3 809

And obviously a 2nd invocation will retrieve the final value from cache.

>>> f() 809

Invalidating a node by calling function helper method.

>>> e.invalidate() >>> f() ..in f ..in e 809

Invalidating a node without arguments if previous call did have arguments won't have any effect.

>>> d.invalidate() >>> f() 809

Whereas with arguments specified exactly as prior call will. Note the sensitivity of argument specification.

>>> d.invalidate(5, y=-3) >>> f() ..in f ..in e ..in d with x=5 and y=-3 809

It is also possible to add a trigger that will be called on completion of a function. This might be used to produce some form of data binding in applications.

def my_trigger(res): print(f"got {res}") >>> c.node_calculated.append(my_trigger) >>> c.invalidate(2, 3) >>> f() call f call e call c with x=2 and y=3 got 206 809

Included is a simple spreadsheet demo. Read more here

• Support watching global variables.
• Support multiprocessing.
• Support asyncio?.

Some similar packages spotted. None of them tested.

• pyungo
• dask
• schedula
• graphkit
• mdf