renew 0.3.1

Gives a reproducible manner to your objects and can serialize them in 100% pythonic format.

Semi-text-pickling in pure python.

If you meet just a few restrictions, you can store classes

state into a python file and import or evaluate it somewhere

else or later on. You can even use it as a database unless

the amount of data is huge.

repr(object)

Does repr stand for “representation” or “reproduction”?

According to python documentation __repr__ functionality has two

separate approaches. From https://docs.python.org/3/library/functions.html#repr (v 3.7.2)

repr(object) Return a string containing a printable representation of an object.

For many types, this function makes an attempt to return a string

that would yield an object with the same value when passed to eval(),

otherwise the representation is a string enclosed in angle brackets

that contains the name of the type of the object together with

additional information often including the name and address of

the object. A class can control what this function returns for

its instances by defining a __repr__() method.

1. reproducible repr:

For several native objects it returns a string that can be used

to reproduce given object, i.e. to create a copy of given object.

a = [1, 3.141559, None, "string"]
statement_str = repr(a)
assert statement_str == '[1, 3.141559, None, "string"]'

You may tell that repr of an object is reproducible if this is meet:

a = [1, 3.14159, None, "string"]
statement_str = repr(a)
assert repr(eval(statement_str)) == statement_str
# if the object implements __eq__ this should be also true:
assert eval(statement_str) == a

2. descriptive repr:

Unfortunately python does not serve the “reproducible repr” out of the box

for types defined by user:

class Car(object):
    def __init__(self, body_type, engine_power):
        self.body_type = body_type
        self.engine_power = engine_power

car = Car("coupe", 124.0)
# repr(car) == '<__main__.Car object at 0x7f0ff6313290>'
# but using renew:

import renew

@renew.make_renew_reprs(namespace="bar")
class ReproducibleCar(object):
    def __init__(self, body_type, engine_power):
        self.body_type = body_type
        self.engine_power = engine_power

car2 = ReproducibleCar("sedan", 110.0)
assert repr(car2) == 'bar.ReproducibleCar("sedan", 110.0)'

The method above is implemented as a decorator, but you can also use a inheritance to get the same result.

import renew

repr_from_cars = renew.make_reproducible(namespace="cars", dependency="that.things")
repr_from_persons = renew.make_reproducible(namespace="persons", dependency="living.things")

class Car(repr_from_cars):
    def __init__(self, body_type, engine_power, fuel, seats, color=None):
        self.body_type = body_type
        self.engine_power = engine_power
        self.fuel = fuel
        self.seats = seats
        self.color = color

class Driver(repr_from_persons):
    def __init__(self, first_name, last_name, *cars):
        self.first_name = first_name
        self.last_name = last_name
        self.cars = cars

car_1 = Car("Truck", 120.0, "diesel", 2)
car_2 = Car("Van", 145.0, "diesel", seats=7, color="silver")
car_3 = Car("Roadster", 210.0, "gasoline", seats=2)

driver_1 = Driver("Blenda", "Klapa", car_1)
driver_2 = Driver("Trytka", "Blotnick", car_2, car_3)

assert repr(driver_1) == "persons.Driver('Blenda', 'Klapa', cars.Car('Truck', 120.0, 'diesel', 2))"
assert repr(driver_2) == """\
persons.Driver(
    'Trytka',
    'Blotnick',
    cars.Car('Van', 145.0, 'diesel', 7, 'silver'),
    cars.Car('Roadster', 210.0, 'gasoline', 2),
)"""

renew.serialize("/tmp/target.py", blenda=driver_1, trytka=driver_2)

The created file looks like this:

#!/usr/bin/env python
# -*- coding: utf-8 -*-

# This file has been created with renew.
# A py-pickling tool: https://pypi.org/project/renew/

from living.things import persons
from that.things import cars

blenda = persons.Driver('Blenda', 'Klapa', cars.Car('Truck', 120.0, 'diesel', 2))

trytka = persons.Driver(
    'Trytka',
    'Blotnick',
    cars.Car('Van', 145.0, 'diesel', 7, 'silver'),
    cars.Car('Roadster', 210.0, 'gasoline', 2),
)

How it works?

Note that ReproducibleCar does not explicitly implement the __repr__, but the renew.reproducible

decorator supplements it (overrides it if any has been defined before).

renew.reproduction inspects constructor’s argument specification

of decorated class and yields a string that tries to be a call statement composed of

• namespace, e.g. your package name (according to desired importing convention)

• given class name

• given class’ attributes values, that have the same names and order as constructor arguments

That forms the only one usage restriction:

The class has to store all the constructor arguments in its attributes with the same name (as in ReproducibleCar definition above).

Limitations

Besides the statement above:

• constructor arguments have to get exactly same name as instance attributes

• plain keyword-arguments of constructors are not (yet) supported (default-arguments work well, however)

• keys of plain dict being “complex” objects are getting a bit ugly layout if repr of given key spans multiple lines.

• renew does not cross-reference objects while serializing.

  Although neither pickle nor marshal does cross-reference, renew most probably could do it but it’s

  hard to tell how to let renew know where and how a chain of objects have to be cross-referenced.

• For ultra-capable meta programming MacroPy: https://pypi.org/project/MacroPy/ would be a better choice.

For full list of features and usage examples, please refer to unit tests, especially tests/test_renew.py.