Could we implement generator using class?

Question

when we look at the Python documentation we could see that generators are always defined using yield statement, but in the Internet we could see that some people are trying to implement generators using classes (eg. here How to write a generator class?).

Here is example generator implementation using classes:

from collections import Generator
class Fib(Generator):
    def __init__(self):
        self.a, self.b = 0, 1        
    def send(self, ignored_arg):
        return_value = self.a
        self.a, self.b = self.b, self.a+self.b
        return return_value
    def throw(self, type=None, value=None, traceback=None):
        raise StopIteration

When we execute it in repl we can see it is not the generator, but ordinary object. It only tries to behave like generator.

>>> x = Fib()
>>> x
<__main__.Fib object at 0x7f05a61eab70>

When we look at PEP 342:

4. Add a close() method for generator-iterators, which raises GeneratorExit at the point where the generator was paused.

I think it is not possible to meet that condition using own implementation with classes.

Am I wrong? Is it really possible to implement real generator using classes?

Answer 1

Generators are simply a type of iterator. From the datamodel documentation:

Generator functions
A function or method which uses the yield statement [...] is called a generator function. Such a function, when called, always returns an iterator object which can be used to execute the body of the function: calling the iterator’s iterator.__next__() method will cause the function to execute until it provides a value using the yield statement. When the function executes a return statement or falls off the end, a StopIteration exception is raised and the iterator will have reached the end of the set of values to be returned.

You can't tell by the repr() output if something is a generator. Python looks for the iterator methods, and you can implement your own send and throw methods on top of those, as you have done.

As such, your implementation works as designed, it is valid iterator:

>>> x = Fib()
>>> next(x)
0
>>> next(x)
1

Without the collections.abc.Generator base, you can also implement your own __iter__ method (this has to return self), and a __next__ method that produces the next value when called or raises StopIteration when done.

The base collections.abc.Generator implementation defines the __next__ method for you and supplies an __iter__ method that simply calls self.send(None).