A generator is a function that emulates the behavior of an iterable object. In general, implementing a generator is shorter and easier than creating an iterable. Several standard functions are implemented as generators, for example range().

Generators are based on the yield keyword. When calling __next__() on a generator, the function is executed until yield is reached, then the value that is associated with yield is returned. At that point, the function “waits” until __next__() is called again, after which it continues until yield is reached again. StopIteration is raised automatically when the function ends.

There is no need to explicitly define __next__() and/or __iter__(). A function is a generator simply because it contains the yield keyword, and the associated iterable object is automatically created by Python, including appropriate implementations for __next__() and __iter__().

def fibo( maxnum ):
    nr1 = 0
    nr2 = 1
    while nr2 <= maxnum:
        nr3 = nr1 + nr2
        nr1 = nr2
        nr2 = nr3
        yield nr1

fseq = fibo( 1000 )
for n in fseq:
    print( n, end=" " )
print()
for n in fseq:
    print( n, end=" " )

Generator expressions

In Chapter 13¹, I introduced the concept of list comprehension. Since any list can be turned into an iterator, and thus into a generator, Python introduced a similar concept for generators, and calls it “generator expressions.” The syntax for a generator expression is the same as for a list comprehension, except that the square brackets are replaced by round brackets.

For example, the following generator expression returns all squares up to 100:

seq = (x*x for x in range( 11 ))
for x in seq:
    print( x, end=" " )

If you just replace the outer two parentheses by square brackets in the generator expression, the code runs with seq being the result of list comprehension. To be absolutely clear about it: with list comprehension the whole list is generated at once, while with a generator expression the items are generated when needed. Thus, in principle a generator expression is preferable, as it saves memory.