14.1. Generator About

Important

• Processes one element at a time

• Does not remember previous element

• Does not know next element

• Can be used only once

• Save memory (does not require more memory for processing large data)

• Values are computed on demand

• No need to store all values in memory

• Uses around 10% more CPU than regular processing

• Typical usage: streams, processing larger than memory files or data

• Cannot use len() as of generators don't have length

• Previous element is overridden by current on next()

• Functions (list, dict, tuple, frozenset, set, sum, all, any, etc) will evaluate generator instantly

14.1.1. Plain Function

• Plain function returns plain object

✘

>>> def run():
...     return 1

✘

>>> type(run)
<class 'function'>

✘

>>> type(run())
<class 'int'>

14.1.2. Generator Function

• Generator function returns generator object

✘

>>> def run():
...     yield 1

✘

>>> type(run)
<class 'function'>

✘

>>> type(run())
<class 'generator'>

14.1.3. Yield vs. Return

• After return function stops

• After yield function pauses

✘

>>> def run():
...     return 1
...     return 2

✘

>>> def run():
...     yield 1
...     yield 2

14.1.4. Lazy Evaluation

• After yield function pauses

• Calling next() resumes function until next yield

• After last yield raises StopIteration

✘

>>> def run():
...     yield 1
...     yield 2

✘

>>> result = run()

✘

>>> next(result)
1

✘

>>> next(result)
2

✘

>>> next(result)
Traceback (most recent call last):
StopIteration

14.1.5. Instant Evaluation

• Using list() will evaluate generator instantly

• Functions (list, tuple, set, dict, sum, min, max, all, any, etc) will evaluate generator instantly

✘

>>> def run():
...     yield 1
...     yield 2

✘

>>> result = run()
>>> tuple(result)
(1, 2)