Generators. Python provides ways to make looping easier. With a generator, we specify what elements are looped over. Generators are reusable—they make code simpler.
There are two ways to specify a generator. With the yield-keyword, we encapsulate a looping mechanism into a method. With an expression, we use the for-keyword to specify simpler generators.
Yield. The term yield has a meaning similar to "produce": it dispatches specified elements. Unlike return, control returns again to the method.
Here The odds() method yields odd elements in an iterable. It has a yield statement in it.
Info The yield statement can return an expression, not just a variable. Try changing "yield a" to "yield a + 1".
Also The yield pattern here is the more powerful form. It can filter and transform elements.
def odds(arg):
# Yield odd elements in the iterable.
for a in arg:
if a % 2 != 0:
yield a
# An input list.
items = [100, 101, 102, 103, 104, 105]
# Display all odd items.
for item in odds(items):
print(item)
# Print copied list.
copy = list(odds(items))
print(copy)101
103
105
[101, 103, 105]
Expression. Sometimes in Python programs we see generator expressions. These return an iterator. Here we multiply each element in the list by 2 with a generator expression.
Info The expression "A" for "B" in "C" uses the generator expression syntax form.
Part 1 The first part, "A," is the result expression. We can use any expression we want.
Part 2 The second, "B," is the element we are acting upon within the collection "C."
# Example integers.
items = [100, 10, 1]
# Use generator expression.
multiplied = list(x * 2 for x in items)
print(multiplied)[200, 20, 2]
Performance. In Python, list comprehensions are similar to generators. Are generators faster? In the Python documentation, we learn that list comprehensions tend to use more memory.
Version 1 This version of the code uses a generator expression. We create a list on each iteration.
Version 2 Here we use a list comprehension. The syntax is similar to the version with a generator in it.
Result I found that a list comprehension was faster in both implementations I tried, Python and PyPy.
So Often a list comprehension is faster despite the increased memory usage that may occur (according to the Python documentation).
import time
data = [1, 2, 3, 4, 5, 6, 7, 8, 9]
print(time.time())
# Version 1: generator.
i = 0
while i < 1000000:
double = list(x * 2 for x in data)
i += 1
print(time.time())
# Version 2: list comprehension.
i = 0
while i < 1000000:
double = [x * 2 for x in data]
i += 1
print(time.time())1404138028.766
1404138031.806 Generator: 3.04 s [Python 3.3]
1404138033.569 List comprehension: 1.76 s
1404137810.806
1404137811.086 Generator: 0.28 s [PyPy 2.3.1]
1404137811.287 List comprehension: 0.20 s
Discussion. In declarative programming (like generators), we specify our requirement with a declaration. Python then does the low-level work to fulfill our request.
And Imperative programming is where you specify individual statements on lines. It often involves traditional loops.
A review. Generators simplify iteration. With them, we create looping abstractions. We then can reuse looping methods throughout programs—this leads to less code and higher-quality logic.
Dot Net Perls is a collection of tested code examples. Pages are continually updated to stay current, with code correctness a top priority.
Sam Allen is passionate about computer languages. In the past, his work has been recommended by Apple and Microsoft and he has studied computers at a selective university in the United States.
This page was last updated on Apr 14, 2023 (edit).
