multiprocessing_generator
A library to prefetch items from a Python generator in the background, using a separate process.
Install (no dependencies):
pip install multiprocessing_generator
Example:
from multiprocessing_generator import ParallelGenerator
def my_generator():
while True:
# ... download something long ...
yield result
with ParallelGenerator(
my_generator(),
max_lookahead=100) as g:
for elem in g:
# ... do some heavy processing on that element ...
Up to 100 elements ahead of what is consumed will be fetched by the generator in the background, which is useful when the producer and the consumer do not use the same resources (for instance network vs. CPU).
The generator handles exceptions and more serious failures transparently.
Distributed under the MIT license.
See also: https://github.com/justheuristic/prefetch_generator