Skip to main content

Multi-worker pipelines with gevent.

Installation

pip install gevent-pipeline

Example

import gevent
import random

from itertools import repeat
from gevent_pipeline import Pipeline

def sample(b):
    r = random.uniform(0, b)
    gevent.sleep(r)
    return r

(Pipeline()
    .from_iter(repeat(1, times=200))
    .map(sample, n_workers=100)
    .filter(lambda x: x < 0.5)
    .fold(max, x0=0, n_workers=50))

Pipeline

Chain together operations with multiple workers for each layer.

Example:

>>> import operator
>>> def only_odd(x):
...     return x & 1
...
>>> def double(x):
...     return 2 * x
...
>>> (Pipeline()
...         .from_iter(range(100))
...         .filter(only_odd, n_workers=10)
...         .map(double, n_workers=8)
...         .fold(operator.add, x0=0, n_workers=5))
5000

The above is functionally equivalent to:

>>> @worker(discard_none=True)
... def only_odd(x):
...     '''Forward only odd numbers to the next layer'''
...     if x & 1:
...        return x
...
>>> @worker()
... def double(x):
...     return 2 * x
...
>>> def load_numbers(q_in, q_out, q_done):
...     for i in range(100):
...         q_out.put(i)
...     q_done.put(None)
...
>>> q_out = ClosableQueue()
>>> p = (Pipeline()
...     .chain_workers(load_numbers)
...     .chain_workers(only_odd, n_workers=10)
...     .chain_workers(double, n_workers=8, q_out=q_out))
>>> sum(i for i in q_out)
5000

There is no guarantee of order:

>>> def f(x):
...     gevent.sleep(random.uniform(0, 0.001))
...     return x
>>> p = Pipeline().from_iter(range(10)).map(f, n_workers=5)
>>> list(p)
[2, 1, 4, 0, 3, 5, 8, 6, 7, 9]

Exceptions in workers

There is a predefined forward_input exception handler, in the event the function raises an exception the handler takes the input to the function and passes it along as if it was the output.

from gevent_pipeline import Pipeline, worker, forward_input

@worker(exception_handler=forward_input)
def f(x):
    if x & 1:
        raise ValueError("oh no!")
        # Will be treated as if it were:
        # return x
    else:
        return 2 * x

p = (Pipeline()
     .from_iter(range(100))
     .chain_workers(f, n_workers=10))

s_odd = sum(range(1, 100, 2))
s_even = sum(2*i for i in range(0, 100, 2))
assert sum(p) == s_odd + s_even

ClosableQueue

Acts like gevent.queue.Queue but in addition has a .close() method which invokes following behavior:

  • Calling .put(item) becomes an error

  • Successive calls to .get() will return whatever remains in the queue and after that StopIteration is returned for every subsequent call

>>> from gevent_pipeline import ClosableQueue
>>> q = ClosableQueue()
>>> q.put('hello')
>>> q.close()
>>> q.get()
'hello'
>>> q.get() is StopIteration
True

Metadata

Release files for gevent-pipeline 0.0.5

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Built distribution (wheel)

Table of built distributions (wheels) for gevent-pipeline 0.0.5
File Interpreter ABI Platform Actions
gevent_pipeline-0.0.5-py34-none-any.whl Python 3.4 none any Details for gevent_pipeline-0.0.5-py34-none-any.whl

Release files / gevent_pipeline-0.0.5-py34-none-any.whl

Download URL gevent_pipeline-0.0.5-py34-none-any.whl
Size 9.1 kB
Tags Python 3.4
SHA-256 checksum
How to use checksums
4979b1e238857d0236442a6401de7d42d5f43aaae8310423271b1ed903a1579a
BLAKE2b-256 checksum
How to use checksums
564e17d9824bff97aa21c6677090bfe174b24be228a631d5955d39d3fb589d8c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No

Security for gevent-pipeline 0.0.5

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page