Pyarallel¶

Simple, explicit parallel execution for Python.

Overview¶

Pyarallel wraps Python's concurrent.futures and asyncio.TaskGroup with better ergonomics: structured error handling, rate limiting, progress callbacks, and async support.

Design principles:

Explicit over implicit — .map() for parallel, normal calls stay normal
No magic — no type-sniffing, no global config, no singletons
Structured errors — all failures collected via ExceptionGroup, never silently lost
Sync + async — mirror APIs, you choose explicitly

Quick Example¶

from pyarallel import parallel_map, RateLimit

# Fan out 200 API calls, rate-limited to 100/min, with progress
results = parallel_map(
    fetch_url,
    urls,
    workers=10,
    rate_limit=RateLimit(100, "minute"),
    on_progress=lambda done, total: print(f"{done}/{total}"),
)

for item in results:        # raises ExceptionGroup if any failed
    process(item)

Or with the decorator:

from pyarallel import parallel

@parallel(workers=4)
def fetch(url):
    return requests.get(url).json()

data = fetch("http://example.com")        # normal call, returns dict
results = fetch.map(["u1", "u2", "u3"])   # parallel, returns ParallelResult

Documentation¶

License¶

MIT — see LICENSE.