Tempo

Asynchronous wrapper for HTTP requests.

Installation

pip install tempo-async

Usage

import tempo

url = 'https://via.placeholder.com'
num_reqs = 100

requests = [ # map in any request parameters: url, query params, HTTP method, etc.
  tempo.RequestConfig(url=url) for _ in range(num_reqs)
]

if __name__ == '__main__':
  tempo.run(requests, rate=10) # fetch 100 cat pictures, 10 a second

Take it a step further by collecting the responses,

import tempo

url = 'https://via.placeholder.com'
num_reqs = 100

### cat picture bucket
catpics = []
###

requests = [ # map in any request parameters: url, query params, HTTP method, etc.
  tempo.RequestConfig(url=url) for _ in range(num_reqs)
]

if __name__ == '__main__':
  tempo.run(requests, collector=catpics, rate=10) # indicated `catpics` should store response
  print(f'Pictures stored in list: {len(catpics)}')

… and adding any number of processing functions to handle responses.

import tempo

url = 'https://via.placeholder.com'
num_reqs = 100
catpics = []

requests = [ # map in any request parameters: url, query params, HTTP method, etc.
  tempo.RequestConfig(url=url) for _ in range(num_reqs)
]

### processors
def say_hi(res) -> None:
  # returns None so does not affect final processed response sent to collectors
  print('Hello cat!')
 
def get_body(res) -> str:
  # since it returns a value, this processor changes the final output of `tempo.run`
  body = res.text
  return body
###

if __name__ == '__main__':
  # process the requests in order of listed processors
  tempo.run(requests, collector=catpics, rate=10, processors=[say_hi, get_body])
  # processors' return values affect output sent to collectors
  print(f'Type of stored result: {type(catpics[0])}') # str, not Response object

Contributing

Submit a pull request! Contributions are welcome!

Please write test coverage for your changes and run tox to test for backwards compatibility among the supported Python versions.

TODOS

Open an issue, create a branch, and submit a PR. (Tests for everything!)

• Handle collection of results as a return value of the run function.
• Decorator for basic async request function accepting iterator of request arguments.
• Accept plain Python dictionaries and JSON in addition to RequestConfig objects for requests mapping.
• Logging.
• Exception handling.
• GitHub Actions.
• Retries with various back-off algorithms and HTTP response header search for 429 causes.
• Generator option for run function.
• Allow collector argument to be an iterable, a function passed a response / processed response object, or a file out. Maybe even stdout.
• Like processors, allow multiple collectors (eg. file, queue).
• Local database for keeping track of requests and their status, for retry and interrupts.
• Asynchronous processor support.
• Handle streaming responses.
• CLI
• Documentation page. Also, good docstrings.
• Test coverage tracking.
• …