Metadata-Version: 2.1
Name: patata
Version: 2.0.0
Summary: Simple lightweight HTTP client to perform GET or POST requests in parallel with multiprocessing and in each process use aiohttp to do them concurrently and maximize the throughtput.
Project-URL: repository, https://github.com/oalfonso-o/patata
Project-URL: documentation, https://github.com/oalfonso-o/patata/blob/main/README.md
Author-email: Oriol Alfonso <oriol@oalfonso.com>
License: MIT License
        
        Copyright (c) 2023 Oriol Alfonso Riba
        
        Permission is hereby granted, free of charge, to any person obtaining a copy
        of this software and associated documentation files (the "Software"), to deal
        in the Software without restriction, including without limitation the rights
        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
        copies of the Software, and to permit persons to whom the Software is
        furnished to do so, subject to the following conditions:
        
        The above copyright notice and this permission notice shall be included in all
        copies or substantial portions of the Software.
        
        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
        SOFTWARE.
License-File: LICENSE
Classifier: Development Status :: 4 - Beta
Classifier: Framework :: AsyncIO
Classifier: Framework :: aiohttp
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Python :: 3 :: Only
Classifier: Programming Language :: Python :: 3.8
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Requires-Python: >=3.8
Requires-Dist: aiohttp<4,>=3.8.4
Requires-Dist: importlib-metadata; python_version < '3.9'
Requires-Dist: pydantic<2,>=1.10.7
Description-Content-Type: text/markdown

# Patata

<p align="center">
    <em>Easy parallel and concurrent GET requests</em>
</p>

<p align="center">
<a href="https://github.com/oalfonso-o/patata/actions?query=workflow%3ACI+event%3Apush+branch%3Amain" target="_blank">
    <img src="https://github.com/oalfonso-o/patata/workflows/CI/badge.svg?event=push&branch=main" alt="Test">
</a>
<a href="https://pypi.org/project/patata" target="_blank">
    <img src="https://img.shields.io/pypi/v/patata?color=%2334D058&label=pypi%20package" alt="Package version">
</a>
<a href="https://pypi.org/project/patata" target="_blank">
    <img src="https://img.shields.io/pypi/pyversions/patata.svg?color=%2334D058" alt="Supported Python versions">
</a>
</p>

The idea of this package is to wrap `multiprocessing` and `async` concurrency and allow the user to perform thousands of requests in parallel and concurrently without having to worry about pools of processes and async event loops.

It only supports GET and POST requests. More methods will be implemented in later versions.

The input is an iterable and the output is a generator. As soon as the requests get answer they are yielded until all the requests of the input are done.

Each element of the iterable is a tuple with `id`, `url` and `data`.
Each element yielded is a tuple with the same `id` and the json response. With the `id` you can
later map the responses.

This is useful for cases where you have a huge amount of requests to perform to an API and you
need to do them as fast as possible.

The client by default detects the number of CPUs available and starts one process per each CPU.
Then chunks the input iterator to provide requests to all the processes.
Each process for each task opens an event loop and performs all those requests concurrently. Once
all requests are awaited, the chunk with all the responses is returned back to the main process.
This is why we can see that our generator is receiving the responses un bulks.

## Install:

From PyPi:
```
pip install patata
```

## Usage:

Use always context manager, for example for GET:

``` python
>>> from patata import Patata
>>> with Patata() as client:
...     responses = client.http("get", [(1, "http://localhost:12345", {}), (2, "http://localhost:12345", {})])
...     for response in responses:
...         print(response)
... 
patata              INFO      Start processing requests with Patata parameters:
patata              INFO        method:             GET
patata              INFO        num_workers:        8
patata              INFO        queue_max_size:     100000
patata              INFO        input_chunk_size:   10000
patata              INFO        pool_submit_size:   1000
(2, {'message': 'Hello world!'})
(1, {'message': 'Hello world!'})
patata              INFO      All requests processed:
patata              INFO        Total requests:     2
patata              INFO        Total time (s):     0.10
patata              INFO        Requests/s:         20.35
```

You can provide a generator to don't blow up the memory:
``` python
>>> from patata import Patata
>>> from collections import deque
>>> 
>>> def mygen():
...     for i in range(100_000):
...          yield (i, "http://localhost:12345", {"key": "value"})
... 
>>> with Patata() as client:
...     responses = client.http("post", mygen())
...     _ = deque(responses)
... 
patata              INFO      Start processing requests with Patata parameters:
patata              INFO        method:             POST
patata              INFO        num_workers:        8
patata              INFO        queue_max_size:     100000
patata              INFO        input_chunk_size:   10000
patata              INFO        pool_submit_size:   1000
patata              INFO      All requests processed:
patata              INFO        Total requests:     100000
patata              INFO        Total time (s):     43.78
patata              INFO        Requests/s:         2283.90
```

## Parameters

You can configure some parameters like the amount of workers or how the client chunks the input:

[patata.Patata](https://github.com/oalfonso-o/patata/blob/main/patata/client.py#L24) parameters:

- `num_workers`:
    - type: int
    - required: False
    - default: os.cpu_count()
    - description: Number of processes to open with multiprocessing
- `queue_max_size`:
    - type: int
    - required: False
    - default: 100.000
    - description: Maximum number of items that can be enqueued. This default number proved to not blow up the memory and to have enough items in the queue to have always work to do with 8 processes. Feel free to adjust it, just watch out the memory usage.
- `input_chunk_size`:
    - type: int
    - required: False
    - default: 10.000
    - description: This is the size of the chunks for the input. We will be reading the input iterator in chunks of this size up to `queue_max_size`.
- `pool_submit_size`:
    - type: int
    - required: False
    - default: 1.000
    - description: Each chunk of `input_chunk_size` will also be chunked to minor chunks of this size before being submited to the pool. The workers will be consuming chunks of this size and each of these chunks will be requested in an event loop.


[patata.Patata.http](https://github.com/oalfonso-o/patata/blob/main/patata/client.py#L42)

Parameters:

- `method`:
    - type: str
    - required: True
    - description: Specify the method of the requests. Valid values: GET, POST.
- `requests`:
    - type: Iterable[Tuple[int, str]]
    - required: True
    - description: Provide the tuples containing the ID of the request and the URL to be requested.

Response: Generator[Tuple[int, str], None, None]. For each input tuple an output tuple will be returned containing the same ID + the JSON of the response.


TODO:
- use pydantic models
- allow specifying custom callbacks after resolving the request, to post-process each response taking benefit of multiprocessing
- flag to disable multiprocessing
- add flag to specify how many requests can fail, this will need to specify also which codes are "ok" or which are "not ok" do decide when to increment this count and decide to stop
- include the missing methods like PUT, DELETE, etc