async demo

async-demo.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "async demo",
      "private_outputs": true,
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyPcsjr1l75LGTZBqErN4YO+",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/gist/thanakijwanavit/d473c84a498a8b37dfdff0a5991a9710/async-demo.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8PhepkTsbfqf"
      },
      "source": [
        "!pip install -q --upgrade nicHelper==0.0.61"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "I2Jtm22SLMGZ"
      },
      "source": [
        "from time import sleep"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QXxXObKnb6U_"
      },
      "source": [
        "def aSlowFunc(input_:str):\n",
        "  sleep(2)\n",
        "  return input_"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Od-I_V45cHz2"
      },
      "source": [
        "# 1. asyncMap (threading backend)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OF10fuS0cFAK"
      },
      "source": [
        "from nicHelper.asyncUtil import asyncThreadMap\n",
        "%time r = asyncThreadMap(aSlowFunc, range(100), threads=100)\n",
        "print(list(r)[:10])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1AsbzuqPkY_A"
      },
      "source": [
        "%%time\n",
        "### multiple inputs\n",
        "input_ = list(zip(range(10), range(1,11)))\n",
        "print(input_)\n",
        "\n",
        "list(asyncThreadMap(lambda x: (lambda x,y: x+y )(x[0],x[1]), input_))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WvMINT9g3vxO"
      },
      "source": [
        "# 2. multiProcessing (use multiple CPU), unlock GIL"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2GxDjBoz34IG"
      },
      "source": [
        "from nicHelper.asyncUtil import asyncProcessMap\n",
        "%time r = asyncProcessMap(aSlowFunc, range(100), threads=100)\n",
        "print(list(r)[:10])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZFAzbIh3ezGi"
      },
      "source": [
        "# 3. use async await"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GNdPTmUIkEEE"
      },
      "source": [
        "# async await helper (cheating)\n",
        "exactly the same as asyncMap but under the hood its using async await instead\n",
        "\n",
        "> its not as fast as you can see, but threads are automatically determined \\n\n",
        "> Note that this thread is subjected to GIl and will only run on 1 core"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "um-mfH2IoCZ_"
      },
      "source": [
        "from nicHelper.asyncUtil import asyncAwaitMap\n",
        "import nest_asyncio\n",
        "nest_asyncio.apply() # unlock nested event loop"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "URVLB_zQkLGP"
      },
      "source": [
        "%time r = asyncAwaitMap(aSlowFunc, range(100))\n",
        "print(r[:10])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "npYZrkpF5nUH"
      },
      "source": [
        "# doing things from scratch"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rn5Kk7JOkLyU"
      },
      "source": [
        "## async await from scratch"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fj8WTLyEe14J"
      },
      "source": [
        "from nicHelper.asyncUtil import async_wrap\n",
        "import asyncio, requests, time, nest_asyncio"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vm_iboT1f5dV"
      },
      "source": [
        "## unlock the nested event loop"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w7J8VJtKO6Yg"
      },
      "source": [
        "nest_asyncio.apply()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "i_o9MtQlgBXA"
      },
      "source": [
        "if you have a non-async function, need to decorate it to convert it into an async function"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DPF_Kq4SLUqg"
      },
      "source": [
        "@async_wrap\n",
        "def aSlowFunc(input_:str):\n",
        "  time.sleep(2)\n",
        "  return input_"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bgFGt-hugYPn"
      },
      "source": [
        "Otherwise you can just write it like this"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "X7QV7PbQL76F"
      },
      "source": [
        "async def asyncSlowFunction(input_):\n",
        "  r = await aSlowFunc(input_)\n",
        "  return r"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2IWT4a6Hgnkx"
      },
      "source": [
        "set up a run loop function (or can call it main)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9QfrrkH6MdXu"
      },
      "source": [
        "async def runLoop():\n",
        "  rtup = (aSlowFunc(i) for i in range (10))\n",
        "  r = await asyncio.gather(*rtup)\n",
        "  return r"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nwvp4fM-NBnm"
      },
      "source": [
        "%%time \n",
        "asyncio.run(runLoop())"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Y06Fg1VY5tG6"
      },
      "source": [
        "# ThreadPool Executor"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KgNqoP5wU_cj"
      },
      "source": [
        "%%time\n",
        "from concurrent.futures import ThreadPoolExecutor\n",
        "def aSlowFunc(input_:str):\n",
        "  time.sleep(2)\n",
        "  return input\n",
        "with ThreadPoolExecutor(100) as tr:\n",
        "  tr.map(aSlowFunc, range(100))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "u7dA32NF5wnN"
      },
      "source": [
        "# ProcessPoolExecutor"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mDueZ8XtzXkQ"
      },
      "source": [
        "%%time\n",
        "from concurrent.futures import ProcessPoolExecutor\n",
        "def aSlowFunc(input_:str):\n",
        "  time.sleep(2)\n",
        "  return input\n",
        "with ProcessPoolExecutor(100) as tr:\n",
        "  tr.map(aSlowFunc, range(100))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OXrES_Rfz-HM"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}