{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Async and Await Primer"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A quick primer on `async`/`await`. Async and await are relatively new features in Python which allow **concurent** programming. They won't make your code magically faster, but may make your code easier to read, maintain and reason about. \n",
    "You will likely hear the terms event-loop, coroutines and many other ones, they will make sens in time. \n",
    "\n",
    "The key thing to remember is that \n",
    " - async-functions can call both sync and async functions.\n",
    " - sync functions can only call sync. \n",
    " - You _must_ always `await <async functions>`\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Event loop\n",
    "\n",
    "It's like the \"One ring\", there shoudl be only one. IPython (and Jupyter) usually already run one.\n",
    "\n",
    "### Bad news\n",
    "If you need to run any code that need to create and manage an event-loop, consult the docs. \n",
    "Typically you can't run a tornado app inside jupyter.\n",
    "\n",
    "### Good news\n",
    "\n",
    "If you don't know/don't care, all is already setup for you. \n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example\n",
    "\n",
    "Let's deactivate enventloop integration and try what is (usually invalid Python)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%autoawait False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from asyncio import sleep"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# does not sleep, need to be awaited\n",
    "print('before sleep')\n",
    "sleep(5)\n",
    "print('after sleep')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "await sleep(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def f():\n",
    "    await sleep(5)\n",
    "f()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "async def f():\n",
    "    print('before...')\n",
    "    await sleep(5)\n",
    "    print('after')\n",
    "### does not call f\n",
    "f()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "await f()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "... back to step beginning. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Autoawait"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Autoawait will _attempt_ to detect async code and run it for you. There are of course limitations (bug report welcome)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%autoawait True"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You will note that any line that start with `%` is invalid Python and are IPython specific syntax. Those are call magics (line-magics with a single `%` sign, cell magics with a double `%%` sign)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print('before')\n",
    "await sleep(5)\n",
    "print('after')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Top level await is now valid syntax. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tpl = 'https://anapioficeandfire.com/api/characters/{}'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%time\n",
    "\n",
    "results = []\n",
    "for i in range(1,50):\n",
    "    import requests\n",
    "    print('.', end='')\n",
    "    r = requests.get(tpl.format(i)).json()['aliases']\n",
    "    print('x', end='')\n",
    "    results.append(r)\n",
    "    \n",
    "for r in results:\n",
    "    print(r)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Moving to asynchronous"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Nothing is perfect; if you get RuntimeErrors with asyncio, you may need to restart your kernel. More during my colleagues aiohttp tutorial this Afternoon"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import aiohttp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "async with aiohttp.ClientSession() as session:\n",
    "    response = await session.get(tpl.format(583))\n",
    "    json = await response.json()\n",
    "    print(json['aliases'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "async def get_char(i, session):\n",
    "        print('.', end='')\n",
    "        response = await session.get(tpl.format(i))\n",
    "        json = await response.json()\n",
    "        print('x', end='')\n",
    "        return json['aliases']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "async with aiohttp.ClientSession() as s:\n",
    "    print(await get_char(1303, s))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tasks = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import asyncio\n",
    "async with aiohttp.ClientSession() as session:\n",
    "    # start \n",
    "    for i in range(1,50):\n",
    "        task = asyncio.ensure_future(get_char(i, session))\n",
    "        tasks.append(task)\n",
    "    results = await asyncio.gather(*tasks)\n",
    "    for r in results:\n",
    "        print(r)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Advance Autoawait usage, Exercise"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Find the documentation for autoawait, and try to make it work with another asynchronous library. For exampe try to ply with [`trio`](https://trio.readthedocs.io/en/latest/), using `trio.sleep`  and `trio.open_nursery` to get several concurent task running, pritning different message regularly and at random intervals. What happen if you use `time.sleep()` instead of `trio.sleep()` ? What hapen if you use `asyncio.sleep()` ?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import trio"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "async def every(n, message):\n",
    "    for i in range(30):\n",
    "        await trio.sleep(n)\n",
    "        print(message)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%autoawait trio"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "async with trio.open_nursery() as nursery:\n",
    "    nursery.start_soon(every, 1, 'Plic')\n",
    "    nursery.start_soon(every, 2, 'Ploc')\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}