{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# TileDB Quickstart Delayed Notebook"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This notebook will walk through usage of TileDB Cloud Delayed APIs\n",
    "\n",
    "We will start off using the high level delayed API and finish with a quick look at the lower level DAG functionality.\n",
    "\n",
    "First let's import the necessary packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import base packages\n",
    "import tiledb\n",
    "import tiledb.cloud\n",
    "from tiledb.cloud.compute import Delayed, DelayedSQL, DelayedArrayUDF\n",
    "import numpy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Introduction To Delayed"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Generic Functions\n",
    "\n",
    "Any python function can be wrapped in a Delayed object making the function executable as a future"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = Delayed(numpy.median)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tiledb.cloud.compute.delayed.Delayed at 0x7fa03c00b978>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# You can see the type is now `Delayed`\n",
    "x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The function can be called with parameters to store it and lazily executed"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tiledb.cloud.compute.delayed.Delayed at 0x7fa03c00b978>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x([1,2,3,4,5])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To force an execution call `compute()`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3.0"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x.compute()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### SQL and Arrays\n",
    "\n",
    "Besides arbitrary python functions, serverless sql queries and array based UDFs can also be called with the delayed API"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>AVG(`a`)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   AVG(`a`)\n",
       "0         2"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# SQL\n",
    "y = DelayedSQL(\"select AVG(`a`) FROM `tiledb://TileDB-Inc/quickstart_sparse`\")\n",
    "\n",
    "# Run query\n",
    "y.compute()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Array\n",
    "z = DelayedArrayUDF(\"tiledb://TileDB-Inc/quickstart_sparse\", lambda x: numpy.average(x[\"a\"]))([(1, 4), (1, 4)])\n",
    "\n",
    "# Run the udf on the array\n",
    "z.compute()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Local Functions\n",
    "\n",
    "Lastly it is also possible to include a generic python function as delayed but have it run locally instead of serverlessly. This is useful for testing or for saving finalized results to your local machine, i.e. saving a image.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.0"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "local = Delayed(numpy.median, local=True)([1,2,3])\n",
    "local.compute()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Task Graphs\n",
    "\n",
    "Delayed objects can be combined into a task graph. Output from one function or query can be passed into another, and dependencies are automatically determined."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Build several delayed objects to build in a graph\n",
    "local = Delayed(lambda x: x * 2, local=True)(100)\n",
    "array_apply = DelayedArrayUDF(\"tiledb://TileDB-Inc/quickstart_sparse\", lambda x: numpy.sum(x[\"a\"]), name=\"array_apply\")([(1, 4), (1, 4)])\n",
    "sql = DelayedSQL(\"select SUM(`a`) as a from `{}`\".format(\"tiledb://TileDB-Inc/quickstart_dense\"), name=\"sql\")\n",
    "\n",
    "# Custom function to use to average all the results we are passing in\n",
    "def mean(local, array_apply, sql):\n",
    "    import numpy\n",
    "    return numpy.mean([local, array_apply, sql.iloc(0)[0]])\n",
    "\n",
    "res = Delayed(func_exec=mean, name=\"node_exec\")(local, array_apply, sql)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A live graph can show the status of the Task Graph"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b0d8ebc128144b84bdaf4f32aef90cb0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "FigureWidget({\n",
       "    'data': [{'hoverinfo': 'none',\n",
       "              'line': {'color': '#888', 'width': 0.5},\n",
       "     …"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "res.visualize(force_plotly=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "114.0"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res.compute()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Advanced Delayed Usage\n",
    "\n",
    "There are several functionalities which are exposed to allow for complex use cases"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Manually specifying dependencies\n",
    "\n",
    "There might be cases were a function relies on another function but does not take it's arguments. An example would be if a function manipulated data on S3 but did not return anything."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# A few base functions:\n",
    "import random\n",
    "node_1 = Delayed(numpy.median, local=True, name=\"node_1\")([1, 2, 3])\n",
    "node_2 = Delayed(lambda x: x * 2, local=True, name=\"node_2\")(node_1)\n",
    "node_3 = Delayed(lambda x: x * 2, local=True, name=\"node_3\")(node_2)\n",
    "\n",
    "nodes_by_name= {'node_1': node_1, 'node_2': node_2, 'node_3': node_3}\n",
    "#Function which sleeps for some time so we can see the graph in different states\n",
    "def f():\n",
    "    import time\n",
    "    import random\n",
    "    x = random.randrange(0, 30)\n",
    "    time.sleep(x)\n",
    "    return x\n",
    "\n",
    "# Randomly add 96 other nodes to the graph. All of these will use the sleep function\n",
    "for i in range(4, 100):\n",
    "    name = \"node_{}\".format(i)\n",
    "    node = Delayed(f, local=True, name=name)()\n",
    "    \n",
    "    dep = random.randrange(1, i-1)\n",
    "    # Randomly set dependency on one other node\n",
    "    node_dep = nodes_by_name[\"node_{}\".format(dep)]\n",
    "    # Force the dependency to be set\n",
    "    node.depends_on(node_dep)\n",
    "    \n",
    "    nodes_by_name[name] = node\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "30e18e102c934783a911215cd88078cb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Visualize(value='{\"nodes\": [\"node_1\", \"node_2\", \"node_4\", \"node_11\", \"node_20\", \"node_61\", \"node_3\", \"node_5\",…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "node_1.visualize()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.0"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "node_99 = nodes_by_name[\"node_99\"]\n",
    "node_99.compute()"
   ]
  },
  {
   "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": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}