{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "DALQzKA3Wh21"
   },
   "source": [
    "# Meta Engines\n",
    "\n",
    "In this notebook we define an oversubscription planning problem and we solve it using a `MetaEngine`.\n",
    "\n",
    "[![Open In GitHub](https://img.shields.io/badge/see-Github-579aca?logo=github)](https:///github.com/aiplan4eu/unified-planning/blob/master/docs/notebooks/06-oversubscription-with-metaengine.ipynb)\n",
    "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/aiplan4eu/unified-planning/blob/master/docs/notebooks/06-oversubscription-with-metaengine.ipynb)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "JnUrQV84Pk4_"
   },
   "source": [
    "### Setup the library"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "WUlNZ8pVPk5A"
   },
   "source": [
    "First, we install unified_planning library and its dependencies from PyPi. Here, we use the `--pre` flag to use the latest development build."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "id": "fhMfqhBqPk5B",
    "tags": [
     "remove_from_CI"
    ]
   },
   "outputs": [],
   "source": [
    "!apt install graphviz graphviz-dev\n",
    "%pip install unified-planning[tamer,plot]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "p1p0OvjqW4g0"
   },
   "source": [
    "### Problem definition\n",
    "\n",
    "We model an oversubscription planning problem."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "id": "nc4DQP3kq-QP"
   },
   "outputs": [],
   "source": [
    "from unified_planning.shortcuts import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "JXDIDNwCm13b"
   },
   "source": [
    "We start the problem modeling defining the `UserType` and the `Fluent`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "id": "nHXzkD2LmWkB"
   },
   "outputs": [],
   "source": [
    "Location = UserType('Location')\n",
    "Robot = UserType('Robot')\n",
    "\n",
    "at = Fluent('at', BoolType(), robot=Robot, location=Location)\n",
    "visited = Fluent('visited', BoolType(), robot=Robot, location=Location)\n",
    "connected = Fluent('connected', BoolType(), l_from=Location, l_to=Location)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "OVUn5VPTmXKY"
   },
   "source": [
    "We define an action `move` that models the movement of a robot between two locations.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "id": "cwXh99K2laqv"
   },
   "outputs": [],
   "source": [
    "move = InstantaneousAction('move', robot=Robot, l_from=Location, l_to=Location)\n",
    "robot = move.parameter('robot')\n",
    "l_from = move.parameter('l_from')\n",
    "l_to = move.parameter('l_to')\n",
    "move.add_precondition(at(robot, l_from))\n",
    "move.add_precondition(connected(l_from, l_to))\n",
    "move.add_effect(at(robot, l_from), False)\n",
    "move.add_effect(at(robot, l_to), True)\n",
    "move.add_effect(visited(robot, l_to), True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "C92YH1nrmRb7"
   },
   "source": [
    "We define the `Object` instances and, after creating the `Problem`, we set the initial values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "id": "MczotRommR_k"
   },
   "outputs": [],
   "source": [
    "r1 = Object('r1', Robot)\n",
    "NLOC = 10\n",
    "locations = [Object('l%s' % i, Location) for i in range(NLOC)]\n",
    "\n",
    "problem = Problem('robot_with_simulated_effects')\n",
    "problem.add_fluent(at, default_initial_value=False)\n",
    "problem.add_fluent(visited, default_initial_value=False)\n",
    "problem.add_fluent(connected, default_initial_value=False)\n",
    "problem.add_action(move)\n",
    "\n",
    "problem.add_object(r1)\n",
    "problem.add_objects(locations)\n",
    "\n",
    "problem.set_initial_value(at(r1, locations[0]), True)\n",
    "problem.set_initial_value(visited(r1, locations[0]), True)\n",
    "for i in range(NLOC - 1):\n",
    "    problem.set_initial_value(connected(locations[i], locations[i+1]), True)\n",
    "problem.set_initial_value(connected(locations[4], locations[8]), True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "57hCstd_ZzK5"
   },
   "source": [
    "Finally, we define the oversubscription goals."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "id": "IyVivfPuZzcd"
   },
   "outputs": [],
   "source": [
    "goals = {}\n",
    "goals[visited(r1, locations[5])] = -5\n",
    "goals[visited(r1, locations[7])] = 4\n",
    "goals[visited(r1, locations[9])] = 10\n",
    "\n",
    "problem.add_quality_metric(up.model.metrics.Oversubscription(goals))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "pXvh83ljlabv"
   },
   "source": [
    "### Solving the problem"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "McV4znupqpkw"
   },
   "source": [
    "We solve the problem using the oversubscription `MetaEngine` with the tamer `Engine`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "id": "bXHWJh2vl5RJ"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "OversubscriptionPlanner[Tamer] returned: SequentialPlan:\n",
      "    move(r1, l0, l1)\n",
      "    move(r1, l1, l2)\n",
      "    move(r1, l2, l3)\n",
      "    move(r1, l3, l4)\n",
      "    move(r1, l4, l8)\n",
      "    move(r1, l8, l9)\n"
     ]
    }
   ],
   "source": [
    "with OneshotPlanner(name='oversubscription[tamer]') as planner:\n",
    "    result = planner.solve(problem)\n",
    "    print(\"%s returned: %s\" % (planner.name, result.plan))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "cUnXZCL4eNyj"
   },
   "source": [
    "To test the oversubscription `MetaEngine`, we update the oversubscription goals to see if it finds a different plan."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "id": "jJGRbo7weLUO"
   },
   "outputs": [],
   "source": [
    "problem.clear_quality_metrics()\n",
    "\n",
    "goals = {}\n",
    "goals[visited(r1, locations[5])] = -5\n",
    "goals[visited(r1, locations[7])] = 6\n",
    "goals[visited(r1, locations[9])] = 10\n",
    "\n",
    "problem.add_quality_metric(up.model.metrics.Oversubscription(goals))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "jpKO5yUKdwlA"
   },
   "source": [
    "To solve the new problem, now we let the system choose the `Engine` to use."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "id": "HBqoZ3-ldyD1",
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "OversubscriptionPlanner[Fast Downward] returned: SequentialPlan:\n",
      "    move(r1, l0, l1)\n",
      "    move(r1, l1, l2)\n",
      "    move(r1, l2, l3)\n",
      "    move(r1, l3, l4)\n",
      "    move(r1, l4, l5)\n",
      "    move(r1, l5, l6)\n",
      "    move(r1, l6, l7)\n",
      "    move(r1, l7, l8)\n",
      "    move(r1, l8, l9)\n"
     ]
    }
   ],
   "source": [
    "with OneshotPlanner(problem_kind=problem.kind) as planner:\n",
    "    result = planner.solve(problem)\n",
    "    plan = result.plan\n",
    "    assert plan is not None\n",
    "    print(\"%s returned: %s\" % (planner.name, result.plan))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from unified_planning.plot import plot_sequential_plan"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ignore the code below, it's used to make this notebook also runnable in the Countinuous Intergation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Redefine the plot package methods imported above to print the plot to a temp file\n",
    "# if the exception \"could not locate runnable browser\" is raised. This usually happens\n",
    "# in the Continuous Integration.\n",
    "\n",
    "from inspect import getmembers, isfunction\n",
    "from unified_planning import plot\n",
    "from functools import partial\n",
    "import os, uuid, tempfile as tf\n",
    "\n",
    "# Define the function that will be executed instead\n",
    "def _function(original_function, *args, **kwargs):\n",
    "    try:\n",
    "        original_function(*args, **kwargs)\n",
    "    except Exception as e:\n",
    "        if \"could not locate runnable browser\" in str(e):\n",
    "            original_function(*args, **kwargs,\n",
    "                filename=f\"{os.path.join(tf.gettempdir(), str(uuid.uuid1()))}.png\"\n",
    "            )\n",
    "        else:\n",
    "            raise e\n",
    "\n",
    "# Iterate over all the functions of the plot package\n",
    "for function_name, function in getmembers(plot, isfunction):\n",
    "    # Override the original function with the new one\n",
    "    globals()[function_name] = partial(_function, function)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "plot_sequential_plan(plan, figsize=(18, 4), node_size=8000, font_size=11)"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "Oversubscription using MetaEngines",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}