dsu-scanner.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from matplotlib import pyplot as plt\n",
    "import time\n",
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def draw_crosses(h, w):\n",
    "    A = np.zeros((h, w))\n",
    "    for i, row in enumerate(A):\n",
    "        for j, cell in enumerate(row):\n",
    "            # crosses\n",
    "            if  -2 < ((j // 4) % (A.shape[0]) - i) < 1 and 1 < i < A.shape[0] - 2:\n",
    "                A[i, j] = 1       \n",
    "            if  0 < (A.shape[0] - ((j // 4) % (A.shape[0]) + i)) < 3 and 1 < i < A.shape[0] - 2:\n",
    "                A[i, j] = 1\n",
    "\n",
    "            # circles\n",
    "            # if (i - A.shape[0] / 2) ** 2 + ((j//4) % (A.shape[0]) - A.shape[0] / 2) ** 2 < .2 * (A.shape[0] ** 2):\n",
    "            #     A[i, j] = 1\n",
    "    return A\n",
    "\n",
    "A = draw_crosses(10, 400)\n",
    "plt.figure(figsize=(15,10))\n",
    "plt.imshow(A, cmap=\"gray\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Context:\n",
    "    def __init__(self, width, row=0):\n",
    "        self.DSU = dict()\n",
    "        self.last = [None] * width\n",
    "        self.width = width\n",
    "        self.row = row\n",
    "    \n",
    "    def eat(self, item):\n",
    "        # print(\"Bounding box ready:\", item.get_bbox())\n",
    "        item.dead = True\n",
    "        t, r, b, l = item.get_bbox()\n",
    "        A[r, t:b+1] = 2\n",
    "        A[l, t:b+1] = 2\n",
    "        A[l:r+1, t] = 2\n",
    "        A[l:r+1, b] = 2\n",
    "    \n",
    "    def reset(self):\n",
    "        for k, v in self.DSU.items():\n",
    "            v.touched = False\n",
    "    \n",
    "    def swallow(self):\n",
    "        for k, v in self.DSU.items():\n",
    "            # process untouched roots - print, draw and remove\n",
    "            if not v.touched and k == v.id(): self.eat(v)\n",
    "\n",
    "        to_remove = set(k for k, v in self.DSU.items() if v.find().dead)                \n",
    "        for k in to_remove: del self.DSU[k]                \n",
    "    \n",
    "class Node:\n",
    "    def __init__(self, col, weight, context):\n",
    "        self.context = context\n",
    "        self.row = context.row\n",
    "        self.col = col\n",
    "        self.weight = weight\n",
    "        \n",
    "        self.touched = True\n",
    "        self.dead = False\n",
    "        \n",
    "        self.top = self.row\n",
    "        self.bottom = self.row\n",
    "        self.left = self.col\n",
    "        self.right = self.col\n",
    "\n",
    "        self.context.DSU[self.id()] = self\n",
    "    \n",
    "    def parent(self):\n",
    "        return self.context.DSU[self.id()]\n",
    "    \n",
    "    def find(self, origin=None):\n",
    "        if origin is None: origin = self\n",
    "        self.touched = True    \n",
    "        self.top = min(origin.top, self.top)\n",
    "        self.bottom = max(origin.bottom, self.bottom)\n",
    "        self.left = min(origin.left, self.left)\n",
    "        self.right = max(origin.right, self.right)\n",
    "        if self.id() != self.parent().id():\n",
    "            self.context.DSU[self.id()] = self.parent().find(origin)\n",
    "        return self.context.DSU[self.id()]\n",
    "\n",
    "    def union(self, other):\n",
    "        i, j = self.find(), other.find()\n",
    "        if i.id() != j.id():\n",
    "            self.context.DSU[i.id()] = j\n",
    "            # update sizes\n",
    "            self.find(other)\n",
    "            other.find(self)\n",
    "\n",
    "    def connected(self, other):\n",
    "        return self.find().id() == other.find().id()\n",
    "\n",
    "    def id(self):\n",
    "        return self.row * self.context.width + self.col\n",
    "    \n",
    "    def get_bbox(self):\n",
    "        return [self.top, self.right, self.bottom, self.left]\n",
    "            \n",
    "    def __str__(self):\n",
    "        return \"[{}-{}, {}-{}][{}{}]\".format(\n",
    "                    self.top, self.bottom, self.left, self.right, \n",
    "                    \"+\" if self.dead else \" \", \n",
    "                    \"T\" if self.touched else \" \")\n",
    "    \n",
    "    def __repr__(self):\n",
    "        return str(self)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def connect(node, neighbours, context):\n",
    "    for neighbour in neighbours:\n",
    "        if neighbour is not None:\n",
    "            node.union(neighbour)\n",
    "    \n",
    "def process(line, context):\n",
    "    assert context.width == len(line), \"Line size don't match\"\n",
    "    context.reset()\n",
    "    r = [None] * context.width\n",
    "    for col in range(context.width):\n",
    "        if line[col] > 0:\n",
    "            r[col] = Node(col, line[col], context)\n",
    "            connectivity = [context.last[col]] if i == 0 else [context.last[col], r[col-1]]\n",
    "            connect(r[col], connectivity, context)\n",
    "    context.last = r\n",
    "    context.row += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "context = Context(A.shape[0])\n",
    "\n",
    "for i in range(A.shape[1]):\n",
    "    process(A[:, i], context)\n",
    "    context.swallow()\n",
    "    \n",
    "process([0] * A.shape[0], context)\n",
    "context.swallow()\n",
    "    \n",
    "plt.figure(figsize=(15,10))\n",
    "plt.imshow(A, cmap=\"gray\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4 / 81 | 8 / 81 | 12 / 81 | 16 / 81 | 20 / 81 | 24 / 81 | 28 / 81 | 32 / 81 | 36 / 81 | 40 / 81 | 44 / 81 | 48 / 81 | 52 / 81 | 56 / 81 | 60 / 81 | 64 / 81 | 68 / 81 | 72 / 81 | 76 / 81 | 80 / 81 | 4 / 81 | 8 / 81 | 12 / 81 | 16 / 81 | 20 / 81 | 24 / 81 | 28 / 81 | 32 / 81 | 36 / 81 | 40 / 81 | 44 / 81 | 48 / 81 | 52 / 81 | 56 / 81 | 60 / 81 | 64 / 81 | 68 / 81 | 72 / 81 | 76 / 81 | 80 / 81 | 4 / 81 | 8 / 81 | 12 / 81 | 16 / 81 | 20 / 81 | 24 / 81 | 28 / 81 | 32 / 81 | 36 / 81 | 40 / 81 | 44 / 81 | 48 / 81 | 52 / 81 | 56 / 81 | 60 / 81 | 64 / 81 | 68 / 81 | 72 / 81 | 76 / 81 | 80 / 81 | "
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiMAAAEKCAYAAADAYVXSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzs3Xd4lFUa9/HvmUmf9EZIAiR0aVICwtKrBQKyYlteXEUpKsqiiALCAoqKumT1BQLIEpDmqwLSdcEFEZClKAICUmNIgJAOCWkzc94/ArPEgAzKZFLuz3V5kXnq/QQhP855zjlKa40QQgghhLMYnF2AEEIIIao3CSNCCCGEcCoJI0IIIYRwKgkjQgghhHAqCSNCCCGEcCoJI0IIIYRwKgkjQgghhHAqCSNCCCGEcCoJI0IIIYRwKhdnF3C94OBgHRUV5ewyhBBCCHEH7N+/P11rHXKr4ypUGImKimLfvn3OLkMIIYQQd4BS6hd7jpNuGiGEEEI4lYQRIYQQQjiVhBEhhBBCOFWFemfkRoqLi0lOTqagoMDZpVR5Hh4eREZG4urq6uxShBBCVCMVPowkJyfj4+NDVFQUSilnl1Nlaa3JyMggOTmZ6OhoZ5cjhBCiGqnw3TQFBQUEBQVJEHEwpRRBQUHSAiWEEKLcVfgwAkgQKSfyfRZCiOrlv//9L6+99hpaa6fWUeG7aYQQQghx55jNZj799FM2btxIcnIy33zzDb1796Znz55Oq6lStIwI+125coW+ffvSuHFjmjZtymuvvVbmmM8//xyllEwwJ4QQ1ciZM2d46qmniI6OZvDgweTl5VG/fn08PT354IMPnFqbhJEqaOzYsRw7dowffviBnTt3smnTJtu+y5cv8+GHH3LPPfc4sUIhhBDlZfv27QwcOJCHHnqIQ4cO2d4NPHz4MAAPP/wwkyZNcmaJEkZuJTExkWbNmtk+e3t7277u1KmT7Tfz+u2dO3emX79+AEyZMoUhQ4bQo0cPGjRowEcffQTAtm3bbMdkZmbi5+fH+++/D0C3bt1KtVpcu3Zubi49e/akdevWNG/enDVr1pSp18vLi+7duwPg5uZG69atSU5Otu2fNGkS48aNw8PD4w98V4QQQlRkVquVZcuWMWTIEJYtW4bZbMZgKPmRHxoaSmhoKI8//jjjxo1j8eLFtG3b1qn1Vqp3Rhz5guWdenlnw4YN5OTk4OfnZ9t28OBBdu/eTV5eHq1ataJv376lznn77bepU6fOLa/t4eHB6tWr8fX1JT09nfbt29O/f/+bfl+ys7NZt24do0ePBuCHH37g7Nmz9OvXzxZ8hBBCVB3nz59nxowZrF+/noyMDB555BEAatasycWLFzGbzcTGxjJp0iRMJpOTq/2fShVGKjqtNdOnT2fChAksXbrUtn3AgAF4enri6elJ9+7d2bNnD/7+/gCkpKSwe/duBg4cWOpagwcPxtPTE4D8/Hzb9SdMmMD27dsxGAykpKSQmppKWFhYmVrMZjOPP/44L774InXr1sVqtTJmzBgWLVrkoKcXQgjhLJs2bWLOnDkkJydjNBo5f/48V65c4ezZs0RERFBcXMwLL7zAX/7yF1sLSUUiYeQOWrFiBd26dSsTDn7dcnH956lTpzJp0iR27dpV6phly5YRExMD/K+bZtmyZaSlpbF//35cXV2Jioq66bwgw4cPp0GDBvztb38DSt4VOXz4MN26dQPgwoUL9O/fn7Vr19ruI4QQovKwWq3MnDmT/fv34+PjQ35+PkajESjpisnOzqZx48b89a9/5e6773Zytb+t4sWj36C1dth/f5TVaiUuLo5x48aV2bdmzRoKCgrIyMhg27Zttr65U6dOkZiYSJ8+fey6R05ODqGhobi6urJ161Z++eXGKzO//vrr5OTk8M9//tO2zc/Pj/T0dBITE0lMTKR9+/YSRIQQohLKzMxkzJgxxMTEsH37dnx9fVFKERoaCpT8PBowYADJycnMnDmzwgcRkJYRu5w5c4ZOnToBJV0m174+dOiQ7Zj8/HwGDRpk6365Xrt27ejbty9JSUlMmjSJ8PBwjh8/zrFjx0hISLC7jsGDBxMbG0tMTAwtW7akcePGZY5JTk5m+vTpNG7cmNatWwMwatQonnnmmdt6ZiGEEBXLnj17ePfddzl16hRGoxGDwUBaWhphYWG2FvcnnniC559/3tZCUlkoZ8+6dr2YmBj967kvjh49yl133eWkiv64KVOm4O3tzdixY51dil0q+/dbCCGqEovFwty5c9mxYweenp4cOnSoVGu+1Wqlfv36DBkyhNjYWCdWemNKqf1a61s2wUvLiBBCCFHBJCUl8c4777B7926aN2+Or68vAIGBgWRkZGCxWGjbti2TJ08mMjLSydX+cRJGHGzKlCnOLkEIIUQlYLVaWbJkCR9//DFZWVkYDAYMBgPp6em2wOHt7U2PHj0YM2YMbm5uTq74zpEwIoQQQjhRSkoK8+bN4/Tp02RnZ5OTk1Nq+G1aWhq+vr7Exsbyl7/8xYmVOo6EESGEEKKcWa1W1q1bx8qVK9m6dSsPPPAAJpMJd3d3Ll68iNVqRWtNp06deO2116hZs6azS3Yoh4YRpVQicBmwAGZ7XmIRQgghqqrTp08zY8YM9u7di9Vq5eDBg2itycjIICgoCK01kZGRDB48mIEDB1bICcocoTxaRrprrdPL4T5CCCFEhaO1ZtWqVWzcuBGlFEePHrW9DxIQEEBmZibZ2dm0atWK4cOHExIS4uySy131iFzVzMSJE6lVq1apxfuu+fTTT2nSpAlNmzatsn2PQghREZw6dYrJkydz1113ERcXh4uLC0aj0TYfldVqpXPnzuzbt48tW7YwceLEahlEwPEtIxr4t1JKA/O01vMdfD8BxMbGMmrUKBo0aFBq+4kTJ3j77bfZuXMnAQEBXLx40UkVCiFE1XRtXpDPP/+cnJwcrly5ws8//0xaWhpNmzYFwN/fn4YNGzJ+/HjbrKnVnaNbRjpqrVsD9wPPK6W6/PoApdRwpdQ+pdS+tLQ0B5dz+xITE2nWrJnt8/WtDZ06deLw4cNltnfu3Jl+/foBJUN7hwwZQo8ePWjQoAEfffQRANu2bbMdk5mZiZ+fn20l3W7dunH95G/Xrp2bm0vPnj1p3bo1zZs3Z82aNTesuX379jd82emjjz7i+eefJyAgAED+EAghxB1y7Ngxhg0bxtChQ9m3bx+XL1/GYDDg7e2Nh4cHxcXFKKUYNGgQy5YtIy4uTv4Ovo5DW0a01ueu/npRKbUaaAds/9Ux84H5UDID662uOW/ePNsP9D+qc+fOxMXF3ZFrXbNhwwZycnLw8/OzbTt48CC7d+8mLy+PVq1a0bdv31LnvP3229SpU+eW1/bw8GD16tX4+vqSnp5O+/bt6d+/f5mF+G7m+PHjAHTs2BGLxcKUKVO47777buPphBBCXG/16tXMmTOHzMxM7r77bjw8PADw9fXl0qVLAEyaNIkXX3zxhl3nooTDwohSygQYtNaXr37dB5jmqPtVBFprpk+fzoQJE1i6dKlt+4ABA/D09MTT05Pu3buzZ88eW59hSkoKu3fvZuDAgaWuNXjwYDw9PYGSdW+uXX/ChAls374dg8FASkoKqampZVYJvhmz2cyJEyfYtm0bycnJdO7cmcOHD99wPR0hhBA3lpuby/vvv8/GjRuxWq1AyWrsGRkZhIWFYTabCQoK4m9/+xv9+/d3crWVgyNbRmoAq6/+q90FWK61/tKB93O6FStW0K1btzLh4NctF9d/njp1KpMmTWLXrl2ljlm2bJltRd1raXrZsmWkpaWxf/9+XF1diYqKoqCgwO76IiMjad++Pa6urkRHR9OoUSNOnDhhW0VYCCHEzR06dIi4uDisViuZmZm2IHJNWloa0dHRjBs3jvDwcCdVWTk5LIxorU8Dd3zd4hEjRjBixIg7fdk/zGq1EhcXx+bNmzlw4ECpfWvWrGH8+PHk5eWxbds23nnnHY4fP86pU6coKiqiT58+ZcLIjeTk5BAaGoqrqytbt27ll19+ua0aH3zwQVasWMGTTz5Jeno6x48fp27durd1DSGEqG7Wrl3LrFmzsFgs1K9fH4Dg4GDOnz+PxWLB39+f4cOH89hjjzm50spLZmC1w5kzZ+jUqRNQ0mVy7etDhw7ZjsnPz2fQoEE37PJo164dffv2JSkpiUmTJhEeHs7x48c5duwYCQkJdtcxePBgYmNjiYmJoWXLljRu3PiGx40bN47ly5dz5coVIiMjeeaZZ5gyZQr33nsv//73v2nSpAlGo5H33nuPoKCg2/lWCCFEtXDlyhXi4uJYs2ZNqRaQoqIi25owDRs2ZMKECbZRMuL3U9cvRexsMTEx+vpRJFD5l7SfMmUK3t7ejB071tml2KWyf7+FEOKPOHv2LHPmzCElJYVTp05RWFhYan9gYCDNmzfn9ddft41MFDenlNpvz+zr0jIihBCi2vvxxx+ZNWsWZ8+epU6dOnh6euLn58fFixfRWuPj42PriqkuU7SXJwkjDjZlyhRnlyCEEOIGioqKiIuLY9WqVZw9e5bz588THh5um2ohKCgIHx8fJk6cSIsWLZxcbdUmYUQIIUS1kpKSwpw5czhy5Ahnz54FICQkhAsXLnDu3DkyMjK45557GDlyJD4+Pk6utnqQMCKEEKJaOHLkCB988AGFhYW4u7sTGhpKamoqRUVFuLi40LdvX1577TX+9Kc/2T2ZpLgzJIwIIYSo0jZv3sz06dPZt28feXl5dO3alUaNGgEQEBCAl5cXEyZMoFWrVk6utPqSMCKEEKJKWrt2LfHx8Vy8eBGlFOHh4Zw4cYLDhw8TGRlJs2bNePfdd0st3yGcQ8JIFbRixQreeust2x++pUuXEhwc7OyyhBCiXGzZsoV//etf+Pj4EBISwrVFWH19fbnvvvsYP348nTp1klExFYj8TlQxZrOZ0aNHs3XrVg4ePEiLFi2YNWuWs8sSQgiH27NnD3369GHixImYTCaUUnh6ehIQEICbmxvjx49n06ZNdOnSRYJIBSO/G7eQmJhIs2bNbJ+vX3WxU6dOHD58uMz2zp07069fP6BkaO+QIUPo0aMHDRo0sK04vG3bNtsxmZmZ+Pn58f777wPQrVs3rp/87dq1c3Nz6dmzJ61bt6Z58+asWbOmTL1aa7TW5OXlobXm0qVLskaCEKJK279/P/fffz/PPvssmZmZWCwWLl68CMDly5d55JFH2LVrFw899JCTKxU3U+m6adatW8f69esB6NevH7GxsaX2f/bZZ2zZsgWAQYMG0bt371L7lyxZwo4dO4CS6dW7dOlyR+vbsGEDOTk5pfogDx48yO7du8nLy6NVq1b07du31Dlvv/22bVz7b/Hw8GD16tX4+vqSnp5O+/bt6d+/f6m3vl1dXYmPj6d58+aYTCYaNGjA7Nmz79wDCiFEBbFr1y5mz55NRkYG6enppf4uTE9Pp0+fPowaNcqJFQp7ScvIHaS1Zvr06UyYMKHU9gEDBuDp6UlwcDDdu3dnz549tn0pKSns3r2bgQMHljpn8ODBtGzZkpYtW5Kfn2+7/oQJE2jRogW9evUiJSWF1NTUUucVFxcTHx/PDz/8wLlz52jRogVvv/22g55YCCHK39GjR3nooYdISEjA29ub8PBwW7eLwWDg2WefZdeuXRJEKpFK1zJSka1YsYJu3boRFhZWavuvx6tf/3nq1KlMmjSpzKq9y5YtIyamZDr/a900y5YtIy0tjf379+Pq6kpUVBQFBQWlzru2YnC9evUAeOSRR3jnnXfuwNMJIYRzHT9+nHHjxpGUlITRaMTX1xc3NzdcXV0JCAjgz3/+M8OGDZP3QSqhShdGYmNjy3TNXO/hhx/m4Ycfvun+IUOGMGTIkDtel9VqJS4ujs2bN9sCwTVr1qxh/Pjx5OXlsW3bNt555x2OHz/OqVOnKCoqok+fPmXCyI3k5OQQGhqKq6srW7du5ZdffilzTEREBEeOHCEtLY2QkBA2b94sC98JISq106dPM27cOE6fPo3BYMBgMKC15vz58wQHB9O0aVPmzJmD0Wh0dqnid6p0YcQZzpw5Q6dOnQDIz8+3fX3o0CHbMfn5+QwaNAh/f/8y57dr146+ffuSlJTEpEmTCA8P5/jx4xw7doyEhAS76xg8eDCxsbHExMTQsmVLGjduXOaY8PBw/v73v9OlSxdcXV2pU6cOixYtus0nFkII50tOTuYf//gHOTk5XLlypVSLh9aaPn36MHr0aAkhVYDSWju7BpuYmBh9/SgSqPxL2k+ZMgVvb2/Gjh3r7FLsUtm/30KIyi8rK4sPPviA//f//p9tkIHZbObIkSMUFRXx6KOPMnbsWOmOqQSUUvu11jG3Ok5aRoQQQlQIO3fu5P/+3//LTz/9xE8//YTWmsaNGxMaGkphYSH9+/fnlVdewcVFfnRVNfI76mBTpkxxdglCCFFhJSYmsnjxYk6cOEFiYiIFBQW4u7sTFBREeno658+fp3v37jz33HMSQqqwSvE7q7WWFRTLQUXqshNCVF2ZmZksX76cL7/8krS0NFq2bInJZMLf358LFy4AJS/j//Of/+Sxxx6Td0KqgQofRjw8PMjIyCAoKEgCiQNprcnIyMDDw8PZpQghqqC8vDzmzp3Lxo0bycrKoqCggKNHj+Lv70/Lli0BCA0NJTs7mz59+jBu3Dj5+6gaqfBhJDIykuTkZNtCR8JxPDw8iIyMdHYZQogqwmq18tlnn7F27Vrc3NwwGAxcunQJg8GAl5cXnp6eZGdnk5mZSYsWLXjiiSfsmo1aVD0VPoy4uroSHR3t7DKEEELY6bvvvmP27NkcOXIEg8FAs2bNcHNzA0pWzs3JycFqtTJs2DBGjx5N3bp1nVyxcLYKH0aEEEJUfIcOHSI+Pp7Dhw9z5coVANvQ28zMTGrWrElRURHBwcE8//zzPPjggzI0V9hIGBFCCPG7ZGZm8uGHH7J//37CwsIwGAwEBQXZwgiUdNUUFhZSt25dRo0ahclkcmLFoqKSMCKEEMJuVquVhQsXsmLFCrKysjAYDLbF6gACAgI4e/Ys7u7u9O7dmxdffJHAwEAnVy0qOgkjQgghbqm4uJj58+ezefNmkpOTgf91w+Tm5lJUVERRURG+vr4sWbJEZnIWt0XCiBBCiJsqKChg8eLFfPvtt5hMJmrUqEFGRgb5+flASUtJkyZNeOKJJ+jevbuTqxWVlcPDiFLKCOwDUrTW/Rx9PyGEEH9cSkoKr7/+Ops3byYlJYXu3bvToEEDAAIDAykqKuKZZ55h4MCB8iKq+MPKo2VkNHAU8C2HewkhhPgDjh8/zptvvsmhQ4cwGo22ySYPHDhA7dq18ff3Z+bMmdSrV8/JlYqqxKFhRCkVCfQFpgMvOfJeQgghfr8DBw4wZ84crFYrRUVFtinYg4OD0VozZswYnnrqKXkZVTiEo9vW/gmMA6w3O0ApNVwptU8ptU9mWRVCiPL1+eef89BDD9G+fXuKi4sxGo34+fnh4eGB1WqlV69eHDt2jJdfflmCiHAYh7WMKKX6ARe11vuVUt1udpzWej4wHyAmJkZWahNCCAezWCzEx8ezZMkSLBYLp0+fprCwkKNHj3L33XeTl5fHo48+yrPPPisr5Ypy4cj/yzoC/ZVSDwAegK9SaqnW+v848J5CCCFuIi8vjzfeeIPk5GSysrKwWCwAhIWFkZWVRXBwMD179mTQoEHyUqooVw4LI1rr8cB4gKstI2MliAghRPnLzs5m8uTJHDlyhHr16mEymXBzc+PixYtYrVbCwsJISEjgnnvucXapopqS9jchhKiikpKSeP3110uNjCkqKrKtoFurVi1ee+012rVr5+xSRTV3yzCilIoGzmutC65+9gRqaK0T7b2J1nobsO33lSiEEOJ2fPnllyxdupSTJ09iNpttI2O01qSkpBAeHs6YMWO4++67nVypECXsaRn5DPjTdZ8tV7e1dUhFQgghbpvZbGb27Nn85z//oUaNGphMJsLCwmxTt1ssFtq1a8cbb7xBaGiok6sVojR7woiL1rro2getdZFSys2BNQkhhLBTcnIy06ZNY+/evRiNRnx9fQkLCwNK5ghJSUmhV69eTJ48GU9PTydXK8SN2RNG0pRS/bXWawGUUgOAdMeWJYQQ4rds3bqVRYsWcejQIQwGg60r5tKlS+Tn51NcXEzTpk2ZM2cObm7y70dRsdkTRkYCy5RSs65+TgaGOK4kIYQQN1JcXMwXX3zBmjVrMJlMeHh4YDKZbIvWWSwW6tSpw9ChQ+nWrZtzixXiNtwyjGitTwHtlVLegNJaX3Z8WUIIIa45duwY06dPZ/v27SQlJZVatM7Pz4+8vDy6devGhAkTCA4OdnK1Qtw+u4f2aq1zHVmIEEKI/7Faraxfv56EhAR++eUXDAaDbZKyn376iXr16mE2mxk8eDCPPvqoTFImKjWZZ0QIISqQgoICFi5cyK5duzCZTLi7u9uCRnBwMBaLhWHDhjF48GAaNWrk5GqFuDMkjAghRAWwd+9e3nvvPf7zn/9w6dIlBg8eDJR0w7i7u1NUVET//v159dVXMZlMTq5WiDvrtsOIUiqGkknQUhxQjxBCVBsWi4WZM2eyYcMGLl++jFIKq9VKcXExJ06coHHjxiilGD9+PP369XN2uUI4zO9pGXkBaKGUOq61fvROFySEEFWd2Wzm73//O0lJSXh5eWEymcjNLXktLzQ0FF9fXx544AGGDBlCrVq1nFytEI5322FEa/1XAKWUz50vRwghqi6z2cy7777LypUrCQ0NpXbt2gCEhISQmpqKi4sLTz/9NGPGjMHFRXrRRfVhz9o0XW60XWu9/c6XI4QQVY/FYuHdd9/l888/RymFUor09HTCwsIwGo0UFxczffp07r33XmeXKoRT2BO9X7nuaw2oq79KGBFCiN9gsViYOnUqp06d4tSpUyilSu3Lz89n4sSJNG/e3IlVCuF89kx6FguglPKgZOZVV2CJg+sSQohKy2q18o9//IPdu3cTHByMt7e3bdE6q9VKdHQ077zzjm3iMiGqu9vplJxPSYtIFiWr9t7nkIqEEKKSslqtzJw5kxUrVqCUws/PzzYjamBgIAaDgRkzZkgIEeJXbieMtNBatwRQSu10UD1CCFHpWK1WPvzwQ5YuXQpg647Jycnh8uXLGI1GXn75ZVq2bOnMMoWosOx5gTXwf1+qAEreGVG/cYoQQlQLWmsWLVrE119/bVu4rqCgACgJKLVr12bSpEncddddTq5UiIrNnpaR/fzvxdXvr27TDqtICCEqOKvVysaNG3njjTfw9/cnKioKgKCgIM6ePUutWrV45513JIQIYSd7XmCNLo9ChBCiorNarcyZM4dFixZx7tw5zp8/T1BQEFFRUVgsFsLCwpg+fTrNmjVzdqlCVCr2dNN8r7VuXR7FCCFERWS1Wlm2bBlffPEFv/zyCwA1atTg4sWL5Obm4uLiwksvvSQL1wnxO9nTTSPvhwghqiWz2cymTZv45JNP8Pb2Jjg4mAsXLlBYWIhSiscee4wZM2YQERHh7FKFqNTsCSONlFIHr/usAK21buGgmoQQwqm++uor5s6dy8GDBzl9+jQdO3akadOmAISFhVFQUMD06dNldIwQd4g9YeQMEOvoQoQQwpmOHj3Khx9+SGJiImlpaQC4u7sDcODAARo0aIC7uztTp06Vd0KEuMPsCSNFWutfHF6JEEKUs4KCAuLj49m4cSNRUVEYDAZq1qxJeno6Wmu8vLyIjIzk8ccfZ9iwYTJZmRAOYk8YecHhVQghRDm61g2TmJiI0WgEIDIyEjc3N1xdXfHx8cFisfDnP/+ZESNG4OHh4eSKhaja7Akjh5RScUDnq5+/AaZprXMcV5YQQtxZ17phLl++TFJSEleuXLEFEYCMjAx8fX3x9fVl9uzZMkeIEOXInjCyEDgMPHL18xAgAfizo4oSQog74Vo3zOrVqwkJCSE4OBiTyURQUBBXrlwBSlbPrVu3LsOHD+fee+91csVCVE/2hJF6WuuHrvs8VSl14FYnXV3ldzvgfvU+n2ut//77yhRCCPtt2rSJ+fPnl+qGyczMtC1a5+/vT2ZmpnTDCFFB2BNG8pVSnbTWOwCUUh2BfDvOKwR6aK1zlVKuwA6l1Cat9e4/UK8QQtzQhQsXmDdvHidOnMBisXD27NlS3TCXL18mOzuboKAgxo0bJ90wQlQg9oSRZ4HFSik/SuYYyQT+equTtNYayL360fXqf7KmjRDijvriiy/49NNPWbNmDY8++igmkwkAT09P8vPzbd0wI0eOpHfv3k6uVghxI/asTXMAuFsp5Xv18yV7L66UMlKy0F59YLbW+r+/t1AhhLimoKCAt956i40bNwJw+PBhCgsLOXPmDA0bNsRisRASEkLXrl2lG0aISuCmYUQpdZ/W+surLSJ/B7pc3W73aBqttQVoqZTyB1YrpZpprQ//6j7DgeEAtWvX/v1PIoSo8hITE5kzZw6pqam2NWIAQkNDOXv2LFprIiIiGDp0KJGRkU6sVAhxOwy/se+tq78mAJcpGU3zCHCJkhE2dtNaZwPbgPtusG++1jpGax0TEhJyO5cVQlQTu3bt4umnn2bAgAHk5OTg4eFBzZo1gZJF7Bo3bsz333/Ptm3bmDx5sgQRISqZ3+qmKbzazVJfa339MF57R9OEAMVa62yllCfQC5jxx8oVQlQXhYWFfPDBB6xcuZKCggIOHTqE0WikVatWuLu74+bmRvPmzRk/fjwNGzZ0drlCiD/gt8LIf4CngNwbjKYx23HtmpS8+GqkpAXmU631+j9asBCiajt16hQzZsygoKCA48ePY7FYcHV1JTAwkIyMDFJTU+nduzcjR47Ey8vL2eUKIe6A3woj04D5QASwXSmVRMlomnzsG01zEGh1J4oUQlR927dv591338Xb2xs/Pz88PT0JCQnh/PnzaK1p2bIl06ZNo0OHDiilnF2uEOIOumkY0VoXAn9VSpmAIEqCyCWtdVZ5FSeEqNqsVivx8fEsW7aMoqIiAAICAvDz8wNKJidzd3dn4sSJtGol/7YRoqqyZ2hvnlIqGmgCeFy6a3WQAAAgAElEQVT7F4nW+mMH1yaEqKIuXLjAm2++ydmzZ0lJSSm1Lysri5ycHEJCQpg6dSphYWFOqlIIUV5uGUaUUn8HulESRjYC9wM7AAkjQojbsn//ft5++218fX1xdXWlRo0apKenU1hYiNYaDw8P/s//+T+MGDECg+G3BvsJIaoSe2ZgHQTcDfygtX5KKVUDWODYsoQQVclnn33GrFmzyM3NRSlF/fr18fPzQylFcHAwVquVV199lU6dOjm7VCGEE9i1No3W2qqUMl+dhfUiUNfBdQkhKjmz2cyiRYtYvnw5ly6VTNx8rZs3NTUVd3d3PDw8iIuLo25d+StFiOrMnjCy7+oMqh9RMrV7LrDHoVUJISqtgoIC4uPj+f777/Hy8iIwMNAWRgCMRiP3338/L774okzTLoQA7HuB9bmrX85VSn0J+F4dtiuEEDanTp3ivffeY/Xq1eTm5vL4448DEBQUxIULF3B3d2fMmDH069fPyZUKISoae1pGUEr9GehEyaq7OwAJI0IIAHbv3s0bb7zB+fPnUUqRnZ1NUVERJ0+eJDo6Gm9vbz744APatGnj7FKFEBWUPaNp5lCy6u6Kq5tGKKV6aa2fd2hlQogKbe3ataxatQo3Nzdyc3Nto19q1KgBwAMPPMCTTz5JaGioM8sUQlQC9rSMdAWaaa01gFJqMXDIoVUJISqkoqIiPv74YxYuXEhubi4dOnQAIDw8nBMnTqC1ZujQoUycOBFXV1cnVyuEqCzsCSM/A7WBa+t110K6aYSoVtLT05k6dSo7duzAYrFw+PBhXFxcaNOmDW5ubri6ujJs2DCGDRsm84MIIW6bPWEkCDiqlLo2gqYt8J1Sai2A1rq/o4oTQjjX/v37+fDDD1FKcfToUYxGI0ajkYCAAHJzczEajQwePJguXbo4u1QhRCVmTxiZ7PAqhBAVypYtW3jvvfcIDg7G29sbgJCQEC5cuIDFYqFv377MmDGDmjVrOrlSIURVYM/Q3m/KoxAhhPOtWrWKf/zjHxQWFgKgtbaFEV9fX2rVqsXUqVPlpVQhxB1l19BeIUTVlpSUxKuvvsqRI0dKvXialZVFQEAA9erVIy4uDi8vLydWKYSoqiSMCFGN5ebmMmPGDN5//30MBgN33XUXUNIi4unpybBhwxgyZIi8lCqEcCgJI0JUQ1arlcWLFzNx4kTOnz8PQNu2bWnQoAF5eXlEREQwe/ZsJ1cphKgu7Jn07AwlM6/aNgFaay0rWwlRCW3fvp0xY8bw/fffl9peu3ZtvL298fb2JioqyjnFCSGqJXvaXi9TMpy3HSWL5LW5+lkIUYmcOnWKhx56iK5du5YKIuHh4SxatAg/Pz/btlatWjmjRCFENWVXR7DWOgPIBCKA/lc/CyEqgZycHMaNG0eTJk1YtWqVbbunpyeTJ0/m+PHjDBgwgJ9++omTJ0+Smppqm1lVCCHKgz3vjJy8OsGZF7AKaK2U6qq1fsqxpQkh/giz2cyCBQuYPHkyaWlppfYNHjyYt99+m1q1agGwc+dOzGYzOTk5REZGYjKZnFGyEKKasieMPArcC1iAf2utLUqphx1blhDij9i8eTMvvfQShw8fLrW9Q4cOxMXFcc8995Tafv1xzZs3L5cahRDiGnsmPSsG1v9q22cOq0gI8bv9/PPPjB07lvXrS/2RpXbt2syYMYNHH30UpVSZ864PI82aNXN4nUIIcT0ZTSNEFZCZmcm0adOYPXs2ZrPZtt1kMjF+/HheeuklPD09b3iu1Wrl6NGjts8SRoQQ5c2ebpo9QA1gObAOKHJoRUIIuxUXFxMfH8+UKVPIysqybVdK8eSTTzJ9+vRbrh+TmppKvXr18PLywmg0EhER4eiyhRCilFuOptFaPwoMBFyBBGCUjKYRwrm01mzYsIHmzZszevToUkGka9eu7Nu3j4ULF9q1kN3Zs2dxc3PjypUrREVF3bAbRwghHMneOZ6tlO6qEUI4yd69e7n33nvp168fP//8s2173bp1WblyJVu3bqV169Y3Pd9sNvPdd98xZcoUdu/eTZMmTejRowcA0dHRDq9fCCF+zZ53RpYB4cAK4EmgSCkVqLXOvMV5tYCPgTBKwsx8rfUHf7hiIaqhgoICPvvsM2bNmsWePXtK7fP19WXSpEm88MILuLu73/B8rTWnT59mz549HDt2jA0bNgAl3Tzt27enXbt2bNmyRSY7E0I4hdL6txs8lFKJ/K9VRGPnC6xKqZpATa3190opH2A/8KDW+sjNzomJidH79u27jfKFqNqSkpKYO3cuCxYsKDNXiMFgYPjw4UydOpXQ0NAbnp+cnExkZCQnTpzg/fffB0omO9uxYwcAHh4ebN68GU9PTzIzMwkMDHTsAwkhqhWl1H6tdcytjrNnaG/U7ylAa30eOH/168tKqaOUzOB60zAihChpxdi6dSuzZs1izZo1WK3WUvvd3Nx47LHHeOWVV2448uXcuXNs376dtWvXcvLkSTZu3Ei9evXw9/cnOzub/Px826J49957Lx4eHgASRIQQTmNPN40X8BJQW2s9XCnVAGiktV5/i1Ovv0YU0Ar47++sU4gq7/Lly3z88cfMnj271FDba2rVqsVzzz3H008/TUhISKl9FouFVatW8c0331BYWMixY8fIz88HSiZAe/zxx+ncuTNZWVm0a9eOBg0aYDDY+8qYEEI4lj1DexMo6WL509XPycBn/GoitJtRSnkDK4G/aa0v3WD/cGA4lEzMJER1c/ToUebMmcPixYu5fPlymf09e/Zk1KhR9OvXDxeX0n9kk5KSWLduHRs2bMDLy4ugoCAMBgNBQUEkJyfj7u5uu2a/fv3K5XmEEOJ22RNG6mmtH1VKPQ6gtc5Xdo79U0q5UhJElmmtV93oGK31fGA+lLwzYl/ZQlRuZrOZ9evXM2vWLL7++usy+729vXnyySd57rnnuOuuu0rty8rKYuPGjWzbto0ffvjBtt3Hx4egoCAAgoODeeKJJ+jTpw/e3t6OfRghhPiD7AkjRUopT66+xKqUqgcU3uqkq4HlX8BRrfXMP1SlEFVEWloaCxYsYO7cuSQlJZXZf9ddd/H8888zZMgQfH19bdu11uzYsYOVK1eSl5dHYWEhR46Ufv3K1dWVkJAQYmNjy6w9I4QQFZk9YeTvwJdAravDfDtSMsT3VjoCQ4BDSqkDV7dN0Fpv/D2FClGZ7dmzh9mzZ/PJJ59QVFR6EmODwcCAAQMYNWoU3bt3LzXp2IULF1i/fj3r16/n3LlztGjRAqPRiKenJ15eXhQWFtKxY0f69+9Pp06dynTjCCFEZWDPaJrNSqnvgfaUDOsdrbVOt+O8HVePF6JaKigo4NNPP2XWrFns3bu3zP7g4GCGDRvGyJEjS70vVVRUxH//+19WrFjB3r17uX74fXZ2NkFBQVgsFgYNGsRf/vIXgoODy+V5hBDCUez9Z1RXoBMlXTWuwGqHVSREJXfkyBFmzJjB/v37uXDhAhkZpVdPaNeuHaNGjeLhhx+2Dau95uuvv2bdunXk5+dTUFBQKoh4e3vTvHlzOnbsSI8ePWQ0jBCiyrBnaO8coD4lM7ACjFBK9dJaP+/QyoSoRHJycvjkk09ISEjg6NGj1K9fH5PJRNeuXdm7dy8XL17kscce4/nnn6dt27a283755Re+/PJLwsPDiY2NxcvLyzYkNzAwEKUU7dq1IzY2lm7dupUJL0IIURXY0zLSFWimr/4TTSm1GDjk0KqEqASsVivbtm0jISGBlStXkp+fj4eHB40bN8bX15fo6GhcXV3p0aMH06dPt62Ge+nSJfbu3csPP/zAJ598AkDDhg2JjY2lVatWrFixAl9fX9q0acPLL79MZGSkMx9TCCEczp4w8jNQG/jl6udawEGHVSREBffLL7+wePFiEhISSExMtG03Go3Ur18fo9FIYWEhRqMRAHd3d5YtW8bYsWPJz8/ntddew2KxoJTCZDKRl5fH8ePHOXXqFPXq1WPixImEhobK6rlCiGrDnjASBBxVSl1bnast8J1Sai2A1rq/o4oToqLIz89n9erVJCQk8PXXX/PrNZ2MRiN9+vQhLS0NrTVGoxF/f3/S0tI4e/Ys+/fvp02bNvTs2ZMGDRpw7NgxtNa0adMGg8HAfffdZ2sBqVGjhjMeUQghnMaeMDLZ4VUIUQFprdm3bx8JCQksX76cnJycMscEBATw+OOPo7XGYrHg7+/P6dOnefPNN2nfvj1//etfbeft3r2bnj170r59eywWC+3ataN169YyKZkQotqzZ2jvN9e+VkpFAAFa68MOrUoIJ7p48SJLly4lISGBw4fL/q+ulKJPnz4MHTqU/v37M3/+fH766SegZBbUBx98kK5duwIwadIknnvuOfLy8jhz5gxaa9q3b0+HDh3K9ZmEEKIis2c0zXvAX4EPgL8AZqXUf7TWYxxdnBDlxWw2s2nTJhISEli3bh1ms7nMMfXq1eOpp57iiSeeoFatWgDs2LGDJUuWUKNGDSIiIvDw8GDcuHG2c5o2bcoXX3xBfn4+NWvWlPdAhBDiBuzpphkINKPkRdaaQDHyAquoIg4fPsySJUv4+OOPuXDhQpn9Xl5ePPzwwwwdOpTOnTuXChOnTp1i4sSJaK25cOECNWrU4K233ioz/0dAQAABAQEOfxYhhKis7Akjl7TWF5VSiVrrAgCl1C3XphGiorJarWzcuJEFCxYQEhLC/v37ycrKKnXMn/70J4YOHcojjzyCj49PqX35+fns3LmTd999l7y8PADCwsJ47733MJlM5fYcQghRVdgTRhorpQ4C9a/+qoC6ji1LiDvvypUrfPzxx8TFxWE0GuncuTPp6ekYDAaaNGlCbm4uAwcOZOjQoTRq1Mh2ntVqxWAwYLVaiYuL4+TJk1itVi5dugSAp6cncXFxBAYGOuvRhBCiUrMnjNx160OEqLjOnz/P7NmziY+PJzMzE5PJxMMPPwyUvCsSEBBAVlYWTzzxBK+//joWi4WcnBy+/fZbdu/ezX//+1+WLl1KjRo1sFqtWK1WAMLDw0lJSeHNN9+kQYMGznxEIYSo1OwZTfPLrY4RoiL68ccfiYuLY/ny5RQXF9u2h4WFkZmZSY0aNQgODsZsNmM0Ghk9ejRWq5VJkyaRkZGBUop9+/YB8N133/Hggw/StGlTTp48Se3atenQoQOdO3emXr16znpEIYSoEmS9cVGlWK1WvvzyS2bOnMnXX39dZn+jRo3w9vYmJSWF7OxsXnjhBQYOHMi5c+fw8fFh69attoXtgoKCbOf9+OOPPPjgg3Tp0oXOnTuXeY9ECCHE7ydhRFQJ+fn5LFmyhLi4OI4dO1Zmf8eOHXnmmWdYtGgRubm5APTp04dBgwYB2IbqHj9+HACDwUBQUBDPP/88HTp0oGHDhgAyQZkQQjiAhBFRqV24cMH2Psi1Fo1rjEYjgwYNYsyYMbRt25ZRo0bZgkhERARjxpSdKmf48OGkpaXh4+ODp6dnuTyDEEJUdxJGRKV08OBB2/sgRUVFpfb5+voybNgwXnjhBerUqUNqaiqvv/46ycnJtnlCpk6dipeXV5nrKqUIDQ0tl2cQQghRQsKIqDSsVitfffUVM2fOZMuWLWX2R0VFMXr0aIYOHYq7uzs7duxg1qxZJCYmEhgYSM2aNfHz86Nly5a0bNnSCU8ghBDiRiSMiAovPz+fpUuXEhcXx9GjR8vs79ChAy+99BL9+/fnhx9+4MMPP2TLli22LhkfHx/bHCDu7u6MHDmyXOsXQgjx2ySMiArr9OnTfPTRRyxYsAB/f3/q1KmDn58fqampZGdn06lTJ/72t79xzz338NVXXzF58mQKCwv59ttvS13n8uXLWCwWGjRowIgRI2SWVCGEqGAkjIgKxWw2s27dOubNm8dXX31l296oUSPbfB6ZmZmcOXOGc+fOUVRUhIuLC9u2baOwsGSVApPJRF5eHhEREdx3333cf//9REVFOeNxhBBC2EHCiKgQkpKSWLBgAf/61784d+4cUDLDqYeHB+fOnSMkJMR27PVzfISGhuLu7k6rVq3YvXs3JpOJhx56iK5du9KiRQtZJVcIISoBCSPCaSwWC5s2bWLevHls3LjRNs06gIuLC6GhoRiNRgICAnj22Wfx9fUlOzubY8eO2bpratasCUDPnj1p06YNTZs2xWg0OuuRhBBC/A4SRkS5O3fuHP/617/46KOPOHv2bJn9NWrUoEuXLpw+fRqA+vXr06tXLwwGAwD33XdfmXNq165N7dq1HVu4EEIIh5AwIsqF1Wpl8+bNzJs3j7Vr12KxWMoc07NnT2JjY0lNTSUjI4NGjRqRkpLCyJEjbUFECCFE1SNhRDhUamoqCQkJfPTRR7aWjusFBwfTv39/goODOXDgAEuWLCE6OprAwEC8vb1p0KABXbp0cULlQgghyouEEXHHaa3Ztm0bc+fOZfXq1aVWzIWSuT569OhBWFgYGRkZ/Pjjj6X2Z2dnExgYiNls5umnn5ZWESGEqOIcFkaUUguBfsBFrXUzR91HVBwZGRksWrSI+fPn2xacu0YpRUREBI0bN8ZoNNoWpvv1wnMBAQF06tSJRo0a0a9fP1mYTgghqgFHtowsAmYBHzvwHsLJtNbs2rWL+Ph4Pv/8c9tcH9d4e3vTpEkT3NzcyM/PJysrC6PRSO3atVFK4e3tTa1atejatSvdunWjefPmMhpGCCGqGYeFEa31dqVUlKOuL5zr8uXLLF26lPj4eA4dOlRmf9OmTenUqRNKKb7//nvy8/Nt+ywWC2azmaioKHr37s0999wjXTFCCFGNyTsj4rYcOnSI+Ph4lixZYlv75Rp3d3eaN2/OiBEjOHfuHCkpKWitqVevHsePHyc0NJQBAwbQr18/IiIinPQEQgghKhqnhxGl1HBgOCDzRFRQhYWFfP7558THx7Nz585S+4xGI+Hh4dStW5fc3FwmTZpE//79+eabb1i+fDlKKWJiYhg5ciR/+tOfcHFx+v9yQgghKhin/2TQWs8H5gPExMRoJ5cjrnP69GnmzZvHwoULSU9PL7XPx8eHbt26ERERgZ+fH1u2bAFg9erV9O/fn3bt2nHp0iU6duxoWzFXCCGEuBGnhxFRsVgsFjZu3Eh8fDxffvklWv8vH7q6uhIQEED37t0ZPHgw69evx2q1cunSJdzc3DCbzfj7+1NQUICnpyexsbFOfBIhhBCVhSOH9q4AugHBSqlk4O9a63856n7ij0lNTWXBggXMnz+fpKQk23YXFxcaN26MUgo3NzcApk6dSqNGjThy5AinT58mKCiI0aNH061bN2rUqOGsRxBCCFFJOXI0zeOOura4M7TWbN++nfj4eFatWlVqcjKTyUTbtm2pW7cuLi4unDx5kpycHADWrVtHo0aNGDhwIAaDgXr16snquEIIIX436aaphnJycliyZAnx8fEcOXLEtt3Dw4PAwEBCQkJwcXHB29vb9sJpcHAwOTk5xMTE0KpVKwAaNmzolPqFEEJULRJGqpFvvvmGJUuW8Mknn5CXlweAv78/TZs2xcvLi8zMzFLHZ2ZmEh4ejtlspkmTJsycOVOG5AohhLjjJIxUcVeuXGHp0qVs2rSJ5ORkiouLCQoKwsvLi8jISNq2bQuA2WwmKyvL9sKqn58fPXv2pHXr1vTq1UuG5AohhHAY+QlTRR0+fJh58+axdetWWrduTWhoKAUFBaSlpRESEmI7rrCwEHd3d1xcXAgKCqJVq1bcf//9dOjQAVdXVyc+gRBCiOpCwkgVkp+fz+eff868efNsk5M1atQIT09PAGrVqkVOTg5FRUW2c7KysqhZsyYtW7bkrbfeIigoyCm1CyGEqL4kjFQBx44dY/78+SxatIisrCyCgoJo0KABJ06c4Oeff6Z58+aEhoYyePBgTCYTX375Jenp6bRp04ZevXoRHBzs7EcQQghRjUkYqaQKCwtZtWoV8+bN45tvvgFKZkVt3LgxJpMJgN69e/Paa6/RqlUrPD098fDwAODuu+92Wt1CCCHEr0kYqWROnjzJ/PnzSUhIKDVFe2hoKA888AA5OTm2ScsCAwPp0qWLvHwqhBCiQpOfUpVAcXExa9asYe7cuXz99ddAyaRkkZGRJCcn4+PjQ2xsLEajkZCQEHx9fWnVqhUjRoyQICKEEKLCk59UFdiZM2f46KOPWLhwIampqbi7u1OzZk2CgoJwd3cHYMCAAbz88sts2rSJH3/8EZPJxCuvvEL9+vWdXL0QQghhHwkjFYzZbGbdunXEx8eza9cuXFxc8PDwoFGjRnh7e5c53s/Pj+joaEaOHGmbpl2CiBBCiMpEwkgFkZKSwpQpU9izZw8uLi4opWjcuPFNj/fx8aFXr162lXENBgMDBgwor3KFEEKIO0bCiBNprfnPf/7DrFmzOHToEP7+/r850ZiLiwsdO3bkgQceoG3btvj6+pZjtUIIIYRjSBhxguzsbBYvXkx8fDypqanUrl0bf3//Use4uroSHh5OZGQkoaGhNGrUiF69euHv78/58+eZNm0aDz30EB06dHDSUwghhBB3hoSRcvT9998zZ84cli9fjsViISIignr16pU65u677+bdd9+96UyoeXl5zJkzh8uXL7No0SJycnK47777yqN8IYQQwiEkjDhYfn4+n376KXPmzGHPnj227XXq1CEwMND2OTAwkGnTptG+ffvfvF5BQYFtuK6bmxtNmzZ1TOFCCCFEOZEw4iAnT55k7ty5JCQkkJubW2o9GKBUt0xsbCwvvfQSPj4+t7xuUFAQr776KgsXLuSee+6hVq1ad7x2IYQQojxJGLmDzGYzGzZsYM6cOfz73/8mLCyMmjVr4urqysGDB3FxceGRRx7hueeeo127drzxxhvce++9t/3eh4eHB88++yxKKQc9iRBCCFF+lNba2TXYxMTE6H379jm7jNt24cIFFixYwPz58zl79qxte5MmTWwr5rZt25Zp06YREhJy29fXWkvwEEIIUekopfZrrWNudZy0jPxOWmu2bt3KP//5Tw4cOEBBQQFpaWm2/UopQkNDuXz5MiaTiY4dO952EMnLy2Pnzp18++23DB06lOjo6Dv9GEIIIYTTSRi5TWlpabz55pts374dAKPRSGhoKFeuXCEtLY2QkBCefvppRowYgY+PD3v27KF79+62FXNvxGq1kpGRQWBgIEaj0bbdYrGwcuVKAN5//32efPJJ2rZt69gHFEIIIcqZhJHbsHv3boYPH46bm1up0ADg5eXF3LlzefLJJ23rxgDcf//9N72e1Wpl4cKF/PTTT1y5coVp06ZRo0YN235fX1+Cg4NJT0/HYDDIJGdCCCGqJAkjt6C15scff2TlypVs2rQJNze3Uvvd3Nzo2LEjsbGxtGvXrlQQAcjJyeHUqVOcOXOGDh06EB4ebttnMBhITU3lypUrQEmry/VhBKB3795YrVZatmxZaiiwEEIIUVVIGLmJixcvsmHDBtatW0dSUlKpfVprYmJieOGFF2jatOlvvly6fPlyDhw4AEBwcHCpMALQtGlTkpKS8PHxKTP8F6Bbt25//GGEEEKICkzCyHWKi4vZvn07a9eu5bvvvsNqtZY5pkuXLowbN46wsDAATpw4wc6dO0lMTKRly5Y8+OCDpY6vW7euLYwkJyff8Hpt2rQhMjJSRswIIYSoliSMUBIo1q5dy6ZNm8jOzi6z32Qy0b17dzp16kSvXr1K7UtPT+e7774DuOFomYYNG1K/fn3q1at3w9lSAwMDpftFCCFEtVbtw8i2bdsYO3bsDffFxMTQuXNn9u7dy/nz59m7d2+ZMHL9cNvz58+XuUZ0dDSvvPLKnS1aCCGEqEKqfRiJiYmhRo0aGAwG3N3dyc/Pp3///vTr14+IiAhyc3Ntw3iTk5Mxm822tWEAQkNDefjhh4mKipKp2YUQQojfwaFhRCl1H/ABYAQWaK3fceT9biU/Px93d3cMBoNtm6enJ7Vq1eLaTLTvvPMOAQEBtv3e3t6EhIRQUFBAVFQUeXl5+Pn52fYbDIYyrSVCCCGEsJ/DwohSygjMBnoDycBepdRarfURR93zZj799FMOHz5MamoqkydPJiIiwrbPaDQSHR3N6dOnAUhJSSkVRgDGjx+Pl5eXvGAqhBBCOIAjW0baASe11qcBlFKfAAOAcg8jGRkZpKamApCYmFgqjEDJujHR0dFERUVRp06dMuebTKZyqVMIIYSojhwZRiKAs9d9Tgbu+fVBSqnhwHCA2rVrO6SQOnXqcODAAQwGww1Hy/To0cMh9xVCCCHErTkyjNyoT6PMEsFa6/nAfChZtdcRhbRr145GjRpRq1atMjOoCiGEEMK5HBlGkoHrh5dEAucceL+bCg4OJjg42Bm3FkIIIcQtGG59yO+2F2iglIpWSrkBjwFrHXg/IYQQQlRCDmsZ0VqblVKjgK8oGdq7UGv9k6PuJ4QQQojKyaHzjGitNwIbHXkPIYQQQvz/9s48WKriisPfT1BcWA1o3CIYl5TiRtCoEYpIgkipGGMJiYm4RBPUJJYxLjGuMSZKGZdoSbkVLkRxI6IVCykViGVAFNkUVFBUFMVdMRgVT/7oM3jfMPO28ObOmzlf1a2593Tfvv3rvjNzprvnnvZNW07TBEEQBEEQNEk4I0EQBEEQ5Eo4I0EQBEEQ5Eo4I0EQBEEQ5IoKAeKqAUlvA6+swyJ7Au+sw/LaC6G7/qhX7aG7vqhX3dB+tW9rZr2aylRVzsi6RtJTZtY/73pUmtBdf9Sr9tBdX9Srbqh97TFNEwRBEARBroQzEgRBEARBrtS6M3J93hXIidBdf9Sr9tBdX9Srbqhx7TW9ZiQIgiAIguqn1kdGgiAIgiCocmrSGZE0VNLzkhZLOivv+rQGSdtIekzSQknPSvqN2y+Q9LqkOb4Ny5xztmt+XtKBGXvJ9vCIyjMlvShpgkdXrgokLZU038JbS/IAAAojSURBVDU+5bZNJU3x+k6R1MPtknS165snqV+mnFGe/0VJozL2b3v5i/1cVV5lQyTtlOnXOZI+knRqLfa5pJslrZC0IGNr8/4td42cdY+RtMi1TZTU3e29Ja3K9PvY1uprrA1z1t7m97akTn682NN7V0bxmnqV0j0ho3mppDlur6k+bxFmVlMbKULwEmA7YANgLrBz3vVqhY4tgH6+3wV4AdgZuAA4vUT+nV1rJ6CPt0GHxtoDuAsY6ftjgdF5687oWQr0LLJdBpzl+2cBl/r+MOAhQMA+wEy3bwq85K89fL+Hpz0J7OvnPAQclLfmEvfxm8C2tdjnwECgH7Cgkv1b7ho56x4CdPT9SzO6e2fzFZXTIn3l2rAKtLf5vQ2cBIz1/ZHAhLx1F6VfDpxXi33ekq0WR0b2Bhab2Utm9hlwJzA85zq1GDNbbmazff9jYCGwVSOnDAfuNLP/mtnLwGJSW5RsD/eqDwDu8fNvAQ5rGzXrjOGkekLD+g4HbrXEDKC7pC2AA4EpZvaemb0PTAGGelpXM/u3pXftrVSf9sHAEjNr7CGA7bbPzWw68F6RuRL9W+4aFaGUbjN72My+8MMZwNaNldFKfeXasGKU6fNyrMt7O9sm9wCDC6MKlaAx3V6PI4E7GiujvfZ5S6hFZ2Qr4LXM8TIa/xKvenxYcU9gpptO8WG3mzPDzOV0l7N/Dfgg8yFYbe1kwMOSnpZ0ots2N7PlkJw1YDO3t1T7Vr5fbK8mRtLwA6oe+rwS/VvuGtXCcaRfswX6SHpG0jRJA9zWGn3V/LnY1vf2mnM8/UPPXw0MAN4ysxcztnro87WoRWeklMfbbv8yJKkzcC9wqpl9BFwHfBPYA1hOGuKD8rpbaq8Wvmtm/YCDgJMlDWwkb01p97nuQ4G73VQvfV6OutAp6RzgC2C8m5YD3zCzPYHTgL9L6krr9FVrm1Ti3q5W7QA/puGPjnro85LUojOyDNgmc7w18EZOdfm/kLQ+yREZb2b3AZjZW2a22sy+BG4gDVtCed3l7O+Qhu06FtmrAjN7w19XABNJOt8qDDP66wrP3lLty2g4FF5V2kkO2Gwzewvqp8+pTP+Wu0auKC2+PRg4yofh8SmKd33/adJaiR1pnb6q/Fys0L295hxP70bzp4vaDK/L4cCEgq0e+rwcteiMzAJ28JXVG5CGuyflXKcW43OJNwELzeyvGXt2zu+HQGGF9iRgpK8c7wPsQFrwVLI9/APvMeAIP38UcH9bamoukjaR1KWwT1rgt4CksfCPiWx9JwFH++rxfYAPfbhyMjBEUg8f/h0CTPa0jyXt4+18NFWi3Wnwa6ke+typRP+Wu0ZuSBoKnAkcamb/ydh7Serg+9uR+velVuor14a5UqF7O9smRwCPFhy+nPk+sMjM1ky/1EOfl6U5q1zb20ZaRfwCyas8J+/6tFLD/qQhtXnAHN+GAbcB890+Cdgic845rvl5Mv8OKdcepBXpT5IWh90NdMpbd6Zec317tlBn0jzvI8CL/rqp2wVc6/rmA/0zZR3n+hYDx2bs/UkffEuAa/AHAOa9ARsD7wLdMraa63OSs7Uc+Jz0C+74SvRvuWvkrHsxaW6/8D4v/PPjR37/zwVmA4e0Vl9jbZiz9ja/t4EN/Xixp2+Xt263jwN+WZS3pvq8JVs8gTUIgiAIglypxWmaIAiCIAjaEeGMBEEQBEGQK+GMBEEQBEGQK+GMBEEQBEGQK+GMBEEQBEGQK+GMBEFQUSRtKOkSSTNUFKk1CIL6pGPTWYIgCNYp1wOPAwPM7PO8KxMEQf7EyEgQtCGSekta5SMAcyS9LGlcJn2c2+ZI+kxST39i4hhJCyTNlzTC8w6S9KDv95S0NFPOAqWAikg6QdIsSXMl3Stp40y+ayS96tdbKam/26cW9hvRMjWT/2JJK31/N0lPKQX3miXpW24f7Lb5SkHQOinFWhpEeljZbEkT/emphfKvlPSE69nb7Xu77Rl/3cntB0p6RNIGRW1zpJe7nqQO3pazlIKx/aK4Lf34dEkXNKbTj8dKWujtt7oZ/b9S0uWSZntde0nq6PUZ5Hn+LOlPTZUVBLVMOCNB0PYsMbM9zGwP4HdFaR2A33paIW7E4aTAYbuTHhk9Ri0L/X2fme1lZrsDC0lPusxe7w9+vadaoQVJmwGDC8dmNs/M+lsK7jUFGCVpQ9ITJkeY2a6kUdjRpKdFbgOc6fb5wPmZ4jcxs/2Ak4Cb3bYIGOjlnwdc4tedTAoJf0Ombt8BTgV+aineyfGkx2DvBewFnOCPF2+xTkm7AvsBu3j7rWpGMZuQ4gz1A6YB51uKHHsMcJ2kHwBDgQubU6cgqFVimiYI8mUj4NMi2/7AHWa2mhQEaxrpi/SjZpbZV9LFQHegMymGS4HOlA8SNl7SKuBV4OeWghSW4lySQ5CNnzOM9Ojp1aRAfzsBL5vZC57lFuBk4B/Aa2Y2LWMvRCemUKaZTZfUVVJ3oAtwi6QdSCES1s/kvwH4PbC5b/eTnLtPPH0IsJukQsySbqR4H58BAyTNcXsvMk5NGZ2rgQ18K+6zcnzJV4HQbgcKAS+flXQb8ACwr5l91szygqAmiZGRIMiXLVk7kmap0N8tYRxwio88XEiKzVGgDyk+RimO8l/880ijC6XoDfQ1sweyRjP7p5n1IQV3PIzyGppyqIrjUxjwR+AxM+sLHEJDPb8mfcl/TopuOho4w0dm8Hr8qjAyZWZ9zOxhT/tXZsTqiqZ0mtlzwF3ACndiNmpCS1P6dgU+IDlRQVDXhDMSBDkhaXvSl95zRUnTgRG+3qEXMJAU4Ku5dAGWS1ofOCpzvW2BLUhBuBrjXdKv/1KcT8NpFSR1yxx+CvQlTa30do0APwOmmdl7wKeSBmTtmfML62P2J02vfEgazXjd04/JXHdLP74IuJzksEwExgNne7bJwGhvCyTtqBQJuinW0ul8CFxVPE0jaVGZctbjq0iyPyEt3EXS4aQpq4HA1T4CFAR1S0zTBEEO+Bfp/cCJJYboJwL7kpwGA84wszd9Yeh+kh4nvXe/7vuQRjz+Rho5OBeYCbxCWpPRxfPMIjkZz0gC2B4YA3zP02/MLNZc48QUsczMphfZDpB0ke+vJEXP/VTSscDdkjr6tcd6nqOBa91BWEJazFrgfUlPAF0z9stI0zSnAY9m8l5Jitq6yvUUuAKYIel24EaSwzdbKdPbpJGbplhLp6T9SNM+w4rsPSk/EvQJsIukp0mOzAjP/xdgsJm9Juka4Cq+CgMfBHVHRO0NghpB0lQzG9SSdEn3mNkRZU6pKJKmAqebWasW1uaFpINJYemvLpG20sw651CtIGhXxMhIENQONzWRflEJW/FaiaCFmNmDTecKgqAxYmQkCIIgCIJciQWsQRAEQRDkSjgjQRAEQRDkSjgjQRAEQRDkSjgjQRAEQRDkSjgjQRAEQRDkSjgjQRAEQRDkyv8AzeDKMJjvZWgAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 648x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#  Cross test\n",
    "repetitions = 5\n",
    "\n",
    "def fillA(W, times):\n",
    "    return draw_crosses(W, W * 4 * times)\n",
    "\n",
    "result = {}\n",
    "maxtimes = 81\n",
    "\n",
    "plt.figure(figsize=(9,4))\n",
    "\n",
    "widths = {24: '-', 16: '-.', 8: ':'}\n",
    "colors = {24: 'black', 16: '#333333', 8: '#666666'}\n",
    "\n",
    "for W, style in widths.items():\n",
    "    for times in range(4, maxtimes, 4):\n",
    "        durations = []\n",
    "        AA = fillA(W, times)\n",
    "        for rep in range(repetitions):\n",
    "            A = AA.copy() \n",
    "            context = Context(A.shape[0])\n",
    "\n",
    "            start = time.perf_counter()\n",
    "\n",
    "            for j in range(A.shape[1]):\n",
    "                process(A[:, j], context)\n",
    "                context.swallow()    \n",
    "            context.swallow()\n",
    "\n",
    "            finish = time.perf_counter()\n",
    "\n",
    "            durations.append(finish - start)\n",
    "\n",
    "        result[A.shape[0] * A.shape[1]] = min(durations)\n",
    "        print(times, \"/\", maxtimes, end=\" | \")\n",
    "\n",
    "    cross_test_results = sorted(result.items()) # sorted by key, return a list of tuples\n",
    "    x, y = zip(*cross_test_results) # unpack a list of pairs into two tuples\n",
    "\n",
    "    plt.xlabel(\"площадь изображения, px\")\n",
    "    plt.ylabel(\"время работы, с\")\n",
    "    plt.plot(x, y, color=colors[W], linewidth=3, linestyle=style, label=\"ширина {}\".format(W))\n",
    "\n",
    "\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4 / 101 | 8 / 101 | 12 / 101 | 16 / 101 | 20 / 101 | 24 / 101 | 28 / 101 | 32 / 101 | 36 / 101 | 40 / 101 | 44 / 101 | 48 / 101 | 52 / 101 | 56 / 101 | 60 / 101 | 64 / 101 | 68 / 101 | 72 / 101 | 76 / 101 | 80 / 101 | 84 / 101 | 88 / 101 | 92 / 101 | 96 / 101 | 100 / 101 | "
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Noise test\n",
    "W = 10\n",
    "repetitions = 50\n",
    "retries = 3\n",
    "\n",
    "def fillA_random(W, times, thr):\n",
    "    A = np.zeros((W, W * times))\n",
    "    for i in range(A.shape[0]):\n",
    "        for j in range(W):\n",
    "            A[i, j] = (random.random() < thr) * 1\n",
    "    return A\n",
    "\n",
    "result = {}\n",
    "maxtimes = 101\n",
    "\n",
    "for times in range(4, maxtimes, 4):\n",
    "    for rep in range(repetitions):\n",
    "        A = fillA_random(W, times, (rep + repetitions) / 3.0 / repetitions)\n",
    "        durations = []\n",
    "        for i in range(retries):\n",
    "            context = Context(A.shape[0])\n",
    "            start = time.perf_counter()\n",
    "            for i in range(A.shape[1]):\n",
    "                process(A[:, i], context)\n",
    "                context.swallow()    \n",
    "            context.swallow()\n",
    "            finish = time.perf_counter()\n",
    "            durations.append(finish - start)\n",
    "        if A.shape[0] * A.shape[1] not in result:\n",
    "            result[A.shape[0] * A.shape[1]] = []\n",
    "        result[A.shape[0] * A.shape[1]].append(min(durations))\n",
    "    print(times, \"/\", maxtimes, end=\" | \")\n",
    "    \n",
    "lists = sorted([(k, sum(v) / repetitions) for k, v in result.items()]) # sorted by key, return a list of tuples\n",
    "\n",
    "x, y = zip(*lists) # unpack a list of pairs into two tuples\n",
    "\n",
    "plt.xlabel(\"площадь изображения, px\")\n",
    "plt.ylabel(\"время работы, с\")\n",
    "plt.plot(x, y, color='black', linewidth=3)\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}