tweet_receiver_spark.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Trabajo de Streaming con Twitter y Spark.\n",
    "\n",
    "### Objetivo: Capturar y procesar tweets con el objetivo de analizar alguna situación, evento, etc actual ...\n",
    "\n",
    "#### Kevin Craig\n",
    "\n",
    "Nos vamos a centrar en Streaming directamente desde Twitter a Spark, sin pasar por MongoDB (aunque podríamos, pero seria Streaming a partir de una fuente de datos estática, y no es lo que buscamos)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Este script se encargará de recibir los tweets enviados mediante el socket anterior a un objeto socketTextStream de spark. \n",
    "\n",
    "*Nota*: Para poder ejecutar este notebook (que es recomendable correrlo todo de una vez, por temas de coordinación temporal con el flujo de tweets) es necesario que esté corriendo de fondo tweet_stream_tweepy.ipynb.\n",
    "\n",
    "Cargamos las librerías y spark."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import time\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import collections\n",
    "import json\n",
    "import pprint\n",
    "import unidecode\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "# Pyspark\n",
    "import findspark\n",
    "findspark.init('/home/kubote/spark/spark-2.2.0-bin-hadoop2.7/')\n",
    "\n",
    "# Librerías derivadas de pyspark\n",
    "from pyspark import SparkContext\n",
    "from pyspark.streaming import StreamingContext\n",
    "from pyspark.sql import SQLContext, SparkSession, Row\n",
    "from pyspark.sql.functions import desc\n",
    "\n",
    "try:\n",
    "\n",
    "    sc.stop()\n",
    "\n",
    "except:\n",
    "\n",
    "    print(\"\")\n",
    "\n",
    "# No queremos excesivos logs\n",
    "def quiet_logs(sc):\n",
    "  logger = sc._jvm.org.apache.log4j\n",
    "  logger.LogManager.getLogger(\"org\"). setLevel( logger.Level.ERROR )\n",
    "  logger.LogManager.getLogger(\"akka\").setLevel( logger.Level.ERROR )\n",
    "\n",
    "\n",
    "# Creamos una instancia de spark\n",
    "sc = SparkContext(\"local[4]\")\n",
    "\n",
    "# Aplicamos el silencio de logs\n",
    "quiet_logs(sc)\n",
    "\n",
    "# Creamos un Streaming object con actualización de 1 segundos. Esto significa que a la hora de capturar\n",
    "# el flujo, cada 1 segundo StreamingContext me va a crear un nuevo batch, es decir, un nuevo \"paquete\"\n",
    "# de tweets.\n",
    "ssc = StreamingContext(sc, 1)\n",
    "\n",
    "# Creamos un SQLContext para poder ejecutar comandos sql en spark.\n",
    "sqlContext = SQLContext(sc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Creamos un Stream directo que conecta con el socket usado en tweet_stream_tweepy.py. Hay que asegurarse de que el host y el puerto sean iguales."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "socket_stream = ssc.socketTextStream(\"127.0.0.1\", 9992)\n",
    "\n",
    "# Ahora seleccionamos la ventana a partir de la cual trabajar. Dado que hemos creado\n",
    "# un StreamingContext de 1 segundo, podemos definir el número de batches como el número\n",
    "# de segundos de la ventana. Usaremos ventanas de 5 minutos cada 5 minutos, de forma\n",
    "# que todos los datos que procesemos cada 5 minutos serán independientes (no tendremos\n",
    "# batches superpuestos). Esto es una prueba, en un experimento se pueden alargar los tiempos según sea conveniente.\n",
    "\n",
    "socket_stream = socket_stream.window(300, 300)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Gestionamos la recepción a través del socket y adaptamos el stream."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Definimos la clase Tweet para seleccionar lo que deseamos del Stream\n",
    "class Tweet(dict):\n",
    "\n",
    "    def __init__(self, tweet_in):\n",
    "\n",
    "        self['followers'] = tweet_in['user']['followers_count']\n",
    "        self['text'] = tweet_in['text']\n",
    "        self['hashtags'] = [x['text'] for x in tweet_in['entities']['hashtags']] if tweet_in['entities']['hashtags'] else None\n",
    "        self['has_hashtag'] = True if tweet_in['entities']['hashtags'] else False\n",
    "\n",
    "lines = socket_stream.map(lambda x: json.loads(x))\n",
    "tweets = lines.map(lambda x: Tweet(x))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Comenzamos a calcular a partir de los tweets. Únicamente usaremos el campo de texto, a pesar de que en teoría podemos disponer de cualquiera, sin embargo, se antoja complicado trabajar, por ejemplo, con el campo \"hashtags\", y muchos otros no contienen información relevante o no contienen información."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# ================ Ahora buscamos obtener estadísticas: =================== #\n",
    "\n",
    "# ====================== Primer análisis: Hashtags =========================#\n",
    "# En primer lugar, una estadística de hashtags que utilice la gente que tweittee sobre cualquiera\n",
    "# de los 5 principales candidatos a presidente del gobierno.\n",
    "\n",
    "# Definimos una tupla nombrada para delimitar los hashtags y su repetición.\n",
    "valores_hash = (\"hashtag\", \"count\")\n",
    "hash_tuple = collections.namedtuple('hash_tuple', valores_hash)\n",
    "\n",
    "# Ahora vamos a transformar nuestro objeto DStream en un dataframe para manipularlo. Spliteamos por espacios.\n",
    "lines = tweets.map(lambda x: x['text'])\n",
    "\n",
    "(lines\n",
    "# Dividimos las lineas por espacios, formando palabras.\n",
    ".flatMap(lambda line: line.split(\" \"))\n",
    "\n",
    "# Filtramos por hashtag en minúsculas.\n",
    ".filter(lambda word: word.lower().startswith(\"#\"))\n",
    "\n",
    "# Mapeamos en forma de tupla con un 1 para contar.\n",
    ".map(lambda word: (word, 1))\n",
    "\n",
    "# Reducimos por palabra del hashtag.\n",
    ".reduceByKey(lambda a, b: a + b)\n",
    "\n",
    "# Transformamos en la tupla nombrada \"Tweet\".\n",
    ".map(lambda rec: hash_tuple(rec[0], rec[1]))\n",
    "\n",
    "# Para cada batch, pasamos el conjunto de tuplas nombradas a dataframes ordenando\n",
    "# por orden descendiente de su repetición.\n",
    ".foreachRDD(lambda rdd: rdd.toDF().sort(desc(\"count\"))\n",
    "\n",
    "# Limitamos la salida a 15 y creamos una tabla temporal para usar comandos SQL en ella.\n",
    ".limit(5).registerTempTable(\"tabla_hashtags\")))\n",
    "\n",
    "\n",
    "# =================== Segundo análisis: popularidad ===================== #\n",
    "# Vamos a medir la popularidad de los 5 principales candidatos a presidente de España en las elecciones\n",
    "# del 28 de Abril. Lo haremos contando el número de veces que sus nombres son tweitteados en relación\n",
    "# al resto.\n",
    "\n",
    "# Definimos una tupla nombrada para delimitar los hashtags y su repetición.\n",
    "valores_pop = (\"candidato\", \"count\")\n",
    "pop_tuple = collections.namedtuple('pop_tuple', valores_pop)\n",
    "\n",
    "# Definimos los candidatos por los que detectar tweets.\n",
    "#candidatos_lower = ['trump', 'clinton', 'obama', 'abascal', 'iglesias', 'sanchez', 'rajoy', 'rivera']\n",
    "candidatos_lower = [['vox', 'abascal'], ['podemos', 'iglesias'], ['psoe', 'sanchez'], ['pp', 'casado'], ['ciudadanos', 'rivera']]\n",
    "\n",
    "# Cremos una función para crear una tabla con las veces que se menciona a cada candidato.\n",
    "def calcular_tabla(candidato):\n",
    "\n",
    "    nombre_tabla = \"tabla_pop_\" + candidato[0]\n",
    "\n",
    "    # Ahora vamos a transformar nuestro objeto DStream en un dataframe para manipularlo. Spliteamos por espacios.\n",
    "    lines = tweets.map(lambda x: x['text'])\n",
    "    (lines\n",
    "    # Dividimos las lineas por espacios, formando palabras.\n",
    "    .flatMap(lambda line: line.replace(\"#\", \"\").replace(\"'\", \"\").replace(\",\", \"\").replace(\".\", \"\").split(\" \"))\n",
    "\n",
    "    # Filtramos por palabras clave en minúsculas y sin acentos.\n",
    "    .map(lambda word: (any(x in unidecode.unidecode(word).lower() for x in candidato), 1))\n",
    "\n",
    "    # Reducimos por True y False, siendo true y false los valores de palabra detectada y no.\n",
    "    .reduceByKey(lambda a, b: a + b)\n",
    "\n",
    "    # Transformamos en la tupla nombrada \"Tweet\".\n",
    "    .map(lambda tupla: pop_tuple(tupla[0], tupla[1]))\n",
    "\n",
    "    # Para cada batch, pasamos el conjunto de tuplas nombradas a dataframes ordenando\n",
    "    # por orden descendiente de su repetición.\n",
    "    .foreachRDD(lambda rdd: rdd.toDF().sort(desc(\"count\"))\n",
    "\n",
    "    # Limitamos la salida a 2 y creamos una tabla temporal para usar comandos SQL en ella.\n",
    "    .limit(2).registerTempTable(nombre_tabla)))\n",
    "\n",
    "# Llamadas para crear cada una de las tablas necesarias con cada batch.\n",
    "for cand in candidatos_lower:\n",
    "    calcular_tabla(cand)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Iniciamos el Stream a través del socket. A partir de este paso, el flujo de datos será contínuo y no se podrá modificar el procesado, por lo que es importante contar con datos antes de empezar a procesar. Es por ello que fijamos un tiempo (delay) a partir del cual comenzar a procesar."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# Iniciamos el stream.\n",
    "ssc.start()\n",
    "\n",
    "# Introducimos un delay de 6 minutos para que se calcule el primer batch y las tablas derivadas.\n",
    "time.sleep(360)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ahora realizaremos las dos tareas objetivo, el estudio de menciones de los candidatos a presidente del gobierno y su evolución temporal, y el estudio de los hashtags utilizados en relación a los mismos."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAD8CAYAAAB3u9PLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAFl9JREFUeJzt3X2QXfV93/H3xyIG7FgYyqIqErZIqzoFxkDYavCQOmlwg9KkEU1DLFobOaWjmuKnOq4H9dE0ow5JM41LG+QyDkYEB0WBOMjY2CGyseMYIxaDAfFgVINBkYLWjh/UZKxE8rd/nJ/KmdVq964e9i7o/Zo5c8/53vM793fu7r2fex7uuakqJEl62bA7IEmaGwwESRJgIEiSGgNBkgQYCJKkxkCQJAEGgiSpMRAkSYCBIElqjht2B6Zz6qmn1pIlS4bdDUl6UXnggQe+UVUjM2kz5wNhyZIljI2NDbsbkvSikuTrM23jLiNJEmAgSJIaA0GSBAwQCElel+Sh3vDdJO9JckqSu5M81W5P7rVZk2RbkieTXNyrn5/kkXbfdUlytFZMkjQz0wZCVT1ZVedW1bnA+cBfAh8DrgY2V9VSYHObJsmZwErgLGA5cH2SeW1x64DVwNI2LD+yqyNJOlQz3WV0EfB/qurrwApgfauvBy5p4yuADVW1p6qeBrYBy5IsBOZX1b3V/SrPzb02kqQhm2kgrARubeMLqmonQLs9rdUXAc/12mxvtUVtfGJdkjQHDBwISV4O/Bzwe9PNOkmtpqhP9lirk4wlGRsfHx+0i5KkwzCTLYSfBr5cVc+36efbbiDa7a5W3w6c3mu3GNjR6osnqR+gqm6oqtGqGh0ZmdEX7SRJh2gmgXAZL+wuAtgErGrjq4A7evWVSY5PcgbdweMtbbfS7iQXtLOLLu+1OaYkmfEgSUfbQJeuSPIK4B8C/6pXvhbYmOQK4FngUoCq2ppkI/AYsBe4qqr2tTZXAjcBJwJ3tUGSNAekO+Fn7hodHa2X2rWMDuUT/1z/O0maW5I8UFWjM2njN5UlSYCBIElqDARJEmAgSJIaA0GSBBgIkqTGQJAkAQaCJKkxECRJgIEgSWoMBEkSYCBIkhoDQZIEGAiSpMZAkCQBBoIkqTEQJEmAgSBJagwESRJgIEiSGgNBkgQMGAhJXp3ktiRPJHk8yRuSnJLk7iRPtduTe/OvSbItyZNJLu7Vz0/ySLvvuiQ5GislSZq5QbcQ/gfwqar6EeAc4HHgamBzVS0FNrdpkpwJrATOApYD1yeZ15azDlgNLG3D8iO0HpKkwzRtICSZD7wR+C2Aqvqrqvo2sAJY32ZbD1zSxlcAG6pqT1U9DWwDliVZCMyvqnurqoCbe20kSUM2yBbCDwPjwEeSPJjkw0leCSyoqp0A7fa0Nv8i4Lle++2ttqiNT6wfIMnqJGNJxsbHx2e0QpKkQzNIIBwH/CiwrqrOA/6CtnvoICY7LlBT1A8sVt1QVaNVNToyMjJAFyVJh2uQQNgObK+q+9r0bXQB8XzbDUS73dWb//Re+8XAjlZfPEldkjQHTBsIVfVnwHNJXtdKFwGPAZuAVa22CrijjW8CViY5PskZdAePt7TdSruTXNDOLrq810aSNGTHDTjfO4GPJnk58DXgl+jCZGOSK4BngUsBqmprko10obEXuKqq9rXlXAncBJwI3NUGSdIckO6En7lrdHS0xsbGht2NI+pQvn4x1/9OkuaWJA9U1ehM2vhNZUkSYCBIkhoDQZIEGAiSpMZAkCQBBoIkqTEQJEmAgSBJagwESRJgIEiSGgNBkgQYCJKkxkCQJAEGgiSpMRAkSYCBIElqDARJEmAgSJIaA0GSBBgIkqTGQJAkAXDcIDMleQbYDewD9lbVaJJTgN8FlgDPAL9YVd9q868Brmjzv6uqPt3q5wM3AScCnwTeXVV15FZn9uWaDLsLknREzGQL4R9U1blVNdqmrwY2V9VSYHObJsmZwErgLGA5cH2Sea3NOmA1sLQNyw9/FSRJR8Lh7DJaAaxv4+uBS3r1DVW1p6qeBrYBy5IsBOZX1b1tq+DmXhtJ0pANGggF/GGSB5KsbrUFVbUToN2e1uqLgOd6bbe32qI2PrF+gCSrk4wlGRsfHx+wi5KkwzHQMQTgwqrakeQ04O4kT0wx72Q71WuK+oHFqhuAGwBGR0df1McYJOnFYqAthKra0W53AR8DlgHPt91AtNtdbfbtwOm95ouBHa2+eJK6JGkOmDYQkrwyyav2jwM/BTwKbAJWtdlWAXe08U3AyiTHJzmD7uDxlrZbaXeSC5IEuLzXRpI0ZIPsMloAfKx7D+c44Heq6lNJ7gc2JrkCeBa4FKCqtibZCDwG7AWuqqp9bVlX8sJpp3e1QZI0B0wbCFX1NeCcSerfBC46SJu1wNpJ6mPA2TPvpiTpaPObypIkwECQJDUGgiQJMBAkSY2BIEkCDARJUmMgSJIAA0GS1BgIkiTAQJAkNQaCJAkwECRJjYEgSQIMBElSM+hPaEpHRPtdjRmp8ldUpdngFoIkCTAQJEmNgSBJAgwESVJjIEiSAANBktQMHAhJ5iV5MMmdbfqUJHcneardntybd02SbUmeTHJxr35+kkfafdflUM5BlCQdFTPZQng38Hhv+mpgc1UtBTa3aZKcCawEzgKWA9cnmdfarANWA0vbsPywei9JOmIGCoQki4GfAT7cK68A1rfx9cAlvfqGqtpTVU8D24BlSRYC86vq3uq+aXRzr40kacgG3UL4IPB+4Pu92oKq2gnQbk9r9UXAc735trfaojY+sX6AJKuTjCUZGx8fH7CLkqTDMW0gJPlZYFdVPTDgMic7LlBT1A8sVt1QVaNVNToyMjLgw0qSDscg1zK6EPi5JP8IOAGYn+QW4PkkC6tqZ9sdtKvNvx04vdd+MbCj1RdPUpckzQHTbiFU1ZqqWlxVS+gOFn+mqt4CbAJWtdlWAXe08U3AyiTHJzmD7uDxlrZbaXeSC9rZRZf32kiShuxwrnZ6LbAxyRXAs8ClAFW1NclG4DFgL3BVVe1rba4EbgJOBO5qg3TMmOmZ1i+lK716pdu5b0aBUFX3APe08W8CFx1kvrXA2knqY8DZM+2kJOno85vKkiTAQJAkNQaCJAkwECRJjYEgSQIMBElSYyBIkgADQZLUGAiSJMBAkCQ1BoIkCTAQJEmNgSBJAgyEAyQzGyTppcJAkCQBBoIkqTEQJEmAgSBJag7nN5Ulco1H1qWXCrcQJEnAAIGQ5IQkW5J8JcnWJNe0+ilJ7k7yVLs9uddmTZJtSZ5McnGvfn6SR9p91yWeuClJc8UgWwh7gJ+sqnOAc4HlSS4ArgY2V9VSYHObJsmZwErgLGA5cH2SeW1Z64DVwNI2LD+C6yJJOgzTBkJ1/m+b/IE2FLACWN/q64FL2vgKYENV7amqp4FtwLIkC4H5VXVvVRVwc6+NJGnIBjqGkGRekoeAXcDdVXUfsKCqdgK029Pa7IuA53rNt7faojY+sS5JmgMGCoSq2ldV5wKL6T7tnz3F7JMdF6gp6gcuIFmdZCzJ2Pj4+CBdlCQdphmdZVRV3wbuodv3/3zbDUS73dVm2w6c3mu2GNjR6osnqU/2ODdU1WhVjY6MjMyki5KkQzTIWUYjSV7dxk8E3gQ8AWwCVrXZVgF3tPFNwMokxyc5g+7g8Za2W2l3kgva2UWX99pIkoZskC+mLQTWtzOFXgZsrKo7k9wLbExyBfAscClAVW1NshF4DNgLXFVV+9qyrgRuAk4E7mqDJGkOmDYQquph4LxJ6t8ELjpIm7XA2knqY8BUxx8kvUj4LfWXHi9dIR0i3xD1UuOlKyRJgIEgSWoMBEkSYCBIkhoDQZIEGAiSpMZAkCQBBoIkqTEQJEmAgSBJagwESRJgIEiSGgNBkgQYCJKkxkCQJAEGgiSpMRAkSYCBIElqDARJEmAgSJKaaQMhyelJPpvk8SRbk7y71U9JcneSp9rtyb02a5JsS/Jkkot79fOTPNLuuy6Jv1IuSXPEIFsIe4Ffrqq/C1wAXJXkTOBqYHNVLQU2t2nafSuBs4DlwPVJ5rVlrQNWA0vbsPwIrosOUzLzQdJLx7SBUFU7q+rLbXw38DiwCFgBrG+zrQcuaeMrgA1Vtaeqnga2AcuSLATmV9W9VVXAzb02kqQhm9ExhCRLgPOA+4AFVbUTutAATmuzLQKe6zXb3mqL2vjEuiRpDhg4EJL8IHA78J6q+u5Us05Sqynqkz3W6iRjScbGx8cH7aIk6TAMFAhJfoAuDD5aVb/fys+33UC0212tvh04vdd8MbCj1RdPUj9AVd1QVaNVNToyMjLoukiHzOMn0mBnGQX4LeDxqvrvvbs2Aava+Crgjl59ZZLjk5xBd/B4S9uttDvJBW2Zl/faSJKG7LgB5rkQeCvwSJKHWu3fAdcCG5NcATwLXApQVVuTbAQeoztD6aqq2tfaXQncBJwI3NUGSdIcMG0gVNUXmHz/P8BFB2mzFlg7SX0MOHsmHZQkzQ6/qSzJYygCDARJUmMgSJIAA0GS1BgIkiTAQJAkNYN8D+HF65BOhZj0ahqSdFhmerX/7hqgs8stBEkSYCBIkhoDQZIEGAiSpMZAkCQBBoIkqTEQJEmAgSBJagwESRJgIEiSGgNBkgQYCJKk5qV9cTtJOgpyzUvzN0TdQpAkAQMEQpIbk+xK8mivdkqSu5M81W5P7t23Jsm2JE8mubhXPz/JI+2+6zLTa8FKko6qQbYQbgKWT6hdDWyuqqXA5jZNkjOBlcBZrc31Sea1NuuA1cDSNkxcpiRpiKYNhKr6PPDnE8orgPVtfD1wSa++oar2VNXTwDZgWZKFwPyqure6X324uddGkjQHHOpB5QVVtROgqnYmOa3VFwFf6s23vdX+uo1PrOto8dfiJM3QkT6oPNm7UE1Rn3whyeokY0nGxsfHj1jnJEkHd6iB8HzbDUS73dXq24HTe/MtBna0+uJJ6pOqqhuqarSqRkdGRg6xizqmJTMbJB1yIGwCVrXxVcAdvfrKJMcnOYPu4PGWtntpd5IL2tlFl/faSDqSZhqGBqKaaY8hJLkV+Ang1CTbgf8MXAtsTHIF8CxwKUBVbU2yEXgM2AtcVVX72qKupDtj6UTgrjZIkuaIaQOhqi47yF0XHWT+tcDaSepjwNkz6p0kadb4TWVJEmAgSJIaA0HSMc1j8C8wECRJgIEgSWoMBEkSYCBIkhoDQZIEGAiSpMZAkCQBBoIkqTEQJEmAgSBJag71JzQlaW6a8bUl/OnY/dxCkCQBBoIkqTEQJEmAgSBJagwESRJgIEiSGgNBkgQMIRCSLE/yZJJtSa6e7ceXJE1uVgMhyTzgN4GfBs4ELkty5mz2QZI0udneQlgGbKuqr1XVXwEbgBWz3AdJ0iRmOxAWAc/1pre3miRpyFI1e9fxSHIpcHFV/cs2/VZgWVW9c8J8q4HVbfJ1wJOz1snhOhX4xrA7MUTH8vofy+sOx/b6H611f21VjcykwWxf3G47cHpvejGwY+JMVXUDcMNsdWquSDJWVaPD7sewHMvrfyyvOxzb6z+X1n22dxndDyxNckaSlwMrgU2z3AdJ0iRmdQuhqvYmeQfwaWAecGNVbZ3NPkiSJjfrv4dQVZ8EPjnbj/siccztJpvgWF7/Y3nd4dhe/zmz7rN6UFmSNHd56QpJEmAgaJYl+Ykkdw67H9JMJHl7kstn2OaeJAOfPTQXXhv+prIkTaOqPjTsPswGtxBmUZJfTfKve9MfSPLLSf5bkkeTPJLkze2+f5Lkj9JZmOSrSf7m8Hr/giRLkjyRZH2Sh5PcluQVSS5K8mBbjxuTHN/mX97m/wLw873lvLLNd39rt6LV35bkD5J8PMnTSd6R5L1tni8lOaXNd26bfjjJx5Kc3OrvSvJYq28YwlO0f/0O9jxd2+vfr7d5X5tkc6ttTvKaVh9Jcnt7ju5PcuGw1udImOI5eaa9Pra04W8PuZ+Xt/59Jclvt9fq+9p9//+Tf5JTkzzTxk9MsqG1+13gxN7y1iUZS7I1yTW9+sFeG8uSfLH9z38xyeta/W1Jfj/Jp5I8leTXem0ua6+9R5P8aqvNS3JT7/3l30y54lXlMEsDcB7wud70Y8Aq4G6603AXAM8CC9v9twDvAO4ELht2/3v9XgIUcGGbvhH4D3SXJfk7rXYz8B7ghFZfCgTYCNzZ5vmvwFva+KuBrwKvBN4GbANeBYwA3wHe3ub7DeA9bfxh4Mfb+H8BPtjGdwDH71/uHHue3k/3zfv0+wd8HFjVxv8F8Adt/HeAH2vjrwEeH/bf/yg8J+8DngH+fatdvv9/ZEh9PKv9jU5t06cAHwDe16bvAUbb+KnAM238vXSn0gO8Htjbm++UdjuvtX/9NK+N+cBxbfxNwO1t/G3A14CTWvuv033Z94fo3jtG6Pb8fAa4BDgfuLu3blO+HtxCmEVV9SBwWpIfSnIO8C3gXODWqtpXVc8DnwP+XmvyTmANsKeqbh1Kpw/uuar6kzZ+C3AR8HRVfbXV1gNvBH6k1Z+q7j/ylt4yfgq4OslDdC+SE+je9AA+W1W7q2qcLhA+3uqPAEuSnET3z/25CY8HXVB8NMlb6F6UwzTxeXoj8D3gw0l+HvjLdt8b6N78AX4b+LE2/ibgf7XnaBMwP8mrZqXnR8/E52T/ut7au33DrPfqBT8J3FZV3wCoqj8fsN0baf/fVfUw3f/hfr+Y5MvAg3SBcyZTvzZOAn4vyaN0H4LO6t23uaq+U1Xfo/tQ+Vq694x7qmq8qvYCH239+Rrww0n+Z5LlwHenWgEDYfbdBvwC8Ga6q71minkXAd8HFiSZa3+rmZyvfLB5A/zTqjq3Da+pqsfbfXt6832/N/19pj/29TN0l1k/H3ggyTCPlU1c97+mu+rv7XSf4D41TbuXAW/oPUeLqmr30enqrJn4nNQk9WGeD59pHn8vL7x3njDhvgPaJTmDbivooqp6PfCJXruDPc6v0H0oOhv4xxMep//a2Ef3epj0faSqvgWcQ/eB6yrgwwd5PMBAGIYNdJfs+AW6cPg88Oa2r2+ELtW3tDexjwD/DHicbnN0LnlNkv2f4i4D/ojuk/v+fb9vpdvaeQI4I8nf6s2736eBdyYJQJLzBn3wqvoO8K0kf7//eC04T6+qz9Ltnnk18IMzXrsjZ+Lz9BBwUnVf0HwP3RYiwBfp/i8A/jnwhTb+h3S7DYHuuMlR7/HRN/E52b+ub+7d3jvrvXrBZrpP9H8DYP8xq55n6D5sQPc63u/zdH87kpxNt1sIut0/fwF8J8kCut+DgalfGycBf9rG3zZAn+8Dfrwd05jXlvW5JKcCL6uq24H/CPzoVAvxLKNZVlVb2yb/n1bVziQfo9s8/grdp4X3V9WfJflPwB9X1R+33QX3J/lE7xP0sD0OrEryv4GngHcDX6LbzD2O7rpVH6qqPemuXvuJJN+ge/Gf3ZbxK8AHgYdbKDwD/OwM+rAK+FCSV9BtGv8S3T7aW9oupQC/UVXfPrxVPSwTn6cPAHcmOaH1b/9BvncBNyb5t8A43brsr/9mkofpXq+fB94+e90/KiY+J+vodo8en+Q+ug+ql03R/qhqr9G1dG+o++h28zzTm+XXgY3prtb8mV59HfCR9rd6CNjSlveVJA8CW+n+T/+k1b83xWvj14D1Sd474TEO1uedSdYAn6X7v/pkVd3Rdk1/pLeHYc1Uy/GbypqxJEvoDn6dPc2sxzSfpwMd7DlpZ+qM7t9vr+Fwl5EkCXALQZLUuIUgSQIMBElSYyBIkgADQZLUGAiSJMBAkCQ1/w+8zaHJaUaVNwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "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": [
    "\n",
    "# Diccionario para crear el dataframe.\n",
    "df_dict = {\"candidatos\": ['vox', 'podemos', 'psoe', 'pp', 'ciudadanos'], \"suma_1\": np.zeros(5), \"suma_2\": np.zeros(5), \"suma_3\": np.zeros(5), \"suma_4\": np.zeros(5), \"suma_5\": np.zeros(5)}\n",
    "\n",
    "# Definimos dos dataframes vacíos con los candidatos y sus menciones y otro con los hashtags y su conteo.\n",
    "df_can = pd.DataFrame(df_dict)\n",
    "df_hashtags_fin = pd.DataFrame({'hashtag': [], 'count': []})\n",
    "\n",
    "# Rellenamos el dataframe anterior de forma acumulativa en 5 ciclos.\n",
    "for a in range(5):\n",
    "\n",
    "    for i in candidatos_lower:\n",
    "\n",
    "        nombre_tabla = \"tabla_pop_\" + i[0]\n",
    "\n",
    "        # Añadimos los valores de True al dataframe creado anteriormente.\n",
    "        top_pop = sqlContext.sql('Select candidato, count from ' + nombre_tabla)\n",
    "        tabla = top_pop.toPandas()\n",
    "\n",
    "        # Nos aseguramos de que exista un valor para añadir, en caso contrario añadimos 0.\n",
    "        if tabla.shape[0] > 1:\n",
    "            val = tabla.iloc[np.where(tabla['candidato'] == True)[0], 1]\n",
    "            val = val[1]\n",
    "        else:\n",
    "            val = 0\n",
    "\n",
    "        # Sumamos a los 0s los valores obtenidos.\n",
    "        df_can.iloc[np.where(df_can['candidatos'] == i[0])[0], a + 1] += val\n",
    "\n",
    "        # Necesitamos sumar las nuevas menciones para mostrar el aumento.\n",
    "        if a > 0:\n",
    "\n",
    "            df_can.iloc[np.where(df_can['candidatos'] == i[0])[0], a + 1] += df_can.iloc[np.where(df_can['candidatos'] == i[0])[0], a]\n",
    "\n",
    "        # Hacemos algo similar con los hashtags, hacemos un merge conservando todos los hashtags diferentes\n",
    "        # y sumando los ya utilizados.\n",
    "        top_hashtags = sqlContext.sql('Select hashtag, count from tabla_hashtags')\n",
    "        df_hashtags = top_hashtags.toPandas()\n",
    "\n",
    "        df_hashtags_fin = df_hashtags_fin.set_index('hashtag').add(df_hashtags.set_index('hashtag'), fill_value=0).reset_index()\n",
    "\n",
    "    # Mostramos la comparativa de menciones en cada iteración.\n",
    "    x = df_can['candidatos']\n",
    "    y = df_can.iloc[:, 1:(a + 2)]\n",
    "\n",
    "    colores = ['red', 'blue', 'green', 'black', 'pink']\n",
    "    for o in range(a + 1):\n",
    "\n",
    "        plt.bar([idx + 0.8*o/(a + 1) for idx in range(len(x))], y.iloc[:, o], color = colores[o], width = 0.8/(a + 1), tick_label = x)\n",
    "    plt.show()\n",
    "\n",
    "    # Introducimos un delay igual al tiempo entre ventanas para que se calcule el nuevo batch y tablas derivadas.\n",
    "    time.sleep(300)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Realizamos los plots finales y los guardamos. También mostramos las tablas obtenidas."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x864 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   candidatos  suma_1  suma_2  suma_3  suma_4  suma_5\n",
      "0         vox  1617.0  3374.0  5120.0  6926.0  8752.0\n",
      "1     podemos  1041.0  2102.0  3202.0  4346.0  5463.0\n",
      "2        psoe   949.0  1952.0  2850.0  3885.0  4939.0\n",
      "3          pp   931.0  1828.0  2858.0  3870.0  4810.0\n",
      "4  ciudadanos   388.0   751.0  1184.0  1632.0  2046.0\n",
      "\n",
      "\n",
      "                    hashtag   count\n",
      "0                      #28A  2350.0\n",
      "1              #Bogotá.\\nLa    18.0\n",
      "2           #EquiparacionYa    60.0\n",
      "3              #EstáPasando    55.0\n",
      "4            #FinCampañaARV   110.0\n",
      "5                 #Iglesias    68.0\n",
      "6                  #Toledo,   195.0\n",
      "7   #TuAbandonoMePuedeMatar    36.0\n",
      "8                      #VOX   370.0\n",
      "9                 #VotaPP🇪🇸    18.0\n",
      "10               #VotaPSOE…    65.0\n",
      "11                     #草彅剛   455.0\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# Plots finales:\n",
    "\n",
    "# de menciones:\n",
    "x = df_can['candidatos']\n",
    "y = df_can.iloc[:, 1:(a + 2)]\n",
    "\n",
    "colores = ['red', 'blue', 'green', 'black', 'pink']\n",
    "plt.figure(figsize = (15, 8))\n",
    "for o in range(a + 1):\n",
    "\n",
    "    plt.bar([idx + 0.8*o/(a + 1) for idx in range(len(x))], y.iloc[:, o], color = colores[o], width = 0.8/(a + 1), tick_label = x)\n",
    "plt.title(\"Número de menciones acumuladas cada 15 segundos por partido.\")\n",
    "plt.savefig('menciones.png')\n",
    "plt.show()\n",
    "\n",
    "# de hashtags.\n",
    "x = df_hashtags_fin['hashtag']\n",
    "y = df_hashtags_fin.iloc[:, 1]\n",
    "\n",
    "plt.figure(figsize = (20, 12))\n",
    "plt.barh([idx for idx in range(len(x))], y, tick_label = x)\n",
    "plt.title(\"Hashtags más utilizados en tweets mencionando a alguno de los 5 partidos principales.\")\n",
    "plt.savefig('hashtags.png')\n",
    "plt.show()\n",
    "\n",
    "print(df_can)\n",
    "\n",
    "print(\"\\n\")\n",
    "\n",
    "print(df_hashtags_fin)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finalizamos deteniendo el streaming."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " \n"
     ]
    }
   ],
   "source": [
    "\n",
    "# Detenemos el streaming.\n",
    "ssc.stop()\n",
    "\n",
    "# Limpiamos el puerto utilizado para poder usarlo inmediatamente después.\n",
    "os.system('kill $(lsof -ti tcp:9992)')\n",
    "os.system('fuser 9992/tcp')\n",
    "\n",
    "print(\" \")\n"
   ]
  }
 ],
 "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
}