{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Word count with PySpark\n",
    "\n",
    "#### Topics covered in this example\n",
    "* Write a file to HDFS, read the file and perform word count on the data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "***\n",
    "\n",
    "## Prerequisites\n",
    "<div class=\"alert alert-block alert-info\">\n",
    "<b>NOTE :</b> In order to execute this notebook successfully as is, please ensure the following prerequisites are completed.</div>\n",
    "\n",
    "* The EMR cluster attached to this notebook should have the `Spark` application installed.\n",
    "* This notebook uses the `PySpark` kernel.\n",
    "***"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Introduction\n",
    "In this example we write a file to hdfs, use pyspark to count the occurrence of each word in the file stored in hdfs and store the results to s3.\n",
    "***"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Setup\n",
    "1. Create a S3 bucket to save your results or use an existing s3 bucket. For example: `s3://EXAMPLE-BUCKET/word-count/`\n",
    "***"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example\n",
    "\n",
    "Create a test data frame with some sample records.\n",
    "We will use the `createDataFrame()` method to create and `printSchema()` method to print out the schema."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "wordsDF = sqlContext.createDataFrame([(\"emr\",), (\"spark\",), (\"example\",), (\"spark\",), (\"pyspark\",), (\"python\",),\n",
    "             (\"example\",), (\"emr\",), (\"example\",), (\"spark\",), (\"pyspark\",), (\"python\",)], [\"words\"])\n",
    "wordsDF.show()\n",
    "wordsDF.printSchema()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Print out the number of unique words so that we can verify this number with the end result count."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "uniqueWordsCount = wordsDF.distinct().groupBy().count().head()[0]\n",
    "print(uniqueWordsCount)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This step only shows an example on how to write to hdfs.\n",
    "You can use an existing file stored in hdfs and read it as shown in the next steps."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "wordsDF.write.csv(\"hdfs:///user/hadoop/test-data.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Read the csv file from hdfs and store in RDD."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "wordsData = sc.textFile(\"hdfs:///user/hadoop/test-data.csv\")\n",
    "wordsData.count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Display the contents of the file."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "wordsData.collect()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Count the occurance of each word and print the count of the result. This should be equal to the number of unique words we found earlier."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "wordsCounts = wordsData.flatMap(lambda line: line.split(\" \")) \\\n",
    "                       .map(lambda word: (word, 1)) \\\n",
    "                       .reduceByKey(lambda a, b: a+b)\n",
    "wordsCounts.count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Display the count for each word."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "wordsCounts.collect()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Save the results to your s3 bucket. The results are stored in the key `word-count` and split based on paritions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "wordsCounts.saveAsTextFile(\"s3://EXAMPLE-BUCKET/word-count\") # Change this to the S3 location that you created in Setup step 1."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "PySpark",
   "language": "",
   "name": "pysparkkernel"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "python",
    "version": 3
   },
   "mimetype": "text/x-python",
   "name": "pyspark",
   "pygments_lexer": "python3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}