/*
  This script parses the mass shootings CSV and makes it easier to deal with.

  case,
  Rite Aid warehouse shooting,

  location,date,summary,
  "Perryman, MD",
  9/20/18,
  "Snochia Moseley, 26, reportedly a disgruntled employee, shot her victims outside the building and on the warehouse floor; she later died from a self-inflicted gunshot at a nearby hospital. (No law enforcement officers responding to her attack fired shots.)",

  fatalities,injured,total_victims,
  3,3,6,

  location_type,
  age_of_shooter,
  Workplace,
  26,

  prior_signs_mental_health_issues,mental_health_details,
  -,-,

  weapons_obtained_legally,where_obtained,weapon_type,weapon_details,
  Yes,-,semiautomatic handgun,Glock 9 mm,

  race,gender,
  Black,F,

  sources,mental_health_sources,sources_additional_age,
  http://www.baltimoresun.com/news/maryland/crime/bs-md-harford-shooting-20180920-story.html; https://www.washingtonpost.com/local/public-safety/maryland-sheriffs-office-says-multiple-victims-in-shooting/2018/09/20/9d2c6464-bcda-11e8-be70-52bd11fe18af_story.html; https://www.cnn.com/2018/09/20/us/maryland-shooting/index.html; https://heavy.com/news/2018/09/snochia-moseley/,-,-,

  latitude,longitude,type,year
  39.455658,-76.208485,Mass,2018
*/

const fs = require('fs')
const parse = require('csv-parse')
const stringify = require('csv-stringify')

const fields = (
  "case," +
  "location,date,summary,fatalities,injured,total_victims," +
  "location_type," +
  "age_of_shooter," +
  "prior_signs_mental_health_issues,mental_health_details," +
  "weapons_obtained_legally,where_obtained,weapon_type,weapon_details," +
  "race,gender," +
  "sources,mental_health_sources,sources_additional_age," +
  "latitude,longitude,type,year").split(',').reduce((a,b,i) => {
  a[b] = i
  return a
}, {})

const headings = ["date", "timestamp", "fatalities", "injured", "total_victims", "age", "case", "weapon_type", "weapon_details"]

function test(row){
  return (
    !!row
    // row[fields.incident_characteristics].indexOf('Samaritan') !== -1
    // // && row[fields.participant_age_group].indexOf('Teen') !== -1
    // && row[fields.participant_age_group].indexOf('Child') !== -1
    // && row[fields.gun_type].indexOf('AR-15') !== -1
  )
}

function apply(row){
  let [m, d, y] = row[fields.date].split('/')
  if (y < 20) y = parseInt(y) + 2000
  return [
    [y, pad(m), pad(d)].join('/'),
    +new Date(y, m, d),
    row[fields.fatalities],
    row[fields.injured],
    row[fields.total_victims],
    row[fields.age_of_shooter],
    row[fields.case],
    row[fields.weapon_type].replace(/\s+$/g, ''),
    row[fields.weapon_details].replace(/\s+$/g, ''),
  ]
}
function pad(n) { return (n < 10) ? ("0" + n) : n }

const input = fs.createReadStream('./data/mass_shootings.csv')
const parser = parse()
const stringifier = stringify()
const output = fs.createWriteStream('./data/mass_shootings_lite.csv')

input.on('readable', function() {
  let buf
  while ((buf = input.read()) !== null) {
    parser.write(buf)
  }
})
input.on('error', function(err){
  console.error('input error', err.message)
})
input.on('finish', function(){
  parser.end()
})

let i = 0
parser.on('readable', function(){
  let row
  while (row = parser.read()) {
    if (i === 0) {
      stringifier.write(headings)
      i += 1
      continue
    }
    if ((++i % 10000) === 0) {
      console.log(i + '...')
    }
    if (test(row)) {
      stringifier.write(apply(row))
    }
  }
})
parser.on('error', function(err){
  console.error('parser error', err.message)
})
parser.on('end', function(){
  stringifier.end()
})

stringifier.on('readable', function(){
  let row
  while(row = stringifier.read()){
    output.write(row)
  }
})
stringifier.on('error', function(err){
  console.error('stringifier error', err.message)
})
stringifier.on('finish', function(){
  output.end()
})