4d_crime.rb

# prototype of crime mapping app

require 'csv'
require 'date'
require 'rubygems'
require 'sparklines'
require 'RMagick'
require 'uuid'

ONLY_HOMICIDES = false
street_parts = /(\d+) B\/O (\d+)(TH|ND|RD) ST (NW|SW|NE|SE)/

start = Time.now
crimes = []

# This loop is probably very slow because
# CSV isn't speedy and neither is date parsing
# strptime and Date.parse are basically identical
# performance-wise

def dateagram(dates, year)
  out = []
  (1..12).each do |month|
    daycounter = Date.strptime("#{month}/1/#{year}", "%m/%d/%Y")
    o = 1
    while daycounter.month == month
      if dates.include? daycounter
        o = 2
        break
      end
      daycounter += 1
    end
    out << o
  end
  id = UUID.new.generate
  line = Sparklines.plot_to_image(out,
    :type => 'whisker',
    :step => 1,
    :whisker_color => "#eeeeee",
    :exception_color => "red",
    :height => 20)
  line.write("graphs/#{id}.png")
  "<img src='graphs/#{id}.png' />"
end

CSV.open('crime_2009a.csv', 'r') do |row|
  type = row[4]
  place = row[7]
  date = Date.strptime(row[2], '%m/%d/%Y')
  stp = street_parts.match(row[7])
  if stp.nil?
    next
  end
  if ONLY_HOMICIDES and row[4] != 'HOMICIDE' ||
    stp[4] != 'NW'
    next
  end
  crimes << [date, stp[1].to_i, stp[2].to_i]
end

grid = {}

crimes.each do |c|
  k = "#{c[2]}-#{c[1]}"
  if grid.has_key?(k)
    grid[k] << c[0]
  else
    grid[k] = [c[0]]
  end
end

sorted_grid = grid.sort_by do |a|
  [a[0].split('-')[0].to_i,a[0].split('-')[1].to_i]
end

street = 0
puts "<table>"
sorted_grid.each do |k, v|
  if k.split('-')[0].to_i != street
    puts "</table><table>"
    puts "<th colspan='2' class='header'>#{k.split('-')[0]}th</th>"
    street = k.split('-')[0].to_i
  end
  puts "<tr><th><span class='block'>#{k.split('-')[1]}</span></th><td>#{dateagram(v, 2009)}</td></tr>"
end

puts "</body></html>"