Make a roster from info pulled from quidditch-reference.com

mwrc_rosters.rb

require 'open-uri'
require 'nokogiri'
require 'pp'
require 'json'

# currently hardcoded for the MWRC event
def pull_players(team_code)
  doc = Nokogiri::HTML(open("http://www.quidditch-reference.com/tournament/160/#{team_code}"))
  rows = doc.xpath("//table//tr[position() > 1]")
  team_name = doc.css("h2")[0].children[3].to_s
  # generate a random team id with any letter or number
  team_id = [('A'..'Z'), ('a'..'z'), (0..9)].map{|a| a.to_a}.flatten.shuffle[0..10].join
  players = []
  
  rows.each do |row|
    player = {}
    player[:number] = row.children[1].children[0].to_s
    n = row.children[3].children[0].children[0].to_s.split(', ')
    player[:fname] = n[1]
    player[:lname] = n[0]
    player[:team_name] = team_name
    player[:team_id] = team_id
    players << player
  end
  players
end

teams = []
root = Nokogiri::HTML(open("http://www.quidditch-reference.com/tournament/160/"))
rows = root.xpath("//table//tr[position() > 1]")

rows.each do |row|
  team = {}
  # these specifically address the structure on quidditch-reference
  team[:name] = row.children[1].children[0].children[0].to_s
  team[:code] = row.children[1].children[2].children[1].attributes['href'].value
  team[:players] = pull_players(team[:code])
  teams << team
end

File.open('mwrc_rosters.json','wb+') do |f|
  # formats with 4space indents instead of 2
  f.write(JSON.pretty_generate(teams,{indent: "    "}))
end