strip-stats/espn_pickem_strip_stats.rb

require 'Nokogiri'
require 'open-uri'
require 'tzinfo'
require 'uri'

require './StripStats.rb'

class EspnPickemStripStats < StripStats
  ###Override method get_data from StripStats class###

  @@season_start_month = 8
  @@timezone_string = 'US/Eastern'

    # Override data getter method and return data extracted from "html_doc" object
  def strip_data(html_source)

    # Load HTML into object
    begin
      if File.exists?(html_source) || html_source =~ URI::regexp 
        html_doc = Nokogiri::HTML(open(html_source))
      else
        html_doc = Nokogiri::HTML(html_source)
      end
    rescue Exception => e
      puts e.message
    end

    # Initialize data attribute
    data = []

    # List of matchup rows in the pickem table
    matchup_rows = html_doc.css('table.pickemTable tr.matchupRow')

    # Loop through rows
    matchup_rows.each do |matchup_row|
      row = {}

      # Matchup id
      row[:matchup_id] = matchup_row.attr('data-matchupid')

      # Game id
      game_href = matchup_row.css('a.matchupLink').attr('href').to_s
      begin
        row[:game_id] = game_href.match(/\?gameId=([\d]+)/)[1]
      rescue
        row[:game_id] = ''
      end

      # Date  
      date = matchup_row.css('div.pickem-date').first.inner_html.strip
      date_parts = date.split(',')
      month_and_day = ''
      if date_parts.length > 1
        month_and_day = date_parts.last.strip
      else
        month_and_day = date_parts.first.strip
      end
      month_and_day_parts = month_and_day.split(' ')
      month = month_and_day_parts.first.strip
      day = month_and_day_parts.last.strip

      # Time
      url = html_doc.xpath("//head/link[@rel='canonical']/@href").to_s
      
      start_year = url.match(/http:\/\/games.espn.com\/nfl-pigskin-pickem\/([\d]+)/)[1].to_i
      year = start_year
      month_int = StripStats.month_int_from_string(month)
      if month_int < @@season_start_month
        year += 1
      end

      time = matchup_row.css('div.pickem-time').first.inner_html.strip
      matches = time.match(/([\d]+):([\d]+)(.+)/)
      hour = matches[1].to_s
      minute = matches[2].to_s
      am_or_pm = matches[3].to_s.upcase
      if am_or_pm == 'AM'
        hour = '0' if hour == '12'
      else
        i_hour = hour.to_i
        hour = (i_hour + 12).to_s if i_hour < 12
      end
      tz = TZInfo::Timezone.get(@@timezone_string)
      utc = tz.local_to_utc(Time.utc(year.to_s, month, day, hour, minute))
      row[:datetime] = utc.strftime('%Y-%m-%d %H:%M:%S %z')

      # Media network
      row[:network] = matchup_row.css('div.tvNetwork').first.inner_html.strip

      # Load divs for each team into array
      div_teams = matchup_row.css('td.teams').first.css('div.pickem-teams')

      # Away team points
      row[:away_team_points] = div_teams.first.css('div.away').first.inner_html.strip

      # Home team points
      row[:home_team_points] = div_teams.last.css('div.home').first.inner_html.strip

      # Away team 3-letter code
      row[:away_team] = div_teams.last.css('button').first.attr('data-f').to_s.strip

      # Home team 3-letter code
      row[:home_team] = div_teams.last.css('button').first.attr('data-u').to_s.strip

      # Spread
      row[:spread] = div_teams.last.css('button').first.attr('data-s').to_s.strip
      
      # Home team confidence %
      divs = matchup_row.css('td.picked').first.css('div.wpwOutsideWrapper')
      row[:confidence] = divs.last.css('span').first.inner_html
      data<< row
    end

    return data
  end
end

# Main program exectuion
URL_TEMPLATE = 'http://games.espn.com/nfl-pigskin-pickem/%d/en/entry?entryID=%d&period=%d'
OUTPUT_FILE = 'out.ruby.csv'
YEAR_ZERO = 2012 #year when first week corresponds to period=1
WEEKS_PER_SEASON = 17
FIELDS = [
  :matchup_id,
  :game_id,
  :datetime,
  :network,
  :away_team,
  :home_team,
  :away_team_points,
  :home_team_points,
  :spread,
  :confidence,
]

File.open(OUTPUT_FILE, 'w') {} #truncate file

# 2015 and 2016
#START_YEAR = 2015
#ENTRY_IDS = [276612, 171981]

# 2016
START_YEAR = 2016
ENTRY_IDS = [171981]

ENTRY_IDS.each_with_index do |entry_id, i|
  year = i + START_YEAR
  start_period = (year - YEAR_ZERO) * WEEKS_PER_SEASON + 1
  end_period = start_period + WEEKS_PER_SEASON - 1
  puts "Stripping ESPN Pick'Em data for %d..." % [year]
  start_period.upto(end_period).each do |period|

    # Get html from URL
    html_source = URL_TEMPLATE % [year, entry_id, period]

    # Get html from file
    #html_source = "period_#{period}.html"
    #next unless File.exists?(html_source)

    espn_pickem_strip_stats = EspnPickemStripStats.new(html_source, OUTPUT_FILE, FIELDS)
    espn_pickem_strip_stats.write_data
    puts "Parsing #{html_source}"
  end
end