As an avid baseball fan, I’ve always been interested in the statistics that surround baseball. More so than in any other sport, baseball is a game ruled by statistics. In this post, I will describe a program that I wrote in the Ruby language to generate box scores for any Major Leage Baseball(MLB) game by using the live XML data provided by MLB. Ruby makes it easy to do using plain Ruby along with just a few libraries.

Live MLB Data
The amount of raw data that MLB makes available is extensive. It includes all of the traditional boxscore statistics plus deep detail on every pitch thrown including its velocity, movement, and location. If you have ever used the Gameday baseball game watcher application than you’ve indirectly been a consumer of this data. All of the data that powers the Gameday application is made available on an MLB server as raw XML files. Also, this is not just historical data. The data is real-time and updated with every pitch thrown during live games. You also have historical data available to you. The data goes back to the 1926 season.

Building a boxscore template with ERb
For those not familiar with ERb, it is a templating engine that allows you to embed Ruby code inside of a document. The Ruby code can then be evaluated at run time and merged with the rest of the document to form a new document. The Rails framework uses ERb to generate HTML content from templates. For the boxscore generator, I used an ERb template to format the layout of the boxscore. This allowed me to keep the presentation seperate from the logic layers of the application. Most Ruby developers are familiar with the ERb templating engine from its usage in Ruby on Rails applications. However, not all developers are aware that ERb can be used very easily outside of a Rails application. ERb is in no way tightly coupled to the Rails framework. ERb is normally included with the standard Ruby distribution. You can use ERb templates by simply requiring ‘erb’ in your Ruby application.

Here is an example of ERb taken from my boxscore template. This section is used to display the linescore of a game:

<div id="linescore">
  <label><%= cities[0] %></label>
  <% (0..8).each do |num| %>
    <%= innings[num][0] %>
  <% end %>
  &nbsp;<b><%= tots[0][0] + '  ' + tots[0][1] + '  ' + tots[0][2] %></b>
  <br/>
  <label><%= cities[1] %></label>
  <% (0..8).each do |num| %>
    <%= innings[num][1] %>
  <% end %>
  &nbsp;<b><%= tots[1][0] + '  ' + tots[1][1] + '  ' + tots[1][2] %></b>
</div>

At the bottom of this post you can find links to download the complete source code for this project including the full boxscore ERb template.

Steps to generate a boxscore
Now that you know about the MLB data source, and ERb, I’ll describe the steps that are used to generate a boxscore in a Ruby application.

• Find the correct Gameday ID for the game you are interested in.
• Retreive the correct boxscore.xml file from gd2.mlb.com
• Parse the boxscore XML to pull out relevant data.
• Merge the data into the boxscore template and save the result.

The sections below describe each of these steps in more detail.

Find the correct Gameday ID for the game you are interested in

Each MLB game can be identified on the gameday server with a gameday ID. The gameday ID is a string that contains the date of the game concatenated with team information. A gameday ID has the following format:

2009_06_09_chnmlb_houmlb_1

Finding the gameday ID for a game you are interested in is done by looking at the subdirectories contained in a directory that corresponds to the game date you are looking for. That directory will contain a subdirectory for each game played on that date. The game ID can be parsed from the name of that game’s subdirectory. For example, if you were interested in the Detroit Tigers game played on June 6, 2009, you would scan through the subdirectories contained in this directotry:

http://gd2.mlb.com/components/game/mlb/year_2009/month_06/day_21/

Within that directory, you would find a subdirectory named:

gid_2009_06_21_milmlb_detmlb_1/

and from that you can find the game id by stripping off the leading ‘gid_’ from the directory name.

Retreive the correct boxscore.xml file from gd2.mlb.com
After you have identified the correct gameday ID, you have to make an HTTP call to retrieve the boxscore.xml file from the gd2.mlb.com server. This is a fairly simple task in Ruby done using the Net::HTTP library.

I have defined a constant to represent the root path for the gameday server.

GD2_MLB_BASE = "http://gd2.mlb.com/components/game"

Below is the ruby code used to retrieve the boxscore.xml file.

def get_boxscore(year, month, day, gid)
  url = "#{GD2_MLB_BASE}/mlb/year_" + year +
             "/month_" + month + "/day_" + day +
             "/gid_"+gid+"/boxscore.xml"
  xml_data = Net::HTTP.get_response(URI.parse(url)).body
  return xml_data
end

Parse the boxscore XML to pull out relevant data
Now that you have the boxscore.xml file, the next thing that you’ll want to do is parse this file to pull out data that you’ll want to use in your formatted boxscore. There are several XML parsing libraries available for ruby. I chose to use the REXML library to do this parsing. REXML is easy to work with and made the task relatively simple to code. In the sample shown below, I am parsing the boxscore XML file to find all of the batter statistics and pushing those into an array of batters. I will then use that array of batters from my boxscore ERb template to display the batters section of the boxscore.

# Returns an array of hashes where each hash holds data
# for a batter whom appeared in the game.  Specify either
# home or away team batters.
def get_batters(home_or_away)
  doc = REXML::Document.new(@xml_data)
  batters = []
  doc.elements.each(
    "boxscore/batting[@team_flag='#{home_or_away}']/batter")
    { |element|
      batter = {}
      batter['name'] = element.attributes['name']
      batter['pos'] = element.attributes['pos']
      batter['ab'] = element.attributes['ab']
      batter['r'] = element.attributes['r']
      batter['bb'] = element.attributes['bb']
      batter['sf'] = element.attributes['sf']
      batter['h'] = element.attributes['h']
      batter['e'] = element.attributes['e']
      batter['d'] = element.attributes['d']
      batter['t'] = element.attributes['t']
      batter['hbp'] = element.attributes['hbp']
      batter['so'] = element.attributes['so']
      batter['hr'] = element.attributes['hr']
      batter['rbi'] = element.attributes['rbi']
      batter['sb'] = element.attributes['sb']
      batter['avg'] = element.attributes['avg']
      batters.push batter
  }
  return batters
end

Merge the data into the boxscore template and save the result
After you have parsed all of the relevant data out of the boxscore XML, you are ready to use that data to build the final boxscore presentation. The code below is a method in the BoxScore class that I created. This method converts the boxscore into a formatted HTML representation. The first few lines call methods which parse the boxscore XML and setup the corect data variables. After that I create a new ERb instance and load the boxscore.html.erb template into it. This is the template that I created to display the boxscore. The next line binds the template to the variables previously defined in this method and returns the result of the template bound with the variables. This produces the resultant boxscore HTML file.

  # Converts the boxscore into a formatted HTML representation.
  def to_html
    cities = get_cities
    innings = get_innings
    tots =  get_linescore_totals
    home_pitchers = get_pitchers('home')
    away_pitchers = get_pitchers('away')
    home_batters = get_batters('home')
    away_batters = get_batters('away')
    home_batters_text = get_batting_text('home')
    away_batters_text = get_batting_text('away')
    game_info = get_game_info

    gameday_info = parse_gameday_id('gid_' + gid)
    template = ERB.new File.new("boxscore.html.erb").read, nil, "%"
    return template.result(binding)
  end

View the results and download the code
Here are links to download the complete sourcecode for my gameday API in ruby. I’ve also provided a link to a boxscore that has been generated using this code.
Full sourcecode
Generated boxscore

MLB copyright notice
If you decide to do something with my api or the MLB data, be sure that you understand the limitations of using the data. MLB provides the data for experimental usage only. You are not permitted to use it for a commercial application. Below is a line taken from their copyright notice covering all of the data available from their gameday server:
“Only individual, non-commercial, non-bulk use of the Materials is permitted and any other use of the Materials is prohibited without prior written authorization from MLBAM.”

Read the original blog entry...