greenmonster 0.1.0 → 0.2.0

data/Readme.markdown

@@ -13,7 +13,7 @@ If you don't want to specify a download location every time you run the spider,
 	# Set games folder location
 	>> Greenmonster.set_games_folder('/Users/geoff/games/')
 
-The spider utility has three public class methods: Spider.pull_game, Spider.pull_day and Spider.pull_days. 
+The spider utility has three public class methods: Spider.pull_game, Spider.pull_day, and Spider.pull_days. 
 
 Spider.pull_game takes a game_id (the folder name of the game on the Gameday server) and a hash of options as arguments. If for some reason the game does not fall in the expected folder for the game's date or sport code, you can add those options to the arguments hash. Other options include :games_folder and :print_games (if false, game IDs are not printed to screen).
 
@@ -28,6 +28,8 @@ Spider.pull_day takes an hash of options as an argument. Greenmonster will creat
     # Pulls all rookie league games for today
     >> Greenmonster::Spider.pull_day({:sport_code => 'rok', :date => Date.today, :games_folder => './home/geoff/games'})
 
+	# Pulls all games in all sport codes for today
+	>> Greenmonster::Spider.pull_day({:all_sport_codes => true, :date => Date.today, :games_folder => './home/geoff/games'})
 
 Spider.pull_days takes a range of dates to process as an argument, plus a hash of arguments to pass to Spider.pull.
 

data/lib/greenmonster/spider.rb

@@ -55,7 +55,7 @@ class Greenmonster::Spider
         copy_gameday_xml(file,paths)
       end
     rescue StandardError => bang
-      puts "Unable to download some data for #{e.attribute('href').value}"  
+      puts "Unable to download some data for #{game_id}"  
     end
     
     return game_id
@@ -80,17 +80,27 @@ class Greenmonster::Spider
       :date => Date.today,
       :sport_code => 'mlb',
     }.merge(args)
-      
+    
+    # If we want all sport codes, set up the array.
+    if args[:all_sport_codes]
+      args[:sport_codes] = %w(aaa aax afa afx asx bbc fps hsb ind int jml nae naf nas nat naw oly rok win)
+    else
+      args[:sport_codes] = [args[:sport_code] || 'mlb']
+    end
+    
     # Iterate through every hyperlink on the page.
     # These links represent the individual game folders
     # for each date. Reject any links that aren't to game
     # folders or that are to what look like backup game
     # folders.
-    (Nokogiri::XML(self.get(gameday_league_and_date_url(args)))/"a").reject{|l| l.attribute('href').value[0,4] != "gid_" or l.attribute('href').value[-5,4] == "_bak"}.each do |e|      
-      self.pull_game(e.attribute('href').value.gsub('/',''),args) 
+    args[:sport_codes].each do |sport_code|
+      args[:sport_code] = sport_code
+      (Nokogiri::XML(self.get(gameday_league_and_date_url(args)))/"a").reject{|l| l.attribute('href').value[0,4] != "gid_" or l.attribute('href').value[-5,4] == "_bak"}.each do |e|      
+        self.pull_game(e.attribute('href').value.gsub('/',''),args) 
+      end
     end
-     
-    return gameday_league_and_date_url(args)
+
+    return args[:sport_code]
   end
 
   ##
@@ -139,8 +149,11 @@ class Greenmonster::Spider
     #   paths: (Hash)
   
     def self.copy_gameday_xml (file_name,paths)
-      open(paths[:localGameFolder] + "#{file_name =~ /inning/ ? 'inning/' : ''}" + file_name, 'w') do |file|
-        file.write(self.get(paths[:mlbGameFolder] + "#{file_name =~ /inning/ ? 'inning/' : ''}" + file_name).body)
+      download = self.get(paths[:mlbGameFolder] + "#{file_name =~ /inning/ ? 'inning/' : ''}" + file_name).body
+      unless download.include?('404 Not Found')
+        open(paths[:localGameFolder] + "#{file_name =~ /inning/ ? 'inning/' : ''}" + file_name, 'w') do |file|
+          file.write(download) 
+        end
       end
     end
   

data/test/test_greenmonster_spider.rb

@@ -60,11 +60,17 @@ class GreenmonsterSpiderTest < MiniTest::Unit::TestCase
   end
   
   def test_pull_games_prior_to_2008
-    Greenmonster::Spider.pull_day({:print_games => false, :games_folder => @local_test_data_location, :date => Date.new(2007,4,15)})
+    Greenmonster::Spider.pull_game('gid_2007_04_15_detmlb_tormlb_1', {:games_folder => @local_test_data_location, :print_games => false})
     
     assert_equal 12, Dir.entries(@local_test_data_location + '/mlb/year_2007/month_04/day_15/gid_2007_04_15_detmlb_tormlb_1/inning/').count
   end
   
+  def test_pull_all_sport_codes_for_day
+    Greenmonster::Spider.pull_day({:all_sport_codes => true, :print_games => false, :games_folder => @local_test_data_location, :date => Date.new(2011,7,1)})
+  
+    assert_equal 9, Dir.entries(@local_test_data_location).count
+  end
+  
   def test_pull_days
     Greenmonster::Spider.pull_days((Date.new(2011,8,4)..Date.new(2011,8,5)), {:print_games => false, :games_folder => @local_test_data_location})
     
@@ -80,7 +86,19 @@ class GreenmonsterSpiderTest < MiniTest::Unit::TestCase
   end
   
   def test_pull_single_game_by_game_id
-    Greenmonster::Spider.pull_game('gid_2011_07_04_tormlb_bos_mlb_1', {:games_folder => @local_test_data_location, :print_games => false})
+    Greenmonster::Spider.pull_game('gid_2011_07_04_tormlb_bosmlb_1', {:games_folder => @local_test_data_location, :print_games => false})
+  end
+  
+  def test_no_exception_raised_if_game_data_not_available
+    assert_output(nil,'') do
+      Greenmonster::Spider.pull_game('gid_2011_07_04_zzzmlb_yyymlb_1', {:games_folder => @local_test_data_location, :print_games => false})
+    end
+  end
+  
+  def test_local_file_not_created_if_remote_file_does_not_exist
+    Greenmonster::Spider.pull_game('gid_2011_07_01_xxxmlb_yyymlb_1', {:games_folder => @local_test_data_location, :print_games => false})
+    
+    assert_equal 3, Dir.entries(@local_test_data_location + '/mlb/year_2011/month_07/day_01/gid_2011_07_01_xxxmlb_yyymlb_1').count
   end
   
   def teardown

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: greenmonster
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.2.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-03-13 00:00:00.000000000 Z
+date: 2012-03-14 00:00:00.000000000 Z
 dependencies: []
 description: A utility for working with MLB Gameday XML data.
 email: geoff.harcourt@gmail.com