hoopscrape 1.0.4

data/lib/hoopscrape/ArrayConversions.rb

@@ -0,0 +1,56 @@
+# Array Extensions for Type Conversion
+class Array
+  # Create Hash Array
+  # @param keys [Array] Symbols to be used as field names
+  # @return [[Hash]] Array<Hash>
+  # @example
+  #   teams   = es.teamList # Array of Team data
+  #   hash_a  = teams.to_hashes
+  #   => [{:t_abbr=>"BOS", :t_name=>"Boston Celtics", :division=>"Atlantic", :conference=>"Eastern"} ... ]
+  def to_hashes(keys = [])
+    return [] if empty?
+    two_d = first.is_a? Array # Check for 2D array
+    keys = checkKeys(keys, two_d ? first.size : size) # Determine keys
+    return [Hash[keys.map.with_index { |key, idx| [key, self[idx]] }]] unless two_d # 1D Array
+    map { |ary| Hash[keys.map.with_index { |key, idx| [key, ary[idx]] }] }          # 2D Array
+  end
+
+  # Create Struct Array
+  # @param keys [Array] Symbols to be used as field names
+  # @return [[Struct]] Array<Struct>
+  # @example
+  #   teams   = es.teamList # Array of Team data
+  #   structs = teams.to_structs
+  #   => [#<struct t_abbr="BOS", t_name="Boston Celtics", division="Atlantic", conference="Eastern"> ... ]
+  def to_structs(keys = [])
+    return [] if empty?
+    keys = checkKeys(keys, first.size)
+    to_hashes(keys).map { |hash| Struct.new(*hash.keys).new(*hash.values) }
+  end
+
+  # Determine default field names
+  # @param keys [[Symbol]] Field Names
+  # @param k_id [Int] Key Identifier
+  # @return [[Symbol]]
+  def checkKeys(keys, k_id)
+    return keys unless keys.empty?
+    [S_BOX_P, S_BOX_T, S_GAME_F, S_GAME_P, S_ROSTER, S_TEAM].each do |default|
+      return default if default.size.eql?(k_id)
+    end
+  end
+
+  # Replace old symbol with new symbol in Array
+  # @param old_sym [Symbol] Symbol to remove
+  # @param new_sym [Symbol] Symbol to add
+  def change_sym!(old_sym, new_sym)
+    map! { |x| x.eql?(old_sym) ? new_sym : x }
+  end
+
+  # Get an NbaBoxscore
+  # @param [Symbol] Format
+  # @return [NbaBoxscore] Boxscore
+  def boxscore(f_mat = nil)
+    return nil unless size == S_GAME_P.size
+    HoopScrape.boxscore(self[8], f_mat)
+  end
+end

data/lib/hoopscrape/Hash.rb

@@ -0,0 +1,9 @@
+# Extend Hash
+class Hash
+  # Get an NbaBoxscore
+  # @param [Symbol] Format
+  # @return [NbaBoxscore] Boxscore
+  def boxscore(f_mat = nil)
+    HoopScrape.boxscore(self[:boxscore_id], (f_mat || :to_hashes))
+  end
+end

data/lib/hoopscrape/Navigator.rb

@@ -0,0 +1,60 @@
+# Array Navigator
+class Navigator
+  # Array of data to Navigate
+  attr_reader :list
+
+  # Store array and initialize cursor
+  def initialize(list)
+    @list    = list
+    @bounds  = [-1, list.size]
+    @cursor  = -1
+  end
+
+  # Return the player located at the current navigation cursor
+  # @return [[Object]] Array Element
+  def curr
+    self[@cursor] if @cursor >= 0
+  end
+
+  # Increments the navigation cursor and return the item at that location
+  # @return (see #curr)
+  def next
+    self[(@cursor += 1)]
+  end
+
+  # Decrements the navigation cursor and return the item at that location
+  # @return (see #curr)
+  def prev
+    self[(@cursor -= 1)]
+  end
+
+  # Updates the navigation cursor if out of bounds. Returns the item at the given location.
+  # Returns the underlying array if no index is given.
+  # @return [Element] Array Element or Array
+  def [](idx = nil)
+    return @list if idx.nil?
+    @cursor = @bounds[0] if @cursor < @bounds[0]
+    @cursor = @bounds[1] if @cursor > @bounds[1]
+    @list[idx] if [inbounds?(idx), !@list.nil?].all?
+  end
+
+  # Checks if the requested index is within the array bounderies
+  def inbounds?(idx)
+    (@bounds[0]..@bounds[1]).cover?(idx)
+  end
+
+  # Updates the cursor and returns the first element of the array
+  def first
+    self[@cursor = 0]
+  end
+
+  # Updates the cursor and returns the last element of the array
+  def last
+    self[@cursor = @list.size - 1]
+  end
+
+  # Returns the size of the underlying array
+  def size
+    @list.size
+  end
+end

data/lib/hoopscrape/NbaBoxScore.rb

@@ -0,0 +1,174 @@
+require 'date'
+require_relative './PrintUtils.rb'
+
+# Access NBA boxscore data
+class NbaBoxScore
+  include NbaUrls
+
+  # @return [String] Game Date
+  attr_reader :gameDate
+
+  # @return [String] Away Team Name
+  attr_reader :awayName
+
+  # @return [Navigator] Away Team Stats Array
+  # @note (see SymbolDefaults::BOX_P)
+  # @see BOX_P
+  attr_reader :awayPlayers
+
+  # @return [[String]] Away Team Combined Stats
+  # @note (see SymbolDefaults::BOX_T)
+  # @see BOX_T
+  attr_reader :awayTotals
+
+  # @return [String] Home Team Name
+  attr_reader :homeName
+
+  # @return [Navigator] Home Team Stats Array
+  # @note (see #awayPlayers)
+  # @see BOX_P
+  attr_reader :homePlayers
+
+  # @return [[String]] Home Team Combined Stats
+  # @note (see #awayTotals)
+  # @see BOX_T
+  attr_reader :homeTotals
+
+  # Boxscore ID
+  attr_reader :id
+  attr_reader :awayScore
+  attr_reader :homeScore
+
+  # Scrape Box Score Data
+  # @param game_id [Integer] Boxscore ID
+  # @example
+  # 	bs = NbaBoxScore.new(400828035)
+  def initialize(args)
+    doc = getNokoDoc(args[:game_id], args[:file])
+    return if doc.nil?
+    @id = args[:game_id].to_s
+    @gameDate = readGameDate(doc)
+    @awayName, @homeName = readTeamNames(doc)
+    return unless @gameDate.index('00:00:00') # Only past games have stats
+    @awayPlayers, @awayTotals, @awayScore = readTeamStats(doc, 'away', args[:format])
+    @homePlayers, @homeTotals, @homeScore = readTeamStats(doc, 'home', args[:format])
+  end
+
+  private
+
+  def getNokoDoc(game_id, file)
+    return Nokogiri::HTML(open(file)) if game_id.nil? # Parse File
+    Nokogiri::HTML(open(boxScoreUrl + game_id.to_s))  # Parse URL
+  end
+
+  # Reads the game date from a Nokogiri::Doc
+  # @param d [Nokogiri::HTML::Document]
+  # @return [String] Game date
+  # @example
+  # 	bs.readGameDate(doc) #=> "Mon, Nov 23"
+  # @note
+  # 	Times will be Local to the system Timezone
+  #
+  def readGameDate(d)
+    date = d.title.split('-')[2].delete(',')
+    time = d.xpath('//span[contains(@class,"game-time")]')[0].text.strip rescue ''
+    time = '00:00:00' if time == 'Final' || time.empty?
+    DateTime.parse(date + ' ' + time).strftime('%Y-%m-%d %H:%M:%S')
+  end
+
+  # Reads the team names from a Nokogiri::Doc
+  # @param d [Nokogiri::HTML::Document]
+  # @return [String, String] Team 1, Team 2
+  # @example
+  # 	bs.readGameDate(doc)
+  #
+  def readTeamNames(d)
+    names = d.xpath('//span[@class="long-name" or @class="short-name"]')
+    away = names[0].text + ' ' + names[1].text
+    home = names[2].text + ' ' + names[3].text
+    [away, home]
+  end
+
+  # Extract Player Stats
+  # @param rows [[Nokogiri::XML::NodeSet]] Cumulative Team Stats
+  # @param tid [String] Team ID
+  # @return [[String]]  Processed Team Stats
+  def processPlayerRows(rows, tid, new_form)
+    result = [] # Extracted Player Data
+    rows.each_with_index do |row, index|
+      curr_row = [tid]	# Team ID
+
+      row.children.each do |cell|	# Process Columns
+        c_val = cell.text.strip
+        case cell.attribute('class').text
+        when 'name'
+          curr_row << cell.children[0].attribute('href').text[%r{id/(\d+)}, 1] # Player ID
+          curr_row << cell.children[0].children[0].text.strip # Player Short Name (i.e. D. Wade)
+          curr_row << cell.children[1].text.strip # Position
+          # binding.pry
+        when 'fg', '3pt', 'ft'
+          # Made-Attempts
+          curr_row += c_val.split('-')
+        else
+          curr_row << c_val
+        end
+      end
+
+      curr_row << (index < 5).to_s  # Check if Starter
+      result << curr_row            # Save processed data
+    end
+    return result.send(new_form, S_BOX_P) unless new_form.nil?
+    result
+  end
+
+  # Extract Team Stats
+  # @param row [[Nokogiri::XML::NodeSet]] Cumulative Team Stats
+  # @param tid [String]  Team ID
+  # @return [[String]]   Processed Team Stats
+  def processTeamRow(row, tid, new_form)
+    result = []
+    row.children.each do |cell|
+      c_val = cell.text.strip
+      case cell.attribute('class').text
+      when 'name'
+        result << tid
+      when 'fg', '3pt', 'ft'
+        # Made-Attempts
+        result += c_val.split('-')
+      else
+        next if c_val.empty?
+        result << c_val
+      end
+    end
+    return [result.send(new_form, S_BOX_T).first, result.last] unless new_form.nil?
+    [result, result.last]
+  end
+
+  # Reads the team stats from a Nokogiri::Doc
+  # @param d [Nokogiri::HTML::Document]
+  # @param id [String] Team selector -> home/away
+  # @return [String] Game date
+  # @example
+  # 	bs.readTeamStats(doc,'away')
+  #
+  def readTeamStats(d, id, new_form)
+    # Extract player tables
+    p_tables = d.xpath('//div[@class="sub-module"]/*/table/tbody')
+
+    if id == 'away'
+      p_tab = p_tables[0, 2]
+      tid = getTid(@awayName)
+    else
+      p_tab = p_tables[2, 4]
+      tid = getTid(@homeName)
+    end
+
+    player_rows = p_tab.xpath('tr[not(@class)]') # Ignore TEAM rows
+    team_row    = p_tab.xpath('tr[@class="highlight"]')[0] # Ignore Percentage row
+
+    player_stats = processPlayerRows(player_rows, tid, new_form)
+    team_totals, team_score = processTeamRow(team_row, tid, new_form)
+
+    [Navigator.new(player_stats), team_totals, team_score]
+  end
+end

data/lib/hoopscrape/NbaPlayer.rb

@@ -0,0 +1,82 @@
+# Read basic bio info from ESPN Player page
+class NbaPlayer
+  include NbaUrls
+
+  # @return [String] Name
+  attr_accessor :name
+  # @return [String] Position
+  attr_accessor :position
+  # @return [Integer] Age
+  attr_accessor :age
+  # @return [String] College
+  attr_accessor :college
+  # @return [Integer] Weight
+  attr_accessor :weight
+  # @return [Integer] Height (ft)
+  attr_accessor :h_ft
+  # @return [Integer] Height (in)
+  attr_accessor :h_in
+
+  # Read Player Data
+  def initialize(espn_player_id, file = '')
+    espn_player_id = espn_player_id.to_s
+    if !espn_player_id.empty?
+      url = playerUrl + espn_player_id
+      doc = Nokogiri::HTML(open(url))
+    else
+      doc = Nokogiri::HTML(open(file)) rescue nil
+    end
+    return if doc.nil?
+
+    readInfo(doc)
+  end
+
+  # alias for h_ft
+  def height_ft
+    @h_ft
+  end
+
+  # alias for h_in
+  def height_in
+    @h_in
+  end
+
+  private
+
+  # Extract basic bio info info class attributes
+  def readInfo(d)
+    @name     = d.xpath("//div[@class='mod-content']/*/h1 | //div[@class='mod-content']/h1")[0].text.strip
+    @position = d.xpath("//ul[@class='general-info']/li")[0].text.gsub(/#\d*\s*/, '')
+    @college  = d.xpath('//span[text() = "College"]/parent::li').text.gsub('College', '')
+
+    height, weight = gatherHeightWeight(d)
+
+    @weight = processWeight(weight)
+    processHeight(height)
+    processAge(d)
+  end
+
+  def processAge(d)
+    /:\s(?<age_num>\d\d)/ =~ d.xpath('//span[text() = "Born"]/parent::li').text
+    @age = age_num.to_i.to_s
+  end
+
+  def gatherHeightWeight(d)
+    h_w = d.xpath("//ul[@class='general-info']/li")[1]
+    h_w.text.split(',') unless h_w.nil?
+  end
+
+  def processWeight(weight)
+    return 0 if weight.nil? || weight.empty?
+    weight.strip.split(' ')[0]
+  end
+
+  def processHeight(height)
+    if !height.nil? && !height.empty?
+      @h_ft, @h_in = height.strip.split('\'')
+      @h_in = @h_in.delete('"').strip
+    else
+      @h_ft = @h_in = 0
+    end
+  end
+end

data/lib/hoopscrape/NbaRoster.rb

@@ -0,0 +1,74 @@
+require_relative './Navigator'
+# Access NBA roster data
+class NbaRoster
+  include NbaUrls
+  include PrintUtils
+
+  # @return [String] Coach Name
+  attr_reader :coach
+
+  # Returns Team Roster
+  # @return [[[String]]] Player List Table ({ROSTER Row Description})
+  # @see ROSTER
+  attr_reader :players
+
+  # Scrape Roster Data
+  # @param team_id [String] Team ID
+  # @example
+  # 	r = NbaRoster.new("UTA")
+  # 	r = NbaRoster.new('', 'test/data/rosterData.html')
+  def initialize(args = {})
+    if args[:team_id]
+      url = formatTeamUrl(args[:team_id], teamRosterUrl) # Generate URL
+      doc = Nokogiri::HTML(open(url)) # Get DOM
+    elsif args[:file]
+      doc = Nokogiri::HTML(open(args[:file]))
+    end
+    return if doc.nil?
+
+    team_id ||= getTid(doc.title.split(/\d{4}/).first.strip)
+    list      = doc.xpath('//div/div/table/tr')
+    p_list    = list[2, list.length - 3] # Get Player Nodes
+
+    @coach   = list[-1].children.first.text.split(':').last.strip # Read Coach Name
+    @players = Navigator.new processPlayerTable(p_list, team_id, args[:format])
+  end
+
+  private
+
+  # Collect Roster Data
+  # @param table [[Nokogiri::XML::NodeSet]] Roster Table
+  # @param team_id [String] Team ID
+  # @return [[[String]]] Processed Roster Data
+  def processPlayerTable(table, team_id, new_form)
+    result = []
+    table.each do |row|
+      tmp = [team_id]	# Start row with Team ID
+      row.children.each_with_index do |cell, cnt|
+        processCell(cell, tmp, cnt)
+      end
+      result << tmp
+    end
+    return result.send(new_form, S_ROSTER) unless new_form.nil?
+    result
+  end
+
+  # Extract and Normalize Player Data
+  def processCell(cell, tmp, cnt)
+    txt = cell.text.chomp.strip
+    case cnt
+    when 0, 2, 3, 5	# 0 Player No, 2 Position, 3 Age, 5 Weight
+      tmp << txt
+    when 1	# Player Name
+      tmp << txt.tr("'", "\'")
+      tmp << cell.children.first.attribute('href').text[%r{id/(\d+)}, 1] # Player ID
+    when 4													  # Player Height
+      tmp.concat(txt.split('-'))
+    when 6 														# College
+      tmp << txt.tr("'", "\'").strip
+    when 7 														# Salary
+      # Remove extraneous symbols & default to 0
+      tmp << txt.delete('$').delete(',').strip.to_i.to_s
+    end
+  end
+end