Chris@909: module CodeRay
Chris@909: module Encoders
Chris@909:   
Chris@909:   # Makes a statistic for the given tokens.
Chris@909:   # 
Chris@909:   # Alias: +stats+
Chris@909:   class Statistic < Encoder
Chris@909:     
Chris@909:     register_for :statistic
Chris@909:     
Chris@909:     attr_reader :type_stats, :real_token_count  # :nodoc:
Chris@909:     
Chris@909:     TypeStats = Struct.new :count, :size  # :nodoc:
Chris@909:     
Chris@909:   protected
Chris@909:     
Chris@909:     def setup options
Chris@909:       super
Chris@909:       
Chris@909:       @type_stats = Hash.new { |h, k| h[k] = TypeStats.new 0, 0 }
Chris@909:       @real_token_count = 0
Chris@909:     end
Chris@909:     
Chris@909:     STATS = <<-STATS  # :nodoc:
Chris@909: 
Chris@909: Code Statistics
Chris@909: 
Chris@909: Tokens            %8d
Chris@909:   Non-Whitespace  %8d
Chris@909: Bytes Total       %8d
Chris@909: 
Chris@909: Token Types (%d):
Chris@909:   type                     count     ratio    size (average)
Chris@909: -------------------------------------------------------------
Chris@909: %s
Chris@909:     STATS
Chris@909:     
Chris@909:     TOKEN_TYPES_ROW = <<-TKR  # :nodoc:
Chris@909:   %-20s  %8d  %6.2f %%   %5.1f
Chris@909:     TKR
Chris@909:     
Chris@909:     def finish options
Chris@909:       all = @type_stats['TOTAL']
Chris@909:       all_count, all_size = all.count, all.size
Chris@909:       @type_stats.each do |type, stat|
Chris@909:         stat.size /= stat.count.to_f
Chris@909:       end
Chris@909:       types_stats = @type_stats.sort_by { |k, v| [-v.count, k.to_s] }.map do |k, v|
Chris@909:         TOKEN_TYPES_ROW % [k, v.count, 100.0 * v.count / all_count, v.size]
Chris@909:       end.join
Chris@909:       @out << STATS % [
Chris@909:         all_count, @real_token_count, all_size,
Chris@909:         @type_stats.delete_if { |k, v| k.is_a? String }.size,
Chris@909:         types_stats
Chris@909:       ]
Chris@909:       
Chris@909:       super
Chris@909:     end
Chris@909:     
Chris@909:   public
Chris@909:     
Chris@909:     def text_token text, kind
Chris@909:       @real_token_count += 1 unless kind == :space
Chris@909:       @type_stats[kind].count += 1
Chris@909:       @type_stats[kind].size += text.size
Chris@909:       @type_stats['TOTAL'].size += text.size
Chris@909:       @type_stats['TOTAL'].count += 1
Chris@909:     end
Chris@909:     
Chris@909:     # TODO Hierarchy handling
Chris@909:     def begin_group kind
Chris@909:       block_token ':begin_group', kind
Chris@909:     end
Chris@909:     
Chris@909:     def end_group kind
Chris@909:       block_token ':end_group', kind
Chris@909:     end
Chris@909:     
Chris@909:     def begin_line kind
Chris@909:       block_token ':begin_line', kind
Chris@909:     end
Chris@909:     
Chris@909:     def end_line kind
Chris@909:       block_token ':end_line', kind
Chris@909:     end
Chris@909:     
Chris@909:     def block_token action, kind
Chris@909:       @type_stats['TOTAL'].count += 1
Chris@909:       @type_stats[action].count += 1
Chris@909:       @type_stats[kind].count += 1
Chris@909:     end
Chris@909:     
Chris@909:   end
Chris@909:   
Chris@909: end
Chris@909: end