Chris@0: module CodeRay
Chris@0: module Encoders
Chris@0: 
Chris@0:   # Makes a statistic for the given tokens.
Chris@0:   class Statistic < Encoder
Chris@0: 
Chris@0:     include Streamable
Chris@0:     register_for :stats, :statistic
Chris@0: 
Chris@0:     attr_reader :type_stats, :real_token_count
Chris@0: 
Chris@0:   protected
Chris@0: 
Chris@0:     TypeStats = Struct.new :count, :size
Chris@0: 
Chris@0:     def setup options
Chris@0:       @type_stats = Hash.new { |h, k| h[k] = TypeStats.new 0, 0 }
Chris@0:       @real_token_count = 0
Chris@0:     end
Chris@0: 
Chris@0:     def generate tokens, options
Chris@0:       @tokens = tokens
Chris@0:       super
Chris@0:     end
Chris@0: 
Chris@0:     def text_token text, kind
Chris@0:       @real_token_count += 1 unless kind == :space
Chris@0:       @type_stats[kind].count += 1
Chris@0:       @type_stats[kind].size += text.size
Chris@0:       @type_stats['TOTAL'].size += text.size
Chris@0:       @type_stats['TOTAL'].count += 1
Chris@0:     end
Chris@0: 
Chris@0:     # TODO Hierarchy handling
Chris@0:     def block_token action, kind
Chris@0:       @type_stats['TOTAL'].count += 1
Chris@0:       @type_stats['open/close'].count += 1
Chris@0:     end
Chris@0: 
Chris@0:     STATS = <<-STATS
Chris@0: 
Chris@0: Code Statistics
Chris@0: 
Chris@0: Tokens            %8d
Chris@0:   Non-Whitespace  %8d
Chris@0: Bytes Total       %8d
Chris@0: 
Chris@0: Token Types (%d):
Chris@0:   type                     count     ratio    size (average)
Chris@0: -------------------------------------------------------------
Chris@0: %s
Chris@0:       STATS
Chris@0: # space                    12007   33.81 %     1.7
Chris@0:     TOKEN_TYPES_ROW = <<-TKR
Chris@0:   %-20s  %8d  %6.2f %%   %5.1f
Chris@0:       TKR
Chris@0: 
Chris@0:     def finish options
Chris@0:       all = @type_stats['TOTAL']
Chris@0:       all_count, all_size = all.count, all.size
Chris@0:       @type_stats.each do |type, stat|
Chris@0:         stat.size /= stat.count.to_f
Chris@0:       end
Chris@0:       types_stats = @type_stats.sort_by { |k, v| [-v.count, k.to_s] }.map do |k, v|
Chris@0:         TOKEN_TYPES_ROW % [k, v.count, 100.0 * v.count / all_count, v.size]
Chris@0:       end.join
Chris@0:       STATS % [
Chris@0:         all_count, @real_token_count, all_size,
Chris@0:         @type_stats.delete_if { |k, v| k.is_a? String }.size,
Chris@0:         types_stats
Chris@0:       ]
Chris@0:     end
Chris@0: 
Chris@0:   end
Chris@0: 
Chris@0: end
Chris@0: end