differ.rb

#diff parsing to side by side differences

class Differ
  attr_reader :rows
  
  LINE_REGEX    = /^((?:---|[><\\]))/
  CONTROL_REGEX = /(\d+(?:,\d+)?)([acd])(\d+(?:,\d+)?)/
  
  def initialize(left_file, right_file)
    @rows  = []
    @left_file    = left_file
    @right_file   = right_file
    @curr_control = {}
    @raw_file     = File.open(@left_file).read.split("\n")
    @raw_file.unshift("") #diffs are 1 indexed
    @index        = 0
    get_diff
    while(@diff.length > 0 && @index <= @raw_file.length)
      build
    end
  end
  
  private
  
  def get_diff
    @diff = %x(diff #{@left_file} #{@right_file} --speed-large-files --ignore-all-space --ignore-blank-lines).split("\n")
  end
  
  def build
    control_string
    advance
    diff_string
  end
  
  def control_string
    return if !@diff.first.match(CONTROL_REGEX)
    p "control"
    
    @index = !@curr_control[:from].nil? ?
              @curr_control[:from].last.to_i+1 : 0
    @diff_line = @diff.shift
    matches = @diff_line.match(CONTROL_REGEX)
    @curr_control = {:from => matches[1].split(","), 
                     :to => matches[3].split(","), 
                     :control => matches[2]}
  end
  
  def diff_string
    return if !@diff.first.match(LINE_REGEX)
    @curr_diff = [[],[]]
    while(!@diff.first.nil? && @diff.first.match(LINE_REGEX))
      p "diff"
      @diff_line = @diff.shift
      case @diff_line[0,1]
      when "<" then @curr_diff[0] << @diff_line[2, @diff_line.length - 1]
      when ">" then @curr_diff[1] << @diff_line[2, @diff_line.length - 1]
      end
    end

    end_range = @curr_diff[0].length > @curr_diff[1].length ? 
                @curr_diff[0].length : @curr_diff[1].length
    (0..end_range-1).each do |index|
      @rows << [@curr_diff[0][index] || "", @curr_diff[1][index] || "", @curr_control[:control]]
    end
  end
  
  def advance
    while(@index < @curr_control[:from].first.to_i)
      p "advance"
      insert_step
    end
  end
  
  def insert_step
    @rows << [@raw_file[@index], @raw_file[@index], ""]
    @index = @index + 1
  end
  
  class << self
    def run(directory=".")
      Differ.new(Dir["#{directory}/*.txt"][0], Dir["#{directory}/*.txt"][1])
    end
  end
end