module SpecIDXML; end

require 'spec_id_xml'
require 'strscan'

class SampleEnzyme
  include SpecIDXML

  attr_accessor :name
  # amino acids after which to cleave
  attr_accessor :cut
  # cleave at 'cut' amino acids UNLESS it is followed by 'no_cut'
  attr_accessor :no_cut
  # 'C' or 'N'
  attr_accessor :sense

  # Currently, recognize: 
  #   trypsin
  # For other enzymes, you must set :cut, :no_cut, :name, and :sense
  def initialize(name=nil)
    @sense = nil
    @cut = nil
    @no_cut = nil
    @name = name
    if @name
      # set the values if we recognize this name
      send(@name.to_sym)
    end
  end

  def trypsin
    @sense = 'C'
    @cut = 'KR'
    @no_cut = 'P'
  end

  def to_pepxml
    element_xml(:sample_enzyme, [:name]) do
      short_element_xml(:specificity, [:cut, :no_cut, :sense])
    end
  end

  # returns all peptides of missed cleavages <= 'missed_cleavages'
  # so 2 missed cleavages will return all no missed cleavage peptides
  # all 1 missed cleavages and all 2 missed cleavages.
  def digest(string, missed_cleavages=0)
    s = StringScanner.new(string)
    no_cut_regex = Regexp.new("[#{@no_cut}]")
    regex = Regexp.new("[#{@cut}]")
    peps = []
    last_pos = 0
    current_pep = ''
    loop do
      if s.eos?
        break
      end
      m = s.scan_until(regex)
      if m  ## found a cut point
        last_pos = s.pos
        # is the next amino acid a no_cut?
        if string[s.pos,1] =~ no_cut_regex 
          current_pep << m
        else
          # cut it 
          current_pep << m
          peps << current_pep
          current_pep = ''
        end
      else  ## didn't find a cut point
        current_pep << string[last_pos..-1]
        peps << current_pep 
        break
      end
    end
    ## LOOP through and grab each set of missed cleavages from num down to 0
    all_sets_of_peps = []
    (0..missed_cleavages).to_a.reverse.map do |num_mc|
      all_sets_of_peps.push( *(get_missed_cleavages(peps, num_mc)) )
    end
    all_sets_of_peps 
  end

  # takes an array of peptides and returns an array containing 'num' missed
  # cleavages
  # DOES NOT contain peptides that contain < num of missed cleavages
  # (i.e., will not return missed cleaveages of 1 or 2 if num == 3
  def get_missed_cleavages(tryptic_peps, num)
    (0...(tryptic_peps.size - num)).to_a.map do |i|
      tryptic_peps[i,num+1].join
    end
  end

  def self.tryptic(string, missed_cleavages=0)
    self.new("trypsin").digest(string, missed_cleavages)
  end

end