require 'bio'
require 'fileutils'

require 'neurohmmer/arg_validators'
require 'neurohmmer/hmmer'
require 'neurohmmer/output'

# Top level module / namespace.
module Neurohmmer
  class <<self
    attr_accessor :opt
    attr_accessor :conf

    def init(opt)
      @opt = ArgumentsValidators.run(opt)
      @conf = {
        hmm_dir: File.expand_path('../../data/hmm', __FILE__),
        raw_data: File.expand_path('../../data/raw_data', __FILE__),
        raw_alignments: File.expand_path('../../data/raw_data/alignments',
                                         __FILE__),
        hmm_output: File.join(@opt[:temp_dir], 'input.hmm_search.out'),
        html_output: "#{@opt[:input_file]}.neurohmmer.html",
        fasta_output: "#{@opt[:input_file]}.neurohmmer.fa"
      }
      init_input
    end

    def run
      Hmmer.search
      hmm_analysis = Hmmer.analyse_output
      Output.to_fasta(hmm_analysis)
      Output.to_html(hmm_analysis)
      remove_temp_dir
    end

    def extract_sequence(id)
      id = id.gsub(/\s+/, '')
      idx = @input_index[id]
      seq = IO.binread(@opt[:input_file], idx[1] - idx[0], idx[0])
      seq.scan(/>([^\n]*)\n([A-Za-z\n\*]*)/)[0]
    end

    private

    def init_input
      FileUtils.mkdir_p(@opt[:temp_dir])
      @opt[:input_file] = translate_input if @opt[:type] == :genetic
      @input_index = index_input_file
    end

    # Translates the input data in all 6 frames
    def translate_input(input = @opt[:input_file])
      translated_file = File.join(@opt[:temp_dir], 'input.translated.fa')
      File.open(translated_file, 'w') do |file|
        Bio::FlatFile.open(Bio::FastaFormat, input).each_entry do |entry|
          (1..6).each do |f|
            file.puts ">#{entry.definition}-frame:#{f}"
            file.puts entry.naseq.translate(f)
          end
        end
      end
      translated_file
    end

    # Indexes the input file - returns a hash in the following format:
    # {seq id: [start byte in file, end byte in file] }
    def index_input_file
      c = IO.binread(@opt[:input_file])
      keys   = c.scan(/>(.*)\n/).flatten
      values = c.enum_for(:scan, /(>[^>]+)/).map { Regexp.last_match.begin(0) }
      index(c, keys, values)
    end

    # A method run from index_input_file that creates a simple hash with the
    # {seq id: [start byte in file, end byte in file] }
    def index(content, keys, values)
      fasta_index = {}
      keys.each_with_index do |k, i|
        id = k[0..115].gsub(/\s+/, '')
        endf  = (i == values.length - 1) ? content.length - 1 : values[i + 1]
        fasta_index[id] = [values[i], endf]
      end
      fasta_index
    end

    def remove_temp_dir
      return unless File.directory?(@opt[:temp_dir])
      FileUtils.rm_rf(@opt[:temp_dir])
    end
  end
end