#!/usr/bin/env ruby
Signal.trap("PIPE", "EXIT")

require "pp"

require "big_simon"

Rya::AbortIf.abort_unless ARGV.count >= 1, "usage: TOP=number consensus_predictions scores_scaled.*.txt > consensus_predictions.txt"

by_program = {}
lines = []

TOP = (ENV["TOP"] || 3).to_i
AT_LEAST = (ENV["AT_LEAST"] || 2).to_i

ARGV.each do |fname|
  scores = {}
  File.open(fname, "rt").each_line.with_index do |line, idx|
    unless idx.zero?
      virus, host, score = line.chomp.split "\t"

      unless scores.has_key? virus
        scores[virus] = []
      end

      scores[virus] << [host, score.to_f]
    end
  end

  scores.sort_by { |virus, _| virus }.each do |virus, host_scores|
    # Lowest score is the best
    top_5 = host_scores.sort_by { |host, score| score }.take(TOP).map(&:first)

    line = [File.basename(fname), virus, top_5]
    lines << line

    # puts line.join "\t"
  end
  # puts
end

lines.each do |line|
  program, virus, all = line
  first = all.first

  unless by_program.has_key? virus
    by_program[virus] = {}
  end

  unless program == "scores_scaled.mean.txt"
    by_program[virus][program] = { first: first, all: all }
  end
end

# These track the number of times a host shows up in the first spot and in the top N spots for that virus for all programs.
first_table = {}
top_host_table = {}

by_program.each do |virus, program_tables|
  first_table[virus] = {}
  top_host_table[virus] = {}

  program_tables.each do |program, top_info|


    first_host = top_info[:first]
    all_top = top_info[:all]

    unless first_table[virus].has_key? first_host
      first_table[virus][first_host] = []
    end


    first_table[virus][first_host] << program

    all_top.each do |top_host|
      unless top_host_table[virus].has_key? top_host
        top_host_table[virus][top_host] = []
      end

      top_host_table[virus][top_host] << program
    end
  end
end

# first_table.each do |virus, host_counts|
#   host_counts.sort_by { |_, programs| programs.count }.reverse.each do |host, programs|
#     STDERR.puts [virus, :best, host, programs.count, programs].join "\t"
#   end
#   STDERR.puts
# end

top_host_table.each do |virus, host_counts|
  puts "\n\n>Virus-------#{virus}"
  host_counts.sort_by { |_, programs| programs.count }.reverse.each do |host, programs|
    if programs.count >= AT_LEAST
      puts [virus, :top_N, host, programs.count, programs].join "\t"
    end
  end
end