#!/usr/bin/env ruby

require 'ostruct'
require 'optparse'
opt = OpenStruct.new

ARGV.options {|opts|
  opts.banner << " tab-file"
  #opts.on(nil, "--report", "run apisReport when done") {|t| opt.report = t}    
  begin
    opts.parse!
  rescue
    STDERR.puts $!.message
    STDERR.puts opts
    exit(1)
  end
  if (ARGV.size != 1)
    STDERR.puts opts
    exit(1)
  end
}

csv, rest = ARGV

sumTax = [Hash.new, Hash.new, Hash.new, Hash.new, Hash.new, Hash.new]
lvTax = ["Kingdom", "Phylum", "Class", "Order", "Family", "Genus"]

total = 0
File.new(csv).each {|line|
  fields = line.chomp.split("\t")
  if (fields[3] =~/TREE2/)
    tax = fields[3]
  else
    tax = fields[2]
  end
  levels = tax.split("|")
  0.upto(4) {|i|
    tax = levels[i + 1]
    tax = "Unclassified" if tax.nil?
    sumTax[i][tax] = 0 if sumTax[i][tax].nil?
    sumTax[i][tax] += 1
  }
  total += 1
}

0.upto(4) {|i|
  printf("%s:\n", lvTax[i])
  sumTax[i].keys.sort {|x, y| sumTax[i][y] <=> sumTax[i][x]}.each {|key|
    printf("%5.1f%% %s\n", sumTax[i][key]*100.0/total, key)
  }
  printf("\n")
}