#!/usr/bin/env ruby

require 'MESH'

class Numeric
  def duration
    secs, millisecs = self.divmod 1
    # secs = self.to_int
    mins = secs / 60
    hours = mins / 60
    days = hours / 24

    if days > 0
      "#{days} days and #{hours % 24} hours"
    elsif hours > 0
      "#{hours} hours and #{mins % 60} minutes"
    elsif mins > 0
      "#{mins} minutes and #{secs % 60} seconds"
    elsif secs >= 0
      "#{(millisecs + secs).round(3)} seconds"
    end
  end
end

def time_this(name, &block)
  # print "#{name}"
  # STDOUT.flush
  start = Time.now.to_f
  result = yield
  finish = Time.now.to_f
  # puts "\t#{(finish - start).duration}"
  finish - start
end

mesh_tree = MESH::Tree.new
mesh_tree.load_translation(:en_gb)
mesh_tree.load_wikipedia

json_str = File.new('./example.json').read
extracted = JSON.parse(json_str)

timings = Hash.new { |h, k| h[k] = [] }

(0..10).each do |i|
  time = time_this('Matching in title') { mesh_tree.match_in_text(extracted['title']) }
  timings[:title_headings] << time
  time = time_this('Matching in description') { mesh_tree.match_in_text(extracted['description']) }
  timings[:description_headings] << time
  time = time_this('Matching in content') { mesh_tree.match_in_text(extracted['content']) }
  timings[:content_headings] << time
end

# (0..10).each do |i|
#   time = time_this('Matching in title') { mesh_tree.match_in_text_2(extracted['title']) }
#   timings[:title_entries_2] << time
#   time = time_this('Matching in description') { mesh_tree.match_in_text_2(extracted['description']) }
#   timings[:description_entries_2] << time
#   time = time_this('Matching in content') { mesh_tree.match_in_text_2(extracted['content']) }
#   timings[:content_entries_2] << time
# end

timings.each do |k, v|
  t = v.map { |e| e.round(3) }.join("\t")
  puts "#{k}\t#{t}"
  avg = v.inject { |sum, el| sum + el }.to_f / v.size
  puts "#{k}\t#{avg.round(3)}"
end

# classifier = MESH::Classifier.new()
# classification = time_this('Classifying from matches') { classifier.classify([
#                                                                                  {weight: 10.0, matches: title_headings},
#                                                                                  {weight: 5.0, matches: description_headings},
#                                                                                  {weight: 1.0, matches: content_headings}
#                                                                              ]) }