require 'nokogiri'
require 'activesupport'
include ActiveSupport::CoreExtensions::Hash
require 'progressbar'
require 'term/ansicolor'
require 'ruby-prof'

class String
  include Term::ANSIColor
end

class Pythia

  attr_accessor :url_store, :index

  def initialize
    puts "Initializing Pythia Search Engine....."
    start = Time.now
    @document_list = YAML.load_file('../db/index/catalogue.yml')
    @index = {}
    puts "loading index"
    @index = load_inverted_index
    puts "Time elapsed to Initialize Pythia Search Engine: #{Time.now - start} seconds"
    @results = {}
  end

  def start
    search
  end

  def search
    puts "Welcome to Pythia"
    puts "Ask anything:"
    while true
      query = gets.chomp.split
      #query = query.map{|word| word.lematize}
      results(query)
    end
  end

  def results(query)
    start = Time.now
    query.each do |keyword|
      str = "#{keyword}"
      docs_ids = @index[keyword]
      if !(docs_ids == [])
        docs_ids.each do |obj|
          find_docs_weight(obj)
          end
        else
          puts "Nothing Found for the word: #{str.green}"
      end
    puts " "
    end
    print_results(query)
    puts "Time elapsed to print results is #{Time.now - start} \n "
    puts "Ask something else..."
  end

  def find_docs_weight(element)
    doc_id = find_document_id(element[0])
    doc_weight = element[1]
     if @results.keys.include?(doc_id)
       @results[doc_id] = @results[doc_id].to_f + doc_weight.to_f
      else
        @results[doc_id] = doc_weight.to_f
     end
   @results
  end

  def print_results(query)
    puts "#{'We found:'.green}#{@results.size} #{'pages for your query:'.green} #{query.join(" ")}"
     @results.sort_by {|key, value| value}.reverse.each do |elem|
       puts "#{'Title:'.green} #{elem[0]}  #{'Weight:'.red} #{elem[1]} "
       end
     clear_results
  end

  def clear_results
    @results = {}
  end

  def find_document_id(num)
    @document_list[num.to_i]
  end

  def load_inverted_index

    start = Time.now
    f = File.open("../db/index/inverted_index.xml","r")
    puts "Time elapsed to load the xml file is #{Time.now - start} seconds"

    start = Time.now
    doc = Nokogiri::XML.parse(f)
    puts "Time elapsed parse the xml file with nokogiri is #{Time.now - start} seconds"

    my_table = doc.search('//lemma').map{ |e| Hash.from_xml(e.to_xml)['lemma']; }

    dictionary = Hash.new{ |h, k| h[k] = [] }
    my_table.each do |value|
      if value["document"].is_a? Array
        value["document"].each do |element|
          doc_id = element["id"]
          weight = element["weight"]
          dictionary["#{value["name"]}"] << [doc_id, weight]
        end
      else
        doc_id = value["document"]["id"]
        weight = value["document"]["weight"]
        dictionary["#{value["name"]}"] << [doc_id, weight]
      end
    end
    dictionary
  end

end

Pythia.new.start