require_relative 'application_nginx_log_parser'

start_time = Time.now

RESULT_FILE = 'nginx.search.results.json'

SEARCH_REGEXP = /search\/questions\?(.*?)((q\=(.*?)\&))/i

def parse_file
  match_results = []

  # Withou second parameter the default log format of nginx will be assumed
  parser = Application::NginxLogParser.new('nginx.access.log')

  # It's possible to use a custom regexp to read line-by-line
  # parser = Application::NginxLogParser.new('nginx.access.log', /(.*)/)

  # parse and find for lines matching an regexp
  parser.parse_matching(SEARCH_REGEXP) do |parsed_line, matches|
    print "\r#{parser.current_line}/#{parser.total_lines} = #{parser.percent_read}%"

    match_results << { search: matches[-1] }.merge(parsed_line)
  end
  # Generate well formated results file for each line of JSON
  File.open(RESULT_FILE , 'wb') do |f|
    f.write(JSON.generate(match_results))
  end
end

def skip_log_file_parse?
  ARGV.include?('-s') || ARGV.include?('--skip-log-parse')
end

parse_file unless skip_log_file_parse?

# group data by term
data = JSON.parse(File.read(RESULT_FILE))
grouped_data = data.group_by {|d| URI.decode(d["search"]) }

total_by_term = grouped_data.map {|data| { search: data[0].gsub(/\+/, ' '), total: data[1].size } }

sorted = total_by_term.sort_by {|data| data[:total] }

# well formated results of search by term
File.open('total_searches.json', 'wb') do |f|
  total_searches = sorted.inject(0) {|total, data| total += data[:total] }

  json = JSON.pretty_generate({ total_searches: total_searches, report: sorted.reverse })
  f.write(json)
end

end_time = Time.now

runtime = (end_time - start_time)

puts "\nExecuted in %s seconds" % runtime