Number of words in reports vs number of words in code

seng_stats.rb

#!/usr/bin/env ruby
DETAILED_MODE = !ARGV[2].nil?

begin
  require 'yomu'
rescue LoadError
  msg = '(Looks like this is you)'
  lab = `hostname`.strip.start_with? 'cs'
  puts <<-EOF
You need the 'yomu' gem to use this script.

Install it on your own computer: #{lab ? '' : msg}

gem install yomu

Install it on a lab computer: #{lab ? msg : ''}

gem install yomu --user-install
  EOF
  exit 2
end

if ARGV.length < 2
  puts <<-EOF
That's not how you do this!

useage: ruby seng_stats.rb <report_folder> <repo_file> <detailed mode?>

report_folder is a folder of files (PDF, DOCX, etc) that you wrote this year.
repo_file is a list of git-clonable URLs (either HTTPS or SSH) one per line
add a third option to enable detailed/ debug mode

If the thing was part of a group project, either rename the files to include the
percentage (my_report_thing50%.docx), or change the repo_file line to have the
percentage after the url (space separated), like 'gitlab.com/me/project.git 50%'
  EOF
  exit 1
end

require 'fileutils'

def col(text, color)
  "\033[#{color.to_i}m#{text}\033[0m"
end

def do_reports(folder)
  Dir[folder + '/*'].map do |file|
    print "Processing #{file}... "
    ratio = 1
    if match = file.match(/(\d{2})%/)
      ratio = match[1].to_i / 100.0
    end
    yom = Yomu.new file
    text = yom.text
    size = text.scan(/\w+/).length
    puts "#{size} * #{ratio}" if DETAILED_MODE
    puts col('done.', 32)
    size * ratio
  end.sum
end

def do_repos(file='repos.txt', dest='repos')
  FileUtils::mkdir_p dest
  File.read(file).lines.map do |line|
    url, percent = line.split /\s+/
    url.strip!
    ratio = (((percent || '').match(/(\d{2})%/) || [])[1] || 100).to_i / 100.0
    puts col("Processing #{url}...", 32)
    path = dest + "/#{url}"
    unless File.exists? path
      system 'git', 'clone', url, path
    end
    size = Dir[path + '/**/*'].map do |file|
      if File.file? file
        content = File.read(file)
        if content.valid_encoding?
          size = content.scan(/\w+/).length
          puts "#{file}: #{size}" if DETAILED_MODE
          size
        else
          0
        end
      else
        0
      end
    end.sum
    puts "#{size} * #{ratio}" if DETAILED_MODE
    puts '... done.'
    size * ratio
  end.sum
end
report_folder = ARGV[0]
repo_file = ARGV[1]

puts col('Processing reports in ' + report_folder, 33)
report_size = do_reports(report_folder)
puts col('Processing git repos from ' + repo_file, 33)
repo_size = do_repos(repo_file)

puts col('Done!', 123)

puts '-' * 50

total = (repo_size + report_size).to_f / 100

puts <<-EOF
Report words: #{report_size}
'Words' in code: #{repo_size}

Ratio: reports / code = #{(report_size.to_f / repo_size).round(4)}

Reports make up #{(report_size / total).round(1)}%
Code makes up #{(repo_size / total).round(1)}%
EOF