script/update-ids

#!/usr/bin/env ruby

require 'bundler/setup'
require 'yaml'
require 'pry'
require 'digest'

def generate_language_id(language)
  Digest::SHA256.hexdigest(language).to_i(16) % (2**30 - 1)
end

data = File.read("lib/linguist/languages.yml")
header = data.match(/\A(?:^#[^\n]*\n|^[ \t]*\n)+/m)[0]
languages = YAML.load(data)


# Do a dry-run to report which languages are missing IDs
if ARGV[0] == "--check"
  missing = []

  languages.each do |name, vals|
    missing << name unless vals.has_key?('language_id')
  end

  list = missing.map {|n| "  - #{n}"}.join("\n")
  case missing.count
  when 0
    puts "No languages were found with missing IDs."
    exit 0
  when 1
    puts "1 language was found with a missing ID:"
    puts "\n#{list}\n\n"
    puts "Run `#{$0}` to generate it."
  else
    puts "#{missing.count} languages were found with missing IDs:"
    puts "\n#{list}\n\n"
    puts "Run `#{$0}` to generate them."
  end
  exit 1

# Generate `language_id' fields for those entries missing them.
else
  updated = []

  languages.each do |name, vals|
    unless vals.has_key?('language_id')
      vals.merge!('language_id' => generate_language_id(name))
      updated << name
    end
  end

  list = updated.map {|n| "  - #{n}"}.join("\n")
  case updated.count
  when 0
    puts "No languages were found with missing IDs."
    exit
  when 1
    puts "Updated 1 language:"
    puts "\n#{list}\n\n"
  else
    puts "Updated #{updated.count} languages:"
    puts "\n#{list}\n\n"
  end

  File.write("lib/linguist/languages.yml", header + YAML.dump(languages))
end