tsv_to_yml.rb 109078513

tsv_to_yml.rb

# frozen_string_literal: true

require 'csv'
require 'yaml'

# check file is tsv or not
def tsv?(filename)
  return raise('File not exist!') unless File.exist?(filename.to_s) # No file!

  content = File.read(filename.to_s, encoding: 'utf-8')
  return raise('Encoding Error!') unless content.valid_encoding? # Wrong encoding!

  filename.include?('.tsv')
end

input_tsv_file = ARGV[0] if tsv?(ARGV[0])
output_yaml_file = ARGV[1] if ARGV[1].to_s.include?('.yml')

def tsv2yaml(input_tsv_file, output_yaml_file)
  # turn tsv into csv & change it to yaml
  csv_table = CSV.table(input_tsv_file, { col_sep: "\t" })
  # change key type from symbol to string to avoid test error
  headers = csv_table.headers.map(&:to_s)

  # Make a array of several hashes(several rows)
  row_hashes = csv_table.map do |row|
    # first () produce [{"name" => "Ken"}, {"gender" => "Male"}]
    # to_s to avoid implicit type error
    # merge several hashes into one hash, produce [{"name" => "Ken", "gender" => "Male"}]
    (headers.map { |header| { header => row[header.to_sym].to_s } }).reduce({}, :merge)
  end

  if output_yaml_file then File.open(output_yaml_file, 'w') { |f| f.puts row_hashes.to_yaml }
  else puts row_hashes.to_yaml end
end

tsv2yaml(input_tsv_file, output_yaml_file)