Futz random-ish input to import_mappings

monkey_futz.rb

#!/usr/bin/env ruby

if ARGV.length != 1
  puts 'Usage: monkey-futz <host>'
  exit 1
end

@host = ARGV.first

TYPES = [:archive,
         :archive_with_tna_url,
         :archive_with_bad_url,
         :redirect,
         :unresolved,
         :header,
         :too_many,
         :offsite_host,
         # :blank, # commented out while it fails
         :dupes,
         :not_header]

@path_num    = 0
@new_url_num = 0

def next_new_url
  @new_url_num += 1
  "http://#{@host}/#{@new_url_num}"
end

def next_path_or_url
  @path_num += 1
  "#{rand * 2 > 1 ? "http://#{@host}" : nil}/#{@path_num}"
end

def random_line
  case type = TYPES.sample

  when :archive_with_tna_url then "#{next_path_or_url},http://webarchive.nationalarchives.gov.uk/20121222101738/http://number10.gov.uk//"
  when :archive_with_bad_url then "#{next_path_or_url},http://[][]__*"
  when :offsite_host         then "http://road.cc/,TNA"
  when :archive              then "#{next_path_or_url},TNA"
  when :redirect             then "#{next_path_or_url},#{next_new_url}"
  when :unresolved           then "#{next_path_or_url},"
  when :header               then "old_url,new_url"
  when :too_many             then "#{next_path_or_url},#{next_new_url},extra-stuff"
  when :blank                then ''
  when :not_header           then "oxd_url,nxw_url,rubbish"
  when :dupes                then u = "#{next_path_or_url},TNA"; ("#{u}\n" * 2).chomp

  else "oops - #{type}"
  end
end

1000.times { puts random_line }