Enable loading of explicit sub-class definitions

If this is active, where would the code fit? It feels like there are a range of possibilities, from internal to external:

generalize the data models
generalize the data ingest machinery
provide a tool to make data of this form look like something we already recognize.

I would lean towards the external end of the spectrum: We're already expecting a lot from our users for data prep, so this doesn't seem out of line to ask them to use a tool we provide. On the other hand, making the whole thing easier wouldn't be bad, either.

Here's a rough draft of an external script:

Given:

Green: 2
Green, Red, Tasty: 1
Red: 1
Red, Tasty: 2

run:

headers = []
rows = []

File.readlines(ARGV.shift).each do |line|
  line.chomp!
  labels_joined, count = line.split /\s*:\s*/
  labels = labels_joined.split /\s*,\s*/
  labels.each do |label|
    headers << label unless headers.include?(label)
  end 
  1.upto(count.to_i) do
    rows << labels
  end
end

puts 'fake_id,' + headers.join(',')
rows.each_with_index do |row, i|
  puts i.to_s + ',' + (headers.map do |header|
    row.include?(header) ? 1 : 0
  end.join(','))
end

to produce:

fake_id,Green,Red,Tasty
0,1,0,0
1,1,0,0
2,1,1,1
3,0,1,0
4,0,1,1
5,0,1,1

(and sorry for the bother if this isn't active.)

VCG / upset

Enable loading of explicit sub-class definitions #83