gistfile1.builder

#State,attribute
#California,sunshine
#California,beaches
#Texas,oil
#Texas,rock_festivals

s = TableFu.new(csv)
s.faceted_by("State").each do |state|
  puts state.row
end

#or even better

s.faceted_by("State").each do |state|
  puts state.row['attribute']
end

Well, if you're faceting then you actually are looping through tf instances so, let me push a fix.

also, what do you want state.row to output? It'll just send and array of tablefu::datum instances.

state.row should output an array of all the rows sorted by the "State" column. You can get that effect now by doing

s = s.faceted_by "State"
s.each do |state|
 rows = state.table
 rows.each do |row|
   puts row
 end
end

though it's a little clunky and hard to actually get at the data for just one common column. My second example in the gist should get you

sunshine
beaches
oil
rock_festivals

Yeah it's always gunna be a bit clunky for the one common column. but you can now do row[key] for all rows. and then you can accomplish this with:

s.faceted_by("State").map { |state|
   state.rows.map { |row| row['attribute'] }
}.flatten
# ['sunshine', 'beaches', 'oil', 'rock festivals']

Of course rather than all these hoops, you can also just grab the original table and loop through the rows[key]