In attempting to parse a html table, I ran across a class that I cannot figure out how to parse. Specifically:

defmodule Embedded do
  @doc """
  Floki can parse class selectors with embedded spaces?
  """
  def embedded() do
    html = """
    <tr class="team">
       <td class="name">
         Boston Bruins
       </td>
       <td class="ot-losses">
                            
       </td>
              
        <td class="diff text-success">
           35
        </td>
    </tr>
    """

    {:ok, parsed} = Floki.parse_document(html)
    IO.puts("parsed=\n#{inspect(parsed)}")

    parsed_td = Floki.find(parsed, "td")
    IO.puts("\nparsed_td=#{inspect(parsed_td)}")

    tds = Floki.find(parsed, "td") |> Floki.attribute("class")
    IO.puts("tds=#{inspect(tds)}")
    # tds=["\nname", "ot-losses", "diff text-success"]

    name = Floki.find(parsed, "td.name") |> Floki.text()
    IO.puts("\nname=#{inspect(name)}")
    # name="\n     Boston Bruins\n   "

    ot_losses = Floki.find(parsed, "td.ot_losses") |> Floki.text()
    IO.puts("\not_losses=#{inspect(ot_losses)}")
    # name="\n     Boston Bruins\n   "

    # FAILS: html clearly shows a value of 35 
    diff = Floki.find(parsed, "td.diff text-success") |> Floki.text()
    IO.puts("\ndiff=#{inspect(diff)}")
    # diff=
  end
end

The name and ot-losses get selected ok.

What Floki function I pull out the text for the "diff text-success?
This Floki function should work all other classes in the table.

I have no control over the html - it was read from a website.
The html fragment is presented straight from the source page.