directory_listing_html_parser.ex

defmodule Weather.NOAA.DirectoryListingHTMLParser do
  @moduledoc """
  Implements utilities to parse NOAA's cycle directory at
  http://nomads.ncep.noaa.gov/cgi-bin/filter_<model-name>.pl
  """

  @doc """
  Takes an html input an returns a list of available cycles.

  Args:
    - html - The HTML input to be parsed, string

  Returns a list of cycle maps (%{date: Timex.Date, cycle: integer})
  """
  def cycles(html) do
    Regex.scan(~r/<a[^<]*>(gfs\.\d*)<\/a>/, html)
    |> Enum.map(&parse_date_and_cycle/1)
  end

  defp parse_date_and_cycle(match) do
    {date, cycle} = extract_date_and_cycle(match)
    %{date: parse_date(date), cycle: parse_cycle(cycle)}
  end

  defp extract_date_and_cycle(match) do
    match
    |> List.last
    |> String.replace("gfs.", "")
    |> String.split_at(8)
  end

  defp parse_date(str) do
    Timex.parse!(str, "{YYYY}{0M}{D}") |> Timex.to_date
  end

  defp parse_cycle(str) do
    String.to_integer(str)
  end
end