prop/lib/microwaveprop/space_weather/swpc_client.ex

defmodule Microwaveprop.SpaceWeather.SwpcClient do
  @moduledoc """
  Client for NOAA SWPC's public JSON services (https://services.swpc.noaa.gov/).

  All three products we care about for HF / Es scoring are free, no-auth
  JSON endpoints:

  - `/json/planetary_k_index_1m.json` — planetary K 1-min cadence
  - `/json/f107_cm_flux.json` — 10.7 cm solar flux (hourly)
  - `/json/goes/primary/xrays-1-day.json` — GOES X-ray flux 1-min, 2 bands

  Each product has its own wire format so each has its own `fetch_*` +
  `parse_*` pair. Parsed rows are shaped for direct insert into the
  corresponding schemas via `Microwaveprop.SpaceWeather` context.
  """

  require Logger

  @base_url "https://services.swpc.noaa.gov"
  @default_timeout_ms 15_000

  # ---------- Kp ----------

  @doc """
  Fetch and parse the planetary K 1-min feed. Returns `{:ok, [row]}` where
  each row is `%{valid_time, kp_index, estimated_kp}`.
  """
  @spec fetch_kp() :: {:ok, [map()]} | {:error, term()}
  def fetch_kp, do: fetch_and_parse("/json/planetary_k_index_1m.json", &parse_kp/1)

  @doc false
  @spec parse_kp(String.t()) :: {:ok, [map()]} | {:error, term()}
  def parse_kp(body) do
    with {:ok, decoded} <- decode_array(body) do
      rows =
        decoded
        |> Enum.map(&parse_kp_row/1)
        |> Enum.reject(&is_nil/1)

      {:ok, rows}
    end
  end

  defp parse_kp_row(%{"time_tag" => t, "kp_index" => kp_index} = row) do
    case parse_naive_iso(t) do
      {:ok, valid_time} ->
        %{
          valid_time: valid_time,
          kp_index: to_integer(kp_index),
          estimated_kp: to_float(Map.get(row, "estimated_kp"))
        }

      :error ->
        nil
    end
  end

  defp parse_kp_row(_), do: nil

  # ---------- F10.7 ----------

  @doc """
  Fetch and parse the F10.7 cm flux feed. Returns `{:ok, [row]}` where
  each row is `%{valid_time, frequency_mhz, flux_sfu, reporting_schedule,
  ninety_day_mean}`.
  """
  @spec fetch_f107() :: {:ok, [map()]} | {:error, term()}
  def fetch_f107, do: fetch_and_parse("/json/f107_cm_flux.json", &parse_f107/1)

  @doc false
  @spec parse_f107(String.t()) :: {:ok, [map()]} | {:error, term()}
  def parse_f107(body) do
    with {:ok, decoded} <- decode_array(body) do
      rows =
        decoded
        |> Enum.map(&parse_f107_row/1)
        |> Enum.reject(&is_nil/1)

      {:ok, rows}
    end
  end

  defp parse_f107_row(%{"time_tag" => t, "flux" => flux} = row) do
    case parse_naive_iso(t) do
      {:ok, valid_time} ->
        %{
          valid_time: valid_time,
          frequency_mhz: to_integer(Map.get(row, "frequency")),
          flux_sfu: to_float(flux),
          reporting_schedule: Map.get(row, "reporting_schedule"),
          ninety_day_mean: to_float(Map.get(row, "ninety_day_mean"))
        }

      :error ->
        nil
    end
  end

  defp parse_f107_row(_), do: nil

  # ---------- GOES X-ray ----------

  @long_band "0.1-0.8nm"

  @doc """
  Fetch and parse the GOES primary 1-day X-ray feed. Only keeps the
  0.1–0.8nm long-wavelength band (the standard SWF / D-layer absorption
  proxy). Returns `{:ok, [row]}` where each row is `%{valid_time,
  satellite, flux_wm2, energy_band, electron_contaminated}`.
  """
  @spec fetch_xrays() :: {:ok, [map()]} | {:error, term()}
  def fetch_xrays, do: fetch_and_parse("/json/goes/primary/xrays-1-day.json", &parse_xrays/1)

  @doc false
  @spec parse_xrays(String.t()) :: {:ok, [map()]} | {:error, term()}
  def parse_xrays(body) do
    with {:ok, decoded} <- decode_array(body) do
      rows =
        decoded
        |> Enum.filter(&long_band?/1)
        |> Enum.map(&parse_xray_row/1)
        |> Enum.reject(&is_nil/1)

      {:ok, rows}
    end
  end

  defp long_band?(%{"energy" => @long_band}), do: true
  defp long_band?(_), do: false

  defp parse_xray_row(%{"time_tag" => t, "flux" => flux} = row) do
    case parse_iso_z(t) do
      {:ok, valid_time} ->
        %{
          valid_time: valid_time,
          satellite: to_integer(Map.get(row, "satellite")),
          flux_wm2: to_float(flux),
          energy_band: Map.get(row, "energy"),
          # SWPC's field name is misspelled "electron_contaminaton" upstream.
          electron_contaminated: !!Map.get(row, "electron_contaminaton")
        }

      :error ->
        nil
    end
  end

  defp parse_xray_row(_), do: nil

  # ---------- Shared HTTP / parsing helpers ----------

  defp fetch_and_parse(path, parser) do
    Microwaveprop.Instrument.span([:swpc, :fetch], %{path: path}, fn ->
      url = @base_url <> path

      case Req.get(url, [receive_timeout: @default_timeout_ms] ++ req_options()) do
        {:ok, %{status: 200, body: body}} ->
          parser.(body)

        {:ok, %{status: status, body: body}} ->
          {:error, "SWPC HTTP #{status}: #{inspect(body, printable_limit: 200)}"}

        {:error, reason} ->
          {:error, "SWPC request failed: #{format_req_error(reason)}"}
      end
    end)
  end

  defp decode_array(body) when is_binary(body) do
    case Jason.decode(body) do
      {:ok, list} when is_list(list) -> {:ok, list}
      {:ok, _other} -> {:error, :not_a_list}
      {:error, reason} -> {:error, reason}
    end
  end

  # Req's JSON auto-decode may already have parsed the body if the
  # upstream set Content-Type: application/json. In that case body is a
  # list, not a binary.
  defp decode_array(list) when is_list(list), do: {:ok, list}
  defp decode_array(_), do: {:error, :not_a_list}

  # SWPC's Kp/F10.7 time_tag is naive ISO ("2026-04-15T13:49:00") —
  # documented UTC but without a Z suffix. Add the Z so DateTime.from_iso8601
  # parses it as UTC.
  defp parse_naive_iso(s) when is_binary(s) do
    case DateTime.from_iso8601(s <> "Z") do
      {:ok, dt, _} -> {:ok, DateTime.truncate(dt, :second)}
      _ -> :error
    end
  end

  defp parse_naive_iso(_), do: :error

  defp parse_iso_z(s) when is_binary(s) do
    case DateTime.from_iso8601(s) do
      {:ok, dt, _} -> {:ok, DateTime.truncate(dt, :second)}
      _ -> :error
    end
  end

  defp parse_iso_z(_), do: :error

  defp to_integer(nil), do: nil
  defp to_integer(n) when is_integer(n), do: n
  defp to_integer(n) when is_float(n), do: trunc(n)

  defp to_integer(s) when is_binary(s) do
    case Integer.parse(s) do
      {n, ""} -> n
      _ -> nil
    end
  end

  defp to_float(nil), do: nil
  defp to_float(n) when is_float(n), do: n
  defp to_float(n) when is_integer(n), do: n * 1.0

  defp to_float(s) when is_binary(s) do
    case Float.parse(s) do
      {f, ""} -> f
      _ -> nil
    end
  end

  defp req_options do
    Application.get_env(:microwaveprop, :swpc_req_options, [])
  end

  # SWPC's CDN occasionally truncates the response mid-stream. Req's
  # auto-JSON-decoder then returns `{:error, %Jason.DecodeError{data: body}}`
  # where `:data` is the full raw body (up to ~650 KB for xrays-1-day).
  # Inspecting that struct at warn level spams the log aggregator. Keep
  # enough context to debug (position + a short preview) and drop the body.
  defp format_req_error(%Jason.DecodeError{position: pos, data: data}) do
    preview =
      data
      |> to_string()
      |> binary_part(0, min(80, byte_size(to_string(data))))

    "Jason.DecodeError at byte #{pos} (truncated upstream response; preview=#{inspect(preview)})"
  end

  defp format_req_error(reason), do: inspect(reason, printable_limit: 200)
end