mead_pdf

defmodule Mead.Markup do
  @moduledoc """
  Converts rendered `~PDF` markup into a `Mead.Node` tree.

  Attribute values are strings with CSS-like grammar:

    * lengths and factors: `"250"`, `"12.5"`
    * `flex_direction`: `"row"` or `"column"`
    * colors: `"#RGB"` or `"#RRGGBB"`
    * booleans (`break_before`, `keep_with_next`, `wrap`, `repeat`):
      `"true"`/`"false"`, or bare presence (`<view repeat>`); also
      `break_before="avoid"`

  Text content is HTML-like: whitespace (including source newlines)
  collapses to single spaces; write `<br/>` for a hard line break.
  """

  alias Mead.Node
  alias Phoenix.LiveView.Rendered

  @number_attrs ~w(gap font_size line_height flex_grow flex_shrink flex_basis border_radius)
  @dimension_attrs ~w(width height min_width min_height max_width max_height)
  @sides_attrs ~w(padding margin border)
  @alignment_attrs ~w(align_items align_self)
  @bool_attrs ~w(break_before keep_with_next wrap repeat repeat_after)

  @doc """
  Accepts a rendered `~PDF` template (or raw markup binary) and returns the
  `Mead.Node` tree. Multiple root elements are wrapped in a `view`.
  """
  @spec to_node(Rendered.t() | iodata()) :: Node.t()
  def to_node(%Rendered{} = rendered) do
    rendered
    |> Phoenix.HTML.Safe.to_iodata()
    |> to_node()
  end

  def to_node(markup) do
    roots =
      markup
      |> IO.iodata_to_binary()
      |> Floki.parse_fragment!()
      |> Enum.reject(&ignorable?/1)
      |> Enum.map(&build/1)

    case roots do
      [root] -> root
      many -> Node.view(many)
    end
  end

  defp ignorable?({:comment, _}), do: true
  defp ignorable?(text) when is_binary(text), do: String.trim(text) == ""
  defp ignorable?(_), do: false

  defp build({"view", attrs, children}) do
    Node.view(build_children(children), parse_attrs(attrs))
  end

  defp build({"table", attrs, children}) do
    {columns, attrs} =
      case List.keytake(attrs, "columns", 0) do
        {{"columns", value}, rest} -> {parse_columns(value), rest}
        nil -> {nil, attrs}
      end

    Node.table(build_children(children), [columns: columns] ++ parse_attrs(attrs))
  end

  defp build({"thead", attrs, children}), do: Node.table_header(build_children(children), parse_attrs(attrs))

  defp build({"tbody", attrs, children}), do: Node.table_body(build_children(children), parse_attrs(attrs))

  defp build({"tfoot", attrs, children}), do: Node.table_footer(build_children(children), parse_attrs(attrs))

  defp build({"tr", attrs, children}), do: Node.table_row(build_children(children), parse_attrs(attrs))

  defp build({"td", attrs, children}), do: Node.table_cell(build_children(children), parse_attrs(attrs))

  defp build({"image", attrs, _children}) do
    case List.keytake(attrs, "src", 0) do
      {{"src", src}, rest} -> Node.image(src, parse_attrs(rest))
      nil -> raise ArgumentError, "<image> requires a src attribute"
    end
  end

  # <barcode value="..." format="qr" ec_level="M" quiet_zone="0" .../>
  defp build({"barcode", attrs, _children}) do
    {value, attrs} =
      case List.keytake(attrs, "value", 0) do
        {{"value", value}, rest} -> {value, rest}
        nil -> raise ArgumentError, "<barcode> requires a value attribute"
      end

    {barcode_attrs, attrs} =
      Enum.reduce(~w(format ec_level quiet_zone), {[], attrs}, fn name, {acc, attrs} ->
        case List.keytake(attrs, name, 0) do
          {{^name, raw}, rest} -> {[{String.to_existing_atom(name), parse_barcode_attr(name, raw)} | acc], rest}
          nil -> {acc, attrs}
        end
      end)

    Node.barcode(value, barcode_attrs ++ parse_attrs(attrs))
  end

  @span_tags ~w(span page_number page_count)

  defp build({"text", attrs, children}) do
    rich? = Enum.any?(children, &match?({tag, _, _} when tag in @span_tags, &1))

    if rich? do
      spans =
        Enum.map(children, fn
          content when is_binary(content) -> Node.span(collapse_ws(content))
          {"br", _, _} -> Node.span("\n")
          {"span", span_attrs, span_children} -> Node.span(span_content(span_children), parse_attrs(span_attrs))
          {"page_number", span_attrs, _} -> Node.page_number(parse_attrs(span_attrs))
          {"page_count", span_attrs, _} -> Node.page_count(parse_attrs(span_attrs))
          {tag, _, _} -> raise ArgumentError, "<text> may only contain character data and <span>, got <#{tag}>"
        end)

      Node.text(spans, parse_attrs(attrs))
    else
      content =
        children
        |> Enum.map(fn
          content when is_binary(content) ->
            collapse_ws(content)

          {"br", _, _} ->
            "\n"

          {tag, _, _} ->
            raise ArgumentError,
                  "<text> may only contain character data and <span>, got <#{tag}>"
        end)
        |> IO.iodata_to_binary()
        |> String.trim(" ")

      Node.text(content, parse_attrs(attrs))
    end
  end

  defp build(text) when is_binary(text) do
    raise ArgumentError,
          "loose text #{inspect(String.trim(text))} must be wrapped in <text>...</text>"
  end

  defp span_content(children) do
    children
    |> Enum.map(fn
      content when is_binary(content) -> collapse_ws(content)
      {"br", _, _} -> "\n"
      {tag, _, _} -> raise ArgumentError, "<span> may only contain character data, got nested <#{tag}>"
    end)
    |> IO.iodata_to_binary()
  end

  # Markup text is HTML-like: ALL whitespace (including source newlines
  # from template indentation) collapses to spaces. Hard line breaks are
  # written as <br/>, which injects the `\n` the engine honors.
  defp collapse_ws(content), do: String.replace(content, ~r/\s+/, " ")

  defp build_children(children) do
    children
    |> Enum.reject(&ignorable?/1)
    |> Enum.map(&build/1)
  end

  # Column tracks: space-separated fixed widths and fractions,
  # e.g. columns="40 1fr 2fr".
  defp parse_columns(value) do
    value
    |> String.split()
    |> Enum.map(fn token ->
      case Float.parse(token) do
        {n, ""} -> n
        {n, "fr"} -> {:fr, n}
        _ -> raise ArgumentError, "invalid column track #{inspect(token)} (use e.g. \"40 1fr 2fr\")"
      end
    end)
  end

  defp parse_attrs(attrs), do: Enum.map(attrs, &parse_attr/1)

  defp parse_attr({name, value}) when name in @number_attrs do
    {String.to_existing_atom(name), parse_number(name, value)}
  end

  # Dimensions: points or CSS-style percentages ("250", "50%").
  defp parse_attr({name, value}) when name in @dimension_attrs do
    parsed =
      case String.split_at(value, -1) do
        {number, "%"} -> {:percent, parse_number(name, number)}
        _ -> parse_number(name, value)
      end

    {String.to_existing_atom(name), parsed}
  end

  # CSS shorthand: "5", "5 10", "5 10 15", "5 10 15 20" (t r b l).
  defp parse_attr({name, value}) when name in @sides_attrs do
    parsed =
      case value |> String.split() |> Enum.map(&parse_number(name, &1)) do
        [uniform] -> uniform
        [v, h] -> {v, h, v, h}
        [t, h, b] -> {t, h, b, h}
        [t, r, b, l] -> {t, r, b, l}
        _ -> raise ArgumentError, "#{name} takes 1-4 values, got #{inspect(value)}"
      end

    {String.to_existing_atom(name), parsed}
  end

  # Explicit value maps: `to_existing_atom` on the value would depend on
  # the atom having been minted elsewhere in the running system.
  @alignment_values %{
    "start" => :start,
    "end" => :end,
    "center" => :center,
    "stretch" => :stretch,
    "baseline" => :baseline
  }

  defp parse_attr({name, value}) when name in @alignment_attrs do
    case @alignment_values do
      %{^value => atom} ->
        {String.to_existing_atom(name), atom}

      _ ->
        raise ArgumentError,
              ~s(#{name} must be "start", "end", "center", "stretch", or "baseline", got #{inspect(value)})
    end
  end

  @justify_values %{
    "start" => :start,
    "end" => :end,
    "center" => :center,
    "space-between" => :space_between,
    "space-around" => :space_around,
    "space-evenly" => :space_evenly
  }

  defp parse_attr({"justify_content", value}) do
    case @justify_values do
      %{^value => atom} ->
        {:justify_content, atom}

      _ ->
        raise ArgumentError,
              ~s(justify_content must be "start", "end", "center", "space-between", ) <>
                ~s("space-around", or "space-evenly", got #{inspect(value)})
    end
  end

  # CSS-style `break_before="avoid"` (see `Mead.Node` docs).
  defp parse_attr({"break_before", "avoid"}), do: {:break_before, :avoid}

  defp parse_attr({name, value}) when name in @bool_attrs do
    parsed =
      case value do
        "true" -> true
        "false" -> false
        # Bare attributes parse as "" or as the attribute's own name.
        "" -> true
        ^name -> true
        other -> raise ArgumentError, "invalid boolean #{name}=#{inspect(other)}"
      end

    {String.to_existing_atom(name), parsed}
  end

  defp parse_attr({"href", value}), do: {:href, value}
  defp parse_attr({"bookmark", value}), do: {:bookmark, value}

  defp parse_attr({"bookmark_level", value}) do
    case Integer.parse(value) do
      {level, ""} when level >= 1 -> {:bookmark_level, level}
      _ -> raise ArgumentError, "bookmark_level must be a positive integer, got #{inspect(value)}"
    end
  end

  defp parse_attr({"font_family", value}), do: {:font_family, value}

  defp parse_attr({"font_weight", value}) do
    case value do
      "bold" -> {:font_weight, 700}
      "normal" -> {:font_weight, 400}
      other -> {:font_weight, parse_number("font_weight", other)}
    end
  end

  defp parse_attr({"italic", value}) do
    case value do
      v when v in ["true", "", "italic"] -> {:italic, true}
      "false" -> {:italic, false}
      other -> raise ArgumentError, "invalid boolean italic=#{inspect(other)}"
    end
  end

  @text_align_values %{"left" => :left, "center" => :center, "right" => :right, "justify" => :justify}

  defp parse_attr({"text_align", value}) when is_map_key(@text_align_values, value),
    do: {:text_align, Map.fetch!(@text_align_values, value)}

  defp parse_attr({"text_align", other}) do
    raise ArgumentError, ~s(text_align must be "left", "center", "right", or "justify", got #{inspect(other)})
  end

  defp parse_attr({"flex_direction", "row"}), do: {:flex_direction, :row}
  defp parse_attr({"flex_direction", "column"}), do: {:flex_direction, :column}

  defp parse_attr({"flex_direction", other}),
    do: raise(ArgumentError, ~s(flex_direction must be "row" or "column", got #{inspect(other)}))

  defp parse_attr({name, value}) when name in ~w(background color border_color) do
    {String.to_existing_atom(name), parse_color(name, value)}
  end

  defp parse_attr({name, _value}) do
    raise ArgumentError,
          "unknown attribute #{inspect(name)}; supported: " <>
            Enum.join(
              @number_attrs ++
                @bool_attrs ++
                @dimension_attrs ++
                @sides_attrs ++
                @alignment_attrs ++
                ~w(justify_content flex_direction background color border_color font_family font_weight italic text_align href bookmark bookmark_level),
              ", "
            )
  end

  defp parse_barcode_attr("format", value) do
    formats = ~w(qr aztec data_matrix pdf417 code128 code39 code93 codabar ean8 ean13 upc_a upc_e itf telepen)

    value in formats ||
      raise ArgumentError, "unknown barcode format #{inspect(value)}; expected one of: #{Enum.join(formats, ", ")}"

    String.to_existing_atom(value)
  end

  defp parse_barcode_attr("ec_level", value), do: value

  defp parse_barcode_attr("quiet_zone", value) do
    case Integer.parse(value) do
      {n, ""} when n >= 0 -> n
      _ -> raise ArgumentError, "quiet_zone must be a non-negative integer, got #{inspect(value)}"
    end
  end

  defp parse_number(name, value) do
    case Float.parse(value) do
      {number, ""} -> number
      _ -> raise ArgumentError, "invalid number #{name}=#{inspect(value)}"
    end
  end

  defp parse_color(_name, "#" <> hex) when byte_size(hex) == 3 do
    [r, g, b] = for <<c::binary-1 <- hex>>, do: String.duplicate(c, 2)
    parse_hex_triplet(r, g, b)
  end

  defp parse_color(_name, "#" <> hex) when byte_size(hex) == 6 do
    <<r::binary-2, g::binary-2, b::binary-2>> = hex
    parse_hex_triplet(r, g, b)
  end

  defp parse_color(name, value), do: raise(ArgumentError, ~s(#{name} must be "#RGB" or "#RRGGBB", got #{inspect(value)}))

  defp parse_hex_triplet(r, g, b) do
    {String.to_integer(r, 16), String.to_integer(g, 16), String.to_integer(b, 16)}
  end
end