pdf_info

defmodule PDFInfo do
  #  For backward compatibility, applications that create PDF 1.4 documents should include the metadata for a document in the document information dictionary as well as in the document’s metadata stream. Applications that support PDF 1.4 or later should check for the existence of a metadata stream and synchronize the information in it with that in the document information dictionary. The Adobe metadata framework provides a date stamp for metadata expressed in the framework. If this date stamp is equal to or later than the document modification date recorded in the document information dictionary, the metadata stream shall be taken as authoritative. If, however, the document modification date recorded in the document information dictionary is later than the metadata stream’s date stamp, the document has likely been saved by an application that is not aware of PDF 1.4 metadata streams. In this case, information stored in the document information dictionary should be taken to override any semantically equivalent items in the metadata stream.
  @moduledoc """
  Extracts all /Info and /Metadata objects from a PDF binary using Regex
  and without any external dependencies.
  """

  @doc """
  Checks if the binary starts with the PDF header.

  Returns `true` if the binary starts with the PDF header.
  Returns `false` otherwise.
  """
  @spec is_pdf?(binary) :: boolean
  def is_pdf?(<<"%PDF">> <> _), do: true

  def is_pdf?(_), do: false

  @doc """
  Extracts PDF version from the PDF header.

  Returns `{:ok, version}` if the PDF header is correct.
  Returns `:error` if the PDF header is incorrect.

  ## Examples

      iex> PDFInfo.pdf_version(binary)
      {:ok, "1.5"}
      iex> PDFInfo.pdf_version("not a pdf")
      :error

  """
  @spec pdf_version(binary) :: {:ok, binary} | :error
  def pdf_version(<<"%PDF-">> <> <<version::binary-size(3)>> <> _) do
    {:ok, version}
  end

  def pdf_version(_) do
    :error
  end

  @doc """
  Returns a list of /Encrypt reference strings.

  ## Examples

      iex> PDFInfo.encrypt_refs(binary)
      ["/Encrypt 52 0 R"]

  """
  @spec encrypt_refs(binary) :: list
  def encrypt_refs(binary) when is_binary(binary) do
    Regex.scan(~r{[/]Encrypt[ ]*[0-9].*?R}, binary)
    |> Enum.flat_map(& &1)
    |> Enum.uniq()
  end

  @doc """
  Returns `true` if PDF has at least one /Encrypt reference.
  Returns `false` if PDF has no /Encrypt reference.
  """
  @spec is_encrypted?(binary) :: boolean
  def is_encrypted?(binary) when is_binary(binary) do
    case encrypt_refs(binary) do
      [] -> false
      _ -> true
    end
  end

  @doc """
  Returns a list of /Info reference strings.

  ## Examples

      iex> PDFInfo.info_refs(binary)
      ["/Info 1 0 R"]

  """
  @spec info_refs(binary) :: list
  def info_refs(binary) when is_binary(binary) do
    Regex.scan(~r{[/]Info[\s0-9]*?R}, binary)
    |> Enum.flat_map(& &1)
    |> Enum.uniq()
  end

  @doc """
  Returns a list of /Metadata reference strings.

  ## Examples

      iex> PDFInfo.metadata_refs(binary)
      ["/Metadata 5 0 R"]

  """
  @spec metadata_refs(binary) :: list
  def metadata_refs(binary) when is_binary(binary) do
    Regex.scan(~r{[/]Metadata[\s0-9]*?R}, binary)
    |> Enum.flat_map(& &1)
    |> Enum.uniq()
  end

  @doc """
  Maps /Info reference strings to objects and parses the objects.

  ## Examples

      iex> PDFInfo.info_objects(binary)
      %{
        "/Info 1 0 R" => [
            %{
                "Author" => "The PostgreSQL Global Development Group",
                "CreationDate" => "D:20200212212756Z",
                ...
            }
        ]
      }

  """
  @spec info_objects(binary) :: map
  def info_objects(binary) when is_binary(binary) do
    binary
    |> raw_info_objects()
    |> Enum.reduce(%{}, fn {info_ref, list}, acc ->
      Map.put(acc, info_ref, Enum.map(list, &parse_info_object/1))
    end)
  end

  @doc """
  Maps /Metadata reference strings to objects and parses the objects.

  ## Examples

      iex> PDFInfo.metadata_objects(binary)
      %{
          "/Metadata 285 0 R" => [
            %{
              {"dc", "format"} => "application/pdf",
              {"pdf", "Producer"} => "Adobe PDF Library 15.0",
              {"xmp", "CreateDate"} => "2018-06-06T17:02:53+02:00",
              {"xmp", "CreatorTool"} => "Acrobat PDFMaker 17 für Word",
              {"xmp", "MetadataDate"} => "2018-06-06T17:03:13+02:00",
              {"xmp", "ModifyDate"} => "2018-06-06T17:03:13+02:00",
              ...
            }
          ]
      }

  """
  @spec metadata_objects(binary) :: map
  def metadata_objects(binary) when is_binary(binary) do
    binary
    |> raw_metadata_objects()
    |> Enum.map(&parse_metadata_object/1)
    |> Enum.reject(&Enum.empty?(&1))
  end

  @doc """
  Maps the /Info reference strings to the raw objects.

  ## Examples

      iex> PDFInfo.raw_info_objects(binary)
      %{"/Info 1 0 R" => ["1 0 obj <<..."]}

  """
  @spec raw_info_objects(binary) :: map
  def raw_info_objects(binary) when is_binary(binary) do
    binary
    |> info_refs()
    |> Enum.reduce(%{}, fn info_ref, acc ->
      obj_id =
        info_ref
        |> String.trim_leading("/Info ")
        |> String.trim_trailing(" R")

      list =
        get_object(binary, obj_id)
        |> Enum.flat_map(& &1)
        |> Enum.uniq()

      Map.put(acc, info_ref, list)
    end)
  end

  @doc """
  Maps the /Metadata reference strings to the raw objects.

  ## Examples

      iex> PDFInfo.raw_metadata_objects(binary)
      ["<x:xmpmeta" <> ...]

  """
  @spec raw_metadata_objects(binary) :: list
  def raw_metadata_objects(binary) when is_binary(binary) do
    Enum.zip(
      Regex.scan(~r{<x:xmpmeta}, binary, return: :index),
      Regex.scan(~r{</x:xmpmeta}, binary, return: :index)
    )
    |> Enum.reduce([], fn
      {[{start_position, _}], [{end_position, _}]}, acc when start_position < end_position ->
        raw_meta =
          binary_part(binary, start_position, end_position - start_position) <> "</x:xmpmeta>"

        [raw_meta | acc]

      _, acc ->
        acc
    end)
  end

  @doc false
  def parse_info_object(string) when is_binary(string) do
    strings = Regex.scan(~r{/(.*?)\s*\((.*?)\)}, string)

    hex = Regex.scan(~r{/([^ /]+)\s*<(.*?)>}s, string)

    Enum.concat(strings, hex)
    |> Enum.reduce([], fn
      [_, key, val], acc ->
        case decode_value(val) do
          {:ok, val} ->
            [{key, val} | acc]

          :error ->
            acc
        end
    end)
    |> Enum.map(fn {key, val} ->
      {key, val |> fix_non_printable() |> fix_octals()}
    end)
    |> Map.new()
  end

  def decode_value("feff" <> base_16_encoded_utf16_big_endian) do
    base_16_encoded_utf16_big_endian
    |> String.replace(~r{[^0-9a-f]}, "")
    |> Base.decode16(case: :lower)
    |> case do
      {:ok, utf16} ->
        endianness = determine_endianness(utf16, :big)

        utf16 = utf16_size_fix(utf16)

        {:ok, :unicode.characters_to_binary(utf16, {:utf16, endianness})}

      :error ->
        :error
    end
  end

  @doc false
  def decode_value(<<254, 255>> <> utf16) do
    endianness = determine_endianness(utf16, :big)

    utf16 = utf16_size_fix(utf16)

    {:ok, :unicode.characters_to_binary(utf16, {:utf16, endianness})}
  end

  def decode_value("\\376\\377" <> rest) do
    # Fix metadata: https://github.com/mozilla/pdf.js/pull/1598/files#diff-7f3b58adf9e7b7e802f63cc9b3855506R7
    rest
    |> String.replace("\\000\\(", "")
    |> String.trim_trailing("\\")
    |> fix_octal_utf16()
    |> case do
      string when is_binary(string) -> {:ok, string}
      error -> {:error, error}
    end
  end

  def decode_value(val) do
    {:ok, val}
  end

  # for info object only
  @doc false
  def fix_non_printable(val) when is_binary(val) do
    case String.printable?(val) do
      false ->
        # Fix for non printable binary like <<252>> = ü
        val |> :binary.bin_to_list() |> :unicode.characters_to_binary()

      true ->
        val
    end
  end

  @doc false
  def utf16_size_fix(binary) do
    case rem(byte_size(binary), 2) do
      0 -> binary
      1 -> binary <> <<0>>
    end
  end

  @doc false
  def fix_octal_utf16(binary) when is_binary(binary) do
    String.split(binary, ~r{\\[0-3][0-7][0-7]}, include_captures: true, trim: true)
    |> Enum.map(&do_fix_octal_utf16/1)
    |> Enum.chunk_every(2)
    |> Enum.map(fn
      [left_byte, right_byte] -> <<left_byte, right_byte>>
      _ -> <<>>
    end)
    |> Enum.join()
    |> :unicode.characters_to_binary({:utf16, :big})
  end

  @octal_first_digit 0..3 |> Enum.map(&to_string/1)
  @octal_rest_digits 0..7 |> Enum.map(&to_string/1)

  @doc false
  def do_fix_octal_utf16("\\" <> <<d1::bytes-size(1)>> <> <<d2::bytes-size(1)>> <> <<d3::bytes-size(1)>>) when d1 in @octal_first_digit and d2 in @octal_rest_digits and d3 in @octal_rest_digits do
    String.to_integer(d1) * 64 + String.to_integer(d2) * 8 + String.to_integer(d3) * 1
  end

  def do_fix_octal_utf16(<<code::utf8>>) do
    code
  end

  @doc false
  def fix_octals(binary) when is_binary(binary) do
    String.split(binary, ~r{\\[0-3][0-7][0-7]}, include_captures: true)
    |> Enum.map(&do_fix_octal/1)
    |> Enum.join()
  end

  @doc false
  def do_fix_octal("\\" <> <<d1::bytes-size(1)>> <> <<d2::bytes-size(1)>> <> <<d3::bytes-size(1)>>) when d1 in @octal_first_digit and d2 in @octal_rest_digits and d3 in @octal_rest_digits do
    code = String.to_integer(d1) * 64 + String.to_integer(d2) * 8 + String.to_integer(d3) * 1

    <<code::utf8>>
  end

  def do_fix_octal(binary) do
    binary
  end

  @doc false
  def parse_metadata_object(string) when is_binary(string) do
    with [xmp] <- Regex.run(~r{<x:xmpmeta.*?</x:xmpmeta}sm, string) do
      ["dc", "pdf", "pdfx", "xap", "xapMM", "xmp", "xmpMM"]
      |> Enum.reduce(%{}, fn tag, acc ->
        list = Regex.scan(~r{<#{tag}:(.*?)>(.*?)</#{tag}:(.*?)>}sm, xmp)

        reduce_metadata(acc, tag, list)
      end)
    else
      _ -> :error
    end
  end

  @doc false
  def reduce_metadata(acc, type, list) do
    list
    |> Enum.reduce(
      acc,
      fn
        [_, key, val, key], acc ->
          # remove rdf tags like <rdf:Alt>, </rdf:Alt>, <rdf:li ... />
          val =
            val
            |> String.replace(~r{<[a-z]+:[a-z]+>}i, " ")
            |> String.replace(~r{<[a-z]+:.+["']>}i, " ")
            |> String.replace(~r{</[a-z]+:[a-z]+>}i, " ")
            |> String.replace(~r{<[a-z]+:.+/>}i, " ")
            |> String.trim()

          case decode_value(val) do
            {:ok, val} ->
              Map.put(acc, {type, key}, val)

            :error ->
              acc
          end

        _, acc ->
          acc
      end
    )
  end

  @doc false
  def get_object(binary, obj_id) when is_binary(binary) and is_binary(obj_id) do
    obj_id = String.replace(obj_id, " ", "\s")

    ~r{[^0-9]#{obj_id}.obj(?s).*?endobj}
    |> Regex.scan(binary)
  end

  # it can have messed up byte order marker
  @doc false
  def determine_endianness(binary, initial_guess) do
    determine_endianness(binary, initial_guess, 0, 0)
  end

  defp determine_endianness(<<>>, _, little_zeros, big_zeros) do
    case {little_zeros, big_zeros} do
      {lz, bz} when lz > bz -> :little
      {lz, bz} when bz >= lz -> :big
    end
  end

  defp determine_endianness(<<0>> <> binary, :little, little_zeros, big_zeros) do
    determine_endianness(binary, :big, little_zeros + 1, big_zeros)
  end

  defp determine_endianness(<<0>> <> binary, :big, little_zeros, big_zeros) do
    determine_endianness(binary, :little, little_zeros, big_zeros + 1)
  end

  defp determine_endianness(<<_>> <> binary, :little, little_zeros, big_zeros) do
    determine_endianness(binary, :big, little_zeros, big_zeros)
  end

  defp determine_endianness(<<_>> <> binary, :big, little_zeros, big_zeros) do
    determine_endianness(binary, :little, little_zeros, big_zeros)
  end
end