ec5591dff434575811698a14a3e9c658250d85ea

defmodule Basenji.Reader.PDFReader do
  @moduledoc false

  import Basenji.Reader

  def format, do: :pdf

  def get_magic_numbers, do: [%{offset: 0, magic: [0x25, 0x50, 0x44, 0x46, 0x2D]}]

  def close(_any), do: :ok

  def get_entries(pdf_file_path, _opts \\ []) do
    with {:ok, %{pages: pages}} <- get_metadata(pdf_file_path) do
      padding = String.length("#{pages}") - 1

      file_entries =
        1..pages
        |> Enum.map(fn idx ->
          %{file_name: "#{String.pad_leading("#{idx}", padding, "0")}.jpeg"}
        end)

      {:ok, %{entries: file_entries}}
    end
  end

  def get_entry_stream!(pdf_file_path, entry) do
    file_name = entry[:file_name]
    {page_num, _rest} = Integer.parse(file_name)

    create_resource(fn ->
      with {:ok, output} <- exec("pdftoppm", ["-f", "#{page_num}", "-singlefile", "-jpeg", "-q", pdf_file_path]) do
        [output |> :binary.bin_to_list()]
      end
    end)
  end

  def read(pdf_file_path, _opts \\ []) do
    with {:ok, %{entries: file_entries}} <- get_entries(pdf_file_path) do
      file_entries =
        file_entries
        |> Enum.map(fn entry ->
          entry
          |> Map.put(:stream_fun, fn -> get_entry_stream!(pdf_file_path, entry) end)
        end)

      {:ok, %{entries: file_entries}}
    end
  end

  def get_metadata(pdf_file_path) do
    with {:ok, output} <- exec("pdfinfo", ["-isodates", pdf_file_path]) do
      metadata =
        String.split(output, "\n")
        |> Map.new(fn line ->
          [k, v] = String.split(line, ":", parts: 2)
          to_metadata(k, v)
        end)

      {:ok, metadata}
    end
  end

  defp to_metadata(k, v) do
    k = k |> String.trim() |> ProperCase.snake_case() |> String.to_atom()
    v = convert_value(k, v |> String.trim())
    {k, v}
  end

  defp convert_value(:creation_date, v), do: NaiveDateTime.from_iso8601!(v)
  defp convert_value(:mod_date, v), do: NaiveDateTime.from_iso8601!(v)
  defp convert_value(:pages, v), do: String.to_integer(v)

  defp convert_value(:filesize, v) do
    {first, _rest} = Integer.parse(v)
    first
  end

  defp convert_value(:pagesize, v) do
    String.split(v, " ")
    |> Enum.reduce([], fn part, acc ->
      Integer.parse(part)
      |> case do
        {coord, _rest} -> [coord | acc]
        _ -> acc
      end
    end)
    |> case do
      [x, y] -> {x, y}
      _ -> v
    end
  end

  defp convert_value(_k, "yes"), do: true
  defp convert_value(_k, "no"), do: false
  defp convert_value(_k, "none"), do: nil
  defp convert_value(_k, v), do: v
end

1	defmodule Basenji.Reader.PDFReader do
2	@moduledoc false
3
4	import Basenji.Reader
5
6	def format, do: :pdf	4✔
7
8	def get_magic_numbers, do: [%{offset: 0, magic: [0x25, 0x50, 0x44, 0x46, 0x2D]}]	502✔
9
10	def close(_any), do: :ok	5✔
11
12	def get_entries(pdf_file_path, _opts \\ []) do
13	with {:ok, %{pages: pages}} <- get_metadata(pdf_file_path) do	503✔
14	padding = String.length("#{pages}") - 1	502✔
15
16	file_entries =	502✔
17	1..pages
18	\|> Enum.map(fn idx ->
19	%{file_name: "#{String.pad_leading("#{idx}", padding, "0")}.jpeg"}	2,008✔
20	end)
21
22	{:ok, %{entries: file_entries}}
23	end
24	end
25
26	def get_entry_stream!(pdf_file_path, entry) do
27	file_name = entry[:file_name]	9✔
28	{page_num, _rest} = Integer.parse(file_name)	9✔
29
30	create_resource(fn ->	9✔
31	with {:ok, output} <- exec("pdftoppm", ["-f", "#{page_num}", "-singlefile", "-jpeg", "-q", pdf_file_path]) do	9✔
32	[output \|> :binary.bin_to_list()]
33	end
34	end)
35	end
36
37	def read(pdf_file_path, _opts \\ []) do
38	with {:ok, %{entries: file_entries}} <- get_entries(pdf_file_path) do	501✔
39	file_entries =	501✔
40	file_entries
41	\|> Enum.map(fn entry ->
42	entry
43	\|> Map.put(:stream_fun, fn -> get_entry_stream!(pdf_file_path, entry) end)	2,004✔
44	end)
45
46	{:ok, %{entries: file_entries}}
47	end
48	end
49
50	def get_metadata(pdf_file_path) do
51	with {:ok, output} <- exec("pdfinfo", ["-isodates", pdf_file_path]) do	503✔
52	metadata =	502✔
53	String.split(output, "\n")
54	\|> Map.new(fn line ->
55	[k, v] = String.split(line, ":", parts: 2)	9,036✔
56	to_metadata(k, v)	9,036✔
57	end)
58
59	{:ok, metadata}
60	end
61	end
62
63	defp to_metadata(k, v) do
64	k = k \|> String.trim() \|> ProperCase.snake_case() \|> String.to_atom()	9,036✔
65	v = convert_value(k, v \|> String.trim())	9,036✔
66	{k, v}
67	end
68
69	defp convert_value(:creation_date, v), do: NaiveDateTime.from_iso8601!(v)	502✔
70	defp convert_value(:mod_date, v), do: NaiveDateTime.from_iso8601!(v)	502✔
71	defp convert_value(:pages, v), do: String.to_integer(v)	502✔
72
73	defp convert_value(:filesize, v) do
74	{first, _rest} = Integer.parse(v)	502✔
75	first	502✔
76	end
77
78	defp convert_value(:pagesize, v) do
79	String.split(v, " ")
80	\|> Enum.reduce([], fn part, acc ->
81	Integer.parse(part)
82	\|> case do	2,008✔
83	{coord, _rest} -> [coord \| acc]	1,004✔
84	_ -> acc	1,004✔
85	end
86	end)
87	\|> case do	502✔
88	[x, y] -> {x, y}	502✔
NEW 89	_ -> v	×
90	end
91	end
92
NEW 93	defp convert_value(_k, "yes"), do: true	×
94	defp convert_value(_k, "no"), do: false	4,016✔
95	defp convert_value(_k, "none"), do: nil	502✔
96	defp convert_value(_k, v), do: v	2,008✔
97	end

camatcode / basenji / ec5591dff434575811698a14a3e9c658250d85ea

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous