20b5d958196753eb8956905585facd08d7d502b0

defmodule Basenji.Reader.PDFReader do
  @moduledoc false

  import Basenji.Reader

  def format, do: :pdf

  def file_extensions, 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")}.jpg"}
        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 ->
          String.split(line, ":", parts: 2)
          |> case do
            [k, v] -> to_metadata(k, v)
            [v] -> to_metadata("unknown_#{System.monotonic_time()}", v)
          end
        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: DateTimeParser.parse!(v)
  defp convert_value(:mod_date, v), do: DateTimeParser.parse!(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 file_extensions, do: ["pdf"]	543✔
9
10	def get_magic_numbers, do: [%{offset: 0, magic: [0x25, 0x50, 0x44, 0x46, 0x2D]}]	541✔
11
12	def close(_any), do: :ok	5✔
13
14	def get_entries(pdf_file_path, _opts \\ []) do
15	with {:ok, %{pages: pages}} <- get_metadata(pdf_file_path) do	544✔
16	padding = String.length("#{pages}") - 1	543✔
17
18	file_entries =	543✔
19	1..pages
20	\|> Enum.map(fn idx ->
21	%{file_name: "#{String.pad_leading("#{idx}", padding, "0")}.jpg"}	2,172✔
22	end)
23
24	{:ok, %{entries: file_entries}}
25	end
26	end
27
28	def get_entry_stream!(pdf_file_path, entry) do
29	file_name = entry[:file_name]	9✔
30	{page_num, _rest} = Integer.parse(file_name)	9✔
31
32	create_resource(fn ->	9✔
33	with {:ok, output} <- exec("pdftoppm", ["-f", "#{page_num}", "-singlefile", "-jpeg", "-q", pdf_file_path]) do	9✔
34	[output \|> :binary.bin_to_list()]
35	end
36	end)
37	end
38
39	def read(pdf_file_path, _opts \\ []) do
40	with {:ok, %{entries: file_entries}} <- get_entries(pdf_file_path) do	542✔
41	file_entries =	542✔
42	file_entries
43	\|> Enum.map(fn entry ->
44	entry
45	\|> Map.put(:stream_fun, fn -> get_entry_stream!(pdf_file_path, entry) end)	2,168✔
46	end)
47
48	{:ok, %{entries: file_entries}}
49	end
50	end
51
52	def get_metadata(pdf_file_path) do
53	with {:ok, output} <- exec("pdfinfo", ["-isodates", pdf_file_path]) do	544✔
54	metadata =	543✔
55	String.split(output, "\n")
56	\|> Map.new(fn line ->
57	String.split(line, ":", parts: 2)
58	\|> case do	9,774✔
59	[k, v] -> to_metadata(k, v)	9,774✔
NEW 60	[v] -> to_metadata("unknown_#{System.monotonic_time()}", v)	×
61	end
62	end)
63
64	{:ok, metadata}
65	end
66	end
67
68	defp to_metadata(k, v) do
69	k = k \|> String.trim() \|> ProperCase.snake_case() \|> String.to_atom()	9,774✔
70	v = convert_value(k, v \|> String.trim())	9,774✔
71	{k, v}
72	end
73
74	defp convert_value(:creation_date, v), do: DateTimeParser.parse!(v)	543✔
75	defp convert_value(:mod_date, v), do: DateTimeParser.parse!(v)	543✔
76	defp convert_value(:pages, v), do: String.to_integer(v)	543✔
77
78	defp convert_value(:filesize, v) do
79	{first, _rest} = Integer.parse(v)	543✔
80	first	543✔
81	end
82
83	defp convert_value(:pagesize, v) do
84	String.split(v, " ")
85	\|> Enum.reduce([], fn part, acc ->
86	Integer.parse(part)
87	\|> case do	2,172✔
88	{coord, _rest} -> [coord \| acc]	1,086✔
89	_ -> acc	1,086✔
90	end
91	end)
92	\|> case do	543✔
93	[x, y] -> {x, y}	543✔
94	_ -> v	×
95	end
96	end
97
98	defp convert_value(_k, "yes"), do: true	×
99	defp convert_value(_k, "no"), do: false	4,344✔
100	defp convert_value(_k, "none"), do: nil	543✔
101	defp convert_value(_k, v), do: v	2,172✔
102	end

camatcode / basenji / 20b5d958196753eb8956905585facd08d7d502b0

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