Sponsored by AppSignal
Would you like to see your link here? Contact us
Notesclub

Discography Prototype :: ID3 Reader

discography_prototype_id3.livemd

Discography Prototype :: ID3 Reader

Mix.install([
  {:id3, git: "https://github.com/ndac-todoroki/elixir-id3.git", branch: "update-deps"}
])

Summary

The purpose of this is to determine how feasible it would be to read tags for MP3s in a known iTunes directory.

The full project should:

Now, Next, Later

  1. Globbing scan for all *.mp3s.
    1. Now: Restrict the scan to a single directory like iTunes?
    2. Later: Other audio formats could be useful.
    3. Store the base path minus the file name?
  2. Read ID3v2 or possibly ID3v1 tags for metadata.
  3. Filter unique metadata for Artist and Album.
    1. Later: Merge records.
  4. Store metadata in database tables.
  5. Background worker checks Spotify or another music API for missing albums.
  6. Create a UI to display artist information.
    1. Absolutely steal from prior art like Apple Music, Amazon Music, Spotify, etc. This should be familiar
  7. Send notifications.
    1. Now: In-app notifications as APIs are slowly polled throughout the month.
    2. Later: Email notifications?
    3. Next: Notification screen to view missing albums by artist.
  8. View artist by missing albums.
    1. Next: Different views like owned, not owned, or possibly wish list?
    2. Now: Mixed views that show what’s owned vs missing? One single view is likely preferred with filtering added later.
  9. Later: Verify albums have all songs. Partial albums may count as incomplete.
    1. This likely changes the whole data format, we may store the full metadata including file name and full path.
    2. I may be entirely wrong about needing the path at any point. My mind says I need it but a scan of the entire iTunes 40gb is stupid fucking quick for what I expect. 5 seconds to read to the end of 40gb of files is impressive.
  10. Later: Merge capabilities
  11. This is likely where paths come in to resolve conflicts.
# From https://benjamintan.io/blog/2014/06/10/elixir-bit-syntax-and-id3/
defmodule ID3Parser do
  def parse(file_name) do
    case File.read(file_name) do
      {:ok, binary} ->
        mp3_byte_size = byte_size(binary) - 128
        <<_::binary-size(mp3_byte_size), id3_tag::binary>> = binary

        <<"TAG", title::binary-size(30), artist::binary-size(30), album::binary-size(30),
          year::binary-size(4), comment::binary-size(30), _rest::binary>> = id3_tag

        IO.puts(title)
        IO.puts(artist)
        IO.puts(album)
        IO.puts(year)
        IO.puts(comment)

      _ ->
        IO.puts("Couldn't open #{file_name}")
    end
  end
end
# From https://github.com/anisiomqs/shoutcast_server/blob/master/lib/mp3_file.ex
defmodule Mp3File do
  def extract_metadata(file) do
    read_file = File.read!(file)
    file_length = byte_size(read_file)
    music_data = file_length - 128
    <<_::binary-size(music_data), id3_section::binary>> = read_file
    id3_section
  end

  defp parse_id3(metadata) do
    <<_::binary-size(3), title::binary-size(30), artist::binary-size(30), album::binary-size(30),
      year::binary-size(4), comment::binary-size(28), _::binary>> = metadata

    %{
      title: sanitize(title),
      artist: sanitize(artist),
      album: sanitize(album),
      year: sanitize(year),
      comment: sanitize(comment)
    }
  end

  defp sanitize(text) do
    not_zero = &amp;(&amp;1 != <<0>>)
    text |> String.graphemes() |> Enum.filter(not_zero) |> to_string |> String.trim()
  end

  def extract_id3(file) do
    metadata = extract_metadata(file)
    parse_id3(metadata)
  end

  def extract_id3_list(folder) do
    folder |> list |> Enum.map(&amp;extract_id3/1)
  end

  def list(folder) do
    folder |> Path.join("**/*.mp3") |> Path.wildcard()
  end
end
directory = "/Users/jbrayton/Music/iTunes/iTunes Media/Music/Periphery/"
files = "**/*.mp3"

path = Path.join(directory, files)

Path.wildcard(path)
|> Enum.map(fn path ->
  # Read tags
  {:ok, tag} = ID3.get_tag(path)
  {path, tag}
end)
|> Enum.uniq_by(fn {_path, tag} ->
  # Filter down to unique artist/album pairs
  tag.artist &amp;&amp; tag.album
end)

# |> Enum.map(fn {_path, tag} ->
#   IO.inspect(tag, label: "Tag")
#   {tag.artist, tag.album}
# end)

# Create Rust cargo config for rustler compilation
# directory = Path.absname(__DIR__)
# IO.inspect(directory)

# directory_path = Path.join(directory, "/native/id3/.cargo")
# File.mkdir_p!(directory_path)
# content = """
# [target.x86_64-apple-darwin]
# rustflags = [
#     "-C", "link-arg=-undefined",
#     "-C", "link-arg=dynamic_lookup",
# ]
# """
# File.write!("#{directory_path}/config", content)

file_name =
  "/Users/jbrayton/Music/iTunes/iTunes Media/Music/Periphery/Juggernaut_ Alpha/1-01 A Black Minute.mp3"

ID3Parser.parse(file_name)
Mp3File.extract_id3(file_name)

# path = Path.join(directory, files)
# Path.wildcard(path)
# |> Enum.map(fn path ->
#   ID3Parser.parse(path)
# end)
# Fails with
# ** (MatchError) no match of right hand side value: "UUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUU"
#     #cell:h57dsjc5diwri2dpbfsovoxuui2lapp3:15: ID3Parser.parse/1
#     #cell:fyltqturndp3qyoheuqz4mi7eey3nvti:52: (file)
#     #cell:fyltqturndp3qyoheuqz4mi7eey3nvti:51: (file)

# Mp3File.extract_id3_list(directory)
# Fails with
# ** (UnicodeConversionError) invalid encoding starting at <<170>>
#     (elixir 1.14.2) lib/list.ex:1103: List.to_string/1
#     #cell:uqbhjfpibw3u7iaxhsufaya45wswaprx:30: Mp3File.sanitize/1
#     #cell:uqbhjfpibw3u7iaxhsufaya45wswaprx:20: Mp3File.parse_id3/1
#     (elixir 1.14.2) lib/enum.ex:1658: Enum."-map/2-lists^map/1-0-"/2
#     (elixir 1.14.2) lib/enum.ex:1658: Enum."-map/2-lists^map/1-0-"/2
#     #cell:fyltqturndp3qyoheuqz4mi7eey3nvti:40: (file)

It’s honestly easier and we get far more information with the ID3 module. Love the No need for an external library. comment from José at https://stackoverflow.com/questions/30417699/how-to-read-and-write-id3v1-and-id3v2-tags-in-elixir#comment48928044_30417699

Do you need an external library? No. Do you get a shitload more done with one? Hell fucking yes dude.

To be absolutely fair, those modules above are very tight but they also only implement a loose v1 spec. ID3v2 is way more involved.

I tried https://github.com/thechangelog/id3vx and really wanted it to win but either I don’t have any ID3v2 files or it required more brainpower than I was willing to throw at it. I did mention implementing the v1 spec on Twitter so it may be worth attempting.

The chart for v1 on https://benjamintan.io/blog/2014/06/10/elixir-bit-syntax-and-id3/ points to https://en.wikipedia.org/wiki/ID3#ID3v1_and_ID3v1.1[5]. It would’ve been nice had he included this considering there is clearly subscript in the image that now doesn’t line up. It’s also worth pointing out that the link to the post about Erlang is at https://legacyblog.citizen428.net/blog/2010/09/04/erlang-bit-syntax-and-id3/.

This is an article from 2014 with comments in 2015 so I should likely be a little more calm. I’m not exactly annoyed or angry, I just expected far less friction than I was met with. I had totally expected that I needed to abandon this as a capstone project but now I believe I can handle this rather cleanly.

directory = "/Users/jbrayton/Music/iTunes/iTunes Media/Music"
files = "/**/**/*.mp3"

path = Path.join(directory, files)

Path.wildcard(path)
|> Enum.map(fn path ->
  # Read tags
  tag =
    case ID3.get_tag(path) do
      {:ok, tag} -> tag
      {:error, _error} -> %ID3.Tag{}
    end

  %{path: path, tag: tag}
end)
|> Enum.uniq_by(fn %{tag: tag} ->
  # Filter down to unique artist/album pairs
  # It would be quicker to use a MapSet to stay unique
  # ETS could work but overkill, why not benchmark it?
  if !is_nil(tag) do
    tag.artist &amp;&amp; tag.album
  end
end)
|> Enum.filter(fn %{tag: tag} ->
  is_nil(tag.album_artist)
end)
|> Enum.sort_by(
  fn %{tag: tag} ->
    tag.artist || tag.album_artist
  end,
  :asc
)

# Benchmark the naive approaches vs MapSet, ETS, or anything that enforces uniqueness.
# List appending may also be beneficial or keyword lists.
# I should have time for this detour provided that's completed by 12/20 at the latest.
# That would give me an entire month of limited time on UI polishing and separating concerns.
defmodule LiveBeats.MP3Stat do
  @moduledoc """
  Decodes MP3s and parses out information.

  MP3 decoding and duration calculation credit to:
  https://shadowfacts.net/2021/mp3-duration/
  """
  import Bitwise
  alias LiveBeats.MP3Stat

  defstruct duration: 0,
            size: 0,
            path: nil,
            title: nil,
            album: nil,
            artist: nil,
            year: 1901,
            genre: nil,
            tags: nil,
            picture: nil

  @declared_frame_ids ~w(AENC APIC ASPI COMM COMR ENCR EQU2 ETCO GEOB GRID LINK MCDI MLLT OWNE PRIV PCNT POPM POSS RBUF RVA2 RVRB SEEK SIGN SYLT SYTC TALB TBPM TCOM TCON TCOP TDEN TDLY TDOR TDRC TDRL TDTG TENC TEXT TFLT TIPL TIT1 TIT2 TIT3 TKEY TLAN TLEN TMCL TMED TMOO TOAL TOFN TOLY TOPE TOWN TPE1 TPE2 TPE3 TPE4 TPOS TPRO TPUB TRCK TRSN TRSO TSOA TSOP TSOT TSRC TSSE TSST TXXX UFID USER USLT WCOM WCOP WOAF WOAR WOAS WORS WPAY WPUB WXXX)

  @v1_l1_bitrates {:invalid, 32, 64, 96, 128, 160, 192, 224, 256, 288, 320, 352, 384, 416, 448,
                   :invalid}
  @v1_l2_bitrates {:invalid, 32, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320, 384,
                   :invalid}
  @v1_l3_bitrates {:invalid, 32, 40, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320,
                   :invalid}
  @v2_l1_bitrates {:invalid, 32, 48, 56, 64, 80, 96, 112, 128, 144, 160, 176, 192, 224, 256,
                   :invalid}
  @v2_l2_l3_bitrates {:invalid, 8, 16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 144, 160,
                      :invalid}

  def to_mmss(duration) when is_integer(duration) do
    hours = div(duration, 60 * 60)
    minutes = div(duration - hours * 60 * 60, 60)
    seconds = rem(duration - hours * 60 * 60 - minutes * 60, 60)

    [minutes, seconds]
    |> Enum.map(fn count -> String.pad_leading("#{count}", 2, ["0"]) end)
    |> Enum.join(":")
  end

  def parse(path) do
    stat = File.stat!(path)
    {tag_info, rest} = parse_tag(File.read!(path))
    duration = parse_frame(rest, 0, 0, 0)

    case duration do
      duration when is_float(duration) and duration > 0 ->
        IO.inspect(tag_info)
        picture = tag_info["APIC"]
        title = Enum.at(tag_info["TIT2"] || [], 0)
        album = Enum.at(tag_info["TALB"] || [], 0)
        artist = Enum.at(tag_info["TPE1"] || [], 0)
        year = Enum.at(tag_info["TDRC"] || [], 0) || Enum.at(tag_info["TYER"] || [], 0)
        genre = Enum.at(tag_info["TCON"] || [], 0)
        seconds = round(duration)

        {:ok,
         %MP3Stat{
           duration: seconds,
           size: stat.size,
           path: path,
           tags: tag_info,
           title: title,
           album: album,
           artist: artist,
           year: year,
           genre: genre,
           picture: picture
         }}

      _other ->
        {:error, :bad_file}
    end
  rescue
    _ -> {:error, :bad_file}
  end

  defp parse_tag(<<
         "ID3",
         major_version::integer,
         _revision::integer,
         _unsynchronized::size(1),
         extended_header::size(1),
         _experimental::size(1),
         _footer::size(1),
         0::size(4),
         tag_size_synchsafe::binary-size(4),
         rest::binary
       >>) do
    tag_size = decode_synchsafe_integer(tag_size_synchsafe)

    {rest, ext_header_size} =
      if extended_header == 1 do
        skip_extended_header(major_version, rest)
      else
        {rest, 0}
      end

    parse_frames(major_version, rest, tag_size - ext_header_size, [])
  end

  defp parse_tag(<<
         _first::integer,
         _second::integer,
         _third::integer,
         rest::binary
       >>) do
    # has no ID3
    {%{}, rest}
  end

  defp parse_tag(_), do: {%{}, ""}

  defp decode_synchsafe_integer(<>), do: bin

  defp decode_synchsafe_integer(binary) do
    binary
    |> :binary.bin_to_list()
    |> Enum.reverse()
    |> Enum.with_index()
    |> Enum.reduce(0, fn {el, index}, acc -> acc ||| el <<< (index * 7) end)
  end

  defp skip_extended_header(3, <<
         ext_header_size::size(32),
         _flags::size(16),
         _padding_size::size(32),
         rest::binary
       >>) do
    remaining_ext_header_size = ext_header_size - 6
    <<_::binary-size(remaining_ext_header_size), rest::binary>> = rest
    {rest, ext_header_size}
  end

  defp skip_extended_header(4, <<
         ext_header_size_synchsafe::size(32),
         1::size(8),
         _flags::size(8),
         rest::binary
       >>) do
    ext_header_size = decode_synchsafe_integer(ext_header_size_synchsafe)
    remaining_ext_header_size = ext_header_size - 6
    <<_::binary-size(remaining_ext_header_size), rest::binary>> = rest
    {rest, ext_header_size}
  end

  defp parse_frames(_, data, tag_length_remaining, frames)
       when tag_length_remaining <= 0 do
    {Map.new(frames), data}
  end

  defp parse_frames(
         major_version,
         <<
           frame_id::binary-size(4),
           frame_size_maybe_synchsafe::binary-size(4),
           0::size(1),
           _tag_alter_preservation::size(1),
           _file_alter_preservation::size(1),
           _read_only::size(1),
           0::size(4),
           _grouping_identity::size(1),
           0::size(2),
           _compression::size(1),
           _encryption::size(1),
           _unsynchronized::size(1),
           _has_data_length_indicator::size(1),
           _unused::size(1),
           rest::binary
         >>,
         tag_length_remaining,
         frames
       ) do
    frame_size =
      case major_version do
        4 ->
          decode_synchsafe_integer(frame_size_maybe_synchsafe)

        3 ->
          <> = frame_size_maybe_synchsafe
          size
      end

    total_frame_size = frame_size + 10
    next_tag_length_remaining = tag_length_remaining - total_frame_size

    result = decode_frame(frame_id, frame_size, rest)

    case result do
      {nil, rest, :halt} ->
        {Map.new(frames), rest}

      {nil, rest, :cont} ->
        parse_frames(major_version, rest, next_tag_length_remaining, frames)

      {new_frame, rest} ->
        parse_frames(major_version, rest, next_tag_length_remaining, [new_frame | frames])
    end
  end

  defp parse_frames(_, data, _, frames) do
    {Map.new(frames), data}
  end

  defp decode_frame("TXXX", frame_size, <>) do
    {description, desc_size, rest} = decode_string(text_encoding, frame_size - 1, rest)
    {value, _, rest} = decode_string(text_encoding, frame_size - 1 - desc_size, rest)
    {{"TXXX", {description, value}}, rest}
  end

  defp decode_frame(
         "COMM",
         frame_size,
         <>
       ) do
    {short_desc, desc_size, rest} = decode_string(text_encoding, frame_size - 4, rest)
    {value, _, rest} = decode_string(text_encoding, frame_size - 4 - desc_size, rest)
    {{"COMM", {language, short_desc, value}}, rest}
  end

  defp decode_frame("APIC", frame_size, <>) do
    {mime_type, mime_len, rest} = decode_string(0, frame_size - 1, rest)

    <> = rest

    {description, desc_len, rest} =
      decode_string(text_encoding, frame_size - 1 - mime_len - 1, rest)

    image_data_size = frame_size - 1 - mime_len - 1 - desc_len
    {image_data, rest} = :erlang.split_binary(rest, image_data_size)

    {{"APIC", {mime_type, picture_type, description, image_data}}, rest}
  end

  defp decode_frame(id, frame_size, rest) do
    cond do
      Regex.match?(~r/^T[0-9A-Z]+$/, id) ->
        decode_text_frame(id, frame_size, rest)

      id in @declared_frame_ids ->
        <<_frame_data::binary-size(frame_size), rest::binary>> = rest
        {nil, rest, :cont}

      true ->
        {nil, rest, :halt}
    end
  end

  defp decode_text_frame(id, frame_size, <>) do
    {strs, rest} = decode_string_sequence(text_encoding, frame_size - 1, rest)
    {{id, strs}, rest}
  end

  defp decode_string_sequence(encoding, max_byte_size, data, acc \\ [])

  defp decode_string_sequence(_, max_byte_size, data, acc) when max_byte_size <= 0 do
    {Enum.reverse(acc), data}
  end

  defp decode_string_sequence(encoding, max_byte_size, data, acc) do
    {str, str_size, rest} = decode_string(encoding, max_byte_size, data)
    decode_string_sequence(encoding, max_byte_size - str_size, rest, [str | acc])
  end

  defp convert_string(encoding, str) when encoding in [0, 3] do
    str
  end

  defp convert_string(1, data) do
    {encoding, bom_length} = :unicode.bom_to_encoding(data)
    {_, string_data} = String.split_at(data, bom_length)
    :unicode.characters_to_binary(string_data, encoding)
  end

  defp convert_string(2, data) do
    :unicode.characters_to_binary(data, {:utf16, :big})
  end

  defp decode_string(encoding, max_byte_size, data) when encoding in [1, 2] do
    {str, rest} = get_double_null_terminated(data, max_byte_size)

    {convert_string(encoding, str), byte_size(str) + 2, rest}
  end

  defp decode_string(encoding, max_byte_size, data) when encoding in [0, 3] do
    case :binary.split(data, <<0>>) do
      [str, rest] when byte_size(str) + 1 <= max_byte_size ->
        {str, byte_size(str) + 1, rest}

      _ ->
        {str, rest} = :erlang.split_binary(data, max_byte_size)
        {str, max_byte_size, rest}
    end
  end

  defp get_double_null_terminated(data, max_byte_size, acc \\ [])

  defp get_double_null_terminated(rest, 0, acc) do
    {acc |> Enum.reverse() |> :binary.list_to_bin(), rest}
  end

  defp get_double_null_terminated(<<0, 0, rest::binary>>, _, acc) do
    {acc |> Enum.reverse() |> :binary.list_to_bin(), rest}
  end

  defp get_double_null_terminated(<>, max_byte_size, acc) do
    next_max_byte_size = max_byte_size - 2
    get_double_null_terminated(rest, next_max_byte_size, [b, a | acc])
  end

  defp parse_frame(
         <<
           0xFF::size(8),
           0b111::size(3),
           version_bits::size(2),
           layer_bits::size(2),
           _protected::size(1),
           bitrate_index::size(4),
           sampling_rate_index::size(2),
           padding::size(1),
           _private::size(1),
           _channel_mode_index::size(2),
           _mode_extension::size(2),
           _copyright::size(1),
           _original::size(1),
           _emphasis::size(2),
           _rest::binary
         >> = data,
         acc,
         frame_count,
         offset
       ) do
    with version when version != :invalid <- lookup_version(version_bits),
         layer when layer != :invalid <- lookup_layer(layer_bits),
         sampling_rate when sampling_rate != :invalid <-
           lookup_sampling_rate(version, sampling_rate_index),
         bitrate when bitrate != :invalid <- lookup_bitrate(version, layer, bitrate_index) do
      samples = lookup_samples_per_frame(version, layer)
      frame_size = get_frame_size(samples, layer, bitrate, sampling_rate, padding)
      frame_duration = samples / sampling_rate
      <<_skipped::binary-size(frame_size), rest::binary>> = data
      parse_frame(rest, acc + frame_duration, frame_count + 1, offset + frame_size)
    else
      _ ->
        <<_::size(8), rest::binary>> = data
        parse_frame(rest, acc, frame_count, offset + 1)
    end
  end

  defp parse_frame(<<_::size(8), rest::binary>>, acc, frame_count, offset) do
    parse_frame(rest, acc, frame_count, offset + 1)
  end

  defp parse_frame(<<>>, acc, _frame_count, _offset) do
    acc
  end

  defp lookup_version(0b00), do: :version25
  defp lookup_version(0b01), do: :invalid
  defp lookup_version(0b10), do: :version2
  defp lookup_version(0b11), do: :version1

  defp lookup_layer(0b00), do: :invalid
  defp lookup_layer(0b01), do: :layer3
  defp lookup_layer(0b10), do: :layer2
  defp lookup_layer(0b11), do: :layer1

  defp lookup_sampling_rate(_version, 0b11), do: :invalid
  defp lookup_sampling_rate(:version1, 0b00), do: 44100
  defp lookup_sampling_rate(:version1, 0b01), do: 48000
  defp lookup_sampling_rate(:version1, 0b10), do: 32000
  defp lookup_sampling_rate(:version2, 0b00), do: 22050
  defp lookup_sampling_rate(:version2, 0b01), do: 24000
  defp lookup_sampling_rate(:version2, 0b10), do: 16000
  defp lookup_sampling_rate(:version25, 0b00), do: 11025
  defp lookup_sampling_rate(:version25, 0b01), do: 12000
  defp lookup_sampling_rate(:version25, 0b10), do: 8000

  defp lookup_bitrate(_version, _layer, 0), do: :invalid
  defp lookup_bitrate(_version, _layer, 0xF), do: :invalid
  defp lookup_bitrate(:version1, :layer1, index), do: elem(@v1_l1_bitrates, index)
  defp lookup_bitrate(:version1, :layer2, index), do: elem(@v1_l2_bitrates, index)
  defp lookup_bitrate(:version1, :layer3, index), do: elem(@v1_l3_bitrates, index)

  defp lookup_bitrate(v, :layer1, index) when v in [:version2, :version25],
    do: elem(@v2_l1_bitrates, index)

  defp lookup_bitrate(v, l, index) when v in [:version2, :version25] and l in [:layer2, :layer3],
    do: elem(@v2_l2_l3_bitrates, index)

  defp lookup_samples_per_frame(:version1, :layer1), do: 384
  defp lookup_samples_per_frame(:version1, :layer2), do: 1152
  defp lookup_samples_per_frame(:version1, :layer3), do: 1152
  defp lookup_samples_per_frame(v, :layer1) when v in [:version2, :version25], do: 384
  defp lookup_samples_per_frame(v, :layer2) when v in [:version2, :version25], do: 1152
  defp lookup_samples_per_frame(v, :layer3) when v in [:version2, :version25], do: 576

  defp get_frame_size(samples, layer, kbps, sampling_rate, padding) do
    sample_duration = 1 / sampling_rate
    frame_duration = samples * sample_duration
    bytes_per_second = kbps * 1000 / 8
    size = floor(frame_duration * bytes_per_second)

    if padding == 1 do
      size + lookup_slot_size(layer)
    else
      size
    end
  end

  defp lookup_slot_size(:layer1), do: 4
  defp lookup_slot_size(l) when l in [:layer2, :layer3], do: 1
end

# file_name =
#   "/Users/jbrayton/Music/iTunes/iTunes Media/Music/Periphery/Juggernaut_ Alpha/1-01 A Black Minute.mp3"

# LiveBeats.MP3Stat.parse(file_name)

directory = "/Users/jbrayton/Music/iTunes/iTunes Media/Music"
directory = "/Users/jbrayton/Music/iTunes/iTunes Media/Music/Periphery"
files = "/**/**/*.mp3"

path = Path.join(directory, files)

Path.wildcard(path)
|> Enum.map(fn path ->
  # Read tags
  case LiveBeats.MP3Stat.parse(path) do
    {:ok, tag} -> tag
    {:error, _error} -> %{}
  end
end)
|> Enum.uniq_by(fn tag ->
  # Filter down to unique artist/album pairs
  # It would be quicker to use a MapSet to stay unique
  # ETS could work but overkill, why not benchmark it?
  if tag != %{} do
    tag.artist &amp;&amp; tag.album
  end
end)
|> Enum.sort_by(
  fn tag ->
    tag.artist
  end,
  :asc
)

# LiveBeats takes about 70+ seconds for everything on my machine
#   this is doing the extra work of determining the duration, which we don't need.
#   I could perform the surgery needed to strip that out.
# There are already issues with this though, like the year possibly being TYER or TDRC
# It's much easier to just use the Rust implementation, especially considering we may make that
#   our "collector".
# Elixir-native parsing isn't a terrible idea eventually, I just don't have the time as a capstone.