Refactored yt-dlp media module

This commit is contained in:
Kieran Eglin 2024-05-28 11:51:19 -07:00
parent 5b1c285670
commit 3b89a346ef
No known key found for this signature in database
GPG key ID: 193984967FCF432D
5 changed files with 84 additions and 34 deletions

View file

@ -51,7 +51,7 @@ defmodule Pinchflat.Media.MediaItem do
livestream
media_id
source_id
upload_date
uploaded_at
short_form_content
)a
@ -130,6 +130,7 @@ defmodule Pinchflat.Media.MediaItem do
~w(__meta__ __struct__ metadata tasks media_items_search_index)a
end
# TODO: refactor
defp update_upload_date_index(%{changes: changes} = changeset) when is_map_key(changes, :upload_date) do
source_id = get_field(changeset, :source_id)
source = Sources.get_source!(source_id)

View file

@ -95,7 +95,11 @@ defmodule Pinchflat.Metadata.MetadataFileHelpers do
def parse_upload_date(upload_date) do
<<year::binary-size(4)>> <> <<month::binary-size(2)>> <> <<day::binary-size(2)>> = upload_date
Date.from_iso8601!("#{year}-#{month}-#{day}")
# TODO: test new
case DateTime.from_iso8601("#{year}-#{month}-#{day}T00:00:00Z") do
{:ok, datetime, _} -> datetime
_ -> raise "Invalid upload date: #{upload_date}"
end
end
@doc """

View file

@ -38,6 +38,7 @@ defmodule Pinchflat.Metadata.NfoBuilder do
end
defp build_for_media_item(metadata) do
# TODO: see how this works post-change
upload_date = MetadataFileHelpers.parse_upload_date(metadata["upload_date"])
# Cribbed from a combination of the Kodi wiki, ytdl-nfo, and ytdl-sub.
# WHO NEEDS A FANCY XML PARSER ANYWAY?!

View file

@ -10,7 +10,7 @@ defmodule Pinchflat.YtDlp.Media do
:original_url,
:livestream,
:short_form_content,
:upload_date,
:uploaded_at,
:duration_seconds
]
@ -21,7 +21,7 @@ defmodule Pinchflat.YtDlp.Media do
:original_url,
:livestream,
:short_form_content,
:upload_date,
:uploaded_at,
:duration_seconds
]
@ -72,7 +72,7 @@ defmodule Pinchflat.YtDlp.Media do
Returns the output template for yt-dlp's indexing command.
"""
def indexing_output_template do
"%(.{id,title,was_live,webpage_url,description,aspect_ratio,duration,upload_date})j"
"%(.{id,title,was_live,webpage_url,description,aspect_ratio,duration,upload_date,timestamp})j"
end
@doc """
@ -90,7 +90,7 @@ defmodule Pinchflat.YtDlp.Media do
livestream: !!response["was_live"],
duration_seconds: response["duration"] && round(response["duration"]),
short_form_content: response["webpage_url"] && short_form_content?(response),
upload_date: response["upload_date"] && MetadataFileHelpers.parse_upload_date(response["upload_date"])
uploaded_at: parse_uploaded_at(response)
}
end
@ -110,6 +110,18 @@ defmodule Pinchflat.YtDlp.Media do
end
end
defp parse_uploaded_at(%{"timestamp" => ts} = response) when is_number(ts) do
case DateTime.from_unix(ts) do
{:ok, datetime} -> datetime
_ -> MetadataFileHelpers.parse_upload_date(response["upload_date"])
end
end
# This field is needed before inserting into the database, but absence
# of this field should fail at insert-time rather than here
defp parse_uploaded_at(%{"upload_date" => nil}), do: nil
defp parse_uploaded_at(response), do: MetadataFileHelpers.parse_upload_date(response["upload_date"])
defp backend_runner do
# This approach lets us mock the command for testing
Application.get_env(:pinchflat, :yt_dlp_runner)

View file

@ -79,7 +79,7 @@ defmodule Pinchflat.YtDlp.MediaTest do
describe "indexing_output_template/0" do
test "contains all the greatest hits" do
assert "%(.{id,title,was_live,webpage_url,description,aspect_ratio,duration,upload_date})j" ==
assert "%(.{id,title,was_live,webpage_url,description,aspect_ratio,duration,upload_date,timestamp})j" ==
Media.indexing_output_template()
end
end
@ -94,7 +94,8 @@ defmodule Pinchflat.YtDlp.MediaTest do
"was_live" => false,
"aspect_ratio" => 1.0,
"duration" => 60,
"upload_date" => "20210101"
"upload_date" => "20210101",
"timestamp" => 1_600_000_000
}
assert %Media{
@ -104,7 +105,7 @@ defmodule Pinchflat.YtDlp.MediaTest do
original_url: "https://www.youtube.com/watch?v=TiZPUDkDYbk",
livestream: false,
short_form_content: false,
upload_date: Date.from_iso8601!("2021-01-01"),
uploaded_at: ~U[2020-09-13 12:26:40Z],
duration_seconds: 60
} == Media.response_to_struct(response)
end
@ -146,34 +147,11 @@ defmodule Pinchflat.YtDlp.MediaTest do
response = %{
"webpage_url" => nil,
"aspect_ratio" => nil,
"duration" => nil
}
assert %Media{short_form_content: nil} = Media.response_to_struct(response)
end
test "parses the upload date" do
response = %{
"webpage_url" => "https://www.youtube.com/watch?v=TiZPUDkDYbk",
"aspect_ratio" => 1.0,
"duration" => 61,
"duration" => nil,
"upload_date" => "20210101"
}
expected_date = Date.from_iso8601!("2021-01-01")
assert %Media{upload_date: ^expected_date} = Media.response_to_struct(response)
end
test "doesn't blow up if upload date is missing" do
response = %{
"webpage_url" => "https://www.youtube.com/watch?v=TiZPUDkDYbk",
"aspect_ratio" => 1.0,
"duration" => 61,
"upload_date" => nil
}
assert %Media{upload_date: nil} = Media.response_to_struct(response)
assert %Media{short_form_content: nil} = Media.response_to_struct(response)
end
test "parses the duration" do
@ -209,4 +187,58 @@ defmodule Pinchflat.YtDlp.MediaTest do
assert %Media{livestream: false} = Media.response_to_struct(response)
end
end
describe "response_to_struct/1 when testing uploaded_at" do
test "parses the upload date from the timestamp if present" do
response = %{
"webpage_url" => "https://www.youtube.com/watch?v=TiZPUDkDYbk",
"aspect_ratio" => 1.0,
"duration" => 61,
"upload_date" => "20210101",
"timestamp" => 1_600_000_000
}
expected_date = ~U[2020-09-13 12:26:40Z]
assert %Media{uploaded_at: ^expected_date} = Media.response_to_struct(response)
end
test "parses the upload date from the uploaded_at if timestamp is present but nil" do
response = %{
"webpage_url" => "https://www.youtube.com/watch?v=TiZPUDkDYbk",
"aspect_ratio" => 1.0,
"duration" => 61,
"upload_date" => "20210101",
"timestamp" => nil
}
expected_date = ~U[2021-01-01 00:00:00Z]
assert %Media{uploaded_at: ^expected_date} = Media.response_to_struct(response)
end
test "parses the upload date from the uploaded_at if timestamp absent" do
response = %{
"webpage_url" => "https://www.youtube.com/watch?v=TiZPUDkDYbk",
"aspect_ratio" => 1.0,
"duration" => 61,
"upload_date" => "20210101"
}
expected_date = ~U[2021-01-01 00:00:00Z]
assert %Media{uploaded_at: ^expected_date} = Media.response_to_struct(response)
end
test "doesn't blow up if upload date is missing" do
response = %{
"webpage_url" => "https://www.youtube.com/watch?v=TiZPUDkDYbk",
"aspect_ratio" => 1.0,
"duration" => 61,
"upload_date" => nil
}
assert %Media{uploaded_at: nil} = Media.response_to_struct(response)
end
end
end