[WIP] getting retention methods in place

This commit is contained in:
Kieran Eglin 2024-04-03 09:15:54 -07:00
parent 5c909cfbfd
commit c696576850
No known key found for this signature in database
GPG key ID: 193984967FCF432D
7 changed files with 135 additions and 2 deletions

View file

@ -46,7 +46,13 @@ config :pinchflat, Oban,
engine: Oban.Engines.Lite, engine: Oban.Engines.Lite,
repo: Pinchflat.Repo, repo: Pinchflat.Repo,
# Keep old jobs for 30 days for display in the UI # Keep old jobs for 30 days for display in the UI
plugins: [{Oban.Plugins.Pruner, max_age: 30 * 24 * 60 * 60}], plugins: [
{Oban.Plugins.Pruner, max_age: 30 * 24 * 60 * 60},
{Oban.Plugins.Cron,
crontab: [
{"@midnight", Pinchflat.Downloading.MediaRetentionWorker}
]}
],
# TODO: consider making this an env var or something? # TODO: consider making this an env var or something?
queues: [ queues: [
default: 10, default: 10,

View file

@ -0,0 +1,21 @@
defmodule Pinchflat.Downloading.MediaRetentionWorker do
@moduledoc false
use Oban.Worker,
queue: :local_metadata,
unique: [period: :infinity, states: [:available, :scheduled, :retryable, :executing]],
tags: ["media_item", "local_metadata"]
require Logger
# TODO: docs
# TODO: test
# TODO: set to a 1 min interval to make sure that cron works
# TODO: update wiki after this is merged
# TODO: remove data backfill worker
# TODO: (other PR - not this one) add way to manually trigger an index of a source AND a pending media download
# of a source
@impl Oban.Worker
def perform(%Oban.Job{}) do
end
end

View file

@ -22,6 +22,22 @@ defmodule Pinchflat.Media do
Repo.all(MediaItem) Repo.all(MediaItem)
end end
# TODO: docs
# TODO: add attr for excluding media items from this list (ie: keep forever)
@doc """
Returns a list of media_items that are cullable based on the retention period
of the source they belong to.
Returns [%MediaItem{}, ...]
"""
def list_cullable_media_items do
MediaQuery.new()
|> MediaQuery.join_sources()
|> MediaQuery.with_media_filepath()
|> MediaQuery.with_passed_retention_period()
|> Repo.all()
end
@doc """ @doc """
Returns a list of pending media_items for a given source, where Returns a list of pending media_items for a given source, where
pending means the `media_filepath` is `nil` AND the media_item pending means the `media_filepath` is `nil` AND the media_item

View file

@ -44,7 +44,7 @@ defmodule Pinchflat.Media.MediaItem do
source_id source_id
upload_date upload_date
short_form_content short_form_content
)a )a
schema "media_items" do schema "media_items" do
# This is _not_ used as the primary key or internally in the database # This is _not_ used as the primary key or internally in the database

View file

@ -17,6 +17,7 @@ defmodule Pinchflat.Media.MediaQuery do
# Prefixes: # Prefixes:
# - for_* - belonging to a certain record # - for_* - belonging to a certain record
# - join_* - for joining on a certain record
# - with_* - for filtering based on full, concrete attributes # - with_* - for filtering based on full, concrete attributes
# - matching_* - for filtering based on partial attributes (e.g. LIKE, regex, full-text search) # - matching_* - for filtering based on partial attributes (e.g. LIKE, regex, full-text search)
# #
@ -31,6 +32,23 @@ defmodule Pinchflat.Media.MediaQuery do
where(query, [mi], mi.source_id == ^source.id) where(query, [mi], mi.source_id == ^source.id)
end end
def join_sources(query) do
from(mi in query, join: s in assoc(mi, :source), as: :sources)
end
def with_passed_retention_period(query) do
where(
query,
[mi, sources],
fragment(
"IFNULL(?, 0) > 0 AND DATETIME('now', '-' || ? || ' day') > ?",
sources.retention_period_days,
sources.retention_period_days,
mi.media_downloaded_at
)
)
end
def with_id(query, id) do def with_id(query, id) do
where(query, [mi], mi.id == ^id) where(query, [mi], mi.id == ^id)
end end

View file

@ -108,6 +108,7 @@ defmodule Pinchflat.Sources.Source do
|> dynamic_default(:custom_name, fn cs -> get_field(cs, :collection_name) end) |> dynamic_default(:custom_name, fn cs -> get_field(cs, :collection_name) end)
|> dynamic_default(:uuid, fn _ -> Ecto.UUID.generate() end) |> dynamic_default(:uuid, fn _ -> Ecto.UUID.generate() end)
|> validate_required(required_fields) |> validate_required(required_fields)
|> validate_number(:retention_period_days, greater_than_or_equal_to: 0)
|> cast_assoc(:metadata, with: &SourceMetadata.changeset/2, required: false) |> cast_assoc(:metadata, with: &SourceMetadata.changeset/2, required: false)
|> unique_constraint([:collection_id, :media_profile_id, :title_filter_regex], error_key: :original_url) |> unique_constraint([:collection_id, :media_profile_id, :title_filter_regex], error_key: :original_url)
end end

View file

@ -38,6 +38,77 @@ defmodule Pinchflat.MediaTest do
end end
end end
describe "list_cullable_media_items/0" do
test "returns media items where the source has a retention period" do
source_one = source_fixture(%{retention_period_days: 2})
source_two = source_fixture(%{retention_period_days: 0})
source_three = source_fixture(%{retention_period_days: nil})
_media_item =
media_item_fixture(%{
source_id: source_two.id,
media_filepath: "/video/#{Faker.File.file_name(:video)}",
media_downloaded_at: now_minus(3, :days)
})
_media_item =
media_item_fixture(%{
source_id: source_three.id,
media_filepath: "/video/#{Faker.File.file_name(:video)}",
media_downloaded_at: now_minus(3, :days)
})
expected_media_item =
media_item_fixture(%{
source_id: source_one.id,
media_filepath: "/video/#{Faker.File.file_name(:video)}",
media_downloaded_at: now_minus(3, :days)
})
assert Media.list_cullable_media_items() == [expected_media_item]
end
test "returns media_items with a media_filepath" do
source = source_fixture(%{retention_period_days: 2})
_media_item =
media_item_fixture(%{
source_id: source.id,
media_filepath: nil,
media_downloaded_at: now_minus(3, :days)
})
expected_media_item =
media_item_fixture(%{
source_id: source.id,
media_filepath: "/video/#{Faker.File.file_name(:video)}",
media_downloaded_at: now_minus(3, :days)
})
assert Media.list_cullable_media_items() == [expected_media_item]
end
test "returns items that have passed their retention period" do
source = source_fixture(%{retention_period_days: 2})
_media_item =
media_item_fixture(%{
source_id: source.id,
media_filepath: "/video/#{Faker.File.file_name(:video)}",
media_downloaded_at: now_minus(2, :days)
})
expected_media_item =
media_item_fixture(%{
source_id: source.id,
media_filepath: "/video/#{Faker.File.file_name(:video)}",
media_downloaded_at: now_minus(3, :days)
})
assert Media.list_cullable_media_items() == [expected_media_item]
end
end
describe "list_pending_media_items_for/1" do describe "list_pending_media_items_for/1" do
test "it returns pending without a filepath for a given source" do test "it returns pending without a filepath for a given source" do
source = source_fixture() source = source_fixture()