Search v1 (#19)

* Adds description to media items; hooks it up to indexing/media downloading

* Added a search method using postgres fulltext search

* Hooked up search functionality to the search form

* Added persistence to the search form when on search page
This commit is contained in:
Kieran 2024-02-13 17:46:14 -08:00 committed by GitHub
parent 850cf2db73
commit 58771ee37b
25 changed files with 261 additions and 25 deletions

View file

@ -47,6 +47,40 @@ defmodule Pinchflat.Media do
|> Repo.all()
end
@doc """
Returns a list of media_items that match the search term. Adds a `matching_search_term`
virtual field to the result set.
Returns [%MediaItem{}, ...].
"""
def search(search_term, opts \\ []) do
limit = Keyword.get(opts, :limit, 50)
from(mi in MediaItem,
where: fragment("searchable @@ websearch_to_tsquery(?)", ^search_term),
select_merge: %{
matching_search_term:
fragment(
"""
ts_headline(
'english',
CONCAT(title, ' ', description),
websearch_to_tsquery(?),
'StartSel=[PF_HIGHLIGHT],StopSel=[/PF_HIGHLIGHT]'
)
""",
^search_term
)
},
order_by: {
:desc,
fragment("ts_rank_cd(searchable, websearch_to_tsquery(?), 0)", ^search_term)
},
limit: ^limit
)
|> Repo.all()
end
@doc """
Gets a single media_item.

View file

@ -13,6 +13,7 @@ defmodule Pinchflat.Media.MediaItem do
@allowed_fields ~w(
title
media_id
description
original_url
livestream
media_downloaded_at
@ -27,6 +28,7 @@ defmodule Pinchflat.Media.MediaItem do
schema "media_items" do
field :title, :string
field :media_id, :string
field :description, :string
field :original_url, :string
field :livestream, :boolean, default: false
field :media_downloaded_at, :utc_datetime
@ -39,6 +41,8 @@ defmodule Pinchflat.Media.MediaItem do
# Will very likely revisit because I can't leave well-enough alone.
field :subtitle_filepaths, {:array, {:array, :string}}, default: []
field :matching_search_term, :string, virtual: true
belongs_to :source, Source
has_one :metadata, MediaMetadata, on_replace: :update

View file

@ -32,6 +32,7 @@ defmodule Pinchflat.MediaClient.Backends.YtDlp.MetadataParser do
defp parse_media_metadata(metadata) do
%{
title: metadata["title"],
description: metadata["description"],
media_filepath: metadata["filepath"]
}
end

View file

@ -14,8 +14,9 @@ defmodule Pinchflat.MediaClient.Backends.YtDlp.VideoCollection do
def get_media_attributes(url, command_opts \\ []) do
runner = Application.get_env(:pinchflat, :yt_dlp_runner)
opts = command_opts ++ [:simulate, :skip_download]
output_template = "%(.{id,title,was_live,original_url,description})j"
case runner.run(url, opts, "%(.{id,title,was_live,original_url})j") do
case runner.run(url, opts, output_template) do
{:ok, output} ->
output
|> String.split("\n", trim: true)

View file

@ -52,7 +52,8 @@ defmodule Pinchflat.Tasks.SourceTasks do
title: media_attrs["title"],
media_id: media_attrs["id"],
original_url: media_attrs["original_url"],
livestream: media_attrs["was_live"]
livestream: media_attrs["was_live"],
description: media_attrs["description"]
}
case Media.create_media_item(attrs) do

View file

@ -24,4 +24,21 @@ defmodule Pinchflat.Utils.StringUtils do
|> Base.encode16(case: :lower)
|> String.slice(0..(length - 1))
end
@doc """
Truncates a string to the given length and adds `...` if the string is longer than the given length.
Will break on a word boundary. Nothing happens if the string is shorter than the given length.
Returns binary()
"""
def truncate(string, length) do
if String.length(string) > length do
string
|> String.slice(0..(length - 1))
|> String.replace(~r/\s+\S*$/, "")
|> Kernel.<>("...")
else
string
end
end
end

View file

@ -89,6 +89,8 @@ defmodule PinchflatWeb do
import PinchflatWeb.CustomComponents.TableComponents
import PinchflatWeb.CustomComponents.ButtonComponents
alias Pinchflat.Utils.StringUtils
# Shortcut for generating JS commands
alias Phoenix.LiveView.JS

View file

@ -2,7 +2,7 @@
<.sidebar />
<div class="relative flex flex-1 flex-col overflow-y-auto overflow-x-hidden">
<.header />
<.header params={@conn.params} />
<main>
<div class="mx-auto max-w-screen-2xl p-4 md:p-6 2xl:p-10">
<.flash_group flash={@flash} />

View file

@ -1,5 +1,5 @@
<header class="sticky top-0 z-999 flex w-full bg-white drop-shadow-1 dark:bg-boxdark dark:drop-shadow-none">
<div class="flex flex-grow items-center justify-end px-4 py-4 shadow-2 md:px-6 2xl:px-11">
<div class="flex flex-grow items-center justify-between lg:justify-end px-4 py-4 shadow-2 md:px-6 2xl:px-11">
<div class="flex items-center gap-2 sm:gap-4 lg:hidden">
<button
class="z-99999 block rounded-sm border border-stroke bg-white p-1.5 shadow-sm dark:border-strokedark dark:bg-boxdark lg:hidden"
@ -7,22 +7,24 @@
>
<.icon name="hero-bars-3" />
</button>
<a class="block flex-shrink-0 lg:hidden" href="#">
<a class="hidden sm:block flex-shrink-0 lg:hidden" href="#">
<h2 class="text-title-md2 font-bold text-white">Pinchflat</h2>
</a>
</div>
<div class="hidden sm:block bg-meta-4 rounded-md">
<%!-- Aspirational (for now) --%>
<div class="bg-meta-4 rounded-md">
<div class="relative">
<span class="absolute left-2 top-1/2 -translate-y-1/2 flex">
<.icon name="hero-magnifying-glass" />
</span>
<input
type="text"
placeholder="Type to search..."
class="w-full bg-transparent pl-9 pr-4 border-0 focus:ring-0 focus:outline-none xl:w-125"
/>
<form action={~p"/search"}>
<input
type="text"
name="q"
value={@params["q"]}
placeholder="Type to search..."
class="w-full bg-transparent pl-9 pr-4 border-0 focus:ring-0 focus:outline-none xl:w-125"
/>
</form>
</div>
</div>
</div>

View file

@ -16,7 +16,7 @@
</.link>
</nav>
</div>
<div class="rounded-sm border border-stroke bg-white px-5 pb-2.5 pt-6 shadow-default dark:border-strokedark dark:bg-boxdark sm:px-7.5 xl:pb-1">
<div class="rounded-sm border border-stroke bg-white px-5 py-5 shadow-default dark:border-strokedark dark:bg-boxdark sm:px-7.5">
<div class="max-w-full overflow-x-auto">
<div class="flex flex-col gap-10 dark:text-white">
<h3 class="mt-14 font-bold text-xl">Relationships</h3>
@ -35,10 +35,10 @@
<.list_items_from_map map={Map.from_struct(@source)} />
<h3 class="font-bold text-xl">Downloaded Media</h3>
<%= if length(@downloaded_media) > 0 do %>
<%= if match?([_|_], @downloaded_media) do %>
<.table rows={@downloaded_media} table_class="text-black dark:text-white">
<:col :let={media_item} label="Title">
<%= String.slice(media_item.title, 0..50) %>...
<%= StringUtils.truncate(media_item.title, 50) %>
</:col>
<:col :let={media_item} label="" class="flex place-content-evenly">
<.link
@ -54,10 +54,10 @@
<% end %>
<h3 class="font-bold text-xl">Pending Media</h3>
<%= if length(@pending_media) > 0 do %>
<%= if match?([_|_], @pending_media) do %>
<.table rows={@pending_media} table_class="text-black dark:text-white">
<:col :let={media_item} label="Title">
<%= String.slice(media_item.title, 0..50) %>...
<%= StringUtils.truncate(media_item.title, 50) %>
</:col>
<:col :let={media_item} label="" class="flex place-content-evenly">
<.link

View file

@ -0,0 +1,12 @@
defmodule PinchflatWeb.Searches.SearchController do
use PinchflatWeb, :controller
alias Pinchflat.Media
def show(conn, params) do
search_term = Map.get(params, "q", "")
search_results = Media.search(search_term)
render(conn, :show, search_term: search_term, search_results: search_results)
end
end

View file

@ -0,0 +1,25 @@
defmodule PinchflatWeb.Searches.SearchHTML do
use PinchflatWeb, :html
embed_templates "search_html/*"
@doc """
Highlight search terms in a string of text based on `[PF_HIGHLIGHT]` and `[/PF_HIGHLIGHT]` tags
"""
attr :text, :string, required: true
def highlight_search_terms(assigns) do
split_string = String.split(assigns.text, ~r{\[PF_HIGHLIGHT\]|\[/PF_HIGHLIGHT\]}, include_captures: true)
assigns = assign(assigns, split_string: split_string)
~H"""
<%= for fragment <- @split_string do %>
<%= render_fragment(fragment) %>
<% end %>
"""
end
defp render_fragment("[PF_HIGHLIGHT]"), do: raw(~s(<span class="font-bold italic">))
defp render_fragment("[/PF_HIGHLIGHT]"), do: raw("</span>")
defp render_fragment(text), do: text
end

View file

@ -0,0 +1,29 @@
<div class="mb-6 flex gap-3 flex-row items-center justify-between">
<h2 class="text-title-md2 font-bold text-black dark:text-white">
<span class="hidden sm:inline">Search </span>Results for "<%= StringUtils.truncate(@search_term, 50) %>"
</h2>
</div>
<div class="rounded-sm border border-stroke bg-white px-5 py-5 shadow-default dark:border-strokedark dark:bg-boxdark sm:px-7.5">
<div class="max-w-full overflow-x-auto">
<div class="flex flex-col gap-10 dark:text-white">
<%= if match?([_|_], @search_results) do %>
<.table rows={@search_results} table_class="text-black dark:text-white">
<:col :let={result} label="Title">
<%= StringUtils.truncate(result.title, 40) %>
</:col>
<:col :let={result} label="Excerpt">
<.highlight_search_terms text={result.matching_search_term} />
</:col>
<:col :let={result} label="" class="flex place-content-evenly">
<.link navigate={~p"/media/#{result.id}"} class="hover:text-secondary duration-200 ease-in-out mx-0.5">
<.icon name="hero-eye" />
</.link>
</:col>
</.table>
<% else %>
<p class="font-bold text-lg text-center text-black dark:text-white">No results found</p>
<% end %>
</div>
</div>
</div>

View file

@ -21,6 +21,7 @@ defmodule PinchflatWeb.Router do
resources "/media_profiles", MediaProfiles.MediaProfileController
resources "/media", Media.MediaItemController, only: [:show]
resources "/search", Searches.SearchController, only: [:show], singleton: true
resources "/sources", MediaSources.SourceController do
resources "/media", Media.MediaItemController, only: [:show]

View file

@ -0,0 +1,9 @@
defmodule Pinchflat.Repo.Migrations.AddDescriptionToMediaItems do
use Ecto.Migration
def change do
alter table(:media_items) do
add :description, :text
end
end
end

View file

@ -0,0 +1,28 @@
defmodule Pinchflat.Repo.Migrations.AddSearchFieldToMediaItems do
use Ecto.Migration
def up do
execute """
ALTER TABLE media_items
ADD COLUMN searchable tsvector
GENERATED ALWAYS AS (
setweight(to_tsvector('english', coalesce(title, '')), 'A') ||
setweight(to_tsvector('english', coalesce(description, '')), 'B')
) STORED;
"""
execute """
CREATE INDEX media_items_searchable_idx ON media_items USING gin(searchable);
"""
end
def down do
execute """
DROP INDEX media_items_searchable_idx;
"""
alter table(:media_items) do
remove :searchable
end
end
end

View file

@ -36,6 +36,12 @@ defmodule Pinchflat.MediaClient.Backends.YtDlp.MediaParserTest do
assert result.title == "Trying to Wheelie Without the Rear Brake"
end
test "it extracts the description", %{metadata: metadata} do
result = Parser.parse_for_media_item(metadata)
assert is_binary(result.description)
end
test "it returns the metadata as a map", %{metadata: metadata} do
result = Parser.parse_for_media_item(metadata)

View file

@ -20,7 +20,7 @@ defmodule Pinchflat.MediaClient.Backends.YtDlp.VideoCollectionTest do
test "it passes the expected default args" do
expect(YtDlpRunnerMock, :run, fn _url, opts, ot ->
assert opts == [:simulate, :skip_download]
assert ot == "%(.{id,title,was_live,original_url})j"
assert ot == "%(.{id,title,was_live,original_url,description})j"
{:ok, ""}
end)

View file

@ -47,7 +47,7 @@ defmodule Pinchflat.MediaClient.SourceDetailsTest do
test "it passes the expected arguments to the backend" do
expect(YtDlpRunnerMock, :run, fn @channel_url, opts, ot ->
assert opts == [:simulate, :skip_download]
assert ot == "%(.{id,title,was_live,original_url})j"
assert ot == "%(.{id,title,was_live,original_url,description})j"
{:ok, ""}
end)

View file

@ -60,7 +60,7 @@ defmodule Pinchflat.MediaClient.VideoDownloaderTest do
test "it sets the media_downloaded_at", %{media_item: media_item} do
assert media_item.media_downloaded_at == nil
assert {:ok, updated_media_item} = VideoDownloader.download_for_media_item(media_item)
assert DateTime.diff(DateTime.utc_now(), updated_media_item.media_downloaded_at) < 1
assert DateTime.diff(DateTime.utc_now(), updated_media_item.media_downloaded_at) < 2
end
test "it extracts the title", %{media_item: media_item} do
@ -68,6 +68,11 @@ defmodule Pinchflat.MediaClient.VideoDownloaderTest do
assert updated_media_item.title == "Trying to Wheelie Without the Rear Brake"
end
test "it extracts the description", %{media_item: media_item} do
assert {:ok, updated_media_item} = VideoDownloader.download_for_media_item(media_item)
assert is_binary(updated_media_item.description)
end
test "it extracts the media_filepath", %{media_item: media_item} do
assert media_item.media_filepath == nil
assert {:ok, updated_media_item} = VideoDownloader.download_for_media_item(media_item)

View file

@ -183,6 +183,37 @@ defmodule Pinchflat.MediaTest do
end
end
describe "search/1" do
setup do
media_item =
media_item_fixture(%{
title: "The quick brown fox",
description: "jumps over the lazy dog"
})
{:ok, %{media_item_id: media_item.id}}
end
test "searches based on title", %{media_item_id: media_item_id} do
assert [%{id: ^media_item_id}] = Media.search("quick")
end
test "searches based on description", %{media_item_id: media_item_id} do
assert [%{id: ^media_item_id}] = Media.search("lazy")
end
test "adds a matching_search_term attribute with the relevant text" do
assert [res] = Media.search("quick")
assert String.contains?(res.matching_search_term, "The [PF_HIGHLIGHT]quick[/PF_HIGHLIGHT] brown fox")
end
test "optionally lets you specify a limit" do
media_item_fixture(%{title: "The small gray dog"})
assert [_] = Media.search("dog", limit: 1)
end
end
describe "get_media_item!/1" do
test "it returns the media_item with given id" do
media_item = media_item_fixture()

View file

@ -63,6 +63,7 @@ defmodule Pinchflat.Tasks.SourceTasksTest do
assert Enum.count(media_items) == 3
assert ["video1", "video2", "video3"] == Enum.map(media_items, & &1.media_id)
assert ["Video 1", "Video 2", "Video 3"] == Enum.map(media_items, & &1.title)
assert ["desc1", "desc2", "desc3"] == Enum.map(media_items, & &1.description)
assert Enum.all?(media_items, fn mi -> mi.original_url end)
assert Enum.all?(media_items, fn %MediaItem{} -> true end)
end
@ -110,7 +111,7 @@ defmodule Pinchflat.Tasks.SourceTasksTest do
SourceTasks.index_media_items(source)
source = Repo.reload!(source)
assert DateTime.diff(DateTime.utc_now(), source.last_indexed_at) < 1
assert DateTime.diff(DateTime.utc_now(), source.last_indexed_at) < 2
end
end

View file

@ -27,4 +27,18 @@ defmodule Pinchflat.Utils.StringUtilsTest do
assert String.length(StringUtils.random_string(64)) == 64
end
end
describe "truncate/2" do
test "truncates a string to the given length and adds ..." do
assert StringUtils.truncate("hello world", 5) == "hello..."
end
test "breaks on a word boundary" do
assert StringUtils.truncate("hello world", 7) == "hello..."
end
test "does not truncate a string shorter than the given length" do
assert StringUtils.truncate("hello", 10) == "hello"
end
end
end

View file

@ -0,0 +1,10 @@
defmodule PinchflatWeb.SearchControllerTest do
use PinchflatWeb.ConnCase
describe "show search" do
test "renders the page", %{conn: conn} do
conn = get(conn, ~p"/search")
assert html_response(conn, 200) =~ "Results"
end
end
end

View file

@ -36,19 +36,22 @@ defmodule Pinchflat.MediaSourceFixtures do
id: "video1",
title: "Video 1",
original_url: "https://example.com/video1",
was_live: false
was_live: false,
description: "desc1"
},
%{
id: "video2",
title: "Video 2",
original_url: "https://example.com/video2",
was_live: true
was_live: true,
description: "desc2"
},
%{
id: "video3",
title: "Video 3",
original_url: "https://example.com/video3",
was_live: false
was_live: false,
description: "desc3"
}
]