hj/pleroma
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Merge branch 'develop' into fix-muted-web-push

Browse source
This commit is contained in:
Mark Felder 2024-05-11 08:06:04 -04:00
commit 16c72d0701
46 changed files with 1096 additions and 619 deletions
Download patch file Download diff file Expand all files Collapse all files

1
lib/pleroma/application.ex
View file

@ -156,6 +156,7 @@ defmodule Pleroma.Application do
build_cachex("web_resp", limit: 2500),
build_cachex("emoji_packs", expiration: emoji_packs_expiration(), limit: 10),
build_cachex("failed_proxy_url", limit: 2500),
build_cachex("failed_media_helper_url", default_ttl: :timer.minutes(15), limit: 2_500),
build_cachex("banned_urls", default_ttl: :timer.hours(24 * 30), limit: 5_000),
build_cachex("chat_message_id_idempotency_key",
expiration: chat_message_id_idempotency_key_expiration(),

49
lib/pleroma/helpers/media_helper.ex
View file

@ -12,6 +12,8 @@ defmodule Pleroma.Helpers.MediaHelper do
require Logger
@cachex Pleroma.Config.get([:cachex, :provider], Cachex)
def missing_dependencies do
Enum.reduce([ffmpeg: "ffmpeg"], [], fn {sym, executable}, acc ->
if Pleroma.Utils.command_available?(executable) do
@ -43,29 +45,40 @@ defmodule Pleroma.Helpers.MediaHelper do
@spec video_framegrab(String.t()) :: {:ok, binary()} | {:error, any()}
def video_framegrab(url) do
with executable when is_binary(executable) <- System.find_executable("ffmpeg"),
false <- @cachex.exists?(:failed_media_helper_cache, url),
{:ok, env} <- HTTP.get(url, [], pool: :media),
{:ok, pid} <- StringIO.open(env.body) do
body_stream = IO.binstream(pid, 1)
result =
Exile.stream!(
[
executable,
"-i",
"pipe:0",
"-vframes",
"1",
"-f",
"mjpeg",
"pipe:1"
],
input: body_stream,
ignore_epipe: true,
stderr: :disable
)
|> Enum.into(<<>>)
task =
Task.async(fn ->
Exile.stream!(
[
executable,
"-i",
"pipe:0",
"-vframes",
"1",
"-f",
"mjpeg",
"pipe:1"
],
input: body_stream,
ignore_epipe: true,
stderr: :disable
)
|> Enum.into(<<>>)
end)
{:ok, result}
case Task.yield(task, 5_000) do
nil ->
Task.shutdown(task)
@cachex.put(:failed_media_helper_cache, url, nil)
{:error, {:ffmpeg, :timeout}}
result ->
{:ok, result}
end
else
nil -> {:error, {:ffmpeg, :command_not_found}}
{:error, _} = error -> error

20
lib/pleroma/html.ex
View file

@ -65,20 +65,16 @@ defmodule Pleroma.HTML do
end
end
@spec extract_first_external_url_from_object(Pleroma.Object.t()) ::
{:ok, String.t()} | {:error, :no_content}
@spec extract_first_external_url_from_object(Pleroma.Object.t()) :: String.t() | nil
def extract_first_external_url_from_object(%{data: %{"content" => content}})
when is_binary(content) do
url =
content
|> Floki.parse_fragment!()
|> Floki.find("a:not(.mention,.hashtag,.attachment,[rel~=\"tag\"])")
|> Enum.take(1)
|> Floki.attribute("href")
|> Enum.at(0)
{:ok, url}
content
|> Floki.parse_fragment!()
|> Floki.find("a:not(.mention,.hashtag,.attachment,[rel~=\"tag\"])")
|> Enum.take(1)
|> Floki.attribute("href")
|> Enum.at(0)
end
def extract_first_external_url_from_object(_), do: {:error, :no_content}
def extract_first_external_url_from_object(_), do: nil
end

6
lib/pleroma/web/activity_pub/activity_pub.ex
View file

@ -147,9 +147,7 @@ defmodule Pleroma.Web.ActivityPub.ActivityPub do
# Splice in the child object if we have one.
activity = Maps.put_if_present(activity, :object, object)
ConcurrentLimiter.limit(Pleroma.Web.RichMedia.Helpers, fn ->
Task.start(fn -> Pleroma.Web.RichMedia.Helpers.fetch_data_for_activity(activity) end)
end)
Pleroma.Web.RichMedia.Card.get_by_activity(activity)
# Add local posts to search index
if local, do: Pleroma.Search.add_to_index(activity)
@ -177,7 +175,7 @@ defmodule Pleroma.Web.ActivityPub.ActivityPub do
id: "pleroma:fakeid"
}
Pleroma.Web.RichMedia.Helpers.fetch_data_for_activity(activity)
Pleroma.Web.RichMedia.Card.get_by_activity(activity)
{:ok, activity}
{:remote_limit_pass, _} ->

4
lib/pleroma/web/activity_pub/side_effects.ex
View file

@ -230,9 +230,7 @@ defmodule Pleroma.Web.ActivityPub.SideEffects do
end
end
ConcurrentLimiter.limit(Pleroma.Web.RichMedia.Helpers, fn ->
Task.start(fn -> Pleroma.Web.RichMedia.Helpers.fetch_data_for_activity(activity) end)
end)
Pleroma.Web.RichMedia.Card.get_by_activity(activity)
Pleroma.Search.add_to_index(Map.put(activity, :object, object))

12
lib/pleroma/web/activity_pub/views/user_view.ex
View file

@ -67,8 +67,13 @@ defmodule Pleroma.Web.ActivityPub.UserView do
def render("user.json", %{user: %User{nickname: nil} = user}),
do: render("service.json", %{user: user})
def render("user.json", %{user: %User{nickname: "internal." <> _} = user}),
do: render("service.json", %{user: user}) |> Map.put("preferredUsername", user.nickname)
def render("user.json", %{user: %User{nickname: "internal." <> _} = user}) do
render("service.json", %{user: user})
|> Map.merge(%{
"preferredUsername" => user.nickname,
"webfinger" => "acct:#{User.full_nickname(user)}"
})
end
def render("user.json", %{user: user}) do
{:ok, _, public_key} = Keys.keys_from_pem(user.keys)
@ -121,7 +126,8 @@ defmodule Pleroma.Web.ActivityPub.UserView do
"discoverable" => user.is_discoverable,
"capabilities" => capabilities,
"alsoKnownAs" => user.also_known_as,
"vcard:bday" => birthday
"vcard:bday" => birthday,
"webfinger" => "acct:#{User.full_nickname(user)}"
}
|> Map.merge(maybe_make_image(&User.avatar_url/2, "icon", user))
|> Map.merge(maybe_make_image(&User.banner_url/2, "image", user))

4
lib/pleroma/web/api_spec/schemas/status.ex
View file

@ -58,6 +58,10 @@ defmodule Pleroma.Web.ApiSpec.Schemas.Status do
format: :uri,
description: "Preview thumbnail"
},
image_description: %Schema{
type: :string,
description: "Alternate text that describes what is in the thumbnail"
},
title: %Schema{type: :string, description: "Title of linked resource"},
description: %Schema{type: :string, description: "Description of preview"}
}

16
lib/pleroma/web/mastodon_api/controllers/status_controller.ex
View file

@ -38,7 +38,6 @@ defmodule Pleroma.Web.MastodonAPI.StatusController do
when action in [
:index,
:show,
:card,
:context,
:show_history,
:show_source
@ -473,21 +472,6 @@ defmodule Pleroma.Web.MastodonAPI.StatusController do
end
end
@doc "GET /api/v1/statuses/:id/card"
@deprecated "https://github.com/tootsuite/mastodon/pull/11213"
def card(
%{assigns: %{user: user}, private: %{open_api_spex: %{params: %{id: status_id}}}} = conn,
_
) do
with %Activity{} = activity <- Activity.get_by_id(status_id),
true <- Visibility.visible_for_user?(activity, user) do
data = Pleroma.Web.RichMedia.Helpers.fetch_data_for_activity(activity)
render(conn, "card.json", data)
else
_ -> render_error(conn, :not_found, "Record not found")
end
end
@doc "GET /api/v1/statuses/:id/favourited_by"
def favourited_by(
%{assigns: %{user: user}, private: %{open_api_spex: %{params: %{id: id}}}} = conn,

27
lib/pleroma/web/mastodon_api/views/status_view.ex
View file

@ -21,6 +21,7 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
alias Pleroma.Web.MastodonAPI.StatusView
alias Pleroma.Web.MediaProxy
alias Pleroma.Web.PleromaAPI.EmojiReactionController
alias Pleroma.Web.RichMedia.Card
import Pleroma.Web.ActivityPub.Visibility, only: [get_visibility: 1, visible_for_user?: 2]
@ -29,9 +30,7 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
# pagination is restricted to 40 activities at a time
defp fetch_rich_media_for_activities(activities) do
Enum.each(activities, fn activity ->
spawn(fn ->
Pleroma.Web.RichMedia.Helpers.fetch_data_for_activity(activity)
end)
spawn(fn -> Card.get_by_activity(activity) end)
end)
end
@ -113,9 +112,7 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
# To do: check AdminAPIControllerTest on the reasons behind nil activities in the list
activities = Enum.filter(opts.activities, & &1)
# Start fetching rich media before doing anything else, so that later calls to get the cards
# only block for timeout in the worst case, as opposed to
# length(activities_with_links) * timeout
# Start prefetching rich media before doing anything else
fetch_rich_media_for_activities(activities)
replied_to_activities = get_replied_to_activities(activities)
quoted_activities = get_quoted_activities(activities)
@ -364,7 +361,11 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
summary = object.data["summary"] || ""
card = render("card.json", Pleroma.Web.RichMedia.Helpers.fetch_data_for_activity(activity))
card =
case Card.get_by_activity(activity) do
%Card{} = result -> render("card.json", result)
_ -> nil
end
url =
if user.local do
@ -567,15 +568,8 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
}
end
def render("card.json", %{rich_media: rich_media, page_url: page_url}) do
page_url_data = URI.parse(page_url)
page_url_data =
if is_binary(rich_media["url"]) do
URI.merge(page_url_data, URI.parse(rich_media["url"]))
else
page_url_data
end
def render("card.json", %Card{fields: rich_media}) do
page_url_data = URI.parse(rich_media["url"])
page_url = page_url_data |> to_string
@ -589,6 +583,7 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
provider_url: page_url_data.scheme <> "://" <> page_url_data.host,
url: page_url,
image: image_url,
image_description: rich_media["image:alt"] || "",
title: rich_media["title"] || "",
description: rich_media["description"] || "",
pleroma: %{

14
lib/pleroma/web/pleroma_api/views/bookmark_folder_view.ex
View file

@ -13,10 +13,8 @@ defmodule Pleroma.Web.PleromaAPI.BookmarkFolderView do
%{
id: folder.id |> to_string(),
name: folder.name,
emoji: get_emoji(folder.emoji),
source: %{
emoji: folder.emoji
}
emoji: folder.emoji,
emoji_url: get_emoji_url(folder.emoji)
}
end
@ -24,18 +22,18 @@ defmodule Pleroma.Web.PleromaAPI.BookmarkFolderView do
render_many(folders, __MODULE__, "show.json", Map.delete(opts, :folders))
end
defp get_emoji(nil) do
defp get_emoji_url(nil) do
nil
end
defp get_emoji(emoji) do
defp get_emoji_url(emoji) do
if Emoji.unicode?(emoji) do
emoji
nil
else
emoji = Emoji.get(emoji)
if emoji != nil do
Endpoint.url() |> URI.merge(emoji.relative_url) |> to_string()
Endpoint.url() |> URI.merge(emoji.file) |> to_string()
else
nil
end

13
lib/pleroma/web/pleroma_api/views/chat/message_reference_view.ex
View file

@ -9,6 +9,7 @@ defmodule Pleroma.Web.PleromaAPI.Chat.MessageReferenceView do
alias Pleroma.User
alias Pleroma.Web.CommonAPI.Utils
alias Pleroma.Web.MastodonAPI.StatusView
alias Pleroma.Web.RichMedia.Card
@cachex Pleroma.Config.get([:cachex, :provider], Cachex)
@ -23,6 +24,12 @@ defmodule Pleroma.Web.PleromaAPI.Chat.MessageReferenceView do
}
}
) do
card =
case Card.get_by_object(object) do
%Card{} = card_data -> StatusView.render("card.json", card_data)
_ -> nil
end
%{
id: id |> to_string(),
content: chat_message["content"],
@ -34,11 +41,7 @@ defmodule Pleroma.Web.PleromaAPI.Chat.MessageReferenceView do
chat_message["attachment"] &&
StatusView.render("attachment.json", attachment: chat_message["attachment"]),
unread: unread,
card:
StatusView.render(
"card.json",
Pleroma.Web.RichMedia.Helpers.fetch_data_for_object(object)
)
card: card
}
|> put_idempotency_key()
end

101
lib/pleroma/web/rich_media/backfill.ex Normal file
View file

@ -0,0 +1,101 @@
# Pleroma: A lightweight social networking server
# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Backfill do
alias Pleroma.Web.RichMedia.Card
alias Pleroma.Web.RichMedia.Parser
alias Pleroma.Web.RichMedia.Parser.TTL
alias Pleroma.Workers.RichMediaExpirationWorker
require Logger
@backfiller Pleroma.Config.get([__MODULE__, :provider], Pleroma.Web.RichMedia.Backfill.Task)
@cachex Pleroma.Config.get([:cachex, :provider], Cachex)
@max_attempts 3
@retry 5_000
def start(%{url: url} = args) when is_binary(url) do
url_hash = Card.url_to_hash(url)
args =
args
|> Map.put(:attempt, 1)
|> Map.put(:url_hash, url_hash)
@backfiller.run(args)
end
def run(%{url: url, url_hash: url_hash, attempt: attempt} = args)
when attempt <= @max_attempts do
case Parser.parse(url) do
{:ok, fields} ->
{:ok, card} = Card.create(url, fields)
maybe_schedule_expiration(url, fields)
if Map.has_key?(args, :activity_id) do
stream_update(args)
end
warm_cache(url_hash, card)
{:error, {:invalid_metadata, fields}} ->
Logger.debug("Rich media incomplete or invalid metadata for #{url}: #{inspect(fields)}")
negative_cache(url_hash)
{:error, :body_too_large} ->
Logger.error("Rich media error for #{url}: :body_too_large")
negative_cache(url_hash)
{:error, {:content_type, type}} ->
Logger.debug("Rich media error for #{url}: :content_type is #{type}")
negative_cache(url_hash)
e ->
Logger.debug("Rich media error for #{url}: #{inspect(e)}")
:timer.sleep(@retry * attempt)
run(%{args | attempt: attempt + 1})
end
end
def run(%{url: url, url_hash: url_hash}) do
Logger.debug("Rich media failure for #{url}")
negative_cache(url_hash, :timer.minutes(15))
end
defp maybe_schedule_expiration(url, fields) do
case TTL.process(fields, url) do
{:ok, ttl} when is_number(ttl) ->
timestamp = DateTime.from_unix!(ttl)
RichMediaExpirationWorker.new(%{"url" => url}, scheduled_at: timestamp)
|> Oban.insert()
_ ->
:ok
end
end
defp stream_update(%{activity_id: activity_id}) do
Pleroma.Activity.get_by_id(activity_id)
|> Pleroma.Activity.normalize()
|> Pleroma.Web.ActivityPub.ActivityPub.stream_out()
end
defp warm_cache(key, val), do: @cachex.put(:rich_media_cache, key, val)
defp negative_cache(key, ttl \\ nil), do: @cachex.put(:rich_media_cache, key, nil, ttl: ttl)
end
defmodule Pleroma.Web.RichMedia.Backfill.Task do
alias Pleroma.Web.RichMedia.Backfill
def run(args) do
Task.Supervisor.start_child(Pleroma.TaskSupervisor, Backfill, :run, [args],
name: {:global, {:rich_media, args.url_hash}}
)
end
end

157
lib/pleroma/web/rich_media/card.ex Normal file
View file

@ -0,0 +1,157 @@
defmodule Pleroma.Web.RichMedia.Card do
use Ecto.Schema
import Ecto.Changeset
import Ecto.Query
alias Pleroma.Activity
alias Pleroma.HTML
alias Pleroma.Object
alias Pleroma.Repo
alias Pleroma.Web.RichMedia.Backfill
alias Pleroma.Web.RichMedia.Parser
@cachex Pleroma.Config.get([:cachex, :provider], Cachex)
@config_impl Application.compile_env(:pleroma, [__MODULE__, :config_impl], Pleroma.Config)
@type t :: %__MODULE__{}
schema "rich_media_card" do
field(:url_hash, :binary)
field(:fields, :map)
timestamps()
end
@doc false
def changeset(card, attrs) do
card
|> cast(attrs, [:url_hash, :fields])
|> validate_required([:url_hash, :fields])
|> unique_constraint(:url_hash)
end
@spec create(String.t(), map()) :: {:ok, t()}
def create(url, fields) do
url_hash = url_to_hash(url)
fields = Map.put_new(fields, "url", url)
%__MODULE__{}
|> changeset(%{url_hash: url_hash, fields: fields})
|> Repo.insert(on_conflict: {:replace, [:fields]}, conflict_target: :url_hash)
end
@spec delete(String.t()) :: {:ok, Ecto.Schema.t()} | {:error, Ecto.Changeset.t()} | :ok
def delete(url) do
url_hash = url_to_hash(url)
@cachex.del(:rich_media_cache, url_hash)
case get_by_url(url) do
%__MODULE__{} = card -> Repo.delete(card)
nil -> :ok
end
end
@spec get_by_url(String.t() | nil) :: t() | nil | :error
def get_by_url(url) when is_binary(url) do
if @config_impl.get([:rich_media, :enabled]) do
url_hash = url_to_hash(url)
@cachex.fetch!(:rich_media_cache, url_hash, fn _ ->
result =
__MODULE__
|> where(url_hash: ^url_hash)
|> Repo.one()
case result do
%__MODULE__{} = card -> {:commit, card}
_ -> {:ignore, nil}
end
end)
else
:error
end
end
def get_by_url(nil), do: nil
@spec get_or_backfill_by_url(String.t(), map()) :: t() | nil
def get_or_backfill_by_url(url, backfill_opts \\ %{}) do
case get_by_url(url) do
%__MODULE__{} = card ->
card
nil ->
backfill_opts = Map.put(backfill_opts, :url, url)
Backfill.start(backfill_opts)
nil
:error ->
nil
end
end
@spec get_by_object(Object.t()) :: t() | nil | :error
def get_by_object(object) do
case HTML.extract_first_external_url_from_object(object) do
nil -> nil
url -> get_or_backfill_by_url(url)
end
end
@spec get_by_activity(Activity.t()) :: t() | nil | :error
# Fake/Draft activity
def get_by_activity(%Activity{id: "pleroma:fakeid"} = activity) do
with %Object{} = object <- Object.normalize(activity, fetch: false),
url when not is_nil(url) <- HTML.extract_first_external_url_from_object(object) do
case get_by_url(url) do
# Cache hit
%__MODULE__{} = card ->
card
# Cache miss, but fetch for rendering the Draft
_ ->
with {:ok, fields} <- Parser.parse(url),
{:ok, card} <- create(url, fields) do
card
else
_ -> nil
end
end
else
_ ->
nil
end
end
def get_by_activity(activity) do
with %Object{} = object <- Object.normalize(activity, fetch: false),
{_, nil} <- {:cached, get_cached_url(object, activity.id)} do
nil
else
{:cached, url} ->
get_or_backfill_by_url(url, %{activity_id: activity.id})
_ ->
:error
end
end
@spec url_to_hash(String.t()) :: String.t()
def url_to_hash(url) do
:crypto.hash(:sha256, url) |> Base.encode16(case: :lower)
end
defp get_cached_url(object, activity_id) do
key = "URL|#{activity_id}"
@cachex.fetch!(:scrubber_cache, key, fn _ ->
url = HTML.extract_first_external_url_from_object(object)
Activity.HTML.add_cache_key_for(activity_id, key)
{:commit, url}
end)
end
end

70
lib/pleroma/web/rich_media/helpers.ex
View file

@ -3,65 +3,13 @@
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Helpers do
alias Pleroma.Activity
alias Pleroma.HTML
alias Pleroma.Object
alias Pleroma.Web.RichMedia.Parser
@cachex Pleroma.Config.get([:cachex, :provider], Cachex)
@config_impl Application.compile_env(:pleroma, [__MODULE__, :config_impl], Pleroma.Config)
@options [
pool: :media,
max_body: 2_000_000,
recv_timeout: 2_000
]
def fetch_data_for_object(object) do
with true <- @config_impl.get([:rich_media, :enabled]),
{:ok, page_url} <-
HTML.extract_first_external_url_from_object(object),
{:ok, rich_media} <- Parser.parse(page_url) do
%{page_url: page_url, rich_media: rich_media}
else
_ -> %{}
end
end
def fetch_data_for_activity(%Activity{data: %{"type" => "Create"}} = activity) do
with true <- @config_impl.get([:rich_media, :enabled]),
%Object{} = object <- Object.normalize(activity, fetch: false) do
if object.data["fake"] do
fetch_data_for_object(object)
else
key = "URL|#{activity.id}"
@cachex.fetch!(:scrubber_cache, key, fn _ ->
result = fetch_data_for_object(object)
cond do
match?(%{page_url: _, rich_media: _}, result) ->
Activity.HTML.add_cache_key_for(activity.id, key)
{:commit, result}
true ->
{:ignore, %{}}
end
end)
end
else
_ -> %{}
end
end
def fetch_data_for_activity(_), do: %{}
alias Pleroma.Config
def rich_media_get(url) do
headers = [{"user-agent", Pleroma.Application.user_agent() <> "; Bot"}]
head_check =
case Pleroma.HTTP.head(url, headers, @options) do
case Pleroma.HTTP.head(url, headers, http_options()) do
# If the HEAD request didn't reach the server for whatever reason,
# we assume the GET that comes right after won't either
{:error, _} = e ->
@ -76,7 +24,7 @@ defmodule Pleroma.Web.RichMedia.Helpers do
:ok
end
with :ok <- head_check, do: Pleroma.HTTP.get(url, headers, @options)
with :ok <- head_check, do: Pleroma.HTTP.get(url, headers, http_options())
end
defp check_content_type(headers) do
@ -92,12 +40,13 @@ defmodule Pleroma.Web.RichMedia.Helpers do
end
end
@max_body @options[:max_body]
defp check_content_length(headers) do
max_body = Keyword.get(http_options(), :max_body)
case List.keyfind(headers, "content-length", 0) do
{_, maybe_content_length} ->
case Integer.parse(maybe_content_length) do
{content_length, ""} when content_length <= @max_body -> :ok
{content_length, ""} when content_length <= max_body -> :ok
{_, ""} -> {:error, :body_too_large}
_ -> :ok
end
@ -106,4 +55,11 @@ defmodule Pleroma.Web.RichMedia.Helpers do
:ok
end
end
defp http_options do
[
pool: :media,
max_body: Config.get([:rich_media, :max_body], 5_000_000)
]
end
end

114
lib/pleroma/web/rich_media/parser.ex
View file

@ -5,134 +5,28 @@
defmodule Pleroma.Web.RichMedia.Parser do
require Logger
@cachex Pleroma.Config.get([:cachex, :provider], Cachex)
@config_impl Application.compile_env(:pleroma, [__MODULE__, :config_impl], Pleroma.Config)
defp parsers do
Pleroma.Config.get([:rich_media, :parsers])
end
def parse(nil), do: {:error, "No URL provided"}
def parse(nil), do: nil
@spec parse(String.t()) :: {:ok, map()} | {:error, any()}
def parse(url) do
with :ok <- validate_page_url(url),
{:ok, data} <- get_cached_or_parse(url),
{:ok, _} <- set_ttl_based_on_image(data, url) do
{:ok, data} <- parse_url(url) do
data = Map.put(data, "url", url)
{:ok, data}
end
end
defp get_cached_or_parse(url) do
case @cachex.fetch(:rich_media_cache, url, fn ->
case parse_url(url) do
{:ok, _} = res ->
{:commit, res}
{:error, reason} = e ->
# Unfortunately we have to log errors here, instead of doing that
# along with ttl setting at the bottom. Otherwise we can get log spam
# if more than one process was waiting for the rich media card
# while it was generated. Ideally we would set ttl here as well,
# so we don't override it number_of_waiters_on_generation
# times, but one, obviously, can't set ttl for not-yet-created entry
# and Cachex doesn't support returning ttl from the fetch callback.
log_error(url, reason)
{:commit, e}
end
end) do
{action, res} when action in [:commit, :ok] ->
case res do
{:ok, _data} = res ->
res
{:error, reason} = e ->
if action == :commit, do: set_error_ttl(url, reason)
e
end
{:error, e} ->
{:error, {:cachex_error, e}}
end
end
defp set_error_ttl(_url, :body_too_large), do: :ok
defp set_error_ttl(_url, {:content_type, _}), do: :ok
# The TTL is not set for the errors above, since they are unlikely to change
# with time
defp set_error_ttl(url, _reason) do
ttl = Pleroma.Config.get([:rich_media, :failure_backoff], 60_000)
@cachex.expire(:rich_media_cache, url, ttl)
:ok
end
defp log_error(url, {:invalid_metadata, data}) do
Logger.debug(fn -> "Incomplete or invalid metadata for #{url}: #{inspect(data)}" end)
end
defp log_error(url, reason) do
Logger.warning(fn -> "Rich media error for #{url}: #{inspect(reason)}" end)
end
@doc """
Set the rich media cache based on the expiration time of image.
Adopt behaviour `Pleroma.Web.RichMedia.Parser.TTL`
## Example
defmodule MyModule do
@behaviour Pleroma.Web.RichMedia.Parser.TTL
def ttl(data, url) do
image_url = Map.get(data, :image)
# do some parsing in the url and get the ttl of the image
# and return ttl is unix time
parse_ttl_from_url(image_url)
end
end
Define the module in the config
config :pleroma, :rich_media,
ttl_setters: [MyModule]
"""
@spec set_ttl_based_on_image(map(), String.t()) ::
{:ok, integer() | :noop} | {:error, :no_key}
def set_ttl_based_on_image(data, url) do
case get_ttl_from_image(data, url) do
ttl when is_number(ttl) ->
ttl = ttl * 1000
case @cachex.expire_at(:rich_media_cache, url, ttl) do
{:ok, true} -> {:ok, ttl}
{:ok, false} -> {:error, :no_key}
end
_ ->
{:ok, :noop}
end
end
defp get_ttl_from_image(data, url) do
[:rich_media, :ttl_setters]
|> Pleroma.Config.get()
|> Enum.reduce({:ok, nil}, fn
module, {:ok, _ttl} ->
module.ttl(data, url)
_, error ->
error
end)
end
def parse_url(url) do
defp parse_url(url) do
with {:ok, %Tesla.Env{body: html}} <- Pleroma.Web.RichMedia.Helpers.rich_media_get(url),
{:ok, html} <- Floki.parse_document(html) do
html
|> maybe_parse()
|> Map.put("url", url)
|> clean_parsed_data()
|> check_parsed_data()
end

13
lib/pleroma/web/rich_media/parser/ttl.ex
View file

@ -4,4 +4,17 @@
defmodule Pleroma.Web.RichMedia.Parser.TTL do
@callback ttl(map(), String.t()) :: integer() | nil
@spec process(map(), String.t()) :: {:ok, integer() | nil}
def process(data, url) do
[:rich_media, :ttl_setters]
|> Pleroma.Config.get()
|> Enum.reduce_while({:ok, nil}, fn
module, acc ->
case module.ttl(data, url) do
ttl when is_number(ttl) -> {:halt, {:ok, ttl}}
_ -> {:cont, acc}
end
end)
end
end

7
lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex
View file

@ -7,7 +7,7 @@ defmodule Pleroma.Web.RichMedia.Parser.TTL.AwsSignedUrl do
@impl true
def ttl(data, _url) do
image = Map.get(data, :image)
image = Map.get(data, "image")
if aws_signed_url?(image) do
image
@ -15,14 +15,15 @@ defmodule Pleroma.Web.RichMedia.Parser.TTL.AwsSignedUrl do
|> format_query_params()
|> get_expiration_timestamp()
else
{:error, "Not aws signed url #{inspect(image)}"}
nil
end
end
defp aws_signed_url?(image) when is_binary(image) and image != "" do
%URI{host: host, query: query} = URI.parse(image)
String.contains?(host, "amazonaws.com") and String.contains?(query, "X-Amz-Expires")
is_binary(host) and String.contains?(host, "amazonaws.com") and
String.contains?(query, "X-Amz-Expires")
end
defp aws_signed_url?(_), do: nil

20
lib/pleroma/web/rich_media/parser/ttl/opengraph.ex Normal file
View file

@ -0,0 +1,20 @@
# Pleroma: A lightweight social networking server
# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Parser.TTL.Opengraph do
@behaviour Pleroma.Web.RichMedia.Parser.TTL
@impl true
def ttl(%{"ttl" => ttl_string}, _url) when is_binary(ttl_string) do
try do
ttl = String.to_integer(ttl_string)
now = DateTime.utc_now() |> DateTime.to_unix()
now + ttl
rescue
_ -> nil
end
end
def ttl(_, _), do: nil
end

1
lib/pleroma/web/router.ex
View file

@ -768,7 +768,6 @@ defmodule Pleroma.Web.Router do
get("/statuses", StatusController, :index)
get("/statuses/:id", StatusController, :show)
get("/statuses/:id/context", StatusController, :context)
get("/statuses/:id/card", StatusController, :card)
get("/statuses/:id/favourited_by", StatusController, :favourited_by)
get("/statuses/:id/reblogged_by", StatusController, :reblogged_by)
get("/statuses/:id/history", StatusController, :show_history)

3
lib/pleroma/workers/receiver_worker.ex
View file

@ -52,7 +52,8 @@ defmodule Pleroma.Workers.ReceiverWorker do
{:error, {:reject, reason}} -> {:cancel, reason}
{:signature, false} -> {:cancel, :invalid_signature}
{:error, {:error, reason = "Object has been deleted"}} -> {:cancel, reason}
e -> e
{:error, _} = e -> e
e -> {:error, e}
end
end
end

15
lib/pleroma/workers/rich_media_expiration_worker.ex Normal file
View file

@ -0,0 +1,15 @@
# Pleroma: A lightweight social networking server
# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Workers.RichMediaExpirationWorker do
alias Pleroma.Web.RichMedia.Card
use Oban.Worker,
queue: :rich_media_expiration
@impl Oban.Worker
def perform(%Job{args: %{"url" => url} = _args}) do
Card.delete(url)
end
end