Page MenuHomePhorge
Files F85712685

No OneTemporary
Actions

Size
7 KB
Referenced Files
None
Subscribers
None

View Options

diff --git a/lib/pleroma/web/rich_media/card.ex b/lib/pleroma/web/rich_media/card.ex
index f49bb2fee..040066f36 100644
--- a/lib/pleroma/web/rich_media/card.ex
+++ b/lib/pleroma/web/rich_media/card.ex
@@ -1,158 +1,162 @@
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
+ if @config_impl.get([:rich_media, :enabled]) do
+ case get_by_url(url) do
+ %__MODULE__{} = card ->
+ card
- nil ->
- backfill_opts = Map.put(backfill_opts, :url, url)
+ nil ->
+ backfill_opts = Map.put(backfill_opts, :url, url)
- Backfill.start(backfill_opts)
+ Backfill.start(backfill_opts)
- nil
+ nil
- :error ->
- nil
+ :error ->
+ nil
+ end
+ else
+ 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 {_, true} <- {:config, @config_impl.get([:rich_media, :enabled])},
%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
diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index 37cf29029..7f6b5d388 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -1,102 +1,106 @@
# 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 do
require Logger
@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: nil
@spec parse(String.t()) :: {:ok, map()} | {:error, any()}
def parse(url) do
- with :ok <- validate_page_url(url),
+ with {_, true} <- {:config, @config_impl.get([:rich_media, :enabled])},
+ :ok <- validate_page_url(url),
{:ok, data} <- parse_url(url) do
data = Map.put(data, "url", url)
{:ok, data}
+ else
+ {:config, _} -> {:error, :rich_media_disabled}
+ e -> e
end
end
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()
|> clean_parsed_data()
|> check_parsed_data()
end
end
defp maybe_parse(html) do
Enum.reduce_while(parsers(), %{}, fn parser, acc ->
case parser.parse(html, acc) do
data when data != %{} -> {:halt, data}
_ -> {:cont, acc}
end
end)
end
defp check_parsed_data(%{"title" => title} = data)
when is_binary(title) and title != "" do
{:ok, data}
end
defp check_parsed_data(data) do
{:error, {:invalid_metadata, data}}
end
defp clean_parsed_data(data) do
data
|> Enum.reject(fn {key, val} ->
not match?({:ok, _}, Jason.encode(%{key => val}))
end)
|> Map.new()
end
@spec validate_page_url(URI.t() | binary()) :: :ok | :error
defp validate_page_url(page_url) when is_binary(page_url) do
validate_tld = @config_impl.get([Pleroma.Formatter, :validate_tld])
page_url
|> Linkify.Parser.url?(validate_tld: validate_tld)
|> parse_uri(page_url)
end
defp validate_page_url(%URI{host: host, scheme: "https"}) do
cond do
Linkify.Parser.ip?(host) ->
:error
host in @config_impl.get([:rich_media, :ignore_hosts], []) ->
:error
get_tld(host) in @config_impl.get([:rich_media, :ignore_tld], []) ->
:error
true ->
:ok
end
end
defp validate_page_url(_), do: :error
defp parse_uri(true, url) do
url
|> URI.parse()
|> validate_page_url
end
defp parse_uri(_, _), do: :error
defp get_tld(host) do
host
|> String.split(".")
|> Enum.reverse()
|> hd
end
end

File Metadata

Mime Type
text/x-diff
Expires
Sat, Sep 19, 12:12 PM (1 d, 23 h)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
1767099
Default Alt Text
(7 KB)

Event Timeline