diff options
Diffstat (limited to 'lib/pleroma/web/rich_media')
| -rw-r--r-- | lib/pleroma/web/rich_media/backfill.ex | 101 | ||||
| -rw-r--r-- | lib/pleroma/web/rich_media/card.ex | 157 | ||||
| -rw-r--r--[-rwxr-xr-x] | lib/pleroma/web/rich_media/helpers.ex | 89 | ||||
| -rw-r--r--[-rwxr-xr-x] | lib/pleroma/web/rich_media/parser.ex | 169 | ||||
| -rw-r--r--[-rwxr-xr-x] | lib/pleroma/web/rich_media/parser/ttl.ex | 15 | ||||
| -rw-r--r--[-rwxr-xr-x] | lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex | 15 | ||||
| -rw-r--r-- | lib/pleroma/web/rich_media/parser/ttl/opengraph.ex | 20 | ||||
| -rw-r--r--[-rwxr-xr-x] | lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex | 0 | ||||
| -rw-r--r--[-rwxr-xr-x] | lib/pleroma/web/rich_media/parsers/o_embed.ex | 0 | ||||
| -rw-r--r--[-rwxr-xr-x] | lib/pleroma/web/rich_media/parsers/ogp.ex | 0 | ||||
| -rw-r--r--[-rwxr-xr-x] | lib/pleroma/web/rich_media/parsers/twitter_card.ex | 0 |
11 files changed, 363 insertions, 203 deletions
diff --git a/lib/pleroma/web/rich_media/backfill.ex b/lib/pleroma/web/rich_media/backfill.ex new file mode 100644 index 0000000..4ec50e1 --- /dev/null +++ b/lib/pleroma/web/rich_media/backfill.ex @@ -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 diff --git a/lib/pleroma/web/rich_media/card.ex b/lib/pleroma/web/rich_media/card.ex new file mode 100644 index 0000000..36a1ae4 --- /dev/null +++ b/lib/pleroma/web/rich_media/card.ex @@ -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 diff --git a/lib/pleroma/web/rich_media/helpers.ex b/lib/pleroma/web/rich_media/helpers.ex index 0488df3..1199944 100755..100644 --- a/lib/pleroma/web/rich_media/helpers.ex +++ b/lib/pleroma/web/rich_media/helpers.ex @@ -3,86 +3,13 @@ # SPDX-License-Identifier: AGPL-3.0-only defmodule Pleroma.Web.RichMedia.Helpers do - alias Pleroma.Activity alias Pleroma.Config - alias Pleroma.HTML - alias Pleroma.Object - alias Pleroma.Web.RichMedia.Parser - - @options [ - pool: :media, - max_body: 2_000_000, - recv_timeout: 2_000 - ] - - @spec validate_page_url(URI.t() | binary()) :: :ok | :error - defp validate_page_url(page_url) when is_binary(page_url) do - validate_tld = Config.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", authority: authority}) - when is_binary(authority) do - cond do - host in Config.get([:rich_media, :ignore_hosts], []) -> - :error - - get_tld(host) in Config.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 - - def fetch_data_for_object(object) do - with true <- Config.get([:rich_media, :enabled]), - {:ok, page_url} <- - HTML.extract_first_external_url_from_object(object), - :ok <- validate_page_url(page_url), - {: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.get([:rich_media, :enabled]), - %Object{} = object <- Object.normalize(activity, fetch: false) do - fetch_data_for_object(object) - else - _ -> %{} - end - end - - def fetch_data_for_activity(_), do: %{} 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 -> @@ -97,7 +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 @@ -113,12 +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 @@ -127,4 +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 diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex index dbe81ea..37cf290 100755..100644 --- a/lib/pleroma/web/rich_media/parser.ex +++ b/lib/pleroma/web/rich_media/parser.ex @@ -5,137 +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"} - - if Pleroma.Config.get(:env) == :test do - @spec parse(String.t()) :: {:ok, map()} | {:error, any()} - def parse(url), do: parse_url(url) - else - @spec parse(String.t()) :: {:ok, map()} | {:error, any()} - def parse(url) do - with {:ok, data} <- get_cached_or_parse(url), - {:ok, _} <- set_ttl_based_on_image(data, url) do - {: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.warn(fn -> "Rich media error for #{url}: #{inspect(reason)}" end) - 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 + def parse(nil), do: nil - Define the module in the config - - config :pleroma, :rich_media, - ttl_setters: [MyModule] - """ - @spec set_ttl_based_on_image(map(), String.t()) :: - {:ok, Integer.t() | :noop} | {:error, :no_key} - def set_ttl_based_on_image(data, url) do - case get_ttl_from_image(data, url) do - {:ok, 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} + @spec parse(String.t()) :: {:ok, map()} | {:error, any()} + def parse(url) do + with :ok <- validate_page_url(url), + {:ok, data} <- parse_url(url) do + data = Map.put(data, "url", url) + {:ok, data} 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 @@ -166,4 +57,46 @@ defmodule Pleroma.Web.RichMedia.Parser do 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 diff --git a/lib/pleroma/web/rich_media/parser/ttl.ex b/lib/pleroma/web/rich_media/parser/ttl.ex index 59d7f87..7e56375 100755..100644 --- a/lib/pleroma/web/rich_media/parser/ttl.ex +++ b/lib/pleroma/web/rich_media/parser/ttl.ex @@ -3,5 +3,18 @@ # SPDX-License-Identifier: AGPL-3.0-only defmodule Pleroma.Web.RichMedia.Parser.TTL do - @callback ttl(Map.t(), String.t()) :: Integer.t() | nil + @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 diff --git a/lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex b/lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex index fa41c16..948c727 100755..100644 --- a/lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex +++ b/lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex @@ -7,25 +7,26 @@ 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 is_aws_signed_url(image) do + if aws_signed_url?(image) do image |> parse_query_params() |> format_query_params() |> get_expiration_timestamp() else - {:error, "Not aws signed url #{inspect(image)}"} + nil end end - defp is_aws_signed_url(image) when is_binary(image) and image != "" do + 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 is_aws_signed_url(_), do: nil + defp aws_signed_url?(_), do: nil defp parse_query_params(image) do %URI{query: query} = URI.parse(image) @@ -45,6 +46,6 @@ defmodule Pleroma.Web.RichMedia.Parser.TTL.AwsSignedUrl do |> Map.get("X-Amz-Date") |> Timex.parse("{ISO:Basic:Z}") - {:ok, Timex.to_unix(date) + String.to_integer(Map.get(params, "X-Amz-Expires"))} + Timex.to_unix(date) + String.to_integer(Map.get(params, "X-Amz-Expires")) end end diff --git a/lib/pleroma/web/rich_media/parser/ttl/opengraph.ex b/lib/pleroma/web/rich_media/parser/ttl/opengraph.ex new file mode 100644 index 0000000..b068896 --- /dev/null +++ b/lib/pleroma/web/rich_media/parser/ttl/opengraph.ex @@ -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 diff --git a/lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex b/lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex index 320a5f5..320a5f5 100755..100644 --- a/lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex +++ b/lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex diff --git a/lib/pleroma/web/rich_media/parsers/o_embed.ex b/lib/pleroma/web/rich_media/parsers/o_embed.ex index 0f30317..0f30317 100755..100644 --- a/lib/pleroma/web/rich_media/parsers/o_embed.ex +++ b/lib/pleroma/web/rich_media/parsers/o_embed.ex diff --git a/lib/pleroma/web/rich_media/parsers/ogp.ex b/lib/pleroma/web/rich_media/parsers/ogp.ex index b7f2b42..b7f2b42 100755..100644 --- a/lib/pleroma/web/rich_media/parsers/ogp.ex +++ b/lib/pleroma/web/rich_media/parsers/ogp.ex diff --git a/lib/pleroma/web/rich_media/parsers/twitter_card.ex b/lib/pleroma/web/rich_media/parsers/twitter_card.ex index cc65372..cc65372 100755..100644 --- a/lib/pleroma/web/rich_media/parsers/twitter_card.ex +++ b/lib/pleroma/web/rich_media/parsers/twitter_card.ex |
