Merge remote-tracking branch 'origin/develop' into matrixmatrix

9 files changed, 117 insertions, 45 deletions

diff --git a/lib/pleroma/web/rich_media/helpers.ex b/lib/pleroma/web/rich_media/helpers.ex
index 752ca9f81..566fc8c8a 100644
--- a/lib/pleroma/web/rich_media/helpers.ex
+++ b/lib/pleroma/web/rich_media/helpers.ex
@@ -9,14 +9,15 @@ defmodule Pleroma.Web.RichMedia.Helpers do
   alias Pleroma.Object
   alias Pleroma.Web.RichMedia.Parser
 
-  @rich_media_options [
+  @options [
     pool: :media,
-    max_body: 2_000_000
+    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 = Pleroma.Config.get([Pleroma.Formatter, :validate_tld])
+    validate_tld = Config.get([Pleroma.Formatter, :validate_tld])
 
     page_url
     |> Linkify.Parser.url?(validate_tld: validate_tld)
@@ -56,7 +57,6 @@ defmodule Pleroma.Web.RichMedia.Helpers do
 
   def fetch_data_for_object(object) do
     with true <- Config.get([:rich_media, :enabled]),
-         false <- object.data["sensitive"] || false,
          {:ok, page_url} <-
            HTML.extract_first_external_url_from_object(object),
          :ok <- validate_page_url(page_url),
@@ -69,7 +69,7 @@ defmodule Pleroma.Web.RichMedia.Helpers do
 
   def fetch_data_for_activity(%Activity{data: %{"type" => "Create"}} = activity) do
     with true <- Config.get([:rich_media, :enabled]),
-         %Object{} = object <- Object.normalize(activity) do
+         %Object{} = object <- Object.normalize(activity, fetch: false) do
       fetch_data_for_object(object)
     else
       _ -> %{}
@@ -78,24 +78,53 @@ defmodule Pleroma.Web.RichMedia.Helpers do
 
   def fetch_data_for_activity(_), do: %{}
 
-  def perform(:fetch, %Activity{} = activity) do
-    fetch_data_for_activity(activity)
-    :ok
-  end
-
   def rich_media_get(url) do
     headers = [{"user-agent", Pleroma.Application.user_agent() <> "; Bot"}]
 
-    options =
-      if Application.get_env(:tesla, :adapter) == Tesla.Adapter.Hackney do
-        Keyword.merge(@rich_media_options,
-          recv_timeout: 2_000,
-          with_body: true
-        )
-      else
-        @rich_media_options
+    head_check =
+      case Pleroma.HTTP.head(url, headers, @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 ->
+          e
+
+        {:ok, %Tesla.Env{status: 200, headers: headers}} ->
+          with :ok <- check_content_type(headers),
+               :ok <- check_content_length(headers),
+               do: :ok
+
+        _ ->
+          :ok
       end
 
-    Pleroma.HTTP.get(url, headers, adapter: options)
+    with :ok <- head_check, do: Pleroma.HTTP.get(url, headers, @options)
+  end
+
+  defp check_content_type(headers) do
+    case List.keyfind(headers, "content-type", 0) do
+      {_, content_type} ->
+        case Plug.Conn.Utils.media_type(content_type) do
+          {:ok, "text", "html", _} -> :ok
+          _ -> {:error, {:content_type, content_type}}
+        end
+
+      _ ->
+        :ok
+    end
+  end
+
+  @max_body @options[:max_body]
+  defp check_content_length(headers) do
+    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
+          {_, ""} -> {:error, :body_too_large}
+          _ -> :ok
+        end
+
+      _ ->
+        :ok
+    end
   end
 end
diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index 5727fda18..d6b54943b 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -1,10 +1,12 @@
 # Pleroma: A lightweight social networking server
-# Copyright © 2017-2020 Pleroma Authors <https://pleroma.social/>
+# Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.Web.RichMedia.Parser do
   require Logger
 
+  @cachex Pleroma.Config.get([:cachex, :provider], Cachex)
+
   defp parsers do
     Pleroma.Config.get([:rich_media, :parsers])
   end
@@ -20,28 +22,61 @@ defmodule Pleroma.Web.RichMedia.Parser do
       with {:ok, data} <- get_cached_or_parse(url),
            {:ok, _} <- set_ttl_based_on_image(data, url) do
         {:ok, data}
-      else
-        {:error, {:invalid_metadata, data}} = e ->
-          Logger.debug(fn -> "Incomplete or invalid metadata for #{url}: #{inspect(data)}" end)
-          e
-
-        error ->
-          Logger.error(fn -> "Rich media error for #{url}: #{inspect(error)}" end)
-          error
       end
     end
 
     defp get_cached_or_parse(url) do
-      case Cachex.fetch!(:rich_media_cache, url, fn _ -> {:commit, parse_url(url)} end) do
-        {:ok, _data} = res ->
-          res
-
-        {:error, _} = e ->
-          ttl = Pleroma.Config.get([:rich_media, :failure_backoff], 60_000)
-          Cachex.expire(:rich_media_cache, url, ttl)
-          e
+      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 """
@@ -73,7 +108,7 @@ defmodule Pleroma.Web.RichMedia.Parser do
       {:ok, ttl} when is_number(ttl) ->
         ttl = ttl * 1000
 
-        case Cachex.expire_at(:rich_media_cache, url, ttl) do
+        case @cachex.expire_at(:rich_media_cache, url, ttl) do
           {:ok, true} -> {:ok, ttl}
           {:ok, false} -> {:error, :no_key}
         end
diff --git a/lib/pleroma/web/rich_media/parser/ttl.ex b/lib/pleroma/web/rich_media/parser/ttl.ex
new file mode 100644
index 000000000..0b7f14fb2
--- /dev/null
+++ b/lib/pleroma/web/rich_media/parser/ttl.ex
@@ -0,0 +1,7 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Web.RichMedia.Parser.TTL do
+  @callback ttl(Map.t(), String.t()) :: Integer.t() | nil
+end
diff --git a/lib/pleroma/web/rich_media/parsers/ttl/aws_signed_url.ex b/lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex
index c5aaea2d4..c7eb267f3 100644
--- a/lib/pleroma/web/rich_media/parsers/ttl/aws_signed_url.ex
+++ b/lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex
@@ -1,7 +1,11 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
 defmodule Pleroma.Web.RichMedia.Parser.TTL.AwsSignedUrl do
   @behaviour Pleroma.Web.RichMedia.Parser.TTL
 
-  @impl Pleroma.Web.RichMedia.Parser.TTL
+  @impl true
   def ttl(data, _url) do
     image = Map.get(data, :image)
 
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 3d577e254..31c3d1e33 100644
--- a/lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex
+++ b/lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex
@@ -1,5 +1,5 @@
 # Pleroma: A lightweight social networking server
-# Copyright © 2017-2020 Pleroma Authors <https://pleroma.social/>
+# Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.Web.RichMedia.Parsers.MetaTagsParser do
diff --git a/lib/pleroma/web/rich_media/parsers/oembed_parser.ex b/lib/pleroma/web/rich_media/parsers/o_embed.ex
index 1fe6729c3..09eabec56 100644
--- a/lib/pleroma/web/rich_media/parsers/oembed_parser.ex
+++ b/lib/pleroma/web/rich_media/parsers/o_embed.ex
@@ -1,5 +1,5 @@
 # Pleroma: A lightweight social networking server
-# Copyright © 2017-2020 Pleroma Authors <https://pleroma.social/>
+# Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.Web.RichMedia.Parsers.OEmbed do
diff --git a/lib/pleroma/web/rich_media/parsers/ogp.ex b/lib/pleroma/web/rich_media/parsers/ogp.ex
index b3b3b059c..d0edf1c88 100644
--- a/lib/pleroma/web/rich_media/parsers/ogp.ex
+++ b/lib/pleroma/web/rich_media/parsers/ogp.ex
@@ -1,5 +1,5 @@
 # Pleroma: A lightweight social networking server
-# Copyright © 2017-2020 Pleroma Authors <https://pleroma.social/>
+# Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.Web.RichMedia.Parsers.OGP do
diff --git a/lib/pleroma/web/rich_media/parsers/ttl/ttl.ex b/lib/pleroma/web/rich_media/parsers/ttl/ttl.ex
deleted file mode 100644
index 6b3ec6d30..000000000
--- a/lib/pleroma/web/rich_media/parsers/ttl/ttl.ex
+++ /dev/null
@@ -1,3 +0,0 @@
-defmodule Pleroma.Web.RichMedia.Parser.TTL do
-  @callback ttl(Map.t(), String.t()) :: {:ok, Integer.t()} | {:error, String.t()}
-end
diff --git a/lib/pleroma/web/rich_media/parsers/twitter_card.ex b/lib/pleroma/web/rich_media/parsers/twitter_card.ex
index 4a04865d2..0adf84159 100644
--- a/lib/pleroma/web/rich_media/parsers/twitter_card.ex
+++ b/lib/pleroma/web/rich_media/parsers/twitter_card.ex
@@ -1,5 +1,5 @@
 # Pleroma: A lightweight social networking server
-# Copyright © 2017-2020 Pleroma Authors <https://pleroma.social/>
+# Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.Web.RichMedia.Parsers.TwitterCard do