X-Git-Url: https://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fparser.ex;h=d6b54943b40ee2ee0289ef648a1f2cdf4a79a3ba;hb=6f83ae27aa924db1da2189cf495bd83cb41ba408;hp=ca592833f3d7574973eb5776126741de3bb42daf;hpb=1a5a7ba6e8a936d3138ca530d3466111e542aef6;p=akkoma

diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index ca592833f..d6b54943b 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -1,8 +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
@@ -10,19 +14,69 @@ defmodule Pleroma.Web.RichMedia.Parser do
   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
-      try do
-        Cachex.fetch!(:rich_media_cache, url, fn _ ->
-          {:commit, parse_url(url)}
-        end)
-        |> set_ttl_based_on_image(url)
-      rescue
-        e ->
-          {:error, "Cachex error: #{inspect(e)}"}
+      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 """
@@ -47,19 +101,26 @@ defmodule Pleroma.Web.RichMedia.Parser do
       config :pleroma, :rich_media,
         ttl_setters: [MyModule]
   """
-  def set_ttl_based_on_image({:ok, data}, url) do
-    with {:ok, nil} <- Cachex.ttl(:rich_media_cache, url),
-         ttl when is_number(ttl) <- get_ttl_from_image(data, url) do
-      Cachex.expire_at(:rich_media_cache, url, ttl * 1000)
-      {:ok, data}
-    else
+  @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, data}
+        {:ok, :noop}
     end
   end
 
   defp get_ttl_from_image(data, url) do
-    Pleroma.Config.get([:rich_media, :ttl_setters])
+    [:rich_media, :ttl_setters]
+    |> Pleroma.Config.get()
     |> Enum.reduce({:ok, nil}, fn
       module, {:ok, _ttl} ->
         module.ttl(data, url)
@@ -69,24 +130,17 @@ defmodule Pleroma.Web.RichMedia.Parser do
     end)
   end
 
-  defp parse_url(url) do
-    try do
-      {:ok, %Tesla.Env{body: html}} = Pleroma.Web.RichMedia.Helpers.rich_media_get(url)
-
+  def 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
-      |> parse_html()
       |> maybe_parse()
       |> Map.put("url", url)
       |> clean_parsed_data()
       |> check_parsed_data()
-    rescue
-      e ->
-        {:error, "Parsing error: #{inspect(e)} #{inspect(__STACKTRACE__)}"}
     end
   end
 
-  defp parse_html(html), do: Floki.parse_document!(html)
-
   defp maybe_parse(html) do
     Enum.reduce_while(parsers(), %{}, fn parser, acc ->
       case parser.parse(html, acc) do
@@ -102,7 +156,7 @@ defmodule Pleroma.Web.RichMedia.Parser do
   end
 
   defp check_parsed_data(data) do
-    {:error, "Found metadata was invalid or incomplete: #{inspect(data)}"}
+    {:error, {:invalid_metadata, data}}
   end
 
   defp clean_parsed_data(data) do