diff options
author | rinpatch <rinpatch@sdf.org> | 2020-11-12 12:34:48 +0000 |
---|---|---|
committer | rinpatch <rinpatch@sdf.org> | 2020-11-12 12:34:48 +0000 |
commit | 1172844ed18d94d84724dc6f11c6e9f72e0ba6ec (patch) | |
tree | 7d48a259e08856ab6db0eba255f20c0c19410463 /lib/pleroma/web/rich_media/parser.ex | |
parent | a0f5e8b27edbe2224d9c2c3997ad5b8ea484244b (diff) | |
parent | b4c6b262d6dc12362f0014a864e8aed6c727c39c (diff) | |
download | pleroma-2.2.0.tar.gz |
Merge branch 'release/2.2.0' into 'stable'v2.2.0
Release/2.2.0
See merge request pleroma/secteam/pleroma!19
Diffstat (limited to 'lib/pleroma/web/rich_media/parser.ex')
-rw-r--r-- | lib/pleroma/web/rich_media/parser.ex | 72 |
1 files changed, 51 insertions, 21 deletions
diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex index 569249f51..c70d2fdba 100644 --- a/lib/pleroma/web/rich_media/parser.ex +++ b/lib/pleroma/web/rich_media/parser.ex @@ -20,31 +20,61 @@ 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 -> - Logger.error(fn -> "Rich media error: #{inspect(error)}" end) 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, :body_too_large} = e -> - e - - {:error, {:content_type, _}} = e -> - e - - # The TTL is not set for the errors above, since they are unlikely to change - # with time - {: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 """ @@ -98,7 +128,7 @@ defmodule Pleroma.Web.RichMedia.Parser do end) end - defp parse_url(url) do + 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 @@ -124,7 +154,7 @@ 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 |