X-Git-Url: http://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fparser.ex;h=c06b0a0f2668338415d592fde838bff42e49e056;hb=0883a706dc376fdfb7de9df1366803e87c8e7c98;hp=b69b2be610a6383ed5f5c97a92f3db4c0fe3ea02;hpb=196cad46f35a63c18d58cd5d982bc4e1f9b0d7c3;p=akkoma
diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index b69b2be61..ef5ead2da 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -1,13 +1,11 @@
# Pleroma: A lightweight social networking server
-# Copyright © 2017-2019 Pleroma Authors
+# Copyright © 2017-2020 Pleroma Authors
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Parser do
- @hackney_options [
+ @options [
pool: :media,
- recv_timeout: 2_000,
- max_body: 2_000_000,
- with_body: true
+ max_body: 2_000_000
]
defp parsers do
@@ -55,8 +53,8 @@ defmodule Pleroma.Web.RichMedia.Parser do
ttl_setters: [MyModule]
"""
def set_ttl_based_on_image({:ok, data}, url) do
- with {:ok, nil} <- Cachex.ttl(:rich_media_cache, url) do
- ttl = get_ttl_from_image(data, url)
+ 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
@@ -77,29 +75,44 @@ defmodule Pleroma.Web.RichMedia.Parser do
end
defp parse_url(url) do
+ opts =
+ if Application.get_env(:tesla, :adapter) == Tesla.Adapter.Hackney do
+ Keyword.merge(@options,
+ recv_timeout: 2_000,
+ with_body: true
+ )
+ else
+ @options
+ end
+
try do
- {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: @hackney_options)
+ {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: opts)
html
+ |> parse_html()
|> maybe_parse()
+ |> Map.put("url", url)
|> clean_parsed_data()
|> check_parsed_data()
rescue
e ->
- {:error, "Parsing error: #{inspect(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
- {:ok, data} -> {:halt, data}
- {:error, _msg} -> {:cont, acc}
+ data when data != %{} -> {:halt, data}
+ _ -> {:cont, acc}
end
end)
end
- defp check_parsed_data(%{title: title} = data) when is_binary(title) and byte_size(title) > 0 do
+ defp check_parsed_data(%{"title" => title} = data)
+ when is_binary(title) and title != "" do
{:ok, data}
end
@@ -110,11 +123,7 @@ defmodule Pleroma.Web.RichMedia.Parser do
defp clean_parsed_data(data) do
data
|> Enum.reject(fn {key, val} ->
- with {:ok, _} <- Jason.encode(%{key => val}) do
- false
- else
- _ -> true
- end
+ not match?({:ok, _}, Jason.encode(%{key => val}))
end)
|> Map.new()
end