diff options
author | kaniini <ariadne@dereferenced.org> | 2019-09-16 07:37:54 +0000 |
---|---|---|
committer | kaniini <ariadne@dereferenced.org> | 2019-09-16 07:37:54 +0000 |
commit | d1a84ceec88ddd1bc2538cae8a4b6e72cde4fe6f (patch) | |
tree | 74967e78cb128070531fc349c9dc1edddfc00e62 | |
parent | 74eba5afe20124646366aace98c2e1b106712108 (diff) | |
parent | b4cf74c1067b866574a63fbd25ccb12cc1fed619 (diff) | |
download | pleroma-d1a84ceec88ddd1bc2538cae8a4b6e72cde4fe6f.tar.gz pleroma-d1a84ceec88ddd1bc2538cae8a4b6e72cde4fe6f.zip |
Merge branch 'feature/optimize_rich_media_parser' into 'develop'
added prepare html for RichMedia.Parser
See merge request pleroma/pleroma!1672
-rw-r--r-- | lib/pleroma/web/rich_media/parser.ex | 6 | ||||
-rw-r--r-- | mix.exs | 2 | ||||
-rw-r--r-- | mix.lock | 2 |
3 files changed, 7 insertions, 3 deletions
diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex index f5f9e358c..c06b0a0f2 100644 --- a/lib/pleroma/web/rich_media/parser.ex +++ b/lib/pleroma/web/rich_media/parser.ex @@ -81,6 +81,7 @@ defmodule Pleroma.Web.RichMedia.Parser do {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: @hackney_options) html + |> parse_html |> maybe_parse() |> Map.put(:url, url) |> clean_parsed_data() @@ -91,6 +92,8 @@ defmodule Pleroma.Web.RichMedia.Parser do end end + defp parse_html(html), do: Floki.parse(html) + defp maybe_parse(html) do Enum.reduce_while(parsers(), %{}, fn parser, acc -> case parser.parse(html, acc) do @@ -100,7 +103,8 @@ defmodule Pleroma.Web.RichMedia.Parser do 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 byte_size(title) > 0 do {:ok, data} end @@ -133,7 +133,7 @@ defmodule Pleroma.Mixfile do {:phoenix_swoosh, "~> 0.2"}, {:gen_smtp, "~> 0.13"}, {:websocket_client, git: "https://github.com/jeremyong/websocket_client.git", only: :test}, - {:floki, "~> 0.20.0"}, + {:floki, "~> 0.23.0"}, {:ex_syslogger, github: "slashmili/ex_syslogger", tag: "1.4.0"}, {:timex, "~> 3.5"}, {:ueberauth, "~> 0.4"}, @@ -34,7 +34,7 @@ "ex_rated": {:hex, :ex_rated, "1.3.3", "30ecbdabe91f7eaa9d37fa4e81c85ba420f371babeb9d1910adbcd79ec798d27", [:mix], [{:ex2ms, "~> 1.5", [hex: :ex2ms, repo: "hexpm", optional: false]}], "hexpm"}, "ex_syslogger": {:git, "https://github.com/slashmili/ex_syslogger.git", "f3963399047af17e038897c69e20d552e6899e1d", [tag: "1.4.0"]}, "excoveralls": {:hex, :excoveralls, "0.11.1", "dd677fbdd49114fdbdbf445540ec735808250d56b011077798316505064edb2c", [:mix], [{:hackney, "~> 1.0", [hex: :hackney, repo: "hexpm", optional: false]}, {:jason, "~> 1.0", [hex: :jason, repo: "hexpm", optional: false]}], "hexpm"}, - "floki": {:hex, :floki, "0.20.4", "be42ac911fece24b4c72f3b5846774b6e61b83fe685c2fc9d62093277fb3bc86", [:mix], [{:html_entities, "~> 0.4.0", [hex: :html_entities, repo: "hexpm", optional: false]}, {:mochiweb, "~> 2.15", [hex: :mochiweb, repo: "hexpm", optional: false]}], "hexpm"}, + "floki": {:hex, :floki, "0.23.0", "956ab6dba828c96e732454809fb0bd8d43ce0979b75f34de6322e73d4c917829", [:mix], [{:html_entities, "~> 0.4.0", [hex: :html_entities, repo: "hexpm", optional: false]}], "hexpm"}, "gen_smtp": {:hex, :gen_smtp, "0.14.0", "39846a03522456077c6429b4badfd1d55e5e7d0fdfb65e935b7c5e38549d9202", [:rebar3], [], "hexpm"}, "gen_stage": {:hex, :gen_stage, "0.14.2", "6a2a578a510c5bfca8a45e6b27552f613b41cf584b58210f017088d3d17d0b14", [:mix], [], "hexpm"}, "gen_state_machine": {:hex, :gen_state_machine, "2.0.5", "9ac15ec6e66acac994cc442dcc2c6f9796cf380ec4b08267223014be1c728a95", [:mix], [], "hexpm"}, |