Merge branch 'oembed_provider' into 'develop'

Opengraph/TwitterCard::summary for statuses and user profiles

See merge request pleroma/pleroma!533

config/config.exs

@@ -208,6 +208,8 @@
   ip: {0, 0, 0, 0},
   port: 9999
 
+config :pleroma, Pleroma.Web.Metadata, providers: [], unfurl_nsfw: false
+
 config :pleroma, :suggestions,
   enabled: false,
   third_party_engine:

docs/config.md

@@ -211,3 +211,9 @@ curl "http://localhost:4000/api/pleroma/admin/invite_token?admin_token=somerando
 * `max_jobs`: The maximum amount of parallel federation jobs running at the same time.
 * `initial_timeout`: The initial timeout in seconds
 * `max_retries`: The maximum number of times a federation job is retried
+
+## Pleroma.Web.Metadata
+* `providers`: a list of metadata providers to enable. Providers availible:
+  * Pleroma.Web.Metadata.Providers.OpenGraph
+  * Pleroma.Web.Metadata.Providers.TwitterCard
+* `unfurl_nsfw`: If set to `true` nsfw attachments will be shown in previews

lib/pleroma/formatter.ex

@@ -43,7 +43,7 @@ def emojify(text) do
 
   def emojify(text, nil), do: text
 
-  def emojify(text, emoji) do
+  def emojify(text, emoji, strip \\ false) do
     Enum.reduce(emoji, text, fn {emoji, file}, text ->
       emoji = HTML.strip_tags(emoji)
       file = HTML.strip_tags(file)
@@ -51,14 +51,24 @@ def emojify(text, emoji) do
       String.replace(
         text,
         ":#{emoji}:",
+        if not strip do
           "<img height='32px' width='32px' alt='#{emoji}' title='#{emoji}' src='#{
             MediaProxy.url(file)
           }' />"
+        else
+          ""
+        end
       )
       |> HTML.filter_tags()
     end)
   end
 
+  def demojify(text) do
+    emojify(text, Emoji.get_all(), true)
+  end
+
+  def demojify(text, nil), do: text
+
   def get_emoji(text) when is_binary(text) do
     Enum.filter(Emoji.get_all(), fn {emoji, _} -> String.contains?(text, ":#{emoji}:") end)
   end
@@ -189,4 +199,16 @@ def finalize({subs, text}) do
       String.replace(result_text, uuid, replacement)
     end)
   end
+
+  def truncate(text, max_length \\ 200, omission \\ "...") do
+    # Remove trailing whitespace
+    text = Regex.replace(~r/([^ \t\r\n])([ \t]+$)/u, text, "\\g{1}")
+
+    if String.length(text) < max_length do
+      text
+    else
+      length_with_omission = max_length - String.length(omission)
+      String.slice(text, 0, length_with_omission) <> omission
+    end
+  end
 end

lib/pleroma/user.ex

@@ -406,6 +406,10 @@ def locked?(%User{} = user) do
     user.info.locked || false
   end
 
+  def get_by_id(id) do
+    Repo.get_by(User, id: id)
+  end
+
   def get_by_ap_id(ap_id) do
     Repo.get_by(User, ap_id: ap_id)
   end
@@ -441,11 +445,33 @@ def get_cached_by_ap_id(ap_id) do
     Cachex.fetch!(:user_cache, key, fn _ -> get_by_ap_id(ap_id) end)
   end
 
+  def get_cached_by_id(id) do
+    key = "id:#{id}"
+
+    ap_id =
+      Cachex.fetch!(:user_cache, key, fn _ ->
+        user = get_by_id(id)
+
+        if user do
+          Cachex.put(:user_cache, "ap_id:#{user.ap_id}", user)
+          {:commit, user.ap_id}
+        else
+          {:ignore, ""}
+        end
+      end)
+
+    get_cached_by_ap_id(ap_id)
+  end
+
   def get_cached_by_nickname(nickname) do
     key = "nickname:#{nickname}"
     Cachex.fetch!(:user_cache, key, fn _ -> get_or_fetch_by_nickname(nickname) end)
   end
 
+  def get_cached_by_nickname_or_id(nickname_or_id) do
+    get_cached_by_id(nickname_or_id) || get_cached_by_nickname(nickname_or_id)
+  end
+
   def get_by_nickname(nickname) do
     Repo.get_by(User, nickname: nickname) ||
       if Regex.match?(~r(@#{Pleroma.Web.Endpoint.host()})i, nickname) do

lib/pleroma/web/metadata.ex

@@ -0,0 +1,40 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Web.Metadata do
+  alias Phoenix.HTML
+
+  def build_tags(params) do
+    Enum.reduce(Pleroma.Config.get([__MODULE__, :providers], []), "", fn parser, acc ->
+      rendered_html =
+        params
+        |> parser.build_tags()
+        |> Enum.map(&to_tag/1)
+        |> Enum.map(&HTML.safe_to_string/1)
+        |> Enum.join()
+
+      acc <> rendered_html
+    end)
+  end
+
+  def to_tag(data) do
+    with {name, attrs, _content = []} <- data do
+      HTML.Tag.tag(name, attrs)
+    else
+      {name, attrs, content} ->
+        HTML.Tag.content_tag(name, content, attrs)
+
+      _ ->
+        raise ArgumentError, message: "make_tag invalid args"
+    end
+  end
+
+  def activity_nsfw?(%{data: %{"sensitive" => sensitive}}) do
+    Pleroma.Config.get([__MODULE__, :unfurl_nsfw], false) == false and sensitive
+  end
+
+  def activity_nsfw?(_) do
+    false
+  end
+end

lib/pleroma/web/metadata/opengraph.ex

@@ -0,0 +1,154 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Web.Metadata.Providers.OpenGraph do
+  alias Pleroma.Web.Metadata.Providers.Provider
+  alias Pleroma.Web.Metadata
+  alias Pleroma.{HTML, Formatter, User}
+  alias Pleroma.Web.MediaProxy
+
+  @behaviour Provider
+
+  @impl Provider
+  def build_tags(%{
+        object: object,
+        url: url,
+        user: user
+      }) do
+    attachments = build_attachments(object)
+    scrubbed_content = scrub_html_and_truncate(object)
+    # Zero width space
+    content =
+      if scrubbed_content != "" and scrubbed_content != "\u200B" do
+        ": “" <> scrubbed_content <> "”"
+      else
+        ""
+      end
+
+    # Most previews only show og:title which is inconvenient. Instagram
+    # hacks this by putting the description in the title and making the
+    # description longer prefixed by how many likes and shares the post
+    # has. Here we use the descriptive nickname in the title, and expand
+    # the full account & nickname in the description. We also use the cute^Wevil
+    # smart quotes around the status text like Instagram, too.
+    [
+      {:meta,
+       [
+         property: "og:title",
+         content: "#{user.name}" <> content
+       ], []},
+      {:meta, [property: "og:url", content: url], []},
+      {:meta,
+       [
+         property: "og:description",
+         content: "#{user_name_string(user)}" <> content
+       ], []},
+      {:meta, [property: "og:type", content: "website"], []}
+    ] ++
+      if attachments == [] or Metadata.activity_nsfw?(object) do
+        [
+          {:meta, [property: "og:image", content: attachment_url(User.avatar_url(user))], []},
+          {:meta, [property: "og:image:width", content: 150], []},
+          {:meta, [property: "og:image:height", content: 150], []}
+        ]
+      else
+        attachments
+      end
+  end
+
+  @impl Provider
+  def build_tags(%{user: user}) do
+    with truncated_bio = scrub_html_and_truncate(user.bio || "") do
+      [
+        {:meta,
+         [
+           property: "og:title",
+           content: user_name_string(user)
+         ], []},
+        {:meta, [property: "og:url", content: User.profile_url(user)], []},
+        {:meta, [property: "og:description", content: truncated_bio], []},
+        {:meta, [property: "og:type", content: "website"], []},
+        {:meta, [property: "og:image", content: attachment_url(User.avatar_url(user))], []},
+        {:meta, [property: "og:image:width", content: 150], []},
+        {:meta, [property: "og:image:height", content: 150], []}
+      ]
+    end
+  end
+
+  defp build_attachments(%{data: %{"attachment" => attachments}}) do
+    Enum.reduce(attachments, [], fn attachment, acc ->
+      rendered_tags =
+        Enum.reduce(attachment["url"], [], fn url, acc ->
+          media_type =
+            Enum.find(["image", "audio", "video"], fn media_type ->
+              String.starts_with?(url["mediaType"], media_type)
+            end)
+
+          # TODO: Add additional properties to objects when we have the data available.
+          # Also, Whatsapp only wants JPEG or PNGs. It seems that if we add a second og:image
+          # object when a Video or GIF is attached it will display that in the Whatsapp Rich Preview.
+          case media_type do
+            "audio" ->
+              [
+                {:meta, [property: "og:" <> media_type, content: attachment_url(url["href"])], []}
+                | acc
+              ]
+
+            "image" ->
+              [
+                {:meta, [property: "og:" <> media_type, content: attachment_url(url["href"])],
+                 []},
+                {:meta, [property: "og:image:width", content: 150], []},
+                {:meta, [property: "og:image:height", content: 150], []}
+                | acc
+              ]
+
+            "video" ->
+              [
+                {:meta, [property: "og:" <> media_type, content: attachment_url(url["href"])], []}
+                | acc
+              ]
+
+            _ ->
+              acc
+          end
+        end)
+
+      acc ++ rendered_tags
+    end)
+  end
+
+  defp scrub_html_and_truncate(%{data: %{"content" => content}} = object) do
+    content
+    # html content comes from DB already encoded, decode first and scrub after
+    |> HtmlEntities.decode()
+    |> String.replace(~r/<br\s?\/?>/, " ")
+    |> HTML.get_cached_stripped_html_for_object(object, __MODULE__)
+    |> Formatter.demojify()
+    |> Formatter.truncate()
+  end
+
+  defp scrub_html_and_truncate(content) when is_binary(content) do
+    content
+    # html content comes from DB already encoded, decode first and scrub after
+    |> HtmlEntities.decode()
+    |> String.replace(~r/<br\s?\/?>/, " ")
+    |> HTML.strip_tags()
+    |> Formatter.demojify()
+    |> Formatter.truncate()
+  end
+
+  defp attachment_url(url) do
+    MediaProxy.url(url)
+  end
+
+  defp user_name_string(user) do
+    "#{user.name} " <>
+      if user.local do
+        "(@#{user.nickname}@#{Pleroma.Web.Endpoint.host()})"
+      else
+        "(@#{user.nickname})"
+      end
+  end
+end

lib/pleroma/web/metadata/provider.ex

@@ -0,0 +1,7 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Web.Metadata.Providers.Provider do
+  @callback build_tags(map()) :: list()
+end

lib/pleroma/web/metadata/twitter_card.ex

@@ -0,0 +1,46 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Web.Metadata.Providers.TwitterCard do
+  alias Pleroma.Web.Metadata.Providers.Provider
+  alias Pleroma.Web.Metadata
+
+  @behaviour Provider
+
+  @impl Provider
+  def build_tags(%{object: object}) do
+    if Metadata.activity_nsfw?(object) or object.data["attachment"] == [] do
+      build_tags(nil)
+    else
+      case find_first_acceptable_media_type(object) do
+        "image" ->
+          [{:meta, [property: "twitter:card", content: "summary_large_image"], []}]
+
+        "audio" ->
+          [{:meta, [property: "twitter:card", content: "player"], []}]
+
+        "video" ->
+          [{:meta, [property: "twitter:card", content: "player"], []}]
+
+        _ ->
+          build_tags(nil)
+      end
+    end
+  end
+
+  @impl Provider
+  def build_tags(_) do
+    [{:meta, [property: "twitter:card", content: "summary"], []}]
+  end
+
+  def find_first_acceptable_media_type(%{data: %{"attachment" => attachment}}) do
+    Enum.find_value(attachment, fn attachment ->
+      Enum.find_value(attachment["url"], fn url ->
+        Enum.find(["image", "audio", "video"], fn media_type ->
+          String.starts_with?(url["mediaType"], media_type)
+        end)
+      end)
+    end)
+  end
+end

lib/pleroma/web/ostatus/ostatus_controller.ex

@@ -7,7 +7,6 @@ defmodule Pleroma.Web.OStatus.OStatusController do
 
   alias Pleroma.{User, Activity, Object}
   alias Pleroma.Web.OStatus.{FeedRepresenter, ActivityRepresenter}
-  alias Pleroma.Repo
   alias Pleroma.Web.{OStatus, Federator}
   alias Pleroma.Web.XML
   alias Pleroma.Web.ActivityPub.ObjectView
@@ -20,7 +19,11 @@ defmodule Pleroma.Web.OStatus.OStatusController do
   def feed_redirect(conn, %{"nickname" => nickname}) do
     case get_format(conn) do
       "html" ->
-        Fallback.RedirectController.redirector(conn, nil)
+        with %User{} = user <- User.get_cached_by_nickname_or_id(nickname) do
+          Fallback.RedirectController.redirector_with_meta(conn, %{user: user})
+        else
+          nil -> {:error, :not_found}
+        end
 
       "activity+json" ->
         ActivityPubController.call(conn, :user)
@@ -136,14 +139,27 @@ def activity(conn, %{"uuid" => uuid}) do
   end
 
   def notice(conn, %{"id" => id}) do
-    with {_, %Activity{} = activity} <- {:activity, Repo.get(Activity, id)},
+    with {_, %Activity{} = activity} <- {:activity, Activity.get_by_id(id)},
          {_, true} <- {:public?, ActivityPub.is_public?(activity)},
          %User{} = user <- User.get_cached_by_ap_id(activity.data["actor"]) do
       case format = get_format(conn) do
         "html" ->
-          conn
+          if activity.data["type"] == "Create" do
-          |> put_resp_content_type("text/html")
+            %Object{} = object = Object.normalize(activity.data["object"])
-          |> send_file(200, Pleroma.Plugs.InstanceStatic.file_path("index.html"))
+
+            Fallback.RedirectController.redirector_with_meta(conn, %{
+              object: object,
+              url:
+                Pleroma.Web.Router.Helpers.o_status_url(
+                  Pleroma.Web.Endpoint,
+                  :notice,
+                  activity.id
+                ),
+              user: user
+            })
+          else
+            Fallback.RedirectController.redirector(conn, nil)
+          end
 
         _ ->
           represent_activity(conn, format, activity, user)

lib/pleroma/web/router.ex

@@ -396,7 +396,11 @@ defmodule Pleroma.Web.Router do
   end
 
   pipeline :ostatus do
-    plug(:accepts, ["xml", "atom", "html", "activity+json"])
+    plug(:accepts, ["html", "xml", "atom", "activity+json"])
+  end
+
+  pipeline :oembed do
+    plug(:accepts, ["json", "xml"])
   end
 
   scope "/", Pleroma.Web do
@@ -414,6 +418,12 @@ defmodule Pleroma.Web.Router do
     post("/push/subscriptions/:id", Websub.WebsubController, :websub_incoming)
   end
 
+  scope "/", Pleroma.Web do
+    pipe_through(:oembed)
+
+    get("/oembed", OEmbed.OEmbedController, :url)
+  end
+
   pipeline :activitypub do
     plug(:accepts, ["activity+json"])
     plug(Pleroma.Web.Plugs.HTTPSignaturePlug)
@@ -501,6 +511,7 @@ defmodule Pleroma.Web.Router do
 
   scope "/", Fallback do
     get("/registration/:token", RedirectController, :registration_page)
+    get("/:maybe_nickname_or_id", RedirectController, :redirector_with_meta)
     get("/*path", RedirectController, :redirector)
 
     options("/*path", RedirectController, :empty)
@@ -509,11 +520,36 @@ defmodule Pleroma.Web.Router do
 
 defmodule Fallback.RedirectController do
   use Pleroma.Web, :controller
+  alias Pleroma.Web.Metadata
+  alias Pleroma.User
 
   def redirector(conn, _params) do
     conn
     |> put_resp_content_type("text/html")
-    |> send_file(200, Pleroma.Plugs.InstanceStatic.file_path("index.html"))
+    |> send_file(200, index_file_path())
+  end
+
+  def redirector_with_meta(conn, %{"maybe_nickname_or_id" => maybe_nickname_or_id} = params) do
+    with %User{} = user <- User.get_cached_by_nickname_or_id(maybe_nickname_or_id) do
+      redirector_with_meta(conn, %{user: user})
+    else
+      nil ->
+        redirector(conn, params)
+    end
+  end
+
+  def redirector_with_meta(conn, params) do
+    {:ok, index_content} = File.read(index_file_path())
+    tags = Metadata.build_tags(params)
+    response = String.replace(index_content, "<!--server-generated-meta-->", tags)
+
+    conn
+    |> put_resp_content_type("text/html")
+    |> send_resp(200, response)
+  end
+
+  def index_file_path do
+    Pleroma.Plugs.InstanceStatic.file_path("index.html")
   end
 
   def registration_page(conn, params) do

mix.exs

@@ -59,6 +59,7 @@ defp deps do
       {:pbkdf2_elixir, "~> 0.12.3"},
       {:trailing_format_plug, "~> 0.0.7"},
       {:html_sanitize_ex, "~> 1.3.0"},
+      {:html_entities, "~> 0.4"},
       {:phoenix_html, "~> 2.10"},
       {:calendar, "~> 0.17.4"},
       {:cachex, "~> 3.0.2"},

test/web/metadata/opengraph_test.exs

@@ -0,0 +1,94 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Web.Metadata.Providers.OpenGraphTest do
+  use Pleroma.DataCase
+  import Pleroma.Factory
+  alias Pleroma.Web.Metadata.Providers.OpenGraph
+
+  test "it renders all supported types of attachments and skips unknown types" do
+    user = insert(:user)
+
+    note =
+      insert(:note, %{
+        data: %{
+          "actor" => user.ap_id,
+          "tag" => [],
+          "id" => "https://pleroma.gov/objects/whatever",
+          "content" => "pleroma in a nutshell",
+          "attachment" => [
+            %{
+              "url" => [
+                %{"mediaType" => "image/png", "href" => "https://pleroma.gov/tenshi.png"}
+              ]
+            },
+            %{
+              "url" => [
+                %{
+                  "mediaType" => "application/octet-stream",
+                  "href" => "https://pleroma.gov/fqa/badapple.sfc"
+                }
+              ]
+            },
+            %{
+              "url" => [
+                %{"mediaType" => "video/webm", "href" => "https://pleroma.gov/about/juche.webm"}
+              ]
+            },
+            %{
+              "url" => [
+                %{
+                  "mediaType" => "audio/basic",
+                  "href" => "http://www.gnu.org/music/free-software-song.au"
+                }
+              ]
+            }
+          ]
+        }
+      })
+
+    result = OpenGraph.build_tags(%{object: note, url: note.data["id"], user: user})
+
+    assert Enum.all?(
+             [
+               {:meta, [property: "og:image", content: "https://pleroma.gov/tenshi.png"], []},
+               {:meta,
+                [property: "og:audio", content: "http://www.gnu.org/music/free-software-song.au"],
+                []},
+               {:meta, [property: "og:video", content: "https://pleroma.gov/about/juche.webm"],
+                []}
+             ],
+             fn element -> element in result end
+           )
+  end
+
+  test "it does not render attachments if post is nsfw" do
+    Pleroma.Config.put([Pleroma.Web.Metadata, :unfurl_nsfw], false)
+    user = insert(:user, avatar: %{"url" => [%{"href" => "https://pleroma.gov/tenshi.png"}]})
+
+    note =
+      insert(:note, %{
+        data: %{
+          "actor" => user.ap_id,
+          "id" => "https://pleroma.gov/objects/whatever",
+          "content" => "#cuteposting #nsfw #hambaga",
+          "tag" => ["cuteposting", "nsfw", "hambaga"],
+          "sensitive" => true,
+          "attachment" => [
+            %{
+              "url" => [
+                %{"mediaType" => "image/png", "href" => "https://misskey.microsoft/corndog.png"}
+              ]
+            }
+          ]
+        }
+      })
+
+    result = OpenGraph.build_tags(%{object: note, url: note.data["id"], user: user})
+
+    assert {:meta, [property: "og:image", content: "https://pleroma.gov/tenshi.png"], []} in result
+
+    refute {:meta, [property: "og:image", content: "https://misskey.microsoft/corndog.png"], []} in result
+  end
+end

test/web/ostatus/ostatus_controller_test.exs

@@ -88,6 +88,7 @@ test "gets an object", %{conn: conn} do
 
     conn =
       conn
+      |> put_req_header("accept", "application/xml")
       |> get(url)
 
     expected =
@@ -114,31 +115,34 @@ test "404s on nonexisting objects", %{conn: conn} do
     |> response(404)
   end
 
+  test "gets an activity in xml format", %{conn: conn} do
+    note_activity = insert(:note_activity)
+    [_, uuid] = hd(Regex.scan(~r/.+\/([\w-]+)$/, note_activity.data["id"]))
+
+    conn
+    |> put_req_header("accept", "application/xml")
+    |> get("/activities/#{uuid}")
+    |> response(200)
+  end
+
   test "404s on deleted objects", %{conn: conn} do
     note_activity = insert(:note_activity)
     [_, uuid] = hd(Regex.scan(~r/.+\/([\w-]+)$/, note_activity.data["object"]["id"]))
     object = Object.get_by_ap_id(note_activity.data["object"]["id"])
 
     conn
+    |> put_req_header("accept", "application/xml")
     |> get("/objects/#{uuid}")
     |> response(200)
 
     Object.delete(object)
 
     conn
+    |> put_req_header("accept", "application/xml")
     |> get("/objects/#{uuid}")
     |> response(404)
   end
 
-  test "gets an activity", %{conn: conn} do
-    note_activity = insert(:note_activity)
-    [_, uuid] = hd(Regex.scan(~r/.+\/([\w-]+)$/, note_activity.data["id"]))
-
-    conn
-    |> get("/activities/#{uuid}")
-    |> response(200)
-  end
-
   test "404s on private activities", %{conn: conn} do
     note_activity = insert(:direct_note_activity)
     [_, uuid] = hd(Regex.scan(~r/.+\/([\w-]+)$/, note_activity.data["id"]))
@@ -154,7 +158,7 @@ test "404s on nonexistent activities", %{conn: conn} do
     |> response(404)
   end
 
-  test "gets a notice", %{conn: conn} do
+  test "gets a notice in xml format", %{conn: conn} do
     note_activity = insert(:note_activity)
 
     conn