Re-fetch user data on salmon decode error.
[akkoma] / lib / pleroma / web / ostatus / ostatus.ex
index d21b9078f4bc81799c7aee86e648de56171ab177..b0d2dda5d2456afeab7d9fbd3620776339ae78b7 100644 (file)
@@ -1,5 +1,15 @@
 defmodule Pleroma.Web.OStatus do
-  alias Pleroma.Web
+  @httpoison Application.get_env(:pleroma, :httpoison)
+
+  import Ecto.Query
+  import Pleroma.Web.XML
+  require Logger
+
+  alias Pleroma.{Repo, User, Web, Object, Activity}
+  alias Pleroma.Web.ActivityPub.ActivityPub
+  alias Pleroma.Web.ActivityPub.Utils
+  alias Pleroma.Web.{WebFinger, Websub}
+  alias Pleroma.Web.OStatus.{FollowHandler, NoteHandler, DeleteHandler}
 
   def feed_path(user) do
     "#{user.ap_id}/feed.atom"
@@ -9,6 +19,308 @@ defmodule Pleroma.Web.OStatus do
     "#{Web.base_url}/push/hub/#{user.nickname}"
   end
 
-  def user_path(user) do
+  def salmon_path(user) do
+    "#{user.ap_id}/salmon"
+  end
+
+  def handle_incoming(xml_string) do
+    with doc when doc != :error <- parse_document(xml_string) do
+      entries = :xmerl_xpath.string('//entry', doc)
+
+      activities = Enum.map(entries, fn (entry) ->
+        {:xmlObj, :string, object_type} = :xmerl_xpath.string('string(/entry/activity:object-type[1])', entry)
+        {:xmlObj, :string, verb} = :xmerl_xpath.string('string(/entry/activity:verb[1])', entry)
+        Logger.debug("Handling #{verb}")
+
+        try do
+          case verb do
+            'http://activitystrea.ms/schema/1.0/delete' ->
+              with {:ok, activity} <- DeleteHandler.handle_delete(entry, doc), do: activity
+            'http://activitystrea.ms/schema/1.0/follow' ->
+              with {:ok, activity} <- FollowHandler.handle(entry, doc), do: activity
+            'http://activitystrea.ms/schema/1.0/share' ->
+              with {:ok, activity, retweeted_activity} <- handle_share(entry, doc), do: [activity, retweeted_activity]
+            'http://activitystrea.ms/schema/1.0/favorite' ->
+              with {:ok, activity, favorited_activity} <- handle_favorite(entry, doc), do: [activity, favorited_activity]
+            _ ->
+              case object_type do
+                'http://activitystrea.ms/schema/1.0/note' ->
+                  with {:ok, activity} <- NoteHandler.handle_note(entry, doc), do: activity
+                'http://activitystrea.ms/schema/1.0/comment' ->
+                  with {:ok, activity} <- NoteHandler.handle_note(entry, doc), do: activity
+                _ ->
+                  Logger.error("Couldn't parse incoming document")
+                  nil
+              end
+          end
+        rescue
+          e ->
+            Logger.error("Error occured while handling activity")
+            Logger.error(xml_string)
+          Logger.error(inspect(e))
+          nil
+        end
+      end)
+      |> Enum.filter(&(&1))
+
+      {:ok, activities}
+    else
+      _e -> {:error, []}
+    end
+  end
+
+  def make_share(entry, doc, retweeted_activity) do
+    with {:ok, actor} <- find_make_or_update_user(doc),
+         %Object{} = object <- Object.get_by_ap_id(retweeted_activity.data["object"]["id"]),
+         id when not is_nil(id) <- string_from_xpath("/entry/id", entry),
+         {:ok, activity, _object} = ActivityPub.announce(actor, object, id, false) do
+      {:ok, activity}
+    end
+  end
+
+  def handle_share(entry, doc) do
+    with {:ok, retweeted_activity} <- get_or_build_object(entry),
+         {:ok, activity} <- make_share(entry, doc, retweeted_activity) do
+      {:ok, activity, retweeted_activity}
+    else
+      e -> {:error, e}
+    end
+  end
+
+  def make_favorite(entry, doc, favorited_activity) do
+    with {:ok, actor} <- find_make_or_update_user(doc),
+         %Object{} = object <- Object.get_by_ap_id(favorited_activity.data["object"]["id"]),
+         id when not is_nil(id) <- string_from_xpath("/entry/id", entry),
+         {:ok, activity, _object} = ActivityPub.like(actor, object, id, false) do
+      {:ok, activity}
+    end
+  end
+
+  def get_or_build_object(entry) do
+    with {:ok, activity} <- get_or_try_fetching(entry) do
+      {:ok, activity}
+    else
+      _e ->
+        with [object] <- :xmerl_xpath.string('/entry/activity:object', entry) do
+          NoteHandler.handle_note(object, object)
+        end
+    end
+  end
+
+  def get_or_try_fetching(entry) do
+    Logger.debug("Trying to get entry from db")
+    with id when not is_nil(id) <- string_from_xpath("//activity:object[1]/id", entry),
+         %Activity{} = activity <- Activity.get_create_activity_by_object_ap_id(id) do
+      {:ok, activity}
+    else e ->
+        Logger.debug("Couldn't get, will try to fetch")
+        with href when not is_nil(href) <- string_from_xpath("//activity:object[1]/link[@type=\"text/html\"]/@href", entry),
+             {:ok, [favorited_activity]} <- fetch_activity_from_url(href) do
+          {:ok, favorited_activity}
+        else e -> Logger.debug("Couldn't find href: #{inspect(e)}")
+        end
+    end
+  end
+
+  def handle_favorite(entry, doc) do
+    with {:ok, favorited_activity} <- get_or_try_fetching(entry),
+         {:ok, activity} <- make_favorite(entry, doc, favorited_activity) do
+      {:ok, activity, favorited_activity}
+    else
+      e -> {:error, e}
+    end
+  end
+
+  def get_attachments(entry) do
+    :xmerl_xpath.string('/entry/link[@rel="enclosure"]', entry)
+    |> Enum.map(fn (enclosure) ->
+      with href when not is_nil(href) <- string_from_xpath("/link/@href", enclosure),
+           type when not is_nil(type) <- string_from_xpath("/link/@type", enclosure) do
+        %{
+          "type" => "Attachment",
+          "url" => [%{
+                       "type" => "Link",
+                       "mediaType" => type,
+                       "href" => href
+                    }]
+        }
+      end
+    end)
+    |> Enum.filter(&(&1))
+  end
+
+  @doc """
+    Gets the content from a an entry.
+  """
+  def get_content(entry) do
+    string_from_xpath("//content", entry)
+  end
+
+  @doc """
+    Get the cw that mastodon uses.
+  """
+  def get_cw(entry) do
+    with scope when not is_nil(scope) <- string_from_xpath("//mastodon:scope", entry),
+         cw when not is_nil(cw) <- string_from_xpath("/*/summary", entry) do
+      cw
+    else _e -> nil
+    end
+  end
+
+  def get_tags(entry) do
+    :xmerl_xpath.string('//category', entry)
+    |> Enum.map(fn (category) -> string_from_xpath("/category/@term", category) end)
+    |> Enum.filter(&(&1))
+    |> Enum.map(&String.downcase/1)
+  end
+
+  def maybe_update(doc, user) do
+    old_data = %{
+      avatar: user.avatar,
+      bio: user.bio,
+      name: user.name,
+      info: user.info
+    }
+
+    with false <- user.local,
+         avatar <- make_avatar_object(doc),
+         bio <- string_from_xpath("//author[1]/summary", doc),
+         name <- string_from_xpath("//author[1]/poco:displayName", doc),
+         info <- Map.put(user.info, "banner", make_avatar_object(doc, "header") || user.info["banner"]),
+         new_data <- %{avatar: avatar || old_data.avatar, name: name || old_data.name, bio: bio || old_data.bio, info: info || old_data.info},
+         false <- new_data == old_data do
+      change = Ecto.Changeset.change(user, new_data)
+      Repo.update(change)
+    else e ->
+      {:ok, user}
+    end
+  end
+
+  def find_make_or_update_user(doc) do
+    uri = string_from_xpath("//author/uri[1]", doc)
+    with {:ok, user} <- find_or_make_user(uri) do
+      maybe_update(doc, user)
+    end
+  end
+
+  def find_or_make_user(uri) do
+    query = from user in User,
+      where: user.ap_id == ^uri
+
+    user = Repo.one(query)
+
+    if is_nil(user) do
+      make_user(uri)
+    else
+      {:ok, user}
+    end
+  end
+
+  def insert_or_update_user(data) do
+    cs = User.remote_user_creation(data)
+    Repo.insert(cs, on_conflict: :replace_all, conflict_target: :nickname)
+  end
+
+  def make_user(uri, update \\ false) do
+    with {:ok, info} <- gather_user_info(uri) do
+      data = %{
+        name: info["name"],
+        nickname: info["nickname"] <> "@" <> info["host"],
+        ap_id: info["uri"],
+        info: info,
+        avatar: info["avatar"],
+        bio: info["bio"]
+      }
+      with false <- update,
+           %User{} = user <- User.get_by_ap_id(data.ap_id) do
+        {:ok, user}
+      else _e -> insert_or_update_user(data)
+      end
+    end
+  end
+
+  # TODO: Just takes the first one for now.
+  def make_avatar_object(author_doc, rel \\ "avatar") do
+    href = string_from_xpath("//author[1]/link[@rel=\"#{rel}\"]/@href", author_doc)
+    type = string_from_xpath("//author[1]/link[@rel=\"#{rel}\"]/@type", author_doc)
+
+    if href do
+      %{
+        "type" => "Image",
+        "url" =>
+          [%{
+              "type" => "Link",
+              "mediaType" => type,
+              "href" => href
+           }]
+      }
+    else
+      nil
+    end
+  end
+
+  def gather_user_info(username) do
+    with {:ok, webfinger_data} <- WebFinger.finger(username),
+         {:ok, feed_data} <- Websub.gather_feed_data(webfinger_data["topic"]) do
+      {:ok, Map.merge(webfinger_data, feed_data) |> Map.put("fqn", username)}
+    else e ->
+      Logger.debug(fn -> "Couldn't gather info for #{username}" end)
+      {:error, e}
+    end
+  end
+
+  # Regex-based 'parsing' so we don't have to pull in a full html parser
+  # It's a hack anyway. Maybe revisit this in the future
+  @mastodon_regex ~r/<link href='(.*)' rel='alternate' type='application\/atom\+xml'>/
+  @gs_regex ~r/<link title=.* href="(.*)" type="application\/atom\+xml" rel="alternate">/
+  @gs_classic_regex ~r/<link rel="alternate" href="(.*)" type="application\/atom\+xml" title=.*>/
+  def get_atom_url(body) do
+    cond do
+      Regex.match?(@mastodon_regex, body) ->
+        [[_, match]] = Regex.scan(@mastodon_regex, body)
+        {:ok, match}
+      Regex.match?(@gs_regex, body) ->
+        [[_, match]] = Regex.scan(@gs_regex, body)
+        {:ok, match}
+      Regex.match?(@gs_classic_regex, body) ->
+        [[_, match]] = Regex.scan(@gs_classic_regex, body)
+        {:ok, match}
+      true ->
+        Logger.debug(fn -> "Couldn't find atom link in #{inspect(body)}" end)
+        {:error, "Couldn't find the atom link"}
+    end
+  end
+
+  def fetch_activity_from_atom_url(url) do
+    with {:ok, %{body: body, status_code: code}} when code in 200..299 <- @httpoison.get(url, [Accept: "application/atom+xml"], follow_redirect: true, timeout: 10000, recv_timeout: 20000) do
+      Logger.debug("Got document from #{url}, handling...")
+      handle_incoming(body)
+    else e -> Logger.debug("Couldn't get #{url}: #{inspect(e)}")
+    end
+  end
+
+  def fetch_activity_from_html_url(url) do
+    Logger.debug("Trying to fetch #{url}")
+    with {:ok, %{body: body}} <- @httpoison.get(url, [], follow_redirect: true, timeout: 10000, recv_timeout: 20000),
+         {:ok, atom_url} <- get_atom_url(body) do
+        fetch_activity_from_atom_url(atom_url)
+    else e -> Logger.debug("Couldn't get #{url}: #{inspect(e)}")
+    end
+  end
+
+  def fetch_activity_from_url(url) do
+    try do
+      with {:ok, activities} when length(activities) > 0 <- fetch_activity_from_atom_url(url) do
+        {:ok, activities}
+      else
+        _e -> with {:ok, activities} <- fetch_activity_from_html_url(url) do
+                {:ok, activities}
+              end
+      end
+    rescue
+      e ->
+        Logger.debug("Couldn't get #{url}: #{inspect(e)}")
+        {:error, "Couldn't get #{url}: #{inspect(e)}"}
+    end
   end
 end