X-Git-Url: http://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fparser.ex;h=5727fda189bf39f4ac501461504ff247f1f86fa7;hb=0a05670b4499d170f6bc0eaecfae0d71be022227;hp=ba8dc6f2a4b0812a9d3a715ce36deefdd921b215;hpb=18234cc44e6bc989e3e3cf15714c54b4fa05b9dd;p=akkoma

diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index ba8dc6f2a..5727fda18 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -1,14 +1,9 @@
 # Pleroma: A lightweight social networking server
-# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# Copyright © 2017-2020 Pleroma Authors <https://pleroma.social/>
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.Web.RichMedia.Parser do
-  @hackney_options [
-    pool: :media,
-    recv_timeout: 2_000,
-    max_body: 2_000_000,
-    with_body: true
-  ]
+  require Logger
 
   defp parsers do
     Pleroma.Config.get([:rich_media, :parsers])
@@ -17,17 +12,34 @@ defmodule Pleroma.Web.RichMedia.Parser do
   def parse(nil), do: {:error, "No URL provided"}
 
   if Pleroma.Config.get(:env) == :test do
+    @spec parse(String.t()) :: {:ok, map()} | {:error, any()}
     def parse(url), do: parse_url(url)
   else
+    @spec parse(String.t()) :: {:ok, map()} | {:error, any()}
     def parse(url) do
-      try do
-        Cachex.fetch!(:rich_media_cache, url, fn _ ->
-          {:commit, parse_url(url)}
-        end)
-        |> set_ttl_based_on_image(url)
-      rescue
-        e ->
-          {:error, "Cachex error: #{inspect(e)}"}
+      with {:ok, data} <- get_cached_or_parse(url),
+           {:ok, _} <- set_ttl_based_on_image(data, url) do
+        {:ok, data}
+      else
+        {:error, {:invalid_metadata, data}} = e ->
+          Logger.debug(fn -> "Incomplete or invalid metadata for #{url}: #{inspect(data)}" end)
+          e
+
+        error ->
+          Logger.error(fn -> "Rich media error for #{url}: #{inspect(error)}" end)
+          error
+      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, _} = e ->
+          ttl = Pleroma.Config.get([:rich_media, :failure_backoff], 60_000)
+          Cachex.expire(:rich_media_cache, url, ttl)
+          e
       end
     end
   end
@@ -35,17 +47,17 @@ defmodule Pleroma.Web.RichMedia.Parser do
   @doc """
   Set the rich media cache based on the expiration time of image.
 
-  Define a module that has `run` function
+  Adopt behaviour `Pleroma.Web.RichMedia.Parser.TTL`
 
   ## Example
 
       defmodule MyModule do
-        def run(data, url) do
+        @behaviour Pleroma.Web.RichMedia.Parser.TTL
+        def ttl(data, url) do
           image_url = Map.get(data, :image)
           # do some parsing in the url and get the ttl of the image
-          # ttl is unix time
-          ttl = parse_ttl_from_url(image_url)  
-          Cachex.expire_at(:rich_media_cache, url, ttl * 1000)
+          # and return ttl is unix time
+          parse_ttl_from_url(image_url)
         end
       end
 
@@ -54,63 +66,68 @@ defmodule Pleroma.Web.RichMedia.Parser do
       config :pleroma, :rich_media,
         ttl_setters: [MyModule]
   """
-  def set_ttl_based_on_image({:ok, data}, url) do
-    case Cachex.ttl(:rich_media_cache, url) do
-      {:ok, nil} ->
-        modules = Pleroma.Config.get([:rich_media, :ttl_setters])
-
-        if Enum.count(modules) > 0 do
-          Enum.each(modules, & &1.run(data, url))
+  @spec set_ttl_based_on_image(map(), String.t()) ::
+          {:ok, Integer.t() | :noop} | {:error, :no_key}
+  def set_ttl_based_on_image(data, url) do
+    case get_ttl_from_image(data, url) do
+      {:ok, ttl} when is_number(ttl) ->
+        ttl = ttl * 1000
+
+        case Cachex.expire_at(:rich_media_cache, url, ttl) do
+          {:ok, true} -> {:ok, ttl}
+          {:ok, false} -> {:error, :no_key}
         end
 
-        {:ok, data}
-
       _ ->
-        {:ok, data}
+        {:ok, :noop}
     end
   end
 
-  def set_ttl_based_on_image(data, _url), do: data
+  defp get_ttl_from_image(data, url) do
+    [:rich_media, :ttl_setters]
+    |> Pleroma.Config.get()
+    |> Enum.reduce({:ok, nil}, fn
+      module, {:ok, _ttl} ->
+        module.ttl(data, url)
 
-  defp parse_url(url) do
-    try do
-      {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: @hackney_options)
+      _, error ->
+        error
+    end)
+  end
 
+  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
       |> maybe_parse()
+      |> Map.put("url", url)
       |> clean_parsed_data()
       |> check_parsed_data()
-    rescue
-      e ->
-        {:error, "Parsing error: #{inspect(e)}"}
     end
   end
 
   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
 
   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
     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