Merge branch 'rich_media_titles_two' into 'develop'

Rich media titles two See merge request pleroma/pleroma!1325

Merge branch 'rich_media_titles_two' into 'develop'
Rich media titles two See merge request pleroma/pleroma!1325
27b9605d · kaniini · 8b7860f1 · 0e415921 · 27b9605d · 27b9605d
Commit 27b9605d authored Jun 23, 2019 by kaniini
4 changed files
--- a/lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex
+++ b/lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex
 defmodule Pleroma.Web.RichMedia.Parsers.MetaTagsParser do
  def parse(html, data, prefix, error_message, key_name, value_name \\ "content") do
-    with elements = [_ | _] <- get_elements(html, key_name, prefix),
-         meta_data =
-           Enum.reduce(elements, data, fn el, acc ->
-             attributes = normalize_attributes(el, prefix, key_name, value_name)
+    meta_data =
+      html
+      |> get_elements(key_name, prefix)
+      |> Enum.reduce(data, fn el, acc ->
+        attributes = normalize_attributes(el, prefix, key_name, value_name)

-             Map.merge(acc, attributes)
-           end) do
-      {:ok, meta_data}
+        Map.merge(acc, attributes)
+      end)
+      |> maybe_put_title(html)
+
+    if Enum.empty?(meta_data) do
+      {:error, error_message}
    else
-      _e -> {:error, error_message}
+      {:ok, meta_data}
    end
  end

@@ -27,4 +31,19 @@ defp normalize_attributes(html_node, prefix, key_name, value_name) do

    %{String.to_atom(data[key_name]) => data[value_name]}
  end
+
+  defp maybe_put_title(%{title: _} = meta, _), do: meta
+
+  defp maybe_put_title(meta, html) when meta != %{} do
+    case get_page_title(html) do
+      "" -> meta
+      title -> Map.put_new(meta, :title, title)
+    end
+  end
+
+  defp maybe_put_title(meta, _), do: meta
+
+  defp get_page_title(html) do
+    Floki.find(html, "title") |> Floki.text()
+  end
 end
--- a/test/fixtures/rich_media/non_ogp_embed.html
+++ b/test/fixtures/rich_media/non_ogp_embed.html
--- a/test/fixtures/rich_media/ogp-missing-title.html
+++ b/test/fixtures/rich_media/ogp-missing-title.html
+<html prefix="og: http://ogp.me/ns#">
+
+<head>
+  <title>The Rock (1996)</title>
+  <meta property="og:type" content="video.movie" />
+  <meta property="og:url" content="http://www.imdb.com/title/tt0117500/" />
+  <meta property="og:image" content="http://ia.media-imdb.com/images/rock.jpg" />
+  <meta property="og:description"
+    content="Directed by Michael Bay. With Sean Connery, Nicolas Cage, Ed Harris, John Spencer.">
+</head>
+
+</html>
--- a/test/web/rich_media/parser_test.exs
+++ b/test/web/rich_media/parser_test.exs
@@ -9,6 +9,21 @@ defmodule Pleroma.Web.RichMedia.ParserTest do
      } ->
        %Tesla.Env{status: 200, body: File.read!("test/fixtures/rich_media/ogp.html")}

+      %{
+        method: :get,
+        url: "http://example.com/non-ogp"
+      } ->
+        %Tesla.Env{status: 200, body: File.read!("test/fixtures/rich_media/non_ogp_embed.html")}
+
+      %{
+        method: :get,
+        url: "http://example.com/ogp-missing-title"
+      } ->
+        %Tesla.Env{
+          status: 200,
+          body: File.read!("test/fixtures/rich_media/ogp-missing-title.html")
+        }
+
      %{
        method: :get,
        url: "http://example.com/twitter-card"
@@ -38,6 +53,11 @@ test "returns error when no metadata present" do
    assert {:error, _} = Pleroma.Web.RichMedia.Parser.parse("http://example.com/empty")
  end

+  test "doesn't just add a title" do
+    assert Pleroma.Web.RichMedia.Parser.parse("http://example.com/non-ogp") ==
+             {:error, "Found metadata was invalid or incomplete: %{}"}
+  end
+
  test "parses ogp" do
    assert Pleroma.Web.RichMedia.Parser.parse("http://example.com/ogp") ==
             {:ok,
@@ -51,6 +71,19 @@ test "parses ogp" do
              }}
  end

+  test "falls back to <title> when ogp:title is missing" do
+    assert Pleroma.Web.RichMedia.Parser.parse("http://example.com/ogp-missing-title") ==
+             {:ok,
+              %{
+                image: "http://ia.media-imdb.com/images/rock.jpg",
+                title: "The Rock (1996)",
+                description:
+                  "Directed by Michael Bay. With Sean Connery, Nicolas Cage, Ed Harris, John Spencer.",
+                type: "video.movie",
+                url: "http://www.imdb.com/title/tt0117500/"
+              }}
+  end
+
  test "parses twitter card" do
    assert Pleroma.Web.RichMedia.Parser.parse("http://example.com/twitter-card") ==
             {:ok,