Generalize WP lazy-loading stripper

lib/frenzy/pipeline/extractor/om_malik.ex

@@ -19,5 +19,6 @@ defmodule Frenzy.Pipeline.Extractor.OmMalik do
       _ ->
         {:error, "no matching elements"}
     end
+    |> Extractor.Util.strip_wp_lazy_loading()
   end
 end

lib/frenzy/pipeline/extractor/util.ex

@@ -0,0 +1,28 @@
+defmodule Frenzy.Pipeline.Extractor.Util do
+  @doc """
+  WordPress Jetpack uses a 1x1 pixel transparent gif in a srcset to keep browsers from loading images
+  by overriding the src attribute. We want to strip those so the images actually load.
+  """
+  @spec strip_wp_lazy_loading(Floki.html_tree()) :: Floki.html_tree()
+  def strip_wp_lazy_loading(tree) do
+    Floki.map(tree, fn
+      {"img", attrs} = el ->
+        class = Enum.find(attrs, fn {k, _} -> k == "class" end)
+
+        if !is_nil(class) && String.contains?(elem(class, 1), "jetpack-lazy-image") do
+          {
+            "img",
+            Enum.filter(attrs, fn
+              {"srcset", _} -> false
+              _ -> true
+            end)
+          }
+        else
+          el
+        end
+
+      el ->
+        el
+    end)
+  end
+end

lib/frenzy/pipeline/extractor/whatever_scalzi.ex

@@ -43,27 +43,7 @@ defmodule Frenzy.Pipeline.Extractor.WhateverScalzi do
                   true
               end)
 
-            article_content
-            |> Floki.map(fn
-              {"img", attrs} = el ->
-                class = Enum.find(attrs, fn {k, _} -> k == "class" end)
-                class = if is_nil(class), do: nil, else: elem(class, 1)
-
-                if !is_nil(class) && String.contains?(class, "jetpack-lazy-image") do
-                  {
-                    "img",
-                    Enum.filter(attrs, fn
-                      {"srcset", _} -> false
-                      _ -> true
-                    end)
-                  }
-                else
-                  el
-                end
-
-              el ->
-                el
-            end)
+            Extractor.Util.strip_wp_lazy_loading(article_content)
 
           _ ->
             nil