lib/frenzy/pipeline/extractor/whatever_scalzi.ex

@@ -18,17 +18,11 @@ defmodule Frenzy.Pipeline.Extractor.WhateverScalzi do
   end
 
   defp get_article_content(html_tree) do
-    # there's no element that contains only the post content
-    # .postarea contains the headline, post content, social media buttons, and comments
-    with [{_tag, _attrs, postarea_children} | _] <- Floki.find(html_tree, ".postarea"),
-         {_before_headline, [_headline | rest]} <-
-           Enum.split_while(postarea_children, fn {tag, _attrs, _children} -> tag != "h1" end),
-         {article_content, _rest} <-
-           Enum.split_while(rest, fn {tag, attrs, _children} ->
-             tag != "div" || !({"id", "jp-post-flair"} in attrs)
-           end) do
-      article_content
-    else
+    case Floki.find(html_tree, "article.post > div.entry-content") do
+      [content_elem | _] ->
+        # remove social media buttons that are included in the .entry-content element
+        Floki.filter_out(content_elem, "div#jp-post-flair")
+
       _ ->
         nil
     end

lib/frenzy/task/fetch_favicon.ex

@@ -25,8 +25,7 @@ defmodule Frenzy.Task.FetchFavicon do
       favicon_url =
         fetch_favicon_url_from_webpage(site_url) || URI.merge(site_url, "/favicon.ico")
 
-      with true <- is_binary(favicon_url),
-           %Feed{favicon_url: old_url} when old_url != favicon_url <- feed,
+      with %Feed{favicon_url: old_url} when old_url != favicon_url <- feed,
            {:ok, favicon_data} <- fetch_favicon_data(favicon_url) do
         changeset =
           Feed.changeset(feed, %{