Compare commits

..

No commits in common. "04ffe0036dd568d1d9a441e590535edd58ce9157" and "5f3be5213254446c83672d1c6796d01c1d20ff6e" have entirely different histories.

2 changed files with 6 additions and 13 deletions

View File

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

View File

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