frenzy/lib/frenzy/pipeline/extractor/daring_fireball.ex

39 lines
855 B
Elixir

defmodule Frenzy.Pipeline.Extractor.DaringFireball do
alias Frenzy.Pipeline.Extractor
@behaviour Extractor
@impl Extractor
def extract(body) do
html_tree = Floki.parse(body)
case get_article_element(html_tree) || get_link_element(html_tree) do
nil ->
{:error, "no matching elements"}
elem ->
{:ok, Floki.raw_html(elem)}
end
end
defp get_article_element(html_tree) do
case Floki.find(html_tree, "div.article") do
[article_elem | _] ->
# articles include extra information in the div.article element
Floki.filter_out(article_elem, "h1, .dateline, #PreviousNext")
_ ->
nil
end
end
defp get_link_element(html_tree) do
case Floki.find(html_tree, "dl.linkedlist dd") do
[dd_elem | _] ->
dd_elem
_ ->
nil
end
end
end