Change extractors to accept/return html trees

lib/frenzy/pipeline/extractor.ex

@@ -1,3 +1,3 @@
 defmodule Frenzy.Pipeline.Extractor do
-  @callback extract(String.t()) :: {:ok, String.t()} | {:error, String.t()}
+  @callback extract(Floki.html_tree()) :: {:ok, Floki.html_tree()} | {:error, String.t()}
 end

lib/frenzy/pipeline/extractor/daring_fireball.ex

@@ -7,15 +7,13 @@ defmodule Frenzy.Pipeline.Extractor.DaringFireball do
   @behaviour Extractor
 
   @impl Extractor
-  def extract(body) do
+  def extract(html_tree) 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)}
+        {:ok, elem}
     end
   end
 

lib/frenzy/pipeline/extractor/erica_sadun.ex

@@ -7,18 +7,14 @@ defmodule Frenzy.Pipeline.Extractor.EricaSadun do
   @behaviour Extractor
 
   @impl Extractor
-  def extract(body) do
+  def extract(html_tree) do
-    html_tree = Floki.parse(body)
-
     case Floki.find(html_tree, ".post-content") do
       [content_elem | _] ->
         # content element includes social media buttons and related posts
-        content =
+        {
-          content_elem
+          :ok,
-          |> Floki.filter_out("div.sharedaddy, div#jp-relatedposts")
+          Floki.filter_out(content_elem, "div.sharedaddy, div#jp-relatedposts")
-          |> Floki.raw_html()
+        }
-
-        {:ok, content}
 
       _ ->
         {:error, "no matching elements"}

lib/frenzy/pipeline/extractor/whatever_scalzi.ex

@@ -7,15 +7,13 @@ defmodule Frenzy.Pipeline.Extractor.WhateverScale do
   @behaviour Extractor
 
   @impl Extractor
-  def extract(body) do
+  def extract(html_tree) do
-    html_tree = Floki.parse(body)
-
     case get_article_content(html_tree) do
       nil ->
         {:error, "no matching elements"}
 
       elem ->
-        {:ok, Floki.raw_html(elem)}
+        {:ok, elem}
     end
   end
 

lib/frenzy/pipeline/scrape_stage.ex

@@ -68,7 +68,16 @@ defmodule Frenzy.Pipeline.ScrapeStage do
         {:ok, Readability.readable_html(article)}
 
       module_name ->
-        apply(String.to_existing_atom("Elixir." <> module_name), :extract, [body])
+        html_tree = Floki.parse(body)
+
+        case apply(String.to_existing_atom("Elixir." <> module_name), :extract, [html_tree]) do
+          {:error, _} = err ->
+            err
+
+          {:ok, html_tree} ->
+            html_tree
+            |> Floki.raw_html()
+        end
     end
   end