From cfd9f7505a91f82dee8cbad730e5c0a87bd705c1 Mon Sep 17 00:00:00 2001
From: Shadowfacts <me@shadowfacts.net>
Date: Thu, 31 Oct 2019 17:38:16 -0400
Subject: [PATCH] Rewrite image URLs without hosts to use the host of the
 article URL

---
 lib/frenzy/pipeline/scrape_stage.ex | 45 +++++++++++++++++++++++------
 1 file changed, 36 insertions(+), 9 deletions(-)

diff --git a/lib/frenzy/pipeline/scrape_stage.ex b/lib/frenzy/pipeline/scrape_stage.ex
index 0caadcd..c6d2363 100644
--- a/lib/frenzy/pipeline/scrape_stage.ex
+++ b/lib/frenzy/pipeline/scrape_stage.ex
@@ -61,23 +61,29 @@ defmodule Frenzy.Pipeline.ScrapeStage do
 
   @spec handle_response(String.t(), HTTPoison.Response.t(), String.t()) ::
           {:ok, String.t()} | {:error, String.t()}
-  defp handle_response(_url, %HTTPoison.Response{status_code: 200, body: body}, extractor) do
+  defp handle_response(url, %HTTPoison.Response{status_code: 200, body: body}, extractor) do
     case extractor do
       "builtin" ->
-        article = Readability.article(body)
-        {:ok, Readability.readable_html(article)}
+        {:ok, Readability.article(body)}
 
       module_name ->
         html_tree = Floki.parse(body)
+        apply(String.to_existing_atom("Elixir." <> module_name), :extract, [html_tree])
+    end
+    |> case do
+      {:ok, html} ->
+        html = Floki.map(html, rewrite_image_urls(URI.parse(url)))
 
-        case apply(String.to_existing_atom("Elixir." <> module_name), :extract, [html_tree]) do
-          {:error, _} = err ->
-            err
+        case extractor do
+          "builtin" ->
+            {:ok, Readability.readable_html(html)}
 
-          {:ok, html_tree} ->
-            html_tree
-            |> Floki.raw_html()
+          _ ->
+            {:ok, Floki.raw_html(html)}
         end
+
+      res ->
+        res
     end
   end
 
@@ -110,4 +116,25 @@ defmodule Frenzy.Pipeline.ScrapeStage do
   defp handle_response(_url, %HTTPoison.Response{} = response, _extractor) do
     {:error, "No handler for response #{inspect(response)}"}
   end
+
+  # Generates a helper function for the article with the given URI that takes an HTML element and,
+  # if it's an <img> element whose src attribute does not have a hostname, adds the hostname and
+  # scheme to the element.
+  defp rewrite_image_urls(%URI{host: host, scheme: scheme}) do
+    fn
+      {"img", [{"src", src} | attrs]} = elem ->
+        case URI.parse(src) do
+          %URI{host: nil, path: path} ->
+            new_src = URI.to_string(%URI{path: path, host: host, scheme: scheme})
+
+            {"img", [{"src", new_src} | attrs]}
+
+          _ ->
+            elem
+        end
+
+      elem ->
+        elem
+    end
+  end
 end