From 4308939726cdbd202f364500dea62bcf6e9bbdb1 Mon Sep 17 00:00:00 2001
From: Shadowfacts <me@shadowfacts.net>
Date: Sun, 1 Sep 2019 16:32:36 -0400
Subject: [PATCH] Add support for parsing multiple links

---
 lib/item.ex               |  3 ++-
 lib/parser/atom.ex        | 25 +++++++++++++++++++++++--
 lib/parser/jsonfeed.ex    |  1 +
 lib/parser/rss2.ex        |  1 +
 lib/parser/rssinjson.ex   |  1 +
 test/parser/atom_test.exs | 17 +++++++++++++++++
 6 files changed, 45 insertions(+), 3 deletions(-)

diff --git a/lib/item.ex b/lib/item.ex
index bc42038..2e77ac7 100644
--- a/lib/item.ex
+++ b/lib/item.ex
@@ -3,11 +3,12 @@ defmodule FeedParser.Item do
   A item in a feed. Has metadata and content from the item.
   """
 
-  defstruct [:guid, :url, :title, :content, :date]
+  defstruct [:guid, :url, :links, :title, :content, :date]
 
   @type t() :: %__MODULE__{
           guid: String.t(),
           url: String.t() | nil,
+          links: [{href :: String.t(), rel :: String.t()} | href :: String.t()],
           title: String.t() | nil,
           content: String.t(),
           date: DateTime.t()
diff --git a/lib/parser/atom.ex b/lib/parser/atom.ex
index c4ff0b1..687b6b8 100644
--- a/lib/parser/atom.ex
+++ b/lib/parser/atom.ex
@@ -47,7 +47,27 @@ defmodule FeedParser.Parser.Atom do
       |> Enum.map(fn entry ->
         id = text('/entry/id/text()', entry)
         title = text('/entry/title/text()', entry)
-        link = attr('/entry/link/@href', entry)
+
+        links =
+          :xmerl_xpath.string('/entry/link', entry)
+          |> Enum.map(fn link ->
+            value = attr('/link/@href', link)
+
+            case attr('/link/@rel', link) do
+              nil -> value
+              rel -> {value, rel}
+            end
+          end)
+
+        url =
+          (Enum.find(links, fn
+             {value, rel} -> rel == "alternate"
+             _value -> false
+           end) || List.first(links))
+          |> case do
+            url when is_binary(url) -> url
+            {url, _rel} -> url
+          end
 
         updated =
           text('/entry/updated/text()', entry)
@@ -62,7 +82,8 @@ defmodule FeedParser.Parser.Atom do
         %FeedParser.Item{
           guid: id,
           title: title,
-          url: link,
+          url: url,
+          links: links,
           content: content,
           date: updated
         }
diff --git a/lib/parser/jsonfeed.ex b/lib/parser/jsonfeed.ex
index ac6375a..55d6348 100644
--- a/lib/parser/jsonfeed.ex
+++ b/lib/parser/jsonfeed.ex
@@ -48,6 +48,7 @@ defmodule FeedParser.Parser.JSONFeed do
         %FeedParser.Item{
           guid: id,
           url: url,
+          links: [url],
           title: title,
           content: content,
           date: date
diff --git a/lib/parser/rss2.ex b/lib/parser/rss2.ex
index b8c9f4f..f965694 100644
--- a/lib/parser/rss2.ex
+++ b/lib/parser/rss2.ex
@@ -63,6 +63,7 @@ defmodule FeedParser.Parser.RSS2 do
           guid: guid,
           title: title,
           url: link,
+          links: [link],
           content: description,
           date: pubDate
         }
diff --git a/lib/parser/rssinjson.ex b/lib/parser/rssinjson.ex
index f4966f7..40603da 100644
--- a/lib/parser/rssinjson.ex
+++ b/lib/parser/rssinjson.ex
@@ -56,6 +56,7 @@ defmodule FeedParser.Parser.RSSInJSON do
         %FeedParser.Item{
           guid: guid,
           url: link,
+          links: [link],
           title: title,
           content: content,
           date: pubDate
diff --git a/test/parser/atom_test.exs b/test/parser/atom_test.exs
index 75f7428..b98f5e0 100644
--- a/test/parser/atom_test.exs
+++ b/test/parser/atom_test.exs
@@ -23,4 +23,21 @@ defmodule FeedParser.Parser.AtomTest do
     assert item.date == ~U[2003-12-13 18:30:02Z]
     assert item.content == "Some text."
   end
+
+  test "parses atom entry with multiple links" do
+    data = File.read!("test/fixtures/atom/multi_url.xml")
+    {true, parsed_data} = Atom.accepts(data, "application/atom+xml")
+    {:ok, %FeedParser.Feed{} = feed} = Atom.parse_feed(parsed_data)
+    [item] = feed.items
+
+    assert item.url ==
+             "https://techcrunch.com/2019/08/30/someone-hacked-jack-dorseys-own-twitter-account/"
+
+    assert item.links == [
+             {"https://techcrunch.com/2019/08/30/someone-hacked-jack-dorseys-own-twitter-account/",
+              "alternate"},
+             {"http://df4.us/rrx", "shorturl"},
+             {"https://daringfireball.net/linked/2019/08/30/dorsey-twitter-account", "related"}
+           ]
+  end
 end