Ignore hashtags and mentions...

...when extracting links.

Plus minor gem updates.

Author: oneiros

Gemfile.lock

@@ -315,7 +315,7 @@ GEM
     ruby-progressbar (1.13.0)
     rubyzip (2.4.1)
     securerandom (0.4.1)
-    selenium-webdriver (4.28.0)
+    selenium-webdriver (4.29.0)
       base64 (~> 0.2)
       logger (~> 1.4)
       rexml (~> 3.2, >= 3.2.5)
@@ -336,7 +336,7 @@ GEM
       base64
     stimulus-rails (1.3.4)
       railties (>= 6.0.0)
-    stringio (3.1.3)
+    stringio (3.1.5)
     tailwindcss-rails (3.3.1)
       railties (>= 7.0.0)
       tailwindcss-ruby (~> 3.0)

app/models/content_object.rb

@@ -31,7 +31,7 @@ class ContentObject < ApplicationRecord
   class << self
     def json_to_attributes(json_object)
       hashtags = json_object["tag"].filter_map { |t| t["name"] if t["type"] == "Hashtag" }
-      links = Nokogiri::HTML5::DocumentFragment.parse(json_object["content"]).css("a[href]").map { |a| a["href"] }
+      links = LinkExtractor.new(json_object).extracted_urls
       {
         object_type: json_object["type"],
         published_at: json_object["published"],

app/models/content_object/link_extractor.rb

@@ -0,0 +1,28 @@
+class ContentObject::LinkExtractor
+  def initialize(content_json)
+    @content_json = content_json
+  end
+
+  def extracted_urls
+    @extracted_urls ||= extract_urls
+  end
+
+  private
+
+  def extract_urls
+    parsed_fragment.css("a[href]:not([rel~=tag]):not(.u-url)").filter_map do |a|
+      a["href"] unless mention?(a["href"])
+    end
+  end
+
+  def parsed_fragment
+    Nokogiri::HTML5::DocumentFragment.parse(@content_json["content"])
+  end
+
+  def mention?(uri)
+    @mentions ||= (@content_json["tag"] || []).filter_map do |tag|
+      tag["href"] if tag["type"] == "Mention"
+    end
+    @mentions.include?(uri)
+  end
+end

test/models/content_object/link_extractor_test.rb

@@ -0,0 +1,29 @@
+require "test_helper"
+
+class ContentObject::LinkExtractorTest < ActiveSupport::TestCase
+  test "#extracted_urls returns an array of links ignoring hashtags and mentions" do
+    content_object = mock_content_object(content:)
+    content_object["tag"] << mention
+
+    extracted_urls = ContentObject::LinkExtractor.new(content_object).extracted_urls
+
+    assert_equal 1, extracted_urls.size
+    assert_equal "https://example.com/home", extracted_urls.first
+  end
+
+  private
+
+  def content
+    <<~HTML
+      <p>Test</p><p><a href="https://fedi.example.com/tags/testtag" class="mention hashtag" rel="tag">#<span>testtag</span></a> test <span class="h-card" translate="no"><a href="https://fedi.example.com/@user1" class="u-url mention">@<span>user1</span></a></span><a href="https://example.com/home" target="_blank" rel="nofollow noopener" translate="no"><span class="invisible">https://</span><span class="">example.com/home</span><span class="invisible"></span></a> test <a href="https://fedi.example.com/@user2">@user2</a>
+    HTML
+  end
+
+  def mention
+    {
+      "type" => "Mention",
+      "href" => "https://fedi.example.com/@user2",
+      "name" => "@[email protected]"
+    }
+  end
+end