From 4f2a6e4205b88f9d6d9c8901edd155c9df051ab9 Mon Sep 17 00:00:00 2001
From: Pablo Brasero <pablo@pablobm.com>
Date: Thu, 20 Nov 2025 11:55:27 +0000
Subject: [PATCH] Move into submodule ahead of new type of scorer

---
 lib/spam_scorer.rb           | 36 +++------------------------------
 lib/spam_scorer/rich_text.rb | 39 ++++++++++++++++++++++++++++++++++++
 test/lib/spam_scorer_test.rb |  6 +++---
 3 files changed, 45 insertions(+), 36 deletions(-)
 create mode 100644 lib/spam_scorer/rich_text.rb

diff --git a/lib/spam_scorer.rb b/lib/spam_scorer.rb
index 61a62ca0f..546a7d824 100644
--- a/lib/spam_scorer.rb
+++ b/lib/spam_scorer.rb
@@ -1,41 +1,11 @@
 # frozen_string_literal: true
 
-class SpamScorer
+module SpamScorer
   SPAMMY_PHRASES = [
     "Business Description:", "Additional Keywords:"
   ].freeze
 
-  def initialize(text)
-    @text = text
+  def self.new_from_rich_text(text)
+    self::RichText.new(text)
   end
-
-  def score
-    link_count = 0
-    link_size = 0
-
-    doc = Nokogiri::HTML(text.to_html)
-
-    if doc.content.empty?
-      link_proportion = 0
-    else
-      doc.xpath("//a").each do |link|
-        link_count += 1
-        link_size += link.content.length
-      end
-
-      link_proportion = link_size.to_f / doc.content.length
-    end
-
-    spammy_phrases = SPAMMY_PHRASES.count do |phrase|
-      doc.content.include?(phrase)
-    end
-
-    ([link_proportion - 0.2, 0.0].max * 200) +
-      (link_count * 40) +
-      (spammy_phrases * 40)
-  end
-
-  private
-
-  attr_reader :text
 end
diff --git a/lib/spam_scorer/rich_text.rb b/lib/spam_scorer/rich_text.rb
new file mode 100644
index 000000000..5f53fb3df
--- /dev/null
+++ b/lib/spam_scorer/rich_text.rb
@@ -0,0 +1,39 @@
+# frozen_string_literal: true
+
+module SpamScorer
+  class RichText
+    def initialize(text)
+      @text = text
+    end
+
+    def score
+      link_count = 0
+      link_size = 0
+
+      doc = Nokogiri::HTML(text.to_html)
+
+      if doc.content.empty?
+        link_proportion = 0
+      else
+        doc.xpath("//a").each do |link|
+          link_count += 1
+          link_size += link.content.length
+        end
+
+        link_proportion = link_size.to_f / doc.content.length
+      end
+
+      spammy_phrases = SPAMMY_PHRASES.count do |phrase|
+        doc.content.include?(phrase)
+      end
+
+      ([link_proportion - 0.2, 0.0].max * 200) +
+        (link_count * 40) +
+        (spammy_phrases * 40)
+    end
+
+    private
+
+    attr_reader :text
+  end
+end
diff --git a/test/lib/spam_scorer_test.rb b/test/lib/spam_scorer_test.rb
index bb1801a53..a8a17a7f6 100644
--- a/test/lib/spam_scorer_test.rb
+++ b/test/lib/spam_scorer_test.rb
@@ -5,19 +5,19 @@ require "test_helper"
 class SpamScorerTest < ActiveSupport::TestCase
   def test_html_spam_score
     r = RichText.new("html", "foo <a href='http://example.com/'>bar</a> baz")
-    scorer = SpamScorer.new(r)
+    scorer = SpamScorer.new_from_rich_text(r)
     assert_equal 55, scorer.score.round
   end
 
   def test_markdown_spam_score
     r = RichText.new("markdown", "foo [bar](http://example.com/) baz")
-    scorer = SpamScorer.new(r)
+    scorer = SpamScorer.new_from_rich_text(r)
     assert_equal 50, scorer.score.round
   end
 
   def test_text_spam_score
     r = RichText.new("text", "foo http://example.com/ bar")
-    scorer = SpamScorer.new(r)
+    scorer = SpamScorer.new_from_rich_text(r)
     assert_equal 141, scorer.score.round
   end
 end
-- 
2.39.5