DEV: refactor ascii slug generation

lib/discourse_modifications/slug_ascii_patch.rb

@@ -0,0 +1,9 @@
+# frozen_string_literal: true
+
+module ::DiscourseModifications
+  module SlugAsciiPatch
+    def ascii_generator(string)
+      super(AnyAscii.transliterate(string.to_s))
+    end
+  end
+end

plugin.rb

@@ -8,7 +8,7 @@
 # url: TODO
 # required_version: 2.7.0
 
-gem 'any_ascii', '0.3.2'
+gem "any_ascii", "0.3.2"
 
 enabled_site_setting :discourse_modifications_enabled
 
@@ -21,19 +21,17 @@ module ::DiscourseModifications
 require_relative "lib/discourse_modifications/engine"
 
 after_initialize do
-  # Code which should run after Rails has finished booting
-
-  # Note: if this file gets too large, consider moving code into separate files in the lib directory
-  # and applying a Initializer pattern to load them.
-
-  Topic.slug_computed_callbacks << ::DiscourseModifications::TopicSlug.method(:slug_for_topic)
+  # Patch Slug.ascii_generator to use AnyAscii before parameterize so that
+  # Unicode characters (including mathematical/stylised variants) transliterate
+  # correctly everywhere Slug.for is called — AR lifecycle hooks, bulk import,
+  # category slugs, etc.
+  Slug.singleton_class.prepend(::DiscourseModifications::SlugAsciiPatch)
 
   # add permalink normalization
   normalizations = SiteSetting.permalink_normalizations
   normalizations = normalizations.blank? ? [] : normalizations.split("|")
-
-  normalizations << ::DiscourseModifications::XF_TOPIC_LINK_NORMALIZATION if normalizations.exclude?(::DiscourseModifications::XF_TOPIC_LINK_NORMALIZATION)
-
+  if normalizations.exclude?(::DiscourseModifications::XF_TOPIC_LINK_NORMALIZATION)
+    normalizations << ::DiscourseModifications::XF_TOPIC_LINK_NORMALIZATION
+  end
   SiteSetting.permalink_normalizations = normalizations.join("|")
-
 end

spec/lib/discourse_modifications/slug_patch_spec.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+RSpec.describe "Slug.ascii_generator AnyAscii patch" do
+  before do
+    allow(SiteSetting).to receive(:slug_generation_method).and_return("ascii")
+    allow(SiteSetting).to receive(:default_locale).and_return("en")
+  end
+
+  it "transliterates mathematical bold/italic Unicode to ASCII" do
+    title = "𝑻𝒉𝒆 𝒔𝒖𝒎𝒎𝒆𝒓 𝒘𝒆 𝒇𝒆𝒍𝒍 𝒊𝒏 𝒍𝒐𝒗𝒆 [𝐇𝐀𝐈𝐊𝐘𝐔]"
+    expect(Slug.for(title)).to eq("the-summer-we-fell-in-love-haikyu")
+  end
+
+  it "still handles plain ASCII titles" do
+    expect(Slug.for("Hello World")).to eq("hello-world")
+  end
+
+  it "still handles accented characters" do
+    expect(Slug.for("Café au lait")).to eq("cafe-au-lait")
+  end
+
+  it "falls back to the default when slug is blank after transliteration" do
+    expect(Slug.for(":smile:")).to eq("topic")
+  end
+
+  it "is not applied when slug_generation_method is encoded" do
+    allow(SiteSetting).to receive(:slug_generation_method).and_return("encoded")
+    expect(Slug.for("Héllo")).to eq("h%C3%A9llo")
+  end
+end