Fill out some examples for Glitch::FilterHelper. #234.

Also add HTML entity decoding to Glitch::FilterHelper, which is needed
to e.g. match "<" to the tag-stripped version of "<p><3</p>" or
"<p>&lt;3</p>".

2 files changed, 55 insertions, 4 deletions

diff --git a/app/models/glitch/filter_helper.rb b/app/models/glitch/filter_helper.rb
index 0dfd69526..11be877c1 100644
--- a/app/models/glitch/filter_helper.rb
+++ b/app/models/glitch/filter_helper.rb
@@ -1,12 +1,16 @@
+require 'htmlentities'
+
 class Glitch::FilterHelper
   include ActionView::Helpers::SanitizeHelper
 
   attr_reader :text_matcher
   attr_reader :tag_matcher
+  attr_reader :entity_decoder
 
   def initialize(receiver_id)
-    @text_matcher = Glitch::KeywordMute.text_matcher_for(receiver_id)
-    @tag_matcher  = Glitch::KeywordMute.tag_matcher_for(receiver_id)
+    @text_matcher   = Glitch::KeywordMute.text_matcher_for(receiver_id)
+    @tag_matcher    = Glitch::KeywordMute.tag_matcher_for(receiver_id)
+    @entity_decoder = HTMLEntities.new
   end
 
   def matches?(status)
@@ -16,8 +20,12 @@ class Glitch::FilterHelper
   private
 
   def matchers_match?(status)
-    text_matcher.matches?(strip_tags(status.text)) ||
-      text_matcher.matches?(strip_tags(status.spoiler_text)) ||
+    text_matcher.matches?(prepare_text(status.text)) ||
+      text_matcher.matches?(prepare_text(status.spoiler_text)) ||
       tag_matcher.matches?(status.tags)
   end
+
+  def prepare_text(text)
+    entity_decoder.decode(strip_tags(text))
+  end
 end
diff --git a/spec/models/glitch/filter_helper_spec.rb b/spec/models/glitch/filter_helper_spec.rb
new file mode 100644
index 000000000..9a808667d
--- /dev/null
+++ b/spec/models/glitch/filter_helper_spec.rb
@@ -0,0 +1,43 @@
+require 'rails_helper'
+
+RSpec.describe Glitch::FilterHelper do
+  describe '#matches?' do
+    let(:alice) { Fabricate(:account, username: 'alice').tap(&:save!) }
+    let(:helper) { Glitch::FilterHelper.new(alice) }
+
+    it 'ignores names of HTML tags in status text' do
+      status = Fabricate(:status, text: '<addr>uh example</addr>')
+      Glitch::KeywordMute.create!(account: alice, keyword: 'addr')
+
+      expect(helper.matches?(status)).to be false
+    end
+
+    it 'ignores properties of HTML tags in status text' do
+      status = Fabricate(:status, text: '<a href="https://www.example.org">uh example</a>')
+      Glitch::KeywordMute.create!(account: alice, keyword: 'href')
+
+      expect(helper.matches?(status)).to be false
+    end
+
+    it 'matches text inside HTML tags' do
+      status = Fabricate(:status, text: '<p>HEY THIS IS SOMETHING ANNOYING</p>')
+      Glitch::KeywordMute.create!(account: alice, keyword: 'annoying')
+
+      expect(helper.matches?(status)).to be true
+    end
+
+    it 'matches < in HTML-stripped text' do
+      status = Fabricate(:status, text: '<p>I <3 oats</p>')
+      Glitch::KeywordMute.create!(account: alice, keyword: '<3')
+
+      expect(helper.matches?(status)).to be true
+    end
+
+    it 'matches &lt; in HTML text' do
+      status = Fabricate(:status, text: '<p>I &lt;3 oats</p>')
+      Glitch::KeywordMute.create!(account: alice, keyword: '<3')
+
+      expect(helper.matches?(status)).to be true
+    end
+  end
+end