From f1f67c46c5b369476090429a46dbc646d772ae25 Mon Sep 17 00:00:00 2001
From: David Yip <yipdw@member.fsf.org>
Date: Sat, 10 Feb 2018 10:32:14 -0600
Subject: [PATCH] Use Html2Text to generate plaintext for keyword mutes.  #236.

This allows us to match URLs inside link hrefs.
---
 app/models/glitch/keyword_mute_helper.rb       | 8 ++------
 spec/models/glitch/keyword_mute_helper_spec.rb | 7 +++++++
 2 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/app/models/glitch/keyword_mute_helper.rb b/app/models/glitch/keyword_mute_helper.rb
index 1b8c64e4ed..6d067947f1 100644
--- a/app/models/glitch/keyword_mute_helper.rb
+++ b/app/models/glitch/keyword_mute_helper.rb
@@ -1,16 +1,12 @@
-require 'htmlentities'
+require 'html2text'
 
 class Glitch::KeywordMuteHelper
-  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)
-    @entity_decoder = HTMLEntities.new
   end
 
   def matches?(status)
@@ -26,6 +22,6 @@ class Glitch::KeywordMuteHelper
   end
 
   def prepare_text(text)
-    entity_decoder.decode(strip_tags(text)).tap { |x| puts x }
+    Html2Text.convert(text)
   end
 end
diff --git a/spec/models/glitch/keyword_mute_helper_spec.rb b/spec/models/glitch/keyword_mute_helper_spec.rb
index 9d09e58da8..b3f991d5b8 100644
--- a/spec/models/glitch/keyword_mute_helper_spec.rb
+++ b/spec/models/glitch/keyword_mute_helper_spec.rb
@@ -39,5 +39,12 @@ RSpec.describe Glitch::KeywordMuteHelper do
 
       expect(helper.matches?(status)).to be true
     end
+
+    it 'matches link hrefs in HTML text' do
+      status = Fabricate(:status, text: '<p><a href="https://example.com/it-was-milk">yep</a></p>')
+      Glitch::KeywordMute.create!(account: alice, keyword: 'milk')
+
+      expect(helper.matches?(status)).to be true
+    end
   end
 end