From 3f2e31800ed9c74fc1b7bc4b475542f0256fefbd Mon Sep 17 00:00:00 2001
From: Christian Schmidt <github@chsc.dk>
Date: Wed, 8 Mar 2023 19:56:41 +0100
Subject: [PATCH] Unescape HTML entities (#24019)

---
 app/lib/plain_text_formatter.rb       |  6 ++-
 spec/lib/plain_text_formatter_spec.rb | 61 ++++++++++++++++++++++++---
 2 files changed, 61 insertions(+), 6 deletions(-)

diff --git a/app/lib/plain_text_formatter.rb b/app/lib/plain_text_formatter.rb
index 08aa29696..6fa2bc5d2 100644
--- a/app/lib/plain_text_formatter.rb
+++ b/app/lib/plain_text_formatter.rb
@@ -18,7 +18,7 @@ class PlainTextFormatter
     if local?
       text
     else
-      strip_tags(insert_newlines).chomp
+      html_entities.decode(strip_tags(insert_newlines)).chomp
     end
   end
 
@@ -27,4 +27,8 @@ class PlainTextFormatter
   def insert_newlines
     text.gsub(NEWLINE_TAGS_RE) { |match| "#{match}\n" }
   end
+
+  def html_entities
+    HTMLEntities.new
+  end
 end
diff --git a/spec/lib/plain_text_formatter_spec.rb b/spec/lib/plain_text_formatter_spec.rb
index c3d0ee630..81e4ae286 100644
--- a/spec/lib/plain_text_formatter_spec.rb
+++ b/spec/lib/plain_text_formatter_spec.rb
@@ -4,7 +4,7 @@ RSpec.describe PlainTextFormatter do
   describe '#to_s' do
     subject { described_class.new(status.text, status.local?).to_s }
 
-    context 'given a post with local status' do
+    context 'when status is local' do
       let(:status) { Fabricate(:status, text: '<p>a text by a nerd who uses an HTML tag in text</p>', uri: nil) }
 
       it 'returns the raw text' do
@@ -12,12 +12,63 @@ RSpec.describe PlainTextFormatter do
       end
     end
 
-    context 'given a post with remote status' do
+    context 'when status is remote' do
       let(:remote_account) { Fabricate(:account, domain: 'remote.test', username: 'bob', url: 'https://remote.test/') }
-      let(:status) { Fabricate(:status, account: remote_account, text: '<p>Hello</p><script>alert("Hello")</script>') }
 
-      it 'returns tag-stripped text' do
-        is_expected.to eq 'Hello'
+      context 'when text contains inline HTML tags' do
+        let(:status) { Fabricate(:status, account: remote_account, text: '<b>Lorem</b> <em>ipsum</em>') }
+
+        it 'strips the tags' do
+          expect(subject).to eq 'Lorem ipsum'
+        end
+      end
+
+      context 'when text contains <p> tags' do
+        let(:status) { Fabricate(:status, account: remote_account, text: '<p>Lorem</p><p>ipsum</p>') }
+
+        it 'inserts a newline' do
+          expect(subject).to eq "Lorem\nipsum"
+        end
+      end
+
+      context 'when text contains a single <br> tag' do
+        let(:status) { Fabricate(:status, account: remote_account, text: 'Lorem<br>ipsum') }
+
+        it 'inserts a newline' do
+          expect(subject).to eq "Lorem\nipsum"
+        end
+      end
+
+      context 'when text contains consecutive <br> tag' do
+        let(:status) { Fabricate(:status, account: remote_account, text: 'Lorem<br><br><br>ipsum') }
+
+        it 'inserts a single newline' do
+          expect(subject).to eq "Lorem\nipsum"
+        end
+      end
+
+      context 'when text contains HTML entity' do
+        let(:status) { Fabricate(:status, account: remote_account, text: 'Lorem &amp; ipsum &#x2764;') }
+
+        it 'unescapes the entity' do
+          expect(subject).to eq 'Lorem & ipsum ❤'
+        end
+      end
+
+      context 'when text contains <script> tag' do
+        let(:status) { Fabricate(:status, account: remote_account, text: 'Lorem <script> alert("Booh!") </script>ipsum') }
+
+        it 'strips the tag and its contents' do
+          expect(subject).to eq 'Lorem ipsum'
+        end
+      end
+
+      context 'when text contains an HTML comment tags' do
+        let(:status) { Fabricate(:status, account: remote_account, text: 'Lorem <!-- Booh! -->ipsum') }
+
+        it 'strips the comment' do
+          expect(subject).to eq 'Lorem ipsum'
+        end
       end
     end
   end