Fixed #2416 - HTML sanitizer blocks email processing because of an endless loop.

lib/html_sanitizer.rb

@@ -1,5 +1,7 @@
 class HtmlSanitizer
   LINKABLE_URL_SCHEMES = URI.scheme_list.keys.map(&:downcase) - ['mailto'] + ['tel']
+  PROCESSING_TIMEOUT = 10
+  UNPROCESSABLE_HTML_MSG = 'This message cannot be displayed due to HTML processing issues. Download the raw message below and open it via an Email client if you still wish to view it.'.freeze
 
 =begin
 
@@ -9,7 +11,8 @@ satinize html string based on whiltelist
 
 =end
 
-  def self.strict(string, external = false)
+  def self.strict(string, external = false, timeout: true)
+    Timeout.timeout(timeout ? PROCESSING_TIMEOUT : nil) do
       @fqdn = Setting.get('fqdn')
 
       # config
@@ -202,6 +205,9 @@ satinize html string based on whiltelist
 
       Loofah.fragment(string).scrub!(scrubber_link).to_s
     end
+  rescue Timeout::Error => e
+    UNPROCESSABLE_HTML_MSG
+  end
 
 =begin
 
@@ -214,7 +220,8 @@ cleanup html string:
 
 =end
 
-  def self.cleanup(string)
+  def self.cleanup(string, timeout: true)
+    Timeout.timeout(timeout ? PROCESSING_TIMEOUT : nil) do
       string.gsub!(/<[A-z]:[A-z]>/, '')
       string.gsub!(%r{</[A-z]:[A-z]>}, '')
       string.delete!("\t")
@@ -230,6 +237,9 @@ cleanup html string:
       string = cleanup_structure(string)
       string
     end
+  rescue Timeout::Error => e
+    UNPROCESSABLE_HTML_MSG
+  end
 
   def self.cleanup_replace_tags(string)
     #return string

spec/lib/html_sanitizer_spec.rb

@@ -183,4 +183,27 @@ RSpec.describe HtmlSanitizer do
       end
     end
   end
+
+  # Issue #2416 - html_sanitizer goes into loop for specific content
+  describe '.strict' do
+    context 'with strings that take a long time (>10s) to parse' do
+      before { allow(Timeout).to receive(:timeout).and_raise(Timeout::Error) }
+
+      it 'returns a timeout error message for the user' do
+        expect(HtmlSanitizer.strict(+'<img src="/some_one.png">', true))
+          .to match(HtmlSanitizer::UNPROCESSABLE_HTML_MSG)
+      end
+    end
+  end
+
+  describe '.cleanup' do
+    context 'with strings that take a long time (>10s) to parse' do
+      before { allow(Timeout).to receive(:timeout).and_raise(Timeout::Error) }
+
+      it 'returns a timeout error message for the user' do
+        expect(HtmlSanitizer.cleanup(+'<img src="/some_one.png">'))
+          .to match(HtmlSanitizer::UNPROCESSABLE_HTML_MSG)
+      end
+    end
+  end
 end