Fixed list parsing

CHANGELOG.md

@@ -1,3 +1,6 @@
+[v#.#.#] ([month] [YYYY])
+  - Add support for nested items in lists
+
 v4.10.0 (September 2023)
   - Update gemspec links
 

lib/acunetix/concerns/cleanup.rb

@@ -7,6 +7,7 @@ def cleanup_html(source)
       result = source.dup
 
       format_table(result)
+      result = format_list(result)
 
       result.gsub!(/"/, '"')
       result.gsub!(/&/, '&')
@@ -34,6 +35,11 @@ def cleanup_html(source)
 
       result.gsub!(/<strong>(.*?)<\/strong>/) { "*#{$1.strip}*" }
       result.gsub!(/<span.*?>(.*?)<\/span>/m){"#{$1.strip}\n"}
+      result.gsub!(/<span.*?>|<\/span>/, '') #repeating again to deal with nested/empty/incomplete span tags
+      result.gsub!(/(&#x\d{0,3};)/m) { "==#{$1.strip}==" }
+
+      # Cleanup lingering <p></p>
+      result.gsub!(/<p.*?>(.*?)<\/p>/m) { $1 }
 
       result
     end
@@ -45,6 +51,26 @@ def cleanup_decimals(source)
       result
     end
 
+    def format_list(str)
+        return str unless str.include?('</ul>') || str.include?('</ol>')
+        deep = 0
+        text = str.split(/<(?<tag>\/?(ul|ol|li)).*?>/)
+        result = ""
+        text.each do |string|
+            string.strip!
+            if (string == 'ul' || string == 'ol')
+                deep = deep + 1
+            elsif (string == '/ul' || string == '/ol')
+                deep = deep - 1
+            elsif (string == 'li')
+                result += "".ljust(deep, "*")
+            elsif (string != '/li' && string.strip != "")
+                result += " " + string + "\n"
+            end
+        end
+        return result
+    end
+
     def format_table(str)
       return unless str.include?('</table>')