Add definedBy validation

Signed-off-by: Afonso Oliveira <[email protected]>

Rakefile

@@ -90,15 +90,37 @@ namespace :test do
   end
   task schema: "gen:arch" do
     validator = Validator.instance
-    puts "Checking arch files against schema.."
+    puts "Checking arch files against schema..."
     arch_files = Dir.glob("#{$root}/arch/**/*.yaml")
     progressbar = ProgressBar.create(total: arch_files.size)
+
+    incorrect_defined_by_files = []
+
     arch_files.each do |f|
       progressbar.increment
       validator.validate(f)
+
+      # Only check definedBy for instructions
+      if f.include?("/inst/")
+        yaml_content = YamlLoader.load(f, permitted_classes: [Date])
+        if defined_by = yaml_content['definedBy']
+          unless validator.valid_defined_by?(defined_by)
+            incorrect_defined_by_files << f
+          end
+        end
+      end
     end
+
+    if incorrect_defined_by_files.any?
+      puts "\nThe following instruction files have incorrectly formatted 'definedBy' fields:"
+      incorrect_defined_by_files.each do |file|
+        puts "  #{Pathname.new(file).relative_path_from($root)}"
+      end
+      exit 1
+    end
+
     Rake::Task["test:insts"].invoke
-    puts "All files validate against their schema"  
+    puts "All files validate against their schema"
   end
   task idl_model: ["gen:arch", "#{$root}/.stamps/arch-gen-_32.stamp", "#{$root}/.stamps/arch-gen-_64.stamp"]  do
     print "Parsing IDL code for RV32..."

lib/validate.rb

@@ -275,4 +275,42 @@ def validate_instruction(path)
       validate_instruction_encoding(inst_name, obj["encoding"]["RV64"])
     end
   end
+  # Validate the format of a definedBy field in an instruction file
+  #
+  # @param defined_by [String, Hash] The definedBy field to validate
+  # @return [Boolean] true if the field format is valid, false otherwise
+  # - String must be non-empty and without commas
+  # - Hash must have single key (anyOf/oneOf/allOf) with array or comma-separated list values
+  def valid_defined_by?(defined_by)
+      return false if defined_by.nil?
+
+      case defined_by
+      when String
+        # Simple string case - shouldn't contain commas
+        !defined_by.include?(',') && !defined_by.strip.empty?
+      when Hash
+        # Only allow one key which must be anyOf, oneOf, or allOf
+        return false unless defined_by.size == 1
+        key = defined_by.keys.first
+        return false unless ['anyOf', 'oneOf', 'allOf'].include?(key)
+
+        value = defined_by[key]
+        case value
+        when Array
+          # Each array element should be a simple string without commas
+          value.all? { |v| v.is_a?(String) && !v.include?(',') && !v.strip.empty? }
+        when String
+          # Also accept the compact form: anyOf: [C,Zca]
+          # Split by comma and/or brackets, strip whitespace and validate
+          elements = value.gsub(/[\[\]]/, '').split(',').map(&:strip)
+          elements.all? { |v| !v.empty? }
+        else
+          false
+        end
+      else
+        false
+      end
+    end
+
 end
+