Add warning for rules mixing positional and named references

Author: ydah

lib/lrama/warnings.rb

@@ -3,6 +3,7 @@
 
 require_relative 'warnings/conflicts'
 require_relative 'warnings/implicit_empty'
+require_relative 'warnings/mixed_references'
 require_relative 'warnings/name_conflicts'
 require_relative 'warnings/redefined_rules'
 require_relative 'warnings/required'
@@ -14,6 +15,7 @@ class Warnings
     def initialize(logger, warnings)
       @conflicts = Conflicts.new(logger, warnings)
       @implicit_empty = ImplicitEmpty.new(logger, warnings)
+      @mixed_references = MixedReferences.new(logger, warnings)
       @name_conflicts = NameConflicts.new(logger, warnings)
       @redefined_rules = RedefinedRules.new(logger, warnings)
       @required = Required.new(logger, warnings)
@@ -24,6 +26,7 @@ def initialize(logger, warnings)
     def warn(grammar, states)
       @conflicts.warn(states)
       @implicit_empty.warn(grammar)
+      @mixed_references.warn(grammar)
       @name_conflicts.warn(grammar)
       @redefined_rules.warn(grammar)
       @required.warn(grammar)

lib/lrama/warnings/mixed_references.rb

@@ -0,0 +1,96 @@
+# rbs_inline: enabled
+# frozen_string_literal: true
+
+module Lrama
+  class Warnings
+    # Warning rationale: Mixing positional and named references in one rule
+    # - It reduces readability and makes semantic actions harder to maintain
+    # - Named references are generally more robust when RHS changes
+    # Scope:
+    # - This warning only targets semantic value references (`$...`)
+    # - It intentionally ignores location references (`@...`),
+    #   index references (`$:...`), and special LHS references (`$$`)
+    class MixedReferences
+      # @rbs (Lrama::Logger logger, bool warnings) -> void
+      def initialize(logger, warnings)
+        @logger = logger
+        @warnings = warnings
+      end
+
+      # @rbs (Lrama::Grammar grammar) -> void
+      def warn(grammar)
+        return unless @warnings
+
+        build_grouped_usage(grammar).each do |rule, usage|
+          next unless usage[:positional] && usage[:named]
+
+          @logger.warn("warning: rule `#{rule.as_comment}` mixes positional and named references; use named references consistently")
+        end
+      end
+
+      private
+
+      # @rbs (Lrama::Grammar grammar) -> Hash[Lrama::Grammar::Rule, { positional: bool, named: bool }]
+      def build_grouped_usage(grammar)
+        grouped_usage = {} #: Hash[Lrama::Grammar::Rule, { positional: bool, named: bool }]
+
+        grammar.rules.each do |rule|
+          next unless (token_code = rule.token_code)
+
+          original_rule = rule.original_rule || rule
+          usage = (grouped_usage[original_rule] ||= { positional: false, named: false })
+          classify_references(token_code.references, token_code.s_value, usage)
+        end
+
+        grouped_usage
+      end
+
+      # @rbs (Array[Lrama::Grammar::Reference] references, String source, { positional: bool, named: bool } usage) -> void
+      def classify_references(references, source, usage)
+        references.each do |ref|
+          if positional_reference?(ref)
+            usage[:positional] = true
+          elsif named_reference?(ref, source)
+            usage[:named] = true
+          end
+        end
+      end
+
+      # @rbs (Lrama::Grammar::Reference ref) -> bool
+      def positional_reference?(ref)
+        return false unless ref.type == :dollar
+        return false if ref.index.nil?
+
+        ref.name.nil?
+      end
+
+      # @rbs (Lrama::Grammar::Reference ref, String source) -> bool
+      def named_reference?(ref, source)
+        return false unless ref.type == :dollar
+
+        return true if !ref.name.nil? && ref.name != "$"
+
+        lhs_alias_reference?(ref, source)
+      end
+
+      # @rbs (Lrama::Grammar::Reference ref, String source) -> bool
+      def lhs_alias_reference?(ref, source)
+        return false unless ref.name == "$"
+
+        lexeme = source[ref.first_column...ref.last_column]
+        return false if lexeme.nil?
+
+        # Keep ignoring special LHS references ($$, $<tag>$), same as reference_to_c.
+        return false if special_lhs_reference_lexeme?(lexeme)
+
+        # Treat remaining `$...` forms as named LHS aliases.
+        true
+      end
+
+      # @rbs (String lexeme) -> bool
+      def special_lhs_reference_lexeme?(lexeme)
+        lexeme == "$$" || (lexeme.start_with?("$<") && lexeme.end_with?("$"))
+      end
+    end
+  end
+end