rule_builder.rb in lrama-0.6.10

- old
+ new

@@ -1,14 +1,17 @@
+# frozen_string_literal: true
+
 module Lrama
   class Grammar
     class RuleBuilder
       attr_accessor :lhs, :line
       attr_reader :lhs_tag, :rhs, :user_code, :precedence_sym
 
-      def initialize(rule_counter, midrule_action_counter, position_in_original_rule_rhs = nil, lhs_tag: nil, skip_preprocess_references: false)
+      def initialize(rule_counter, midrule_action_counter, parameterizing_rule_resolver, position_in_original_rule_rhs = nil, lhs_tag: nil, skip_preprocess_references: false)
         @rule_counter = rule_counter
         @midrule_action_counter = midrule_action_counter
+        @parameterizing_rule_resolver = parameterizing_rule_resolver
         @position_in_original_rule_rhs = position_in_original_rule_rhs
         @skip_preprocess_references = skip_preprocess_references
 
         @lhs = nil
         @lhs_tag = lhs_tag
@@ -17,30 +20,24 @@
         @precedence_sym = nil
         @line = nil
         @rules = []
         @rule_builders_for_parameterizing_rules = []
         @rule_builders_for_derived_rules = []
-        @rule_builders_for_inline_rules = []
         @parameterizing_rules = []
-        @inline_rules = []
         @midrule_action_rules = []
       end
 
       def add_rhs(rhs)
-        if !@line
-          @line = rhs.line
-        end
+        @line ||= rhs.line
 
         flush_user_code
 
         @rhs << rhs
       end
 
       def user_code=(user_code)
-        if !@line
-          @line = user_code&.line
-        end
+        @line ||= user_code&.line
 
         flush_user_code
 
         @user_code = user_code
       end
@@ -53,24 +50,47 @@
 
       def complete_input
         freeze_rhs
       end
 
-      def setup_rules(parameterizing_rule_resolver)
+      def setup_rules
         preprocess_references unless @skip_preprocess_references
-        if rhs.any? { |token| parameterizing_rule_resolver.find_inline(token) }
-          resolve_inline(parameterizing_rule_resolver)
-        else
-          process_rhs(parameterizing_rule_resolver)
-        end
+        process_rhs
         build_rules
       end
 
       def rules
-        @parameterizing_rules + @inline_rules + @midrule_action_rules + @rules
+        @parameterizing_rules + @midrule_action_rules + @rules
       end
 
+      def has_inline_rules?
+        rhs.any? { |token| @parameterizing_rule_resolver.find_inline(token) }
+      end
+
+      def resolve_inline_rules
+        resolved_builders = []
+        rhs.each_with_index do |token, i|
+          if (inline_rule = @parameterizing_rule_resolver.find_inline(token))
+            inline_rule.rhs_list.each do |inline_rhs|
+              rule_builder = RuleBuilder.new(@rule_counter, @midrule_action_counter, @parameterizing_rule_resolver, lhs_tag: lhs_tag)
+              if token.is_a?(Lexer::Token::InstantiateRule)
+                resolve_inline_rhs(rule_builder, inline_rhs, i, Binding.new(inline_rule, token.args))
+              else
+                resolve_inline_rhs(rule_builder, inline_rhs, i)
+              end
+              rule_builder.lhs = lhs
+              rule_builder.line = line
+              rule_builder.precedence_sym = precedence_sym
+              rule_builder.user_code = replace_inline_user_code(inline_rhs, i)
+              resolved_builders << rule_builder
+            end
+            break
+          end
+        end
+        resolved_builders
+      end
+
       private
 
       def freeze_rhs
         @rhs.freeze
       end
@@ -80,35 +100,29 @@
       end
 
       def build_rules
         tokens = @replaced_rhs
 
-        if tokens
-          rule = Rule.new(
-            id: @rule_counter.increment, _lhs: lhs, _rhs: tokens, lhs_tag: lhs_tag, token_code: user_code,
-            position_in_original_rule_rhs: @position_in_original_rule_rhs, precedence_sym: precedence_sym, lineno: line
-          )
-          @rules = [rule]
-          @parameterizing_rules = @rule_builders_for_parameterizing_rules.map do |rule_builder|
-            rule_builder.rules
-          end.flatten
-          @midrule_action_rules = @rule_builders_for_derived_rules.map do |rule_builder|
-            rule_builder.rules
-          end.flatten
-          @midrule_action_rules.each do |r|
-            r.original_rule = rule
-          end
-        else
-          @inline_rules = @rule_builders_for_inline_rules.map do |rule_builder|
-            rule_builder.rules
-          end.flatten
+        rule = Rule.new(
+          id: @rule_counter.increment, _lhs: lhs, _rhs: tokens, lhs_tag: lhs_tag, token_code: user_code,
+          position_in_original_rule_rhs: @position_in_original_rule_rhs, precedence_sym: precedence_sym, lineno: line
+        )
+        @rules = [rule]
+        @parameterizing_rules = @rule_builders_for_parameterizing_rules.map do |rule_builder|
+          rule_builder.rules
+        end.flatten
+        @midrule_action_rules = @rule_builders_for_derived_rules.map do |rule_builder|
+          rule_builder.rules
+        end.flatten
+        @midrule_action_rules.each do |r|
+          r.original_rule = rule
         end
       end
 
       # rhs is a mixture of variety type of tokens like `Ident`, `InstantiateRule`, `UserCode` and so on.
       # `#process_rhs` replaces some kind of tokens to `Ident` so that all `@replaced_rhs` are `Ident` or `Char`.
-      def process_rhs(parameterizing_rule_resolver)
+      def process_rhs
         return if @replaced_rhs
 
         @replaced_rhs = []
 
         rhs.each_with_index do |token, i|
@@ -116,44 +130,44 @@
           when Lrama::Lexer::Token::Char
             @replaced_rhs << token
           when Lrama::Lexer::Token::Ident
             @replaced_rhs << token
           when Lrama::Lexer::Token::InstantiateRule
-            parameterizing_rule = parameterizing_rule_resolver.find_rule(token)
+            parameterizing_rule = @parameterizing_rule_resolver.find_rule(token)
             raise "Unexpected token. #{token}" unless parameterizing_rule
 
             bindings = Binding.new(parameterizing_rule, token.args)
             lhs_s_value = lhs_s_value(token, bindings)
-            if (created_lhs = parameterizing_rule_resolver.created_lhs(lhs_s_value))
+            if (created_lhs = @parameterizing_rule_resolver.created_lhs(lhs_s_value))
               @replaced_rhs << created_lhs
             else
               lhs_token = Lrama::Lexer::Token::Ident.new(s_value: lhs_s_value, location: token.location)
               @replaced_rhs << lhs_token
-              parameterizing_rule_resolver.created_lhs_list << lhs_token
+              @parameterizing_rule_resolver.created_lhs_list << lhs_token
               parameterizing_rule.rhs_list.each do |r|
-                rule_builder = RuleBuilder.new(@rule_counter, @midrule_action_counter, lhs_tag: token.lhs_tag || parameterizing_rule.tag)
+                rule_builder = RuleBuilder.new(@rule_counter, @midrule_action_counter, @parameterizing_rule_resolver, lhs_tag: token.lhs_tag || parameterizing_rule.tag)
                 rule_builder.lhs = lhs_token
                 r.symbols.each { |sym| rule_builder.add_rhs(bindings.resolve_symbol(sym)) }
                 rule_builder.line = line
                 rule_builder.precedence_sym = r.precedence_sym
                 rule_builder.user_code = r.resolve_user_code(bindings)
                 rule_builder.complete_input
-                rule_builder.setup_rules(parameterizing_rule_resolver)
+                rule_builder.setup_rules
                 @rule_builders_for_parameterizing_rules << rule_builder
               end
             end
           when Lrama::Lexer::Token::UserCode
             prefix = token.referred ? "@" : "$@"
             tag = token.tag || lhs_tag
             new_token = Lrama::Lexer::Token::Ident.new(s_value: prefix + @midrule_action_counter.increment.to_s)
             @replaced_rhs << new_token
 
-            rule_builder = RuleBuilder.new(@rule_counter, @midrule_action_counter, i, lhs_tag: tag, skip_preprocess_references: true)
+            rule_builder = RuleBuilder.new(@rule_counter, @midrule_action_counter, @parameterizing_rule_resolver, i, lhs_tag: tag, skip_preprocess_references: true)
             rule_builder.lhs = new_token
             rule_builder.user_code = token
             rule_builder.complete_input
-            rule_builder.setup_rules(parameterizing_rule_resolver)
+            rule_builder.setup_rules
 
             @rule_builders_for_derived_rules << rule_builder
           else
             raise "Unexpected token. #{token}"
           end
@@ -170,31 +184,14 @@
           end
         end
         "#{token.rule_name}_#{s_values.join('_')}"
       end
 
-      def resolve_inline(parameterizing_rule_resolver)
+      def resolve_inline_rhs(rule_builder, inline_rhs, index, bindings = nil)
         rhs.each_with_index do |token, i|
-          if inline_rule = parameterizing_rule_resolver.find_inline(token)
-            inline_rule.rhs_list.each_with_index do |inline_rhs|
-              rule_builder = RuleBuilder.new(@rule_counter, @midrule_action_counter, lhs_tag: lhs_tag, skip_preprocess_references: true)
-              resolve_inline_rhs(rule_builder, inline_rhs, i)
-              rule_builder.lhs = lhs
-              rule_builder.line = line
-              rule_builder.user_code = replace_inline_user_code(inline_rhs, i)
-              rule_builder.complete_input
-              rule_builder.setup_rules(parameterizing_rule_resolver)
-              @rule_builders_for_inline_rules << rule_builder
-            end
-          end
-        end
-      end
-
-      def resolve_inline_rhs(rule_builder, inline_rhs, index)
-        rhs.each_with_index do |token, i|
           if index == i
-            inline_rhs.symbols.each { |sym| rule_builder.add_rhs(sym) }
+            inline_rhs.symbols.each { |sym| rule_builder.add_rhs(bindings.nil? ? sym : bindings.resolve_symbol(sym)) }
           else
             rule_builder.add_rhs(token)
           end
         end
       end
@@ -202,10 +199,15 @@
       def replace_inline_user_code(inline_rhs, index)
         return user_code if inline_rhs.user_code.nil?
         return user_code if user_code.nil?
 
         code = user_code.s_value.gsub(/\$#{index + 1}/, inline_rhs.user_code.s_value)
+        user_code.references.each do |ref|
+          next if ref.index.nil? || ref.index <= index # nil is a case for `$$`
+          code = code.gsub(/\$#{ref.index}/, "$#{ref.index + (inline_rhs.symbols.count-1)}")
+          code = code.gsub(/@#{ref.index}/, "@#{ref.index + (inline_rhs.symbols.count-1)}")
+        end
         Lrama::Lexer::Token::UserCode.new(s_value: code, location: user_code.location)
       end
 
       def numberize_references
         # Bison n'th component is 1-origin
@@ -236,12 +238,9 @@
                 end
               end
             end
 
             if ref.number
-              # TODO: When Inlining is implemented, for example, if `$1` is expanded to multiple RHS tokens,
-              #       `$2` needs to access `$2 + n` to actually access it. So, after the Inlining implementation,
-              #       it needs resolves from number to index.
               ref.index = ref.number
             end
 
             # TODO: Need to check index of @ too?
             next if ref.type == :at