Merge pull request ruby#485 from ydah/nterm-support

Add support for %type declarations using `%nterm` in Nonterminal Symbols

lib/lrama/grammar.rb

@@ -30,7 +30,7 @@ class Grammar
                   :after_shift, :before_reduce, :after_reduce, :after_shift_error_token, :after_pop_stack,
                   :symbols_resolver, :types, :rules, :rule_builders, :sym_to_rules, :no_stdlib, :locations
 
-    def_delegators "@symbols_resolver", :symbols, :nterms, :terms, :add_nterm, :add_term,
+    def_delegators "@symbols_resolver", :symbols, :nterms, :terms, :add_nterm, :add_term, :find_term_by_s_value,
                                         :find_symbol_by_number!, :find_symbol_by_id!, :token_to_symbol,
                                         :find_symbol_by_s_value!, :fill_symbol_number, :fill_nterm_type,
                                         :fill_printer, :fill_destructor, :fill_error_token, :sort_by_number!

lib/lrama/grammar/symbols/resolver.rb

@@ -57,6 +57,10 @@ def add_nterm(id:, alias_name: nil, tag: nil)
           nterm
         end
 
+        def find_term_by_s_value(s_value)
+          terms.find { |s| s.id.s_value == s_value }
+        end
+
         def find_symbol_by_s_value(s_value)
           symbols.find { |s| s.id.s_value == s_value }
         end

lib/lrama/lexer.rb

@@ -16,6 +16,7 @@ class Lexer
       %union
       %token
       %type
+      %nterm
       %left
       %right
       %nonassoc

parser.y

@@ -115,6 +115,18 @@ rule
                             }
                           }
                         }
+                    | "%nterm" symbol_declarations
+                        {
+                          val[1].each {|hash|
+                            hash[:tokens].each {|id|
+                              if @grammar.find_term_by_s_value(id.s_value)
+                                on_action_error("symbol #{id.s_value} redeclared as a nonterminal", id)
+                              else
+                                @grammar.add_type(id: id, tag: hash[:tag])
+                              end
+                            }
+                          }
+                        }
                     | "%left" token_declarations_for_precedence
                         {
                           val[1].each {|hash|

spec/lrama/diagnostics_spec.rb

@@ -29,6 +29,7 @@
           %token tEQEQ  "=="
 
           %type <i> class /* comment for class */
+          %nterm <str> string /* comment for string */
 
           %nonassoc tEQEQ
           %left  tPLUS tMINUS '>'

spec/lrama/grammar/code_spec.rb

@@ -176,9 +176,9 @@
         %type <rule2> rule2
         %type <rule3> rule3
         %type <rule4> rule4
-        %type <rule5> rule5
-        %type <rule6> rule6
-        %type <rule7> rule7
+        %nterm <rule5> rule5
+        %nterm <rule6> rule6
+        %nterm <rule7> rule7
 
         %%
 

spec/lrama/parser_spec.rb

@@ -3710,6 +3710,32 @@ class : keyword_class tSTRING keyword_end { code 1 }
           ERROR
         end
       end
+
+      context "when pass the terminal symbol to `%nterm`" do
+        it "raises an error" do
+          y = <<~INPUT
+            %{
+            // Prologue
+            %}
+            
+            %token EOI 0 "EOI"
+            %nterm EOI
+            
+            %%
+            
+            program: /* empty */
+                   ;
+          INPUT
+
+          parser = Lrama::Parser.new(y, "error_messages/parse.y")
+
+          expect { parser.parse }.to raise_error(ParseError, <<~ERROR)
+            error_messages/parse.y:6:7: symbol EOI redeclared as a nonterminal
+            %nterm EOI
+                   ^^^
+          ERROR
+        end
+      end
     end
   end