Merge pull request #310 from yui-knk/simplify_normalize_rules

Simplify normalize rules
ruby · Dec 24, 2023 · 95909d0 · 95909d0
2 parents c30569b + ed59e31
commit 95909d0
Show file tree

Hide file tree

Showing 6 changed files with 69 additions and 112 deletions.
diff --git a/lib/lrama/grammar.rb b/lib/lrama/grammar.rb
@@ -359,29 +359,8 @@ def append_special_symbols
       @accept_symbol = term
     end
 
-    # 1. Add $accept rule to the top of rules
-    # 2. Extract action in the middle of RHS into new Empty rule
-    # 3. Append id and extract action then create Rule
-    #
-    # Bison 3.8.2 uses different orders for symbol number and rule number
-    # when a rule has actions in the middle of a rule.
-    #
-    # For example,
-    #
-    # `program: $@1 top_compstmt`
-    #
-    # Rules are ordered like below,
-    #
-    # 1 $@1: ε
-    # 2 program: $@1 top_compstmt
-    #
-    # Symbols are ordered like below,
-    #
-    # 164 program
-    # 165 $@1
-    #
     def normalize_rules
-      # 1. Add $accept rule to the top of rules
+      # Add $accept rule to the top of rules
       accept = @accept_symbol
       eof = @eof_symbol
       lineno = @rule_builders.first ? @rule_builders.first.line : 0
@@ -390,24 +369,10 @@ def normalize_rules
       setup_rules
 
       @rule_builders.each do |builder|
-        # Extract actions in the middle of RHS into new rules.
-        builder.midrule_action_rules.each do |rule|
-          @rules << rule
-        end
-
         builder.rules.each do |rule|
-          add_nterm(id: rule._lhs)
-          @rules << rule
-        end
-
-        builder.parameterizing_rules.each do |rule|
           add_nterm(id: rule._lhs, tag: rule.lhs_tag)
           @rules << rule
         end
-
-        builder.midrule_action_rules.each do |rule|
-          add_nterm(id: rule._lhs)
-        end
       end
 
       @rules.sort_by!(&:id)

diff --git a/lib/lrama/grammar/rule_builder.rb b/lib/lrama/grammar/rule_builder.rb
@@ -57,16 +57,8 @@ def setup_rules(parameterizing_resolver)
         build_rules
       end
 
-      def parameterizing_rules
-        @parameterizing_rules
-      end
-
-      def midrule_action_rules
-        @midrule_action_rules
-      end
-
       def rules
-        @rules
+        @parameterizing_rules + @midrule_action_rules + @rules
       end
 
       private

diff --git a/sig/lrama/grammar/rule_builder.rbs b/sig/lrama/grammar/rule_builder.rbs
@@ -25,8 +25,6 @@ module Lrama
       def precedence_sym=: (Lexer::Token user_code) -> void
       def complete_input: () -> void
       def setup_rules: (Grammar::ParameterizingRuleResolver parameterizing_resolver) -> void
-      def parameterizing_rules: () -> Array[Rule]
-      def midrule_action_rules: () -> Array[Rule]
       def rules: () -> Array[Rule]
 
       private

diff --git a/spec/lrama/grammar/rule_builder_spec.rb b/spec/lrama/grammar/rule_builder_spec.rb
@@ -347,16 +347,18 @@ class : keyword_class tSTRING tSTRING keyword_end { $class = $tSTRING; }
       rule_builder.complete_input
       rule_builder.setup_rules(Lrama::Grammar::ParameterizingRuleResolver.new)
 
-      rule = rule_builder.rules.first
-      rules = rule_builder.midrule_action_rules
-
-      expect(rules.count).to eq 2
-      expect(rules[0]._lhs.s_value).to eq '@1'
-      expect(rules[0].token_code.s_value).to eq '$1'
-      expect(rules[0].original_rule).to eq rule
-      expect(rules[1]._lhs.s_value).to eq '$@2'
-      expect(rules[1].token_code.s_value).to eq '$2 + $3'
-      expect(rules[1].original_rule).to eq rule
+      rules = rule_builder.rules
+      midrule_1 = rules.find {|rule| rule._lhs.s_value == "@1"}
+      midrule_2 = rules.find {|rule| rule._lhs.s_value == "$@2"}
+      rule = rules.find {|rule| rule._lhs.s_value == "class"}
+
+      expect(rules.count).to eq 3
+      expect(midrule_1._lhs.s_value).to eq '@1'
+      expect(midrule_1.token_code.s_value).to eq '$1'
+      expect(midrule_1.original_rule).to eq rule
+      expect(midrule_2._lhs.s_value).to eq '$@2'
+      expect(midrule_2.token_code.s_value).to eq '$2 + $3'
+      expect(midrule_2.original_rule).to eq rule
     end
   end
 

diff --git a/spec/lrama/parser_spec.rb b/spec/lrama/parser_spec.rb
@@ -567,10 +567,10 @@
 
         expect(grammar.nterms.sort_by(&:number)).to match_symbols([
           Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 5, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "option_number"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "option_number_alias"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 4, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "option_number"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "option_number_alias"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 4, nullable: true),
         ])
 
         expect(grammar.rules).to eq([
@@ -656,10 +656,10 @@
 
         expect(grammar.nterms.sort_by(&:number)).to match_symbols([
           Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 5, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "option_number"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "option_number_alias"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 4, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "option_number"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "option_number_alias"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 4, nullable: true),
         ])
 
         expect(grammar.rules).to eq([
@@ -744,8 +744,8 @@
 
         expect(grammar.nterms.sort_by(&:number)).to match_symbols([
           Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 1, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "option_bar"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 2, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "option_bar"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 1, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 2, nullable: false),
         ])
 
         expect(grammar.rules).to eq([
@@ -804,10 +804,10 @@
 
         expect(grammar.nterms.sort_by(&:number)).to match_symbols([
           Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 5, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "option_number"), alias_name: nil, number: 7, tag: T::Tag.new(s_value: "<i>"), term: false, token_id: 2, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "option_number_alias"), alias_name: nil, number: 9, tag: T::Tag.new(s_value: "<i>"), term: false, token_id: 4, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "option_number"), alias_name: nil, number: 6, tag: T::Tag.new(s_value: "<i>"), term: false, token_id: 1, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "option_number_alias"), alias_name: nil, number: 8, tag: T::Tag.new(s_value: "<i>"), term: false, token_id: 3, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 4, nullable: true),
         ])
 
         expect(grammar.rules).to eq([
@@ -896,10 +896,10 @@
 
         expect(grammar.nterms.sort_by(&:number)).to match_symbols([
           Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 5, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "nonempty_list_number"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "nonempty_list_number_alias"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 4, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "nonempty_list_number"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "nonempty_list_number_alias"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 4, nullable: false),
         ])
 
         expect(grammar.rules).to eq([
@@ -990,10 +990,10 @@
 
         expect(grammar.nterms.sort_by(&:number)).to match_symbols([
           Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 5, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "list_number"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "list_number_alias"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 4, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "list_number"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "list_number_alias"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 4, nullable: true),
         ])
 
         expect(grammar.rules).to eq([
@@ -1079,9 +1079,9 @@
         grammar = Lrama::Parser.new(y, path).parse
 
         expect(grammar.nterms.sort_by(&:number)).to match_symbols([
-          Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number:  5, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number:  6, tag: nil, term: false, token_id: 1, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "separated_nonempty_list_number"), alias_name: nil, number:  7, tag: nil, term: false, token_id: 2, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 5, tag: nil, term: false, token_id: 0, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "separated_nonempty_list_number"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: false),
         ])
 
         expect(grammar.rules).to eq([
@@ -1141,10 +1141,10 @@
         grammar = Lrama::Parser.new(y, path).parse
 
         expect(grammar.nterms.sort_by(&:number)).to match_symbols([
-          Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number:  5, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number:  6, tag: nil, term: false, token_id: 1, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "separated_list_number"), alias_name: nil, number:  7, tag: nil, term: false, token_id: 2, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "separated_nonempty_list_number"), alias_name: nil, number:  8, tag: nil, term: false, token_id: 3, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 5, tag: nil, term: false, token_id: 0, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "separated_list_number"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "separated_nonempty_list_number"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: true),
         ])
 
         expect(grammar.rules).to eq([
@@ -1225,8 +1225,8 @@
 
         expect(grammar.nterms.sort_by(&:number)).to match_symbols([
           Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 1, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "defined_option_number"), alias_name: nil, number: 8, tag: T::Tag.new(s_value: "<i>"), term: false, token_id: 2, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "defined_option_number"), alias_name: nil, number: 7, tag: T::Tag.new(s_value: "<i>"), term: false, token_id: 1, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 2, nullable: true),
           Sym.new(id: T::Ident.new(s_value: "multi_args_number_string"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 3, nullable: false),
           Sym.new(id: T::Ident.new(s_value: "multi_args_number_number"), alias_name: nil, number: 10, tag: nil, term: false, token_id: 4, nullable: false),
           Sym.new(id: T::Ident.new(s_value: "pair_number_string"), alias_name: nil, number: 11, tag: nil, term: false, token_id: 5, nullable: false),
@@ -1386,10 +1386,10 @@
 
         expect(grammar.nterms.sort_by(&:number)).to match_symbols([
           Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 5, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "option_number"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "nested_option_number"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: true),
-          Sym.new(id: T::Ident.new(s_value: "nested_nested_option_number"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 4, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "option_number"), alias_name: nil, number: 6, tag: nil, term: false, token_id: 1, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "nested_option_number"), alias_name: nil, number: 7, tag: nil, term: false, token_id: 2, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "nested_nested_option_number"), alias_name: nil, number: 8, tag: nil, term: false, token_id: 3, nullable: true),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 9, tag: nil, term: false, token_id: 4, nullable: true),
           Sym.new(id: T::Ident.new(s_value: "multi_option_number_string"), alias_name: nil, number: 10, tag: nil, term: false, token_id: 5, nullable: true),
           Sym.new(id: T::Ident.new(s_value: "nested_multi_option_number"), alias_name: nil, number: 11, tag: nil, term: false, token_id: 6, nullable: true),
           Sym.new(id: T::Ident.new(s_value: "nested_multi_option_string"), alias_name: nil, number: 12, tag: nil, term: false, token_id: 7, nullable: true)
@@ -1700,11 +1700,11 @@ class : keyword_class { code 1 } tSTRING { code 2 } keyword_end { code 3 }
       grammar = Lrama::Parser.new(y, "parse.y").parse
 
       expect(grammar.nterms.sort_by(&:number)).to match_symbols([
-        Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 11, tag: nil,                                 term: false, token_id: 0, nullable: false),
-        Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 12, tag: nil,                                 term: false, token_id: 1, nullable: false),
-        Sym.new(id: T::Ident.new(s_value: "class"),   alias_name: nil, number: 13, tag: T::Tag.new(s_value: "<i>"), term: false, token_id: 2, nullable: false),
-        Sym.new(id: T::Ident.new(s_value: "$@1"),     alias_name: nil, number: 14, tag: nil,                                 term: false, token_id: 3, nullable: true),
-        Sym.new(id: T::Ident.new(s_value: "$@2"),     alias_name: nil, number: 15, tag: nil,                                 term: false, token_id: 4, nullable: true),
+        Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number: 11, tag: nil,                        term: false, token_id: 0, nullable: false),
+        Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number: 12, tag: nil,                        term: false, token_id: 1, nullable: false),
+        Sym.new(id: T::Ident.new(s_value: "$@1"),     alias_name: nil, number: 13, tag: nil,                        term: false, token_id: 2, nullable: true),
+        Sym.new(id: T::Ident.new(s_value: "$@2"),     alias_name: nil, number: 14, tag: nil,                        term: false, token_id: 3, nullable: true),
+        Sym.new(id: T::Ident.new(s_value: "class"),   alias_name: nil, number: 15, tag: T::Tag.new(s_value: "<i>"), term: false, token_id: 4, nullable: false),
       ])
       expect(grammar.rules).to eq([
         Rule.new(

diff --git a/spec/lrama/states_spec.rb b/spec/lrama/states_spec.rb
@@ -993,46 +993,46 @@ class    go to state 5
 
     $default  reduce using rule 1 ($@1)
 
-    program  go to state 1
-    $@1      go to state 2
+    $@1      go to state 1
+    program  go to state 2
 
 
 State 1
 
-    0 $accept: program • "EOI"
+    2 program: $@1 • expr
+    3 expr: • tNUMBER
 
-    "EOI"  shift, and go to state 3
+    tNUMBER  shift, and go to state 3
 
+    expr  go to state 4
 
-State 2
 
-    2 program: $@1 • expr
-    3 expr: • tNUMBER
+State 2
 
-    tNUMBER  shift, and go to state 4
+    0 $accept: program • "EOI"
 
-    expr  go to state 5
+    "EOI"  shift, and go to state 5
 
 
 State 3
 
-    0 $accept: program "EOI" •
+    3 expr: tNUMBER •
 
-    $default  accept
+    $default  reduce using rule 3 (expr)
 
 
 State 4
 
-    3 expr: tNUMBER •
+    2 program: $@1 expr •
 
-    $default  reduce using rule 3 (expr)
+    $default  reduce using rule 2 (program)
 
 
 State 5
 
-    2 program: $@1 expr •
+    0 $accept: program "EOI" •
 
-    $default  reduce using rule 2 (program)
+    $default  accept
 
 
         STR