Merge pull request #197 from ydah/parameterizing_rules_2

Support another syntactic sugar for parameterizing rules
ruby · Nov 5, 2023 · cb9f693 · cb9f693
2 parents ed65667 + 206f2e9
commit cb9f693
Show file tree

Hide file tree

Showing 8 changed files with 437 additions and 284 deletions.
diff --git a/lib/lrama/lexer.rb b/lib/lrama/lexer.rb
@@ -8,7 +8,7 @@ class Lexer
     attr_accessor :status
     attr_accessor :end_symbol
 
-    SYMBOLS = %w(%{ %} %% { } \[ \] : \| ;)
+    SYMBOLS = ['%{', '%}', '%%', '{', '}', '\[', '\]', '\(', '\)', ':', '\|', ';']
     PERCENT_TOKENS = %w(
       %union
       %token
@@ -29,6 +29,11 @@ class Lexer
       %empty
       %code
     )
+    PARAMETERIZING_TOKENS = [
+      'option\(',
+      'nonempty_list\(',
+      'list\('
+    ]
 
     def initialize(text)
       @scanner = StringScanner.new(text)
@@ -89,6 +94,8 @@ def lex_token
         return [@scanner.matched, @scanner.matched]
       when @scanner.scan(/#{PERCENT_TOKENS.join('|')}/)
         return [@scanner.matched, @scanner.matched]
+      when @scanner.scan(/#{PARAMETERIZING_TOKENS.join('|')}/)
+        return [@scanner.matched, @scanner.matched]
       when @scanner.scan(/[\?\+\*]/)
         return [@scanner.matched, @scanner.matched]
       when @scanner.scan(/<\w+>/)

diff --git a/lib/lrama/lexer/token/parameterizing.rb b/lib/lrama/lexer/token/parameterizing.rb
@@ -3,15 +3,15 @@ class Lexer
     class Token
       class Parameterizing < Token
         def option?
-          self.s_value == "?"
+          %w(option ?).include?(self.s_value)
         end
 
         def nonempty_list?
-          self.s_value == "+"
+          %w(nonempty_list +).include?(self.s_value)
         end
 
         def list?
-          self.s_value == "*"
+          %w(list *).include?(self.s_value)
         end
       end
     end

diff --git a/lib/lrama/parser.rb b/lib/lrama/parser.rb
diff --git a/parser.y b/parser.y
@@ -320,6 +320,11 @@ rule
             token = Lrama::Lexer::Token::Parameterizing.new(s_value: val[1])
             result = val[0].append(token)
           }
+     | parameterizing_prefix rhs ")"
+        {
+          token = Lrama::Lexer::Token::Parameterizing.new(s_value: val[0].chop)
+          result = val[1].append(token)
+        }
      | rhs "{"
          {
            if @prec_seen
@@ -345,6 +350,10 @@ rule
            @prec_seen = true
          }
 
+  parameterizing_prefix: "option("
+                       | "nonempty_list("
+                       | "list("
+
   parameterizing_suffix: "?"
                        | "+"
                        | "*"

diff --git a/spec/fixtures/parameterizing_rules/list.y b/spec/fixtures/parameterizing_rules/list.y
@@ -14,10 +14,14 @@ static int yyerror(YYLTYPE *loc, const char *str);
 }
 
 %token <i> number
+%token <i> number_alias
 
 %%
 
-program         : number*
+program         : list(number)
+                ;
+
+alias           : number_alias*
                 ;
 
 %%

diff --git a/spec/fixtures/parameterizing_rules/nonempty_list.y b/spec/fixtures/parameterizing_rules/nonempty_list.y
@@ -14,10 +14,14 @@ static int yyerror(YYLTYPE *loc, const char *str);
 }
 
 %token <i> number
+%token <i> number_alias
 
 %%
 
-program         : number+
+program         : nonempty_list(number)
+                ;
+
+alias           : number_alias+
                 ;
 
 %%

diff --git a/spec/fixtures/parameterizing_rules/option.y b/spec/fixtures/parameterizing_rules/option.y
@@ -14,10 +14,14 @@ static int yyerror(YYLTYPE *loc, const char *str);
 }
 
 %token <i> number
+%token <i> number_alias
 
 %%
 
-program         : number?
+program         : option(number)
+                ;
+
+alias           : number_alias?
                 ;
 
 %%

diff --git a/spec/lrama/parser_spec.rb b/spec/lrama/parser_spec.rb
@@ -559,8 +559,9 @@
         grammar = Lrama::Parser.new(y, path).parse
 
         expect(grammar.nterms.sort_by(&:number)).to eq([
-          Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number:  5, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number:  6, tag: nil, term: false, token_id: 1, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number:  7, tag: nil, term: false, token_id: 0, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number:  8, tag: nil, term: false, token_id: 1, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number:  9, tag: nil, term: false, token_id: 2, nullable: false),
         ])
 
         expect(grammar.rules).to eq([
@@ -574,7 +575,7 @@
             token_code: nil,
             nullable: false,
             precedence_sym: grammar.find_symbol_by_s_value!("YYEOF"),
-            lineno: 20,
+            lineno: 21,
           ),
           Rule.new(
             id: 1,
@@ -585,7 +586,7 @@
             token_code: nil,
             nullable: false,
             precedence_sym: grammar.find_symbol_by_s_value!("option_number"),
-            lineno: 20,
+            lineno: 21,
           ),
           Rule.new(
             id: 2,
@@ -594,7 +595,7 @@
             token_code: nil,
             nullable: true,
             precedence_sym: nil,
-            lineno: 20,
+            lineno: 21,
           ),
           Rule.new(
             id: 3,
@@ -605,7 +606,35 @@
             token_code: nil,
             nullable: false,
             precedence_sym: grammar.find_symbol_by_s_value!("number"),
-            lineno: 20,
+            lineno: 21,
+          ),
+          Rule.new(
+            id: 4,
+            lhs: grammar.find_symbol_by_s_value!("alias"),
+            rhs: [
+              grammar.find_symbol_by_s_value!("option_number_alias"),
+            ],
+            nullable: false,
+            precedence_sym: grammar.find_symbol_by_s_value!("option_number_alias"),
+            lineno: 24,
+          ),
+          Rule.new(
+            id: 5,
+            lhs: grammar.find_symbol_by_s_value!("option_number_alias"),
+            rhs: [],
+            nullable: true,
+            precedence_sym: nil,
+            lineno: 24,
+          ),
+          Rule.new(
+            id: 6,
+            lhs: grammar.find_symbol_by_s_value!("option_number_alias"),
+            rhs: [
+              grammar.find_symbol_by_s_value!("number_alias"),
+            ],
+            nullable: false,
+            precedence_sym: grammar.find_symbol_by_s_value!("number_alias"),
+            lineno: 24,
           ),
         ])
       end
@@ -616,8 +645,9 @@
         grammar = Lrama::Parser.new(y, path).parse
 
         expect(grammar.nterms.sort_by(&:number)).to eq([
-          Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number:  5, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number:  6, tag: nil, term: false, token_id: 1, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number:  7, tag: nil, term: false, token_id: 0, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number:  8, tag: nil, term: false, token_id: 1, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number:  9, tag: nil, term: false, token_id: 2, nullable: false),
         ])
 
         expect(grammar.rules).to eq([
@@ -631,7 +661,7 @@
             token_code: nil,
             nullable: false,
             precedence_sym: grammar.find_symbol_by_s_value!("YYEOF"),
-            lineno: 20,
+            lineno: 21,
           ),
           Rule.new(
             id: 1,
@@ -642,7 +672,7 @@
             token_code: nil,
             nullable: false,
             precedence_sym: grammar.find_symbol_by_s_value!("nonempty_list_number"),
-            lineno: 20,
+            lineno: 21,
           ),
           Rule.new(
             id: 2,
@@ -653,7 +683,7 @@
             token_code: nil,
             nullable: false,
             precedence_sym: grammar.find_symbol_by_s_value!("number"),
-            lineno: 20,
+            lineno: 21,
           ),
           Rule.new(
             id: 3,
@@ -665,7 +695,38 @@
             token_code: nil,
             nullable: false,
             precedence_sym: grammar.find_symbol_by_s_value!("number"),
-            lineno: 20,
+            lineno: 21,
+          ),
+          Rule.new(
+            id: 4,
+            lhs: grammar.find_symbol_by_s_value!("alias"),
+            rhs: [
+              grammar.find_symbol_by_s_value!("nonempty_list_number_alias"),
+            ],
+            nullable: false,
+            precedence_sym: grammar.find_symbol_by_s_value!("nonempty_list_number_alias"),
+            lineno: 24,
+          ),
+          Rule.new(
+            id: 5,
+            lhs: grammar.find_symbol_by_s_value!("nonempty_list_number_alias"),
+            rhs: [
+              grammar.find_symbol_by_s_value!("number_alias"),
+            ],
+            nullable: false,
+            precedence_sym: grammar.find_symbol_by_s_value!("number_alias"),
+            lineno: 24,
+          ),
+          Rule.new(
+            id: 6,
+            lhs: grammar.find_symbol_by_s_value!("nonempty_list_number_alias"),
+            rhs: [
+              grammar.find_symbol_by_s_value!("nonempty_list_number_alias"),
+              grammar.find_symbol_by_s_value!("number_alias"),
+            ],
+            nullable: false,
+            precedence_sym: grammar.find_symbol_by_s_value!("number_alias"),
+            lineno: 24,
           ),
         ])
       end
@@ -676,8 +737,9 @@
         grammar = Lrama::Parser.new(y, path).parse
 
         expect(grammar.nterms.sort_by(&:number)).to eq([
-          Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number:  5, tag: nil, term: false, token_id: 0, nullable: false),
-          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number:  6, tag: nil, term: false, token_id: 1, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "$accept"), alias_name: nil, number:  7, tag: nil, term: false, token_id: 0, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "program"), alias_name: nil, number:  8, tag: nil, term: false, token_id: 1, nullable: false),
+          Sym.new(id: T::Ident.new(s_value: "alias"), alias_name: nil, number:  9, tag: nil, term: false, token_id: 2, nullable: false),
         ])
 
         expect(grammar.rules).to eq([
@@ -691,7 +753,7 @@
             token_code: nil,
             nullable: false,
             precedence_sym: grammar.find_symbol_by_s_value!("YYEOF"),
-            lineno: 20,
+            lineno: 21,
           ),
           Rule.new(
             id: 1,
@@ -702,7 +764,7 @@
             token_code: nil,
             nullable: false,
             precedence_sym: grammar.find_symbol_by_s_value!("list_number"),
-            lineno: 20,
+            lineno: 21,
           ),
           Rule.new(
             id: 2,
@@ -711,7 +773,7 @@
             token_code: nil,
             nullable: true,
             precedence_sym: nil,
-            lineno: 20,
+            lineno: 21,
           ),
           Rule.new(
             id: 3,
@@ -723,7 +785,36 @@
             token_code: nil,
             nullable: false,
             precedence_sym: grammar.find_symbol_by_s_value!("number"),
-            lineno: 20,
+            lineno: 21,
+          ),
+          Rule.new(
+            id: 4,
+            lhs: grammar.find_symbol_by_s_value!("alias"),
+            rhs: [
+              grammar.find_symbol_by_s_value!("list_number_alias"),
+            ],
+            nullable: false,
+            precedence_sym: grammar.find_symbol_by_s_value!("list_number_alias"),
+            lineno: 24,
+          ),
+          Rule.new(
+            id: 5,
+            lhs: grammar.find_symbol_by_s_value!("list_number_alias"),
+            rhs: [],
+            nullable: true,
+            precedence_sym: nil,
+            lineno: 24,
+          ),
+          Rule.new(
+            id: 6,
+            lhs: grammar.find_symbol_by_s_value!("list_number_alias"),
+            rhs: [
+              grammar.find_symbol_by_s_value!("list_number_alias"),
+              grammar.find_symbol_by_s_value!("number_alias"),
+            ],
+            nullable: false,
+            precedence_sym: grammar.find_symbol_by_s_value!("number_alias"),
+            lineno: 24,
           ),
         ])
       end