Merge pull request #285 from ydah/support-user-define

Support user define parameterizing rules

Steepfile

@@ -8,6 +8,9 @@ target :lib do
   check "lib/lrama/grammar/code.rb"
   check "lib/lrama/grammar/counter.rb"
   check "lib/lrama/grammar/error_token.rb"
+  check "lib/lrama/grammar/parameterizing_rule_builder.rb"
+  check "lib/lrama/grammar/parameterizing_rule_resolver.rb"
+  check "lib/lrama/grammar/parameterizing_rule_rhs_builder.rb"
   check "lib/lrama/grammar/parameterizing_rules"
   check "lib/lrama/grammar/percent_code.rb"
   check "lib/lrama/grammar/precedence.rb"

lib/lrama/grammar.rb

@@ -8,6 +8,10 @@
 require "lrama/grammar/reference"
 require "lrama/grammar/rule"
 require "lrama/grammar/rule_builder"
+require "lrama/grammar/parameterizing_rule_builder"
+require "lrama/grammar/parameterizing_rule_resolver"
+require "lrama/grammar/parameterizing_rule_rhs_builder"
+require "lrama/grammar/parameterizing_rule"
 require "lrama/grammar/symbol"
 require "lrama/grammar/type"
 require "lrama/grammar/union"
@@ -36,6 +40,7 @@ def initialize(rule_counter)
       @rule_builders = []
       @rules = []
       @sym_to_rules = {}
+      @parameterizing_resolver = ParameterizingRuleResolver.new
       @empty_symbol = nil
       @eof_symbol = nil
       @error_symbol = nil
@@ -129,6 +134,10 @@ def add_rule_builder(builder)
       @rule_builders << builder
     end
 
+    def add_parameterizing_rule_builder(builder)
+      @parameterizing_resolver.add_parameterizing_rule_builder(builder)
+    end
+
     def prologue_first_lineno=(prologue_first_lineno)
       @aux.prologue_first_lineno = prologue_first_lineno
     end
@@ -310,7 +319,7 @@ def compute_first_set
 
     def setup_rules
       @rule_builders.each do |builder|
-        builder.setup_rules
+        builder.setup_rules(@parameterizing_resolver)
       end
     end
 

lib/lrama/grammar/parameterizing_rule.rb

@@ -0,0 +1,6 @@
+module Lrama
+  class Grammar
+    class ParameterizingRule < Struct.new(:rules, :token, keyword_init: true)
+    end
+  end
+end

lib/lrama/grammar/parameterizing_rule_builder.rb

@@ -0,0 +1,45 @@
+module Lrama
+  class Grammar
+    class ParameterizingRuleBuilder
+      attr_reader :name, :args, :rhs
+
+      def initialize(name, args, rhs)
+        @name = name
+        @args = args
+        @rhs = rhs
+        @required_args_count = args.count
+      end
+
+      def build_rules(token, rule_counter, lhs_tag, line)
+        validate_argument_number!(token)
+        lhs = lhs_token(token)
+        rules = []
+        @rhs.each do |rhs|
+          rules << Rule.new(id: rule_counter.increment, _lhs: lhs, _rhs: [rhs_token(token, rhs)].compact, lhs_tag: lhs_tag, token_code: rhs.user_code, precedence_sym: rhs.precedence_sym, lineno: line)
+        end
+        ParameterizingRule.new(rules: rules, token: lhs)
+      end
+
+      private
+
+      def validate_argument_number!(token)
+        unless @required_args_count == token.args.count
+          raise "Invalid number of arguments. expect: #{@required_args_count} actual: #{token.args.count}"
+        end
+      end
+
+      def lhs_token(token)
+        Lrama::Lexer::Token::Ident.new(s_value: "#{name}_#{token.args.map(&:s_value).join('_')}")
+      end
+
+      def rhs_token(token, rhs)
+        return nil unless rhs.symbol
+        term = rhs.symbol
+        @args.each_with_index do |arg, index|
+          term = token.args[index] if arg.s_value == rhs.symbol.s_value
+        end
+        term
+      end
+    end
+  end
+end

lib/lrama/grammar/parameterizing_rule_resolver.rb

@@ -0,0 +1,24 @@
+module Lrama
+  class Grammar
+    class ParameterizingRuleResolver
+      def initialize
+        @parameterizing_rule_builders = []
+      end
+
+      def add_parameterizing_rule_builder(builder)
+        @parameterizing_rule_builders << builder
+      end
+
+      def defined?(name)
+        @parameterizing_rule_builders.any? { |builder| builder.name == name }
+      end
+
+      def build_rules(token, rule_counter, lhs_tag, line)
+        builder = @parameterizing_rule_builders.select { |b| b.name == token.s_value }.last
+        raise "Unknown parameterizing rule #{token.s_value} at line #{token.line}" unless builder
+
+        builder.build_rules(token, rule_counter, lhs_tag, line)
+      end
+    end
+  end
+end

lib/lrama/grammar/parameterizing_rule_rhs_builder.rb

@@ -0,0 +1,13 @@
+module Lrama
+  class Grammar
+    class ParameterizingRuleRhsBuilder
+      attr_accessor :symbol, :user_code, :precedence_sym
+
+      def initialize
+        @symbol = nil
+        @user_code = nil
+        @precedence_sym = nil
+      end
+    end
+  end
+end

lib/lrama/grammar/rule_builder.rb

@@ -51,9 +51,9 @@ def complete_input
         freeze_rhs
       end
 
-      def setup_rules
+      def setup_rules(parameterizing_resolver)
         preprocess_references unless @skip_preprocess_references
-        process_rhs
+        process_rhs(parameterizing_resolver)
         build_rules
       end
 
@@ -97,7 +97,7 @@ def build_rules
 
       # rhs is a mixture of variety type of tokens like `Ident`, `Parameterizing`, `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
+      def process_rhs(parameterizing_resolver)
         return if @replaced_rhs
 
         @replaced_rhs = []
@@ -110,11 +110,16 @@ def process_rhs
           when Lrama::Lexer::Token::Ident
             @replaced_rhs << token
           when Lrama::Lexer::Token::Parameterizing
-            parameterizing = ParameterizingRules::Builder.new(token, @rule_counter, @lhs_tag, user_code, precedence_sym, line)
-            parameterizing.build.each do |r|
-              @parameterizing_rules << r
+            if parameterizing_resolver.defined?(token.s_value)
+              parameterizing = parameterizing_resolver.build_rules(token, @rule_counter, @lhs_tag, line)
+              @parameterizing_rules = @parameterizing_rules + parameterizing.rules
+              @replaced_rhs << parameterizing.token
+            else
+              # TODO: Delete when the standard library will defined as a grammar file.
+              parameterizing = ParameterizingRules::Builder.new(token, @rule_counter, @lhs_tag, user_code, precedence_sym, line)
+              @parameterizing_rules = @parameterizing_rules + parameterizing.build
+              @replaced_rhs << parameterizing.build_token
             end
-            @replaced_rhs << parameterizing.build_token
           when Lrama::Lexer::Token::UserCode
             prefix = token.referred ? "@" : "$@"
             new_token = Lrama::Lexer::Token::Ident.new(s_value: prefix + @midrule_action_counter.increment.to_s)
@@ -124,7 +129,7 @@ def process_rhs
             rule_builder.lhs = new_token
             rule_builder.user_code = token
             rule_builder.complete_input
-            rule_builder.setup_rules
+            rule_builder.setup_rules(parameterizing_resolver)
 
             @rule_builders_for_derived_rules << rule_builder
           else

lib/lrama/lexer.rb

@@ -28,6 +28,7 @@ class Lexer
       %error-token
       %empty
       %code
+      %rule
     )
 
     def initialize(text)