wip

Author: yui-knk

Steepfile

@@ -7,6 +7,7 @@ target :lib do
   check "lib/lrama/grammar/code/printer_code.rb"
   check "lib/lrama/grammar/code.rb"
   check "lib/lrama/grammar/counter.rb"
+  check "lib/lrama/grammar/parser_state.rb"
   check "lib/lrama/grammar/percent_code.rb"
   check "lib/lrama/grammar/precedence.rb"
   check "lib/lrama/grammar/printer.rb"
@@ -15,6 +16,9 @@ target :lib do
   check "lib/lrama/lexer/token/char.rb"
   check "lib/lrama/lexer/token/ident.rb"
   check "lib/lrama/lexer/token/parameterizing.rb"
+  check "lib/lrama/lexer/token/parser_state_pop.rb"
+  check "lib/lrama/lexer/token/parser_state_push.rb"
+  check "lib/lrama/lexer/token/parser_state_set.rb"
   check "lib/lrama/lexer/token/tag.rb"
   check "lib/lrama/lexer/token/user_code.rb"
   check "lib/lrama/lexer/location.rb"

lib/lrama/grammar.rb

@@ -2,6 +2,7 @@
 require "lrama/grammar/code"
 require "lrama/grammar/counter"
 require "lrama/grammar/error_token"
+require "lrama/grammar/parser_state"
 require "lrama/grammar/percent_code"
 require "lrama/grammar/precedence"
 require "lrama/grammar/printer"
@@ -16,7 +17,7 @@
 module Lrama
   # Grammar is the result of parsing an input grammar file
   class Grammar
-    attr_reader :percent_codes, :eof_symbol, :error_symbol, :undef_symbol, :accept_symbol, :aux
+    attr_reader :percent_codes, :parser_states, :eof_symbol, :error_symbol, :undef_symbol, :accept_symbol, :aux
     attr_accessor :union, :expect,
                   :printers, :error_tokens,
                   :lex_param, :parse_param, :initial_action,
@@ -31,6 +32,7 @@ def initialize(rule_counter)
       @percent_codes = []
       @printers = []
       @error_tokens = []
+      @parser_states = []
       @symbols = []
       @types = []
       @rule_builders = []
@@ -58,6 +60,10 @@ def add_error_token(ident_or_tags:, token_code:, lineno:)
       @error_tokens << ErrorToken.new(ident_or_tags: ident_or_tags, token_code: token_code, lineno: lineno)
     end
 
+    def add_parser_state(state_id, state_list)
+      @parser_states << ParserState.new(state_id: state_id, state_list: state_list)
+    end
+
     def add_term(id:, alias_name: nil, tag: nil, token_id: nil, replace: false)
       if token_id && (sym = @symbols.find {|s| s.token_id == token_id })
         if replace
@@ -186,7 +192,7 @@ def find_symbol_by_id(id)
     end
 
     def find_symbol_by_id!(id)
-      find_symbol_by_id(id) || (raise "Symbol not found: #{id}")
+      find_symbol_by_id(id) || (raise "Symbol not found: #{id.s_value}")
     end
 
     def find_symbol_by_number!(number)

lib/lrama/grammar/parser_state.rb

@@ -0,0 +1,123 @@
+module Lrama
+  class Grammar
+    class ParserState
+      attr_reader :state_id, :state_list
+
+      def initialize(state_id:, state_list:)
+        @state_id = state_id
+        @state_list = state_list
+      end
+
+      def enum_definition
+        <<~ENUM
+          enum #{enum_name}
+          {
+            #{enum_body}
+          };
+          typedef enum #{enum_name} #{enum_type};
+
+          static const char *const #{enum_name_table_name}[] = {
+            #{int_to_name.join(", ")}
+          };
+
+          YY_ATTRIBUTE_UNUSED
+          static const char *
+          #{enum_name}_name (#{enum_type} num)
+          {
+            return #{enum_name_table_name}[num];
+          }
+
+          # define #{state_name_macro}(value) #{enum_name}_name (value)
+          # define #{current_state_name_macro} #{state_name_macro} (*#{stack_prefix}_p)
+        ENUM
+      end
+
+      def state_name_macro
+        "YY_STATE_#{state_name.upcase}_NAME"
+      end
+
+      def current_state_name_macro
+        "YY_CURRENT_STATE_#{state_name.upcase}_NAME"
+      end
+
+      def states_functions
+        # TODO: Stack check for push
+        <<~FUNC
+          # define YYPUSH_STATE_#{state_name.upcase}(value) \\
+            do \\
+              { \\
+                YYDPRINTF ((stderr, "Push %s to #{state_name}\\n", #{state_name_macro} (value))); \\
+                *++#{stack_prefix}_p = value; \\
+              } \\
+            while (0)
+
+          # define YYPOP_STATE_#{state_name.upcase}() \\
+            do \\
+              { \\
+                YYDPRINTF ((stderr, "Pop #{state_name}\\n")); \\
+                if (#{stack_prefix}_p != #{stack_prefix}) \\
+                  { \\
+                    #{stack_prefix}_p -= 1; \\
+                  } \\
+                else \\
+                  { \\
+                    YYDPRINTF ((stderr, "Try to pop empty #{state_name} stack\\n")); \\
+                  } \\
+              } \\
+            while (0)
+
+          # define YYSET_STATE_#{state_name.upcase}(value) \\
+            do \\
+              { \\
+                YYDPRINTF ((stderr, "Set %s to #{state_name}\\n", #{state_name_macro} (value))); \\
+                *#{stack_prefix}_p = value; \\
+              } \\
+            while (0)
+
+          # define YY_STATE_#{state_name.upcase} #{stack_prefix}_p
+        FUNC
+      end
+
+      def states_stacks
+        <<~STACKS
+          /* The parser state stack (#{stack_prefix}): array, bottom, top.  */
+          int #{stack_prefix}_a[YYINITDEPTH];
+          int *#{stack_prefix} = #{stack_prefix}_a;
+          int *#{stack_prefix}_p = #{stack_prefix};
+        STACKS
+      end
+
+      def state_name
+        state_id.s_value
+      end
+
+      def enum_name
+        "yyparser_state_#{state_name}"
+      end
+
+      def enum_type
+        "#{enum_name}_t"
+      end
+
+      def enum_body
+        state_list.map do |state|
+          state.s_value
+        end.join(",\n  ")
+      end
+
+      def int_to_name
+        state_list.map do |state|
+           "\"#{state.s_value}\""
+        end << "YY_NULLPTR"
+      end
+
+      def enum_name_table_name
+        "#{enum_name}_names"
+      end
+
+      def stack_prefix
+        "yyparser_state_#{state_name}"
+      end
+    end
+  end
+end

lib/lrama/grammar/rule_builder.rb

@@ -117,6 +117,12 @@ def process_rhs
           when Lrama::Lexer::Token::Parameterizing
             @parameterizing_rules = ParameterizingRules::Builder.new(token, @rule_counter, lhs, user_code, precedence_sym, line).build
             @replaced_rhs << token
+          when Lrama::Lexer::Token::ParserStatePop
+            process_parser_state_token(token, "parser_state_pop_", "YYPOP_STATE_#{token.s_value.upcase}();", i)
+          when Lrama::Lexer::Token::ParserStatePush
+            process_parser_state_token(token, "parser_state_push_", "YYPUSH_STATE_#{token.s_value.upcase}(#{token.state.s_value});", i)
+          when Lrama::Lexer::Token::ParserStateSet
+            process_parser_state_token(token, "parser_state_set_", "YYSET_STATE_#{token.s_value.upcase}(#{token.state.s_value});", i)
           when Lrama::Lexer::Token::UserCode
             prefix = token.referred ? "@" : "$@"
             new_token = Lrama::Lexer::Token::Ident.new(s_value: prefix + @midrule_action_counter.increment.to_s)
@@ -135,6 +141,20 @@ def process_rhs
         end
       end
 
+      def process_parser_state_token(token, prefix, code, position_in_original_rule_rhs)
+        new_token = Lrama::Lexer::Token::Ident.new(s_value: prefix + token.s_value + @midrule_action_counter.increment.to_s)
+        user_code = Lrama::Lexer::Token::UserCode.new(s_value: code, location: token.location)
+
+        @replaced_rhs << new_token
+        rule_builder = RuleBuilder.new(@rule_counter, @midrule_action_counter, position_in_original_rule_rhs, skip_preprocess_references: true)
+        rule_builder.lhs = new_token
+        rule_builder.user_code = user_code
+        rule_builder.complete_input
+        rule_builder.setup_rules
+
+        @rule_builders_for_derived_rules << rule_builder
+      end
+
       def numberize_references
         # Bison n'th component is 1-origin
         (rhs + [user_code]).compact.each.with_index(1) do |token, i|

lib/lrama/lexer.rb

@@ -28,6 +28,10 @@ class Lexer
       %error-token
       %empty
       %code
+      %parser-state
+      %push
+      %pop
+      %set
     )
 
     def initialize(text)

lib/lrama/lexer/token.rb

@@ -40,5 +40,8 @@ def last_column
 require 'lrama/lexer/token/char'
 require 'lrama/lexer/token/ident'
 require 'lrama/lexer/token/parameterizing'
+require 'lrama/lexer/token/parser_state_pop'
+require 'lrama/lexer/token/parser_state_push'
+require 'lrama/lexer/token/parser_state_set'
 require 'lrama/lexer/token/tag'
 require 'lrama/lexer/token/user_code'

lib/lrama/lexer/token/parser_state_pop.rb

@@ -0,0 +1,8 @@
+module Lrama
+  class Lexer
+    class Token
+      class ParserStatePop < Token
+      end
+    end
+  end
+end

lib/lrama/lexer/token/parser_state_push.rb

@@ -0,0 +1,9 @@
+module Lrama
+  class Lexer
+    class Token
+      class ParserStatePush < Token
+        attr_accessor :state
+      end
+    end
+  end
+end

lib/lrama/lexer/token/parser_state_set.rb

@@ -0,0 +1,9 @@
+module Lrama
+  class Lexer
+    class Token
+      class ParserStateSet < Token
+        attr_accessor :state
+      end
+    end
+  end
+end

lib/lrama/output.rb

@@ -358,6 +358,24 @@ def percent_code(name)
       end.join
     end
 
+    def parser_states_enums
+      @grammar.parser_states.map do |ps|
+        ps.enum_definition
+      end
+    end
+
+    def parser_states_stacks
+      @grammar.parser_states.map do |ps|
+        ps.states_stacks
+      end
+    end
+
+    def parser_states_functions
+      @grammar.parser_states.map do |ps|
+        ps.states_functions
+      end
+    end
+
     private
 
     def eval_template(file, path)