racc 1.6.0 → 1.8.0

data/ext/racc/cparse/cparse.c

@@ -7,8 +7,6 @@
     This library is free software.
     You can distribute/modify this program under the same terms of ruby.
 
-    $originalId: cparse.c,v 1.8 2006/07/06 11:39:46 aamine Exp $
-
 */
 
 #include <ruby.h>
@@ -24,7 +22,7 @@
                         Important Constants
 ----------------------------------------------------------------------- */
 
-#define RACC_VERSION "1.4.15"
+#define RACC_VERSION STRINGIZE(RACC_INFO_VERSION)
 
 #define DEFAULT_TOKEN -1
 #define ERROR_TOKEN    1
@@ -72,12 +70,8 @@ static ID id_d_e_pop;
 #  define LONG2NUM(i) INT2NUM(i)
 #endif
 
-#ifndef HAVE_RB_ARY_SUBSEQ
-#  define rb_ary_subseq(ary, beg, len) rb_ary_new4(len, RARRAY_PTR(ary) + beg)
-#endif
-
-static ID value_to_id _((VALUE v));
-static inline long num_to_long _((VALUE n));
+static ID value_to_id(VALUE v);
+static inline long num_to_long(VALUE n);
 
 static ID
 value_to_id(VALUE v)
@@ -101,8 +95,8 @@ num_to_long(VALUE n)
                         Parser Stack Interfaces
 ----------------------------------------------------------------------- */
 
-static VALUE get_stack_tail _((VALUE stack, long len));
-static void cut_stack_tail _((VALUE stack, long len));
+static VALUE get_stack_tail(VALUE stack, long len);
+static void cut_stack_tail(VALUE stack, long len);
 
 static VALUE
 get_stack_tail(VALUE stack, long len)
@@ -191,27 +185,27 @@ struct cparse_params {
                         Parser Main Routines
 ----------------------------------------------------------------------- */
 
-static VALUE racc_cparse _((VALUE parser, VALUE arg, VALUE sysdebug));
-static VALUE racc_yyparse _((VALUE parser, VALUE lexer, VALUE lexmid,
-                             VALUE arg, VALUE sysdebug));
-
-static void call_lexer _((struct cparse_params *v));
-static VALUE lexer_i _((RB_BLOCK_CALL_FUNC_ARGLIST(block_args, data)));
-
-static VALUE assert_array _((VALUE a));
-static long assert_integer _((VALUE n));
-static VALUE assert_hash _((VALUE h));
-static VALUE initialize_params _((VALUE vparams, VALUE parser, VALUE arg,
-                                 VALUE lexer, VALUE lexmid));
-static void cparse_params_mark _((void *ptr));
-static size_t cparse_params_memsize _((const void *ptr));
-
-static void parse_main _((struct cparse_params *v,
-                         VALUE tok, VALUE val, int resume));
-static void extract_user_token _((struct cparse_params *v,
-                                  VALUE block_args, VALUE *tok, VALUE *val));
-static void shift _((struct cparse_params* v, long act, VALUE tok, VALUE val));
-static int reduce _((struct cparse_params* v, long act));
+static VALUE racc_cparse(VALUE parser, VALUE arg, VALUE sysdebug);
+static VALUE racc_yyparse(VALUE parser, VALUE lexer, VALUE lexmid,
+			  VALUE arg, VALUE sysdebug);
+
+static void call_lexer(struct cparse_params *v);
+static VALUE lexer_i(RB_BLOCK_CALL_FUNC_ARGLIST(block_args, data));
+
+static VALUE assert_array(VALUE a);
+static long assert_integer(VALUE n);
+static VALUE assert_hash(VALUE h);
+static VALUE initialize_params(VALUE vparams, VALUE parser, VALUE arg,
+			       VALUE lexer, VALUE lexmid);
+static void cparse_params_mark(void *ptr);
+static size_t cparse_params_memsize(const void *ptr);
+
+static void parse_main(struct cparse_params *v,
+		       VALUE tok, VALUE val, int resume);
+static void extract_user_token(struct cparse_params *v,
+			       VALUE block_args, VALUE *tok, VALUE *val);
+static void shift(struct cparse_params* v, long act, VALUE tok, VALUE val);
+static int reduce(struct cparse_params* v, long act);
 static rb_block_call_func reduce0;
 
 #ifdef DEBUG
@@ -280,28 +274,11 @@ racc_yyparse(VALUE parser, VALUE lexer, VALUE lexmid, VALUE arg, VALUE sysdebug)
     return v->retval;
 }
 
-#ifdef HAVE_RB_BLOCK_CALL
 static void
 call_lexer(struct cparse_params *v)
 {
     rb_block_call(v->lexer, v->lexmid, 0, NULL, lexer_i, v->value_v);
 }
-#else
-static VALUE
-lexer_iter(VALUE data)
-{
-    struct cparse_params *v = rb_check_typeddata(data, &cparse_params_type);
-
-    rb_funcall(v->lexer, v->lexmid, 0);
-    return Qnil;
-}
-
-static void
-call_lexer(struct cparse_params *v)
-{
-    rb_iterate(lexer_iter, v->value_v, lexer_i, v->value_v);
-}
-#endif
 
 static VALUE
 lexer_i(RB_BLOCK_CALL_FUNC_ARGLIST(block_args, data))

data/ext/racc/cparse/extconf.rb

@@ -1,9 +1,8 @@
-# frozen_string_literal: false
+# frozen_string_literal: true
 #
 
 require 'mkmf'
+require_relative '../../../lib/racc/info'
 
-have_func('rb_block_call')
-have_func('rb_ary_subseq')
-
+$defs << "-D""RACC_INFO_VERSION=#{Racc::VERSION}"
 create_makefile 'racc/cparse'

data/lib/racc/grammar.rb

@@ -27,6 +27,7 @@ module Racc
       @rules   = []  # :: [Rule]
       @start   = nil
       @n_expected_srconflicts = nil
+      @error_on_expect_mismatch = nil
       @prec_table = []
       @prec_table_closed = false
       @closed = false
@@ -36,6 +37,7 @@ module Racc
     attr_reader :start
     attr_reader :symboltable
     attr_accessor :n_expected_srconflicts
+    attr_accessor :error_on_expect_mismatch
 
     def [](x)
       @rules[x]
@@ -335,14 +337,14 @@ module Racc
         if action
           idbase = "#{type}@#{id}-#{@seqs[type] += 1}"
           target = _wrap(idbase, "#{idbase}-core", action)
-          _regist("#{idbase}-core", &block)
+          _register("#{idbase}-core", &block)
         else
-          target = _regist("#{type}@#{id}", &block)
+          target = _register("#{type}@#{id}", &block)
         end
         @grammar.intern(target)
       end
 
-      def _regist(target_name)
+      def _register(target_name)
         target = target_name.intern
         unless _added?(@grammar.intern(target))
           yield(target).each_rule do |rule|
@@ -787,6 +789,81 @@ module Racc
   end
 
 
+  class OptionMark
+    def initialize(lineno)
+      @lineno = lineno
+    end
+
+    def name
+      '?'
+    end
+
+    alias inspect name
+
+    attr_reader :lineno
+  end
+
+
+  class ManyMark
+    def initialize(lineno)
+      @lineno = lineno
+    end
+
+    def name
+      '*'
+    end
+
+    alias inspect name
+
+    attr_reader :lineno
+  end
+
+
+  class Many1Mark
+    def initialize(lineno)
+      @lineno = lineno
+    end
+
+    def name
+      '+'
+    end
+
+    alias inspect name
+
+    attr_reader :lineno
+  end
+
+
+  class GroupStartMark
+    def initialize(lineno)
+      @lineno = lineno
+    end
+
+    def name
+      '('
+    end
+
+    alias inspect name
+
+    attr_reader :lineno
+  end
+
+
+  class GroupEndMark
+    def initialize(lineno)
+      @lineno = lineno
+    end
+
+    def name
+      ')'
+    end
+
+    alias inspect name
+
+    attr_reader :lineno
+  end
+
+
   class Prec
     def initialize(symbol, lineno)
       @symbol = symbol
@@ -993,10 +1070,6 @@ module Racc
         @to_s = '$end'
         @serialized = 'false'
         @string = false
-      when ErrorSymbolValue
-        @to_s = 'error'
-        @serialized = 'Object.new'
-        @string = false
       else
         raise ArgumentError, "unknown symbol value: #{value.class}"
       end

data/lib/racc/grammarfileparser.rb

@@ -76,6 +76,9 @@ module Racc
                         raise CompileError, "`expect' seen twice"
                       end
                       @grammar.n_expected_srconflicts = num
+                    }\
+                  | seq(:ERROR_ON_EXPECT_MISMATCH) {|*|
+                      @grammar.error_on_expect_mismatch = true
                     }
 
     g.convdef     = seq(:symbol, :STRING) {|sym, code|
@@ -133,6 +136,21 @@ module Racc
                   | seq("|") {|*|
                       OrMark.new(@scanner.lineno)
                     }\
+                  | seq("?") {|*|
+                      OptionMark.new(@scanner.lineno)
+                    }\
+                  | seq("*") {|*|
+                      ManyMark.new(@scanner.lineno)
+                    }\
+                  | seq("+") {|*|
+                      Many1Mark.new(@scanner.lineno)
+                    }\
+                  | seq("(") {|*|
+                      GroupStartMark.new(@scanner.lineno)
+                    }\
+                  | seq(")") {|*|
+                      GroupEndMark.new(@scanner.lineno)
+                    }\
                   | seq("=", :symbol) {|_, sym|
                       Prec.new(sym, @scanner.lineno)
                     }\
@@ -210,27 +228,114 @@ module Racc
     end
 
     def add_rule_block(list)
-      sprec = nil
       target = list.shift
       case target
-      when OrMark, UserAction, Prec
+      when OrMark, OptionMark, ManyMark, Many1Mark, GroupStartMark, GroupEndMark, UserAction, Prec
         raise CompileError, "#{target.lineno}: unexpected symbol #{target.name}"
       end
+      enum = list.each.with_index
+      _, sym, idx = _add_rule_block(target, enum)
+      if idx
+        # sym is Racc::GroupEndMark
+        raise "#{sym.lineno}: unexpected symbol ')' at pos=#{idx}"
+      end
+    end
+
+    def _add_rule_block(target, enum)
+      rules = [] # [ [seqs, sprec], .. ]
       curr = []
-      list.each do |i|
-        case i
+      sprec = nil
+      while (sym, idx = enum.next rescue nil)
+        case sym
         when OrMark
-          add_rule target, curr, sprec
+          rules << [curr, sprec]
           curr = []
           sprec = nil
+        when OptionMark
+          curr << _add_option_rule(curr.pop)
+        when ManyMark
+          curr << _add_many_rule(curr.pop)
+        when Many1Mark
+          curr << _add_many1_rule(curr.pop)
+        when GroupStartMark
+          curr << _add_group_rule(enum)
+        when GroupEndMark
+          rules << [curr, sprec]
+          return rules, sym, idx
         when Prec
           raise CompileError, "'=<prec>' used twice in one rule" if sprec
-          sprec = i.symbol
+          sprec = sym.symbol
         else
-          curr.push i
+          curr.push sym
         end
       end
-      add_rule target, curr, sprec
+      rules << [curr, sprec]
+      rules.each do |syms, sprec|
+        add_rule target, syms, sprec
+      end
+      nil
+    end
+
+
+    def _add_option_rule(prev)
+      @option_rule_registry ||= {}
+      target = @option_rule_registry[prev.to_s]
+      return target if target
+      target = _gen_target_name("option", prev)
+      @option_rule_registry[prev.to_s] = target
+      act = UserAction.empty
+      @grammar.add Rule.new(target, [], act)
+      @grammar.add Rule.new(target, [prev], act)
+      target
+    end
+
+    def _add_many_rule(prev)
+      @many_rule_registry ||= {}
+      target = @many_rule_registry[prev.to_s]
+      return target if target
+      target = _gen_target_name("many", prev)
+      @many_rule_registry[prev.to_s] = target
+      src = SourceText.new("result = val[1] ? val[1].unshift(val[0]) : val", __FILE__, __LINE__)
+      act = UserAction.source_text(src)
+      @grammar.add Rule.new(target, [], act)
+      @grammar.add Rule.new(target, [prev, target], act)
+      target
+    end
+
+    def _add_many1_rule(prev)
+      @many1_rule_registry ||= {}
+      target = @many1_rule_registry[prev.to_s]
+      return target if target
+      target = _gen_target_name("many1", prev)
+      @many1_rule_registry[prev.to_s] = target
+      src = SourceText.new("result = val[1] ? val[1].unshift(val[0]) : val", __FILE__, __LINE__)
+      act = UserAction.source_text(src)
+      @grammar.add Rule.new(target, [prev], act)
+      @grammar.add Rule.new(target, [prev, target], act)
+      target
+    end
+
+    def _add_group_rule(enum)
+      target = @grammar.intern("-temp-group", true)
+      rules, _ = _add_rule_block(target, enum)
+      target_name = rules.map{|syms, sprec| syms.join("-")}.join("|")
+      @group_rule_registry ||= {}
+      unless target = @group_rule_registry[target_name]
+        target = @grammar.intern("-group@#{target_name}", true)
+        @group_rule_registry[target_name] = target
+        src = SourceText.new("result = val", __FILE__, __LINE__)
+        act = UserAction.source_text(src)
+        rules.each do |syms, sprec|
+          rule = Rule.new(target, syms, act)
+          rule.specified_prec = sprec
+          @grammar.add rule
+        end
+      end
+      target
+    end
+
+    def _gen_target_name(type, sym)
+      @grammar.intern("-#{type}@#{sym.value}", true)
     end
 
     def add_rule(target, list, sprec)
@@ -260,9 +365,9 @@ module Racc
       _, *blocks = *@scanner.epilogue.split(/^----/)
       blocks.each do |block|
         header, *body = block.lines.to_a
-        label0, pathes = *header.sub(/\A-+/, '').split('=', 2)
+        label0, paths = *header.sub(/\A-+/, '').split('=', 2)
         label = canonical_label(label0)
-        (pathes ? pathes.strip.split(' ') : []).each do |path|
+        (paths ? paths.strip.split(' ') : []).each do |path|
           add_user_code label, SourceText.new(File.read(path), path, 1)
         end
         add_user_code label, SourceText.new(body.join(''), @filename, line + 1)
@@ -391,6 +496,7 @@ module Racc
       'options'  => :OPTION,
       'start'    => :START,
       'expect'   => :EXPECT,
+      'error_on_expect_mismatch' => :ERROR_ON_EXPECT_MISMATCH,
       'class'    => :CLASS,
       'rule'     => :RULE,
       'end'      => :END

data/lib/racc/info.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 #--
 #
 #
@@ -11,7 +12,7 @@
 #++
 
 module Racc
-  VERSION   = '1.6.0'
+  VERSION   = '1.8.0'
   Version = VERSION
   Copyright = 'Copyright (c) 1999-2006 Minero Aoki'
 end

data/lib/racc/parser-text.rb

@@ -1,6 +1,5 @@
 module Racc
   PARSER_TEXT = <<'__end_of_file__'
-# frozen_string_literal: false
 #--
 # Copyright (c) 1999-2006 Minero Aoki
 #
@@ -12,17 +11,23 @@ module Racc
 # without restriction.
 #++
 
-require 'racc/info'
+unless $".find {|p| p.end_with?('/racc/info.rb')}
+$".push "#{__dir__}/racc/info.rb"
+
+module Racc
+  VERSION   = '1.8.0'
+  Version = VERSION
+  Copyright = 'Copyright (c) 1999-2006 Minero Aoki'
+end
 
-unless defined?(NotImplementedError)
-  NotImplementedError = NotImplementError # :nodoc:
 end
 
+
 module Racc
   class ParseError < StandardError; end
 end
 unless defined?(::ParseError)
-  ParseError = Racc::ParseError
+  ParseError = Racc::ParseError # :nodoc:
 end
 
 # Racc is a LALR(1) parser generator.
@@ -55,10 +60,12 @@ end
 # [-v, --verbose]
 #   verbose mode. create +filename+.output file, like yacc's y.output file.
 # [-g, --debug]
-#   add debug code to parser class. To display debuggin information,
+#   add debug code to parser class. To display debugging information,
 #   use this '-g' option and set @yydebug true in parser class.
 # [-E, --embedded]
 #   Output parser which doesn't need runtime files (racc/parser.rb).
+# [-F, --frozen]
+#   Output parser which declares frozen_string_literals: true
 # [-C, --check-only]
 #   Check syntax of racc grammar file and quit.
 # [-S, --output-status]
@@ -262,11 +269,11 @@ module Racc
     #     def next_token
     #       @q.shift
     #     end
-    class_eval %{
+    class_eval <<~RUBY, __FILE__, __LINE__ + 1
     def do_parse
       #{Racc_Main_Parsing_Routine}(_racc_setup(), false)
     end
-    }
+    RUBY
 
     # The method to fetch next token.
     # If you use #do_parse method, you must implement #next_token.
@@ -324,11 +331,11 @@ module Racc
     #
     # RECEIVER#METHOD_ID is a method to get next token.
     # It must 'yield' the token, which format is [TOKEN-SYMBOL, VALUE].
-    class_eval %{
+    class_eval <<~RUBY, __FILE__, __LINE__ + 1
     def yyparse(recv, mid)
       #{Racc_YY_Parse_Method}(recv, mid, _racc_setup(), false)
     end
-    }
+    RUBY
 
     def _racc_yyparse_rb(recv, mid, arg, c_debug)
       action_table, action_check, action_default, action_pointer,
@@ -537,7 +544,7 @@ module Racc
     #
     # If this method returns, parsers enter "error recovering mode".
     def on_error(t, val, vstack)
-      raise ParseError, sprintf("\nparse error on value %s (%s)",
+      raise ParseError, sprintf("parse error on value %s (%s)",
                                 val.inspect, token_to_str(t) || '?')
     end
 

data/lib/racc/parser.rb

@@ -1,4 +1,4 @@
-# frozen_string_literal: false
+# frozen_string_literal: true
 #--
 # Copyright (c) 1999-2006 Minero Aoki
 #
@@ -12,15 +12,11 @@
 
 require 'racc/info'
 
-unless defined?(NotImplementedError)
-  NotImplementedError = NotImplementError # :nodoc:
-end
-
 module Racc
   class ParseError < StandardError; end
 end
 unless defined?(::ParseError)
-  ParseError = Racc::ParseError
+  ParseError = Racc::ParseError # :nodoc:
 end
 
 # Racc is a LALR(1) parser generator.
@@ -53,10 +49,12 @@ end
 # [-v, --verbose]
 #   verbose mode. create +filename+.output file, like yacc's y.output file.
 # [-g, --debug]
-#   add debug code to parser class. To display debuggin information,
+#   add debug code to parser class. To display debugging information,
 #   use this '-g' option and set @yydebug true in parser class.
 # [-E, --embedded]
 #   Output parser which doesn't need runtime files (racc/parser.rb).
+# [-F, --frozen]
+#   Output parser which declares frozen_string_literals: true
 # [-C, --check-only]
 #   Check syntax of racc grammar file and quit.
 # [-S, --output-status]
@@ -260,11 +258,11 @@ module Racc
     #     def next_token
     #       @q.shift
     #     end
-    class_eval %{
+    class_eval <<~RUBY, __FILE__, __LINE__ + 1
     def do_parse
       #{Racc_Main_Parsing_Routine}(_racc_setup(), false)
     end
-    }
+    RUBY
 
     # The method to fetch next token.
     # If you use #do_parse method, you must implement #next_token.
@@ -322,11 +320,11 @@ module Racc
     #
     # RECEIVER#METHOD_ID is a method to get next token.
     # It must 'yield' the token, which format is [TOKEN-SYMBOL, VALUE].
-    class_eval %{
+    class_eval <<~RUBY, __FILE__, __LINE__ + 1
     def yyparse(recv, mid)
       #{Racc_YY_Parse_Method}(recv, mid, _racc_setup(), false)
     end
-    }
+    RUBY
 
     def _racc_yyparse_rb(recv, mid, arg, c_debug)
       action_table, action_check, action_default, action_pointer,
@@ -535,7 +533,7 @@ module Racc
     #
     # If this method returns, parsers enter "error recovering mode".
     def on_error(t, val, vstack)
-      raise ParseError, sprintf("\nparse error on value %s (%s)",
+      raise ParseError, sprintf("parse error on value %s (%s)",
                                 val.inspect, token_to_str(t) || '?')
     end
 

data/lib/racc/parserfilegenerator.rb

@@ -45,6 +45,7 @@ module Racc
       bool_attr :convert_line
       bool_attr :convert_line_all
       bool_attr :embed_runtime
+      bool_attr :frozen_strings
       bool_attr :make_executable
       attr_accessor :interpreter
 
@@ -64,6 +65,7 @@ module Racc
         self.convert_line = true
         self.convert_line_all = false
         self.embed_runtime = false
+        self.frozen_strings = false
         self.make_executable = false
         self.interpreter = nil
       end
@@ -122,6 +124,7 @@ module Racc
     end
 
     def notice
+      line %q[# frozen_string_literal: true] if @params.frozen_strings?
       line %q[#]
       line %q[# DO NOT MODIFY!!!!]
       line %Q[# This file is automatically generated by Racc #{Racc::Version}]
@@ -135,8 +138,8 @@ module Racc
 
     def embed_library(src)
       line %[###### #{src.filename} begin]
-      line %[unless $".index '#{src.filename}']
-      line %[$".push '#{src.filename}']
+      line %[unless $".find {|p| p.end_with?('/#{src.filename}')}]
+      line %[$".push "\#{__dir__}/#{src.filename}"]
       put src, @params.convert_line?
       line %[end]
       line %[###### #{src.filename} end]
@@ -310,8 +313,10 @@ module Racc
           racc_reduce_n,
           racc_use_result_var ]
       End
+      line "Ractor.make_shareable(Racc_arg) if defined?(Ractor)"
       line
       string_list 'Racc_token_to_s_table', table.token_to_s_table
+      line "Ractor.make_shareable(Racc_token_to_s_table) if defined?(Ractor)"
       line
       line "Racc_debug_parser = #{table.debug_parser}"
       line

data/lib/racc/state.rb

@@ -73,6 +73,10 @@ module Racc
           (n_srconflicts() != @grammar.n_expected_srconflicts)
     end
 
+    def should_error_on_expect_mismatch?
+      should_report_srconflict? && @grammar.error_on_expect_mismatch
+    end
+
     def srconflict_exist?
       n_srconflicts() != 0
     end

data/lib/racc/statetransitiontable.rb

@@ -216,7 +216,7 @@ module Racc
         end
         i = ii
       end
-      Regexp.compile(map, 'n')
+      Regexp.compile(map, Regexp::NOENCODING)
     end
 
     def set_table(entries, dummy, tbl, chk, ptr)