lrama 0.5.11 → 0.6.0

data/lib/lrama/grammar.rb

@@ -8,6 +8,10 @@ require "lrama/grammar/printer"
 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 @@ module Lrama
       @rule_builders = []
       @rules = []
       @sym_to_rules = {}
+      @parameterizing_resolver = ParameterizingRuleResolver.new
       @empty_symbol = nil
       @eof_symbol = nil
       @error_symbol = nil
@@ -69,7 +74,7 @@ module Lrama
         return sym
       end
 
-      if sym = @symbols.find {|s| s.id == id }
+      if (sym = @symbols.find {|s| s.id == id })
         return sym
       end
 
@@ -129,6 +134,10 @@ module Lrama
       @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 @@ module Lrama
 
     def setup_rules
       @rule_builders.each do |builder|
-        builder.setup_rules
+        builder.setup_rules(@parameterizing_resolver)
       end
     end
 
@@ -350,56 +359,21 @@ module Lrama
       @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
-      accept = @accept_symbol
-      eof = @eof_symbol
+      # Add $accept rule to the top of rules
       lineno = @rule_builders.first ? @rule_builders.first.line : 0
-      @rules << Rule.new(id: @rule_counter.increment, _lhs: accept.id, _rhs: [@rule_builders.first.lhs, eof.id], token_code: nil, lineno: lineno)
+      @rules << Rule.new(id: @rule_counter.increment, _lhs: @accept_symbol.id, _rhs: [@rule_builders.first.lhs, @eof_symbol.id], token_code: nil, lineno: lineno)
 
       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)
     end
 
     # Collect symbols from rules

data/lib/lrama/lexer/grammar_file.rb

@@ -0,0 +1,21 @@
+module Lrama
+  class Lexer
+    class GrammarFile
+      attr_reader :path, :text
+
+      def initialize(path, text)
+        @path = path
+        @text = text
+      end
+
+      def ==(other)
+        self.class == other.class &&
+        self.path == other.path
+      end
+
+      def lines
+        @lines ||= text.split("\n")
+      end
+    end
+  end
+end

data/lib/lrama/lexer/location.rb

@@ -1,9 +1,10 @@
 module Lrama
   class Lexer
     class Location
-      attr_reader :first_line, :first_column, :last_line, :last_column
+      attr_reader :grammar_file, :first_line, :first_column, :last_line, :last_column
 
-      def initialize(first_line:, first_column:, last_line:, last_column:)
+      def initialize(grammar_file:, first_line:, first_column:, last_line:, last_column:)
+        @grammar_file = grammar_file
         @first_line = first_line
         @first_column = first_column
         @last_line = last_line
@@ -12,11 +13,85 @@ module Lrama
 
       def ==(other)
         self.class == other.class &&
+        self.grammar_file == other.grammar_file &&
         self.first_line == other.first_line &&
         self.first_column == other.first_column &&
         self.last_line == other.last_line &&
         self.last_column == other.last_column
       end
+
+      def partial_location(left, right)
+        offset = -first_column
+        new_first_line = -1
+        new_first_column = -1
+        new_last_line = -1
+        new_last_column = -1
+
+        _text.each.with_index do |line, index|
+          new_offset = offset + line.length + 1
+
+          if offset <= left && left <= new_offset
+            new_first_line = first_line + index
+            new_first_column = left - offset
+          end
+
+          if offset <= right && right <= new_offset
+            new_last_line = first_line + index
+            new_last_column = right - offset
+          end
+
+          offset = new_offset
+        end
+
+        Location.new(
+          grammar_file: grammar_file,
+          first_line: new_first_line, first_column: new_first_column,
+          last_line: new_last_line, last_column: new_last_column
+        )
+      end
+
+      def to_s
+        "#{path} (#{first_line},#{first_column})-(#{last_line},#{last_column})"
+      end
+
+      def generate_error_message(error_message)
+        <<~ERROR.chomp
+          #{path}:#{first_line}:#{first_column}: #{error_message}
+          #{line_with_carets}
+        ERROR
+      end
+
+      def line_with_carets
+        <<~TEXT
+          #{text}
+          #{carets}
+        TEXT
+      end
+
+      private
+
+      def path
+        grammar_file.path
+      end
+
+      def blanks
+        (text[0...first_column] or raise "#{first_column} is invalid").gsub(/[^\t]/, ' ')
+      end
+
+      def carets
+        blanks + '^' * (last_column - first_column)
+      end
+
+      def text
+        @text ||= _text.join("\n")
+      end
+
+      def _text
+        @_text ||=begin
+          range = (first_line - 1)...last_line
+          grammar_file.lines[range] or raise "#{range} is invalid"
+        end
+      end
     end
   end
 end

data/lib/lrama/lexer/token/instantiate_rule.rb

@@ -0,0 +1,18 @@
+module Lrama
+  class Lexer
+    class Token
+      class InstantiateRule < Token
+        attr_accessor :args
+
+        def initialize(s_value:, alias_name: nil, location: nil, args: [])
+          super s_value: s_value, alias_name: alias_name, location: location
+          @args = args
+        end
+
+        def rule_name
+          s_value
+        end
+      end
+    end
+  end
+end

data/lib/lrama/lexer/token/user_code.rb

@@ -35,27 +35,27 @@ module Lrama
           # It need to wrap an identifier with brackets to use ".-" for identifiers
           when scanner.scan(/\$(<[a-zA-Z0-9_]+>)?\$/) # $$, $<long>$
             tag = scanner[1] ? Lrama::Lexer::Token::Tag.new(s_value: scanner[1]) : nil
-            return Lrama::Grammar::Reference.new(type: :dollar, name: "$", ex_tag: tag, first_column: start, last_column: scanner.pos - 1)
+            return Lrama::Grammar::Reference.new(type: :dollar, name: "$", ex_tag: tag, first_column: start, last_column: scanner.pos)
           when scanner.scan(/\$(<[a-zA-Z0-9_]+>)?(\d+)/) # $1, $2, $<long>1
             tag = scanner[1] ? Lrama::Lexer::Token::Tag.new(s_value: scanner[1]) : nil
-            return Lrama::Grammar::Reference.new(type: :dollar, index: Integer(scanner[2]), ex_tag: tag, first_column: start, last_column: scanner.pos - 1)
+            return Lrama::Grammar::Reference.new(type: :dollar, index: Integer(scanner[2]), ex_tag: tag, first_column: start, last_column: scanner.pos)
           when scanner.scan(/\$(<[a-zA-Z0-9_]+>)?([a-zA-Z_][a-zA-Z0-9_]*)/) # $foo, $expr, $<long>program (named reference without brackets)
             tag = scanner[1] ? Lrama::Lexer::Token::Tag.new(s_value: scanner[1]) : nil
-            return Lrama::Grammar::Reference.new(type: :dollar, name: scanner[2], ex_tag: tag, first_column: start, last_column: scanner.pos - 1)
-          when scanner.scan(/\$(<[a-zA-Z0-9_]+>)?\[([a-zA-Z_.][-a-zA-Z0-9_.]*)\]/) # $expr.right, $expr-right, $<long>program (named reference with brackets)
+            return Lrama::Grammar::Reference.new(type: :dollar, name: scanner[2], ex_tag: tag, first_column: start, last_column: scanner.pos)
+          when scanner.scan(/\$(<[a-zA-Z0-9_]+>)?\[([a-zA-Z_.][-a-zA-Z0-9_.]*)\]/) # $[expr.right], $[expr-right], $<long>[expr.right] (named reference with brackets)
             tag = scanner[1] ? Lrama::Lexer::Token::Tag.new(s_value: scanner[1]) : nil
-            return Lrama::Grammar::Reference.new(type: :dollar, name: scanner[2], ex_tag: tag, first_column: start, last_column: scanner.pos - 1)
+            return Lrama::Grammar::Reference.new(type: :dollar, name: scanner[2], ex_tag: tag, first_column: start, last_column: scanner.pos)
 
           # @ references
           # It need to wrap an identifier with brackets to use ".-" for identifiers
           when scanner.scan(/@\$/) # @$
-            return Lrama::Grammar::Reference.new(type: :at, name: "$", first_column: start, last_column: scanner.pos - 1)
+            return Lrama::Grammar::Reference.new(type: :at, name: "$", first_column: start, last_column: scanner.pos)
           when scanner.scan(/@(\d+)/) # @1
-            return Lrama::Grammar::Reference.new(type: :at, index: Integer(scanner[1]), first_column: start, last_column: scanner.pos - 1)
+            return Lrama::Grammar::Reference.new(type: :at, index: Integer(scanner[1]), first_column: start, last_column: scanner.pos)
           when scanner.scan(/@([a-zA-Z][a-zA-Z0-9_]*)/) # @foo, @expr (named reference without brackets)
-            return Lrama::Grammar::Reference.new(type: :at, name: scanner[1], first_column: start, last_column: scanner.pos - 1)
-          when scanner.scan(/@\[([a-zA-Z_.][-a-zA-Z0-9_.]*)\]/) # @expr.right, @expr-right  (named reference with brackets)
-            return Lrama::Grammar::Reference.new(type: :at, name: scanner[1], first_column: start, last_column: scanner.pos - 1)
+            return Lrama::Grammar::Reference.new(type: :at, name: scanner[1], first_column: start, last_column: scanner.pos)
+          when scanner.scan(/@\[([a-zA-Z_.][-a-zA-Z0-9_.]*)\]/) # @[expr.right], @[expr-right]  (named reference with brackets)
+            return Lrama::Grammar::Reference.new(type: :at, name: scanner[1], first_column: start, last_column: scanner.pos)
           end
         end
       end

data/lib/lrama/lexer/token.rb

@@ -1,6 +1,6 @@
 require 'lrama/lexer/token/char'
 require 'lrama/lexer/token/ident'
-require 'lrama/lexer/token/parameterizing'
+require 'lrama/lexer/token/instantiate_rule'
 require 'lrama/lexer/token/tag'
 require 'lrama/lexer/token/user_code'
 

data/lib/lrama/lexer.rb

@@ -1,4 +1,5 @@
 require "strscan"
+require "lrama/lexer/grammar_file"
 require "lrama/lexer/location"
 require "lrama/lexer/token"
 
@@ -28,10 +29,12 @@ module Lrama
       %error-token
       %empty
       %code
+      %rule
     )
 
-    def initialize(text)
-      @scanner = StringScanner.new(text)
+    def initialize(grammar_file)
+      @grammar_file = grammar_file
+      @scanner = StringScanner.new(grammar_file.text)
       @head_column = @head = @scanner.pos
       @head_line = @line = 1
       @status = :initial
@@ -57,8 +60,9 @@ module Lrama
 
     def location
       Location.new(
+        grammar_file: @grammar_file,
         first_line: @head_line, first_column: @head_column,
-        last_line: @line, last_column: column
+        last_line: line, last_column: column
       )
     end
 
@@ -78,8 +82,7 @@ module Lrama
         end
       end
 
-      @head_line = line
-      @head_column = column
+      reset_first_position
 
       case
       when @scanner.eos?
@@ -117,6 +120,8 @@ module Lrama
     def lex_c_code
       nested = 0
       code = ''
+      reset_first_position
+
       while !@scanner.eos? do
         case
         when @scanner.scan(/{/)
@@ -140,12 +145,12 @@ module Lrama
           @line += @scanner.matched.count("\n")
         when @scanner.scan(/'.*?'/)
           code += %Q(#{@scanner.matched})
+        when @scanner.scan(/[^\"'\{\}\n]+/)
+          code += @scanner.matched
+        when @scanner.scan(/#{Regexp.escape(@end_symbol)}/)
+          code += @scanner.matched
         else
-          if @scanner.scan(/[^\"'\{\}\n#{@end_symbol}]+/)
-            code += @scanner.matched
-          else
-            code += @scanner.getch
-          end
+          code += @scanner.getch
         end
       end
       raise ParseError, "Unexpected code: #{code}."
@@ -166,9 +171,14 @@ module Lrama
       end
     end
 
+    def reset_first_position
+      @head_line = line
+      @head_column = column
+    end
+
     def newline
       @line += 1
-      @head = @scanner.pos + 1
+      @head = @scanner.pos
     end
   end
 end