Apply named references

ruby · Sep 26, 2023 · 8b2526e · 8b2526e
1 parent 3972ed8
commit 8b2526e
Show file tree

Hide file tree

Showing 6 changed files with 274 additions and 241 deletions.
diff --git a/lib/lrama/command.rb b/lib/lrama/command.rb
@@ -9,8 +9,6 @@ def run(argv)
       text = options.y.read
       options.y.close if options.y != STDIN
       grammar = Lrama::NewParser.new(text).parse
-      pp Lrama::Parser.new(text).parse
-      pp grammar
       states = Lrama::States.new(grammar, warning, trace_state: (options.trace_opts[:automaton] || options.trace_opts[:closure]))
       states.compute
       context = Lrama::Context.new(states)

diff --git a/lib/lrama/grammar.rb b/lib/lrama/grammar.rb
@@ -309,27 +309,39 @@ def nterms
     def extract_references
       @_rules.each do |lhs, rhs, _|
         rhs.each_with_index do |token, index|
-          next if token.class == Lrama::Grammar::Symbol || token.type != Lrama::Lexer::Token::User_code
+        next if token.class == Lrama::Grammar::Symbol || token.type != Lrama::Lexer::Token::User_code
 
           scanner = StringScanner.new(token.s_value)
           references = []
 
           while !scanner.eos? do
             start = scanner.pos
             case
+            # $ references
+            # 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.new(type: Lrama::Lexer::Token::Tag, s_value: scanner[1]) : nil
               references << [:dollar, "$", tag, start, scanner.pos - 1]
             when scanner.scan(/\$(<[a-zA-Z0-9_]+>)?(\d+)/) # $1, $2, $<long>1
               tag = scanner[1] ? Lrama::Lexer::Token.new(type: Lrama::Lexer::Token::Tag, s_value: scanner[1]) : nil
               references << [:dollar, Integer(scanner[2]), tag, start, scanner.pos - 1]
-            when scanner.scan(/\$(<[a-zA-Z0-9_]+>)?([a-zA-Z_.][-a-zA-Z0-9_.]*)/) # $foo, $expr, $<long>program
+            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.new(type: Lrama::Lexer::Token::Tag, s_value: scanner[1]) : nil
               references << [:dollar, scanner[2], tag, start, scanner.pos - 1]
+            when scanner.scan(/\$(<[a-zA-Z0-9_]+>)?\[([a-zA-Z_.][-a-zA-Z0-9_.]*)\]/) # $expr.right, $expr-right, $<long>program (named reference with brackets)
+              tag = scanner[1] ? Lrama::Lexer::Token.new(type: Lrama::Lexer::Token::Tag, s_value: scanner[1]) : nil
+              references << [:dollar, scanner[2], tag, start, scanner.pos - 1]
+
+            # @ references
+            # It need to wrap an identifier with brackets to use ".-" for identifiers
             when scanner.scan(/@\$/) # @$
               references << [:at, "$", nil, start, scanner.pos - 1]
-            when scanner.scan(/@(\d)+/) # @1
+            when scanner.scan(/@(\d+)/) # @1
               references << [:at, Integer(scanner[1]), nil, start, scanner.pos - 1]
+            when scanner.scan(/@([a-zA-Z][a-zA-Z0-9_]*)/) # @foo, @expr (named reference without brackets)
+              references << [:at, scanner[1], nil, start, scanner.pos - 1]
+            when scanner.scan(/@\[([a-zA-Z_.][-a-zA-Z0-9_.]*)\]/) # @expr.right, @expr-right  (named reference with brackets)
+              references << [:at, scanner[1], nil, start, scanner.pos - 1]
             else
               scanner.getch
             end

diff --git a/lib/lrama/new_lexer.rb b/lib/lrama/new_lexer.rb
@@ -16,7 +16,6 @@ def initialize(text)
     def next_token
       case @status
       when :initial
-        pp @line
         lex_token
       when :c_declaration
         lex_c_code
@@ -51,6 +50,10 @@ def lex_token
           return [@scanner.matched, @scanner.matched]
         when @scanner.scan(/}/)
           return [@scanner.matched, @scanner.matched]
+        when @scanner.scan(/\[/)
+          return [@scanner.matched, @scanner.matched]
+        when @scanner.scan(/\]/)
+          return [@scanner.matched, @scanner.matched]
         when @scanner.scan(/:/)
           return [@scanner.matched, @scanner.matched]
         when @scanner.scan(/\|/)