Accept escaped quotes in token aliases

lib/lrama/lexer.rb

@@ -133,7 +133,7 @@ def lex_token
         return [:CHARACTER, Lrama::Lexer::Token::Char.new(s_value: @scanner.matched, location: location)]
       when @scanner.scan(/'\\\\'|'\\b'|'\\t'|'\\f'|'\\r'|'\\n'|'\\v'|'\\13'/)
         return [:CHARACTER, Lrama::Lexer::Token::Char.new(s_value: @scanner.matched, location: location)]
-      when @scanner.scan(/".*?"/)
+      when @scanner.scan(/"(?:\\.|[^"\\\n])*"/)
         return [:STRING, Lrama::Lexer::Token::Str.new(s_value: %Q(#{@scanner.matched}), location: location)]
       when @scanner.scan(/\d+/)
         return [:INTEGER, Lrama::Lexer::Token::Int.new(s_value: Integer(@scanner.matched), location: location)]

spec/lrama/lexer_spec.rb

@@ -431,6 +431,31 @@
     MSG
   end
 
+  [%q("\""), %q("a\"b"), %q("\\\\"), %q("\n")].each do |literal|
+    it "lexes the escaped string alias #{literal.inspect}" do
+      grammar_file = Lrama::Lexer::GrammarFile.new("alias.y", "%token QUOTE #{literal} NEXT")
+      lexer = Lrama::Lexer.new(grammar_file)
+
+      expect(lexer.next_token).to eq(['%token', token_class::Token.new(s_value: '%token')])
+      expect(lexer.next_token).to eq([:IDENTIFIER, token_class::Ident.new(s_value: 'QUOTE')])
+      token = lexer.next_token
+
+      expect(token).to eq([:STRING, token_class::Str.new(s_value: literal)])
+      expect(token[1].location).to eq Lrama::Lexer::Location.new(grammar_file: grammar_file, first_line: 1, first_column: 13, last_line: 1, last_column: 13 + literal.length)
+      expect(lexer.next_token).to eq([:IDENTIFIER, token_class::Ident.new(s_value: 'NEXT')])
+      expect(lexer.next_token).to be_nil
+    end
+  end
+
+  it 'rejects unterminated or multiline string aliases' do
+    [%q("\"), "\"a\nb\"", "\"a\\\nb\""].each do |literal|
+      grammar_file = Lrama::Lexer::GrammarFile.new("alias.y", literal)
+      lexer = Lrama::Lexer.new(grammar_file)
+
+      expect { lexer.next_token }.to raise_error(ParseError, /Unexpected token/)
+    end
+  end
+
   context 'unexpected_c_code.y' do
     it do
       grammar_file = Lrama::Lexer::GrammarFile.new("invalid.y", "@invalid")

spec/lrama/parser_spec.rb

@@ -118,6 +118,21 @@
   end
 
   describe '#parse' do
+    it 'resolves an escaped string alias to its declared token' do
+      y = <<~'GRAMMAR'
+        %token QUOTE "\""
+        %%
+        start: "\"";
+      GRAMMAR
+      grammar = Lrama::Parser.new(y, "alias.y").parse
+      grammar.prepare
+      grammar.validate!
+
+      quote = grammar.find_symbol_by_s_value!("QUOTE")
+      expect(quote.alias_name).to eq(%q("\""))
+      expect(grammar.rules.last.rhs).to eq([quote])
+    end
+
     it "no_union" do
       path = "common/no_union.y"
       y = File.read(fixture_path(path))