- Fixed dsym unescaping. (mvz)

[git-p4: depot-paths = "//src/ruby_parser/dev/": change = 13581]

lib/ruby_lexer.rb

@@ -232,7 +232,7 @@ def possibly_escape_string text, check
     content = match[1]
 
     if text =~ check then
-      content.gsub(ESC) { unescape $1 }
+      unescape_string content
     else
       content.gsub(/\\\\/, "\\").gsub(/\\\'/, "'")
     end
@@ -590,9 +590,7 @@ def process_simple_string text
     orig_line = lineno
     self.lineno += text.count("\n")
 
-    str = text[1..-2]
-      .gsub(ESC) { unescape($1).b.force_encoding Encoding::UTF_8 }
-    str = str.b unless str.valid_encoding?
+    str = unescape_string text[1..-2]
 
     result EXPR_END, :tSTRING, str, orig_line
   end
@@ -817,6 +815,15 @@ def space_vs_beginning space_type, beg_type, fallback
     end
   end
 
+  def unescape_string str
+    str = str.gsub(ESC) { unescape($1).b.force_encoding Encoding::UTF_8 }
+    if str.valid_encoding?
+      str
+    else
+      str.b
+    end
+  end
+
   def unescape s
     r = ESCAPES[s]
 

test/test_ruby_lexer.rb

@@ -3009,6 +3009,19 @@ def test_yylex_symbol_double_interp
                 :tSTRING_END,     "\"",     EXPR_LIT)
   end
 
+  def test_yylex_symbol_double_escape_octal
+    setup_lexer ":\"Variet\\303\\240\""
+
+    adv = @lex.next_token
+    act_token, act_value = adv
+    act_value = act_value.first
+
+    assert_equal :tSYMBOL, act_token
+    assert_match EXPR_LIT, @lex.lex_state
+    # Force comparison of encodings
+    assert_equal "Varietà", act_value
+  end
+
   def test_yylex_symbol_single
     assert_lex3(":'symbol'",
                 nil,

test/test_ruby_parser.rb

@@ -826,6 +826,13 @@ def test_dsym_to_sym
     assert_parse rb, pt
   end
 
+  def test_dsym_esc_to_sym
+    rb = ':"Variet\303\240"'
+    pt = s(:lit, :Varietà)
+
+    assert_parse rb, pt
+  end
+
   def test_empty
     refute_parse ""
   end