RubyLex.read_escape doesn't discriminate whether it is lexing a ' or " string. For most escapes the existing code
just happens to work; however it always consumes at least one character after \c. This means in '\c', RubyLex misses
the end of the string and keeps lexing.

This patch makes read_escape only treat \\ and \' as escapes when lexing a ' (or %q) string.

read_escape arguably should ungetc the \ when the following character isn't \ or ' when lexing a ' or %q string, because
a single \ isn't an escape, however I haven't implemented this.

Date: 2007-06-13 15:29
Sender: Dominic Cooney

That looks fine. Sorry for so much reformatting in the
original patch. I assumed sticking a return in the middle of
the function was bad style.

Date: 2007-06-13 06:32
Sender: Ryan Davis

I've reworked the diff into the following. Does it still look
acceptable to you?

Index: lib/irb/ruby-lex.rb
=================================================================
==
--- lib/irb/ruby-lex.rb	(revision 12516)
+++ lib/irb/ruby-lex.rb	(working copy)
@@ -1099,6 +1099,16 @@
   end
   
   def read_escape
+    if @ltype == "'" then
+      case ch = getc
+      when "\\", "'"
+        # do nothing
+      else
+        ungetc ch
+      end
+      return
+    end
+
     case ch = getc
     when "\n", "\r", "\f"
     when "\\", "n", "t", "r",
"f", "v", "a", "e",
"b", "s" #"

Date: 2007-06-13 06:11
Sender: Ryan Davis

I really wish you'd only addressed the problem and not the whitespace
in this patch...