Ticket #9646: fix_for_slicing_by_regexp.20071015.diff

activesupport/test/multibyte_handler_test.rb

@@ -160 +160 @@
-    assert_equal "d Блå ffi", @handler.slice(@string, 3, 7), "Unicode characters have to be returned"
-    assert_equal "A", @handler.slice(@string, 0, 1), "Slicing from an offset should return characters"
-    assert_equal " Блå ffi ", @handler.slice(@string, 4..10), "Unicode characters have to be returned"
+    assert_equal "ffi бла", @handler.slice(@string, /ffi бла/u), "Slicing on Regexps should be supported"
+    assert_equal "ffi бла", @handler.slice(@string, /ffi \w\wа/u), "Slicing on Regexps should be supported"
+    assert_equal nil, @handler.slice(@string, /unknown/u), "Slicing on Regexps with no match should return nil"
+    assert_equal "ffi бла", @handler.slice(@string, /(ffi бла)/u,1), "Slicing on Regexps with a match group should be supported"
+    assert_equal nil, @handler.slice(@string, /(ffi)/u,2), "Slicing with a Regexp and asking for an invalid match group should return nil"
-    assert_equal "", @handler.slice(@string, 7..6), "Range is empty, should return an empty string"
-    assert_raise(ActiveSupport::Multibyte::Handlers::EncodingError) { @handler.slice(@bytestring, 2..3) }
-    assert_raise(TypeError, "With 2 args, should raise TypeError for non-Numeric or Regexp first argument") { @handler.slice(@string, 2..3, 1) }

activesupport/lib/active_support/multibyte/handlers/utf8_handler.rb

@@ -255 +255 @@
-        elsif args[0].kind_of? Range
-          cps = u_unpack(str).slice(*args)
-          cps.nil? ? nil : cps.pack('U*')
+        elsif args[0].kind_of? Regexp
+          str.slice(*args)
-        elsif args.size == 1 && args[0].kind_of?(Numeric)
-          u_unpack(str)[args[0]]
-        else