# HG changeset patch
# User Paul Boddie <paul@boddie.org.uk>
# Date 1481665708 -3600
# Node ID 87ba8cc4108297dc58f0595d20f212293ba1728c
# Parent  0be168e888a9d4d66b363354585afd5e222182ba
Make sure that Unicode objects result from text-only concatenation.

diff -r 0be168e888a9 -r 87ba8cc41082 lib/__builtins__/unicode.py
--- a/lib/__builtins__/unicode.py	Tue Dec 13 22:46:52 2016 +0100
+++ b/lib/__builtins__/unicode.py	Tue Dec 13 22:48:28 2016 +0100
@@ -77,6 +77,29 @@
         else:
             return op(other.__data__, self.__data__)
 
+    def _convert(self, result, other):
+
+        "Convert 'result' to a Unicode object if 'other' already is."
+
+        if _isinstance(other, utf8string):
+            return utf8string(result, self.encoding)
+        else:
+            return result
+
+    def __iadd__(self, other):
+
+        "Return a string combining this string with 'other'."
+
+        return self._convert(self._binary_op(str_add, other), other)
+
+    __add__ = __iadd__
+
+    def __radd__(self, other):
+
+        "Return a string combining this string with 'other'."
+
+        return self._convert(self._binary_op_rev(str_add, other), other)
+
     def encode(self, encoding=None):
 
         """
diff -r 0be168e888a9 -r 87ba8cc41082 tests/unicode.py
--- a/tests/unicode.py	Tue Dec 13 22:46:52 2016 +0100
+++ b/tests/unicode.py	Tue Dec 13 22:48:28 2016 +0100
@@ -14,35 +14,56 @@
 u = unicode("æøå", "ISO-8859-1")
 print u                             # æøå
 print u.encode("ISO-8859-1")        # æøå
+print u.encoding                    # ISO-8859-1
 
 # Explicitly from Unicode literals.
 
 u2 = u"æøå"
 print u2                            # æøå
 print u2.encode("ISO-8859-1")       # æøå
+print u2.encoding                   # ISO-8859-1
 
 # Implicitly from string literals.
 
 #u3 = "æøå"
 #print u3                            # æøå
 #print u3.encode("ISO-8859-1")       # æøå
+#print u3.encoding                   # ISO-8859-1
 
 # Combine bytes and text.
 # The text should be decoded.
 
 su = s + u
 print su                            # ÆØÅæøå
+print su.__class__                  # __builtins__.str.string
 
 # Combine text and bytes.
 # The text should be decoded.
 
 us = u + s
 print us                            # æøåÆØÅ
+print us.__class__                  # __builtins__.str.string
+
+# Combine text and text.
+
+uu2 = u + u2
+print uu2                           # æøå
+print uu2.__class__                 # __builtins__.unicode.utf8string
+print uu2.encoding                  # ISO-8859-1
 
 # Inspect and update the encoding of stdout.
+# Note that su and us are byte strings and are not recoded.
 
 print sys.stdout.encoding           # None
+
 sys.stdout.encoding = "ISO-8859-1"
+print sys.stdout.encoding           # ISO-8859-1
 print u                             # æøå
 print su                            # ÆØÅæøå
 print us                            # æøåÆØÅ
+
+sys.stdout.encoding = "UTF-8"
+print sys.stdout.encoding           # UTF-8
+print u                             # æøå
+print su                            # ÆØÅæøå
+print us                            # æøåÆØÅ