Restore Python 2.1 StringIO.py behaviour: support concatenating

Unicode string snippets to larger Unicode strings.

This fix should also go into Python 2.2.1.

Lib/StringIO.py

@@ -28,7 +28,7 @@ Notes:
   bytes that occupy space in the buffer.
 - There's a simple test set (see end of this file).
 """
-
+import types
 try:
     from errno import EINVAL
 except ImportError:
@@ -38,8 +38,10 @@ __all__ = ["StringIO"]
 
 class StringIO:
     def __init__(self, buf = ''):
-        # Force self.buf to be a string
+        # Force self.buf to be a string or unicode
-        self.buf = str(buf)
+        if type(buf) is not types.UnicodeType:
+            buf = str(buf)
+        self.buf = buf
         self.len = len(buf)
         self.buflist = []
         self.pos = 0
@@ -135,7 +137,8 @@ class StringIO:
         if self.closed:
             raise ValueError, "I/O operation on closed file"
         if not s: return
-        # Force s to be a string
+        # Force s to be a string or unicode
+        if type(s) is not types.UnicodeType:
             s = str(s)
         if self.pos > self.len:
             self.buflist.append('\0'*(self.pos - self.len))

Lib/test/test_StringIO.py

@@ -71,6 +71,21 @@ class TestGenericStringIO(unittest.TestCase):
 class TestStringIO(TestGenericStringIO):
     MODULE = StringIO
 
+    def test_unicode(self):
+
+        # The StringIO module also supports concatenating Unicode
+        # snippets to larger Unicode strings. This is tested by this
+        # method. Note that cStringIO does not support this extension.
+        
+        f = self.MODULE.StringIO()
+        f.write(self._line[:6])
+        f.seek(3)
+        f.write(unicode(self._line[20:26]))
+        f.write(unicode(self._line[52]))
+        s = f.getvalue()
+        self.assertEqual(s, unicode('abcuvwxyz!'))
+        self.assertEqual(type(s), types.UnicodeType)
+
 class TestcStringIO(TestGenericStringIO):
     MODULE = cStringIO