Reset internal buffers when seek() is called. This fixes SF bug #1156259.
diff --git a/Lib/codecs.py b/Lib/codecs.py
index b4103fb..092da0c 100644
--- a/Lib/codecs.py
+++ b/Lib/codecs.py
@@ -356,7 +356,17 @@
from decoding errors.
"""
- pass
+ self.bytebuffer = ""
+ self.charbuffer = u""
+ self.atcr = False
+
+ def seek(self, offset, whence):
+ """ Set the input stream's current position.
+
+ Resets the codec buffers used for keeping state.
+ """
+ self.reset()
+ self.stream.seek(offset, whence)
def next(self):
diff --git a/Lib/encodings/utf_16.py b/Lib/encodings/utf_16.py
index a33581c..95abb05 100644
--- a/Lib/encodings/utf_16.py
+++ b/Lib/encodings/utf_16.py
@@ -31,6 +31,13 @@
class StreamReader(codecs.StreamReader):
+ def reset(self):
+ codecs.StreamReader.reset(self)
+ try:
+ del self.decode
+ except AttributeError:
+ pass
+
def decode(self, input, errors='strict'):
(object, consumed, byteorder) = \
codecs.utf_16_ex_decode(input, errors, 0, False)
diff --git a/Lib/test/test_codecs.py b/Lib/test/test_codecs.py
index 01d8955..e6dba34 100644
--- a/Lib/test/test_codecs.py
+++ b/Lib/test/test_codecs.py
@@ -755,6 +755,21 @@
decodedresult += reader.read()
self.assertEqual(decodedresult, s, "%r != %r (encoding=%r)" % (decodedresult, s, encoding))
+ def test_seek(self):
+ # all codecs should be able to encode these
+ s = u"%s\n%s\n" % (100*u"abc123", 100*u"def456")
+ for encoding in all_unicode_encodings:
+ if encoding == "idna": # FIXME: See SF bug #1163178
+ continue
+ if encoding in broken_unicode_with_streams:
+ continue
+ reader = codecs.getreader(encoding)(StringIO.StringIO(s.encode(encoding)))
+ for t in xrange(5):
+ # Test that calling seek resets the internal codec state and buffers
+ reader.seek(0, 0)
+ line = reader.readline()
+ self.assertEqual(s[:len(line)], line)
+
class BasicStrTest(unittest.TestCase):
def test_basics(self):
s = "abc123"