Backport checkin:

Fix and test for SF bug #1098990: codec readline() splits lines apart.

diff --git a/Lib/codecs.py b/Lib/codecs.py
index f6d480cc9826102f791cc00fe7add15bdbcc30ec..b283925e01fa2a608420f8922739f23472530c55 100644 (file)
--- a/Lib/codecs.py
+++ b/Lib/codecs.py
@@ -322,9 +322,9 @@ class StreamReader(Codec):
                         line = line0withend
                     else:
                         line = line0withoutend
-                break
+                    break
             # we didn't get anything or this was our only try
-            elif not data or size is not None:
+            if not data or size is not None:
                 if line and not keepends:
                     line = line.splitlines(False)[0]
                 break

diff --git a/Lib/test/test_codecs.py b/Lib/test/test_codecs.py
index 36c40401770b199e4ec9befbb5686d7198f3656c..da6891f85d8ba3788eaff6cb6ac266e6adf09e01 100644 (file)
--- a/Lib/test/test_codecs.py
+++ b/Lib/test/test_codecs.py
@@ -108,6 +108,36 @@ class ReadTest(unittest.TestCase):
         writer.write(u"foo\r\n")
         self.assertEqual(reader.readline(keepends=True), u"foo\r\n")
 
+    def test_bug1098990_a(self):
+        s1 = u"xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy\r\n"
+        s2 = u"offending line: ladfj askldfj klasdj fskla dfzaskdj fasklfj laskd fjasklfzzzzaa%whereisthis!!!\r\n"
+        s3 = u"next line.\r\n"
+
+        s = (s1+s2+s3).encode(self.encoding)
+        stream = StringIO.StringIO(s)
+        reader = codecs.getreader(self.encoding)(stream)
+        self.assertEqual(reader.readline(), s1)
+        self.assertEqual(reader.readline(), s2)
+        self.assertEqual(reader.readline(), s3)
+        self.assertEqual(reader.readline(), u"")
+
+    def test_bug1098990_b(self):
+        s1 = u"aaaaaaaaaaaaaaaaaaaaaaaa\r\n"
+        s2 = u"bbbbbbbbbbbbbbbbbbbbbbbb\r\n"
+        s3 = u"stillokay:bbbbxx\r\n"
+        s4 = u"broken!!!!badbad\r\n"
+        s5 = u"againokay.\r\n"
+
+        s = (s1+s2+s3+s4+s5).encode(self.encoding)
+        stream = StringIO.StringIO(s)
+        reader = codecs.getreader(self.encoding)(stream)
+        self.assertEqual(reader.readline(), s1)
+        self.assertEqual(reader.readline(), s2)
+        self.assertEqual(reader.readline(), s3)
+        self.assertEqual(reader.readline(), s4)
+        self.assertEqual(reader.readline(), s5)
+        self.assertEqual(reader.readline(), u"")
+
 class UTF16Test(ReadTest):
     encoding = "utf-16"