bpo-24214: Fixed the UTF-8 and UTF-16 incremental decoders. (GH-14304)

* The UTF-8 incremental decoders fails now fast if encounter
  a sequence that can't be handled by the error handler.
* The UTF-16 incremental decoders with the surrogatepass error
  handler decodes now a lone low surrogate with final=False.
This commit is contained in:
Serhiy Storchaka 2019-06-25 11:54:18 +03:00 committed by GitHub
parent 9fe42b49c7
commit 894263ba80
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 36 additions and 6 deletions

View File

@ -429,11 +429,18 @@ class ReadTest(MixInCheckStateHandling):
def test_incremental_surrogatepass(self):
# Test incremental decoder for surrogatepass handler:
# see issue #24214
# High surrogate
data = '\uD901'.encode(self.encoding, 'surrogatepass')
for i in range(1, len(data)):
dec = codecs.getincrementaldecoder(self.encoding)('surrogatepass')
self.assertEqual(dec.decode(data[:i]), '')
self.assertEqual(dec.decode(data[i:], True), '\uD901')
# Low surrogate
data = '\uDC02'.encode(self.encoding, 'surrogatepass')
for i in range(1, len(data)):
dec = codecs.getincrementaldecoder(self.encoding)('surrogatepass')
self.assertEqual(dec.decode(data[:i]), '')
self.assertEqual(dec.decode(data[i:]), '\uDC02')
class UTF32Test(ReadTest, unittest.TestCase):
@ -874,6 +881,23 @@ class UTF8Test(ReadTest, unittest.TestCase):
with self.assertRaises(UnicodeDecodeError):
b"abc\xed\xa0z".decode(self.encoding, "surrogatepass")
def test_incremental_errors(self):
# Test that the incremental decoder can fail with final=False.
# See issue #24214
cases = [b'\x80', b'\xBF', b'\xC0', b'\xC1', b'\xF5', b'\xF6', b'\xFF']
for prefix in (b'\xC2', b'\xDF', b'\xE0', b'\xE0\xA0', b'\xEF',
b'\xEF\xBF', b'\xF0', b'\xF0\x90', b'\xF0\x90\x80',
b'\xF4', b'\xF4\x8F', b'\xF4\x8F\xBF'):
for suffix in b'\x7F', b'\xC0':
cases.append(prefix + suffix)
cases.extend((b'\xE0\x80', b'\xE0\x9F', b'\xED\xA0\x80',
b'\xED\xBF\xBF', b'\xF0\x80', b'\xF0\x8F', b'\xF4\x90'))
for data in cases:
with self.subTest(data=data):
dec = codecs.getincrementaldecoder(self.encoding)()
self.assertRaises(UnicodeDecodeError, dec.decode, data)
class UTF7Test(ReadTest, unittest.TestCase):
encoding = "utf-7"

View File

@ -0,0 +1,2 @@
Improved support of the surrogatepass error handler in the UTF-8 and UTF-16
incremental decoders.

View File

@ -207,7 +207,7 @@ STRINGLIB(utf8_decode)(const char **inptr, const char *end,
goto InvalidContinuation1;
} else if (ch == 0xF4 && ch2 >= 0x90) {
/* invalid sequence
\xF4\x90\x80\80- -- 110000- overflow */
\xF4\x90\x80\x80- -- 110000- overflow */
goto InvalidContinuation1;
}
if (!IS_CONTINUATION_BYTE(ch3)) {
@ -573,10 +573,10 @@ STRINGLIB(utf16_decode)(const unsigned char **inptr, const unsigned char *e,
}
/* UTF-16 code pair: */
if (q >= e)
goto UnexpectedEnd;
if (!Py_UNICODE_IS_HIGH_SURROGATE(ch))
goto IllegalEncoding;
if (q >= e)
goto UnexpectedEnd;
ch2 = (q[ihi] << 8) | q[ilo];
q += 2;
if (!Py_UNICODE_IS_LOW_SURROGATE(ch2))

View File

@ -4945,11 +4945,15 @@ unicode_decode_utf8(const char *s, Py_ssize_t size,
endinpos = startinpos + 1;
break;
case 2:
case 3:
case 4:
if (s == end || consumed) {
if (consumed && (unsigned char)s[0] == 0xED && end - s == 2
&& (unsigned char)s[1] >= 0xA0 && (unsigned char)s[1] <= 0xBF)
{
/* Truncated surrogate code in range D800-DFFF */
goto End;
}
/* fall through */
case 3:
case 4:
errmsg = "invalid continuation byte";
startinpos = s - starts;
endinpos = startinpos + ch - 1;