bpo-24214: Fixed the UTF-8 incremental decoder. (GH-12603) (GH-12627)

miss-islington · serhiy-storchaka · serhiy-storchaka · commit bd48280cb665 · 2019-03-30T15:52:41.000+02:00
The bug occurred when the encoded surrogate character is passed to the incremental decoder in two chunks. (cherry picked from commit 7a465cb) Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
diff --git a/Lib/test/test_codecs.py b/Lib/test/test_codecs.py
@@ -401,6 +401,15 @@ def test_lone_surrogates(self):
             self.assertEqual(test_sequence.decode(self.encoding, "backslashreplace"),
                              before + backslashreplace + after)
 
+    def test_incremental_surrogatepass(self):
+        # Test incremental decoder for surrogatepass handler:
+        # see issue #24214
+        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')
+
 
 class UTF32Test(ReadTest, unittest.TestCase):
     encoding = "utf-32"
diff --git a/Misc/NEWS.d/next/Core and Builtins/2019-03-28-15-22-45.bpo-24214.tZ6lYU.rst b/Misc/NEWS.d/next/Core and Builtins/2019-03-28-15-22-45.bpo-24214.tZ6lYU.rst
@@ -0,0 +1,2 @@
+Fixed support of the surrogatepass error handler in the UTF-8 incremental
+decoder.
diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
@@ -4890,6 +4890,9 @@ PyUnicode_DecodeUTF8Stateful(const char *s,
         case 2:
         case 3:
         case 4:
+            if (s == end || consumed) {
+                goto End;
+            }
             errmsg = "invalid continuation byte";
             startinpos = s - starts;
             endinpos = startinpos + ch - 1;

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,2 @@`
	`1`	`+Fixed support of the surrogatepass error handler in the UTF-8 incremental`
	`2`	`+decoder.`