Issue #28541: Improve test coverage for encoding detection in json library.

serhiy-storchaka · serhiy-storchaka · commit c4a35daa9771 · 2016-10-30T23:00:01.000+02:00
Original patch by Eric Appelt.
diff --git a/Lib/json/__init__.py b/Lib/json/__init__.py
@@ -257,7 +257,8 @@ def detect_encoding(b):
             return 'utf-16-be' if b[1] else 'utf-32-be'
         if not b[1]:
             # XX 00 00 00 - utf-32-le
-            # XX 00 XX XX - utf-16-le
+            # XX 00 00 XX - utf-16-le
+            # XX 00 XX -- - utf-16-le
             return 'utf-16-le' if b[2] or b[3] else 'utf-32-le'
     elif len(b) == 2:
         if not b[0]:
diff --git a/Lib/test/test_json/test_unicode.py b/Lib/test/test_json/test_unicode.py
@@ -65,6 +65,19 @@ def test_bytes_decode(self):
             self.assertEqual(self.loads(bom + encoded), data)
             self.assertEqual(self.loads(encoded), data)
         self.assertRaises(UnicodeDecodeError, self.loads, b'["\x80"]')
+        # RFC-7159 and ECMA-404 extend JSON to allow documents that
+        # consist of only a string, which can present a special case
+        # not covered by the encoding detection patterns specified in
+        # RFC-4627 for utf-16-le (XX 00 XX 00).
+        self.assertEqual(self.loads('"\u2600"'.encode('utf-16-le')),
+                         '\u2600')
+        # Encoding detection for small (<4) bytes objects
+        # is implemented as a special case. RFC-7159 and ECMA-404
+        # allow single codepoint JSON documents which are only two
+        # bytes in utf-16 encodings w/o BOM.
+        self.assertEqual(self.loads(b'5\x00'), 5)
+        self.assertEqual(self.loads(b'\x007'), 7)
+        self.assertEqual(self.loads(b'57'), 57)
 
     def test_object_pairs_hook_with_unicode(self):
         s = '{"xkd":1, "kcw":2, "art":3, "hxm":4, "qrt":5, "pad":6, "hoy":7}'
