Issue python#17909: Accept binary input in json.loads

ncoghlan · ncoghlan · commit b161562f72a2 · 2016-09-10T20:16:18.000+10:00
json.loads (and hence json.load) now support binary input
encoded as UTF-8, UTF-16 or UTF-32.

Patch by Serhiy Storchaka.
diff --git a/Doc/library/json.rst b/Doc/library/json.rst
@@ -268,8 +268,9 @@ Basic Usage
 
 .. function:: loads(s, *, encoding=None, cls=None, object_hook=None, parse_float=None, parse_int=None, parse_constant=None, object_pairs_hook=None, **kw)
 
-   Deserialize *s* (a :class:`str` instance containing a JSON document) to a
-   Python object using this :ref:`conversion table <json-to-py-table>`.
+   Deserialize *s* (a :class:`str`, :class:`bytes` or :class:`bytearray`
+   instance containing a JSON document) to a Python object using this
+   :ref:`conversion table <json-to-py-table>`.
 
    The other arguments have the same meaning as in :func:`load`, except
    *encoding* which is ignored and deprecated.
diff --git a/Doc/whatsnew/3.6.rst b/Doc/whatsnew/3.6.rst
@@ -680,6 +680,14 @@ restriction that :class:`importlib.machinery.BuiltinImporter` and
 :term:`path-like object`.
 
 
+json
+----
+
+:func:`json.load` and :func:`json.loads` now support binary input.  Encoded
+JSON should be represented using either UTF-8, UTF-16, or UTF-32.
+(Contributed by Serhiy Storchaka in :issue:`17909`.)
+
+
 os
 --
 
diff --git a/Lib/json/__init__.py b/Lib/json/__init__.py
@@ -105,6 +105,7 @@
 
 from .decoder import JSONDecoder, JSONDecodeError
 from .encoder import JSONEncoder
+import codecs
 
 _default_encoder = JSONEncoder(
     skipkeys=False,
@@ -240,6 +241,35 @@ def dumps(obj, *, skipkeys=False, ensure_ascii=True, check_circular=True,
 _default_decoder = JSONDecoder(object_hook=None, object_pairs_hook=None)
 
 
+def detect_encoding(b):
+    bstartswith = b.startswith
+    if bstartswith((codecs.BOM_UTF32_BE, codecs.BOM_UTF32_LE)):
+        return 'utf-32'
+    if bstartswith((codecs.BOM_UTF16_BE, codecs.BOM_UTF16_LE)):
+        return 'utf-16'
+    if bstartswith(codecs.BOM_UTF8):
+        return 'utf-8-sig'
+
+    if len(b) >= 4:
+        if not b[0]:
+            # 00 00 -- -- - utf-32-be
+            # 00 XX -- -- - utf-16-be
+            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
+            return 'utf-16-le' if b[2] or b[3] else 'utf-32-le'
+    elif len(b) == 2:
+        if not b[0]:
+            # 00 XX - utf-16-be
+            return 'utf-16-be'
+        if not b[1]:
+            # XX 00 - utf-16-le
+            return 'utf-16-le'
+    # default
+    return 'utf-8'
+
+
 def load(fp, *, cls=None, object_hook=None, parse_float=None,
         parse_int=None, parse_constant=None, object_pairs_hook=None, **kw):
     """Deserialize ``fp`` (a ``.read()``-supporting file-like object containing
@@ -270,8 +300,8 @@ def load(fp, *, cls=None, object_hook=None, parse_float=None,
 
 def loads(s, *, encoding=None, cls=None, object_hook=None, parse_float=None,
         parse_int=None, parse_constant=None, object_pairs_hook=None, **kw):
-    """Deserialize ``s`` (a ``str`` instance containing a JSON
-    document) to a Python object.
+    """Deserialize ``s`` (a ``str``, ``bytes`` or ``bytearray`` instance
+    containing a JSON document) to a Python object.
 
     ``object_hook`` is an optional function that will be called with the
     result of any object literal decode (a ``dict``). The return value of
@@ -307,12 +337,16 @@ def loads(s, *, encoding=None, cls=None, object_hook=None, parse_float=None,
     The ``encoding`` argument is ignored and deprecated.
 
     """
-    if not isinstance(s, str):
-        raise TypeError('the JSON object must be str, not {!r}'.format(
-                            s.__class__.__name__))
-    if s.startswith(u'\ufeff'):
-        raise JSONDecodeError("Unexpected UTF-8 BOM (decode using utf-8-sig)",
-                              s, 0)
+    if isinstance(s, str):
+        if s.startswith('\ufeff'):
+            raise JSONDecodeError("Unexpected UTF-8 BOM (decode using utf-8-sig)",
+                                  s, 0)
+    else:
+        if not isinstance(s, (bytes, bytearray)):
+            raise TypeError('the JSON object must be str, bytes or bytearray, '
+                            'not {!r}'.format(s.__class__.__name__))
+        s = s.decode(detect_encoding(s), 'surrogatepass')
+
     if (cls is None and object_hook is None and
             parse_int is None and parse_float is None and
             parse_constant is None and object_pairs_hook is None and not kw):
diff --git a/Lib/test/test_json/test_decode.py b/Lib/test/test_json/test_decode.py
@@ -72,10 +72,8 @@ def test_invalid_escape(self):
 
     def test_invalid_input_type(self):
         msg = 'the JSON object must be str'
-        for value in [1, 3.14, b'bytes', b'\xff\x00', [], {}, None]:
+        for value in [1, 3.14, [], {}, None]:
             self.assertRaisesRegex(TypeError, msg, self.loads, value)
-        with self.assertRaisesRegex(TypeError, msg):
-            self.json.load(BytesIO(b'[1,2,3]'))
 
     def test_string_with_utf8_bom(self):
         # see #18958
diff --git a/Lib/test/test_json/test_unicode.py b/Lib/test/test_json/test_unicode.py
@@ -1,3 +1,4 @@
+import codecs
 from collections import OrderedDict
 from test.test_json import PyTest, CTest
 
@@ -52,9 +53,18 @@ def test_bytes_encode(self):
         self.assertRaises(TypeError, self.dumps, [b"hi"])
 
     def test_bytes_decode(self):
-        self.assertRaises(TypeError, self.loads, b'"hi"')
-        self.assertRaises(TypeError, self.loads, b'["hi"]')
-
+        for encoding, bom in [
+                ('utf-8', codecs.BOM_UTF8),
+                ('utf-16be', codecs.BOM_UTF16_BE),
+                ('utf-16le', codecs.BOM_UTF16_LE),
+                ('utf-32be', codecs.BOM_UTF32_BE),
+                ('utf-32le', codecs.BOM_UTF32_LE),
+            ]:
+            data = ["a\xb5\u20ac\U0001d120"]
+            encoded = self.dumps(data).encode(encoding)
+            self.assertEqual(self.loads(bom + encoded), data)
+            self.assertEqual(self.loads(encoded), data)
+        self.assertRaises(UnicodeDecodeError, self.loads, b'["\x80"]')
 
     def test_object_pairs_hook_with_unicode(self):
         s = '{"xkd":1, "kcw":2, "art":3, "hxm":4, "qrt":5, "pad":6, "hoy":7}'
diff --git a/Misc/NEWS b/Misc/NEWS
@@ -135,6 +135,9 @@ Core and Builtins
 Library
 -------
 
+- Issue #17909: ``json.load`` and ``json.loads`` now support binary input
+  encoded as UTF-8, UTF-16 or UTF-32. Patch by Serhiy Storchaka.
+
 - Issue #27137: the pure Python fallback implementation of ``functools.partial``
   now matches the behaviour of its accelerated C counterpart for subclassing,
   pickling and text representation purposes. Patch by Emanuel Barry and
