PYTHON-5121 - Use canonical Extended JSON for BSON binary vector spec…

bson/json_util.py

@@ -507,6 +507,29 @@ def loads(s: Union[str, bytes, bytearray], *args: Any, **kwargs: Any) -> Any:
     return json.loads(s, *args, **kwargs)
 
 
+def load(fp: Any, *args: Any, **kwargs: Any) -> Any:
+    """Helper function that wraps :func:`json.load`.
+
+    Automatically passes the object_hook for BSON type conversion.
+
+    Raises ``TypeError``, ``ValueError``, ``KeyError``, or
+    :exc:`~bson.errors.InvalidId` on invalid MongoDB Extended JSON.
+
+    :param json_options: A :class:`JSONOptions` instance used to modify the
+        decoding of MongoDB Extended JSON types. Defaults to
+        :const:`DEFAULT_JSON_OPTIONS`.
+
+    .. versionadded:: 4.12
+    """
+    json_options = kwargs.pop("json_options", DEFAULT_JSON_OPTIONS)
+    # Execution time optimization if json_options.document_class is dict
+    if json_options.document_class is dict:
+        kwargs["object_hook"] = lambda obj: object_hook(obj, json_options)
+    else:
+        kwargs["object_pairs_hook"] = lambda pairs: object_pairs_hook(pairs, json_options)
+    return json.load(fp, *args, **kwargs)
+
+
 def _json_convert(obj: Any, json_options: JSONOptions = DEFAULT_JSON_OPTIONS) -> Any:
     """Recursive helper method that converts BSON types so they can be
     converted into json.

test/bson_binary_vector/float32.json

@@ -32,7 +32,7 @@
     {
       "description": "Infinity Vector FLOAT32",
       "valid": true,
-      "vector": ["-inf", 0.0, "inf"],
+      "vector": [{"$numberDouble": "-Infinity"}, 0.0, {"$numberDouble": "Infinity"} ],
       "dtype_hex": "0x27",
       "dtype_alias": "FLOAT32",
       "padding": 0,

test/test_bson_binary_vector.py

@@ -21,7 +21,7 @@
 from pathlib import Path
 from test import unittest
 
-from bson import decode, encode
+from bson import decode, encode, json_util
 from bson.binary import Binary, BinaryVectorDtype
 
 _TEST_PATH = Path(__file__).parent / "bson_binary_vector"
@@ -62,9 +62,6 @@ def run_test(self):
                 cB_exp = binascii.unhexlify(canonical_bson_exp.encode("utf8"))
                 decoded_doc = decode(cB_exp)
                 binary_obs = decoded_doc[test_key]
-                # Handle special float cases like '-inf'
-                if dtype_exp in [BinaryVectorDtype.FLOAT32]:
-                    vector_exp = [float(x) for x in vector_exp]
 
                 # Test round-tripping canonical bson.
                 self.assertEqual(encode(decoded_doc), cB_exp, description)
@@ -104,7 +101,7 @@ def run_test(self):
 def create_tests():
     for filename in _TEST_PATH.glob("*.json"):
         with codecs.open(str(filename), encoding="utf-8") as test_file:
-            test_method = create_test(json.load(test_file))
+            test_method = create_test(json_util.load(test_file))
         setattr(TestBSONBinaryVector, "test_" + filename.stem, test_method)