Added bson_corpus tests. Needs more, and review of json_util

Author: caseyclements

bson/json_util.py

@@ -610,17 +610,20 @@ def _parse_canonical_binary(doc: Any, json_options: JSONOptions) -> Union[Binary
 
 
 def _parse_canonical_binary_vector(doc: Any, dummy0: Any) -> BinaryVector:
-    binary = doc["$binaryVector"]
-    b64 = binary["base64"]
-    dtype = getattr(DTYPES, binary["dtype"])
-    padding = binary["padding"]
+    if "dtype" in doc:
+        b64 = doc["$binaryVector"]
+    else:
+        doc = doc["$binaryVector"]
+        b64 = doc["base64"]
+    dtype = getattr(DTYPES, doc["dtype"])
+    padding = doc["padding"]
     if not isinstance(b64, str):
         raise TypeError(f"$binaryVector base64 must be a string: {doc}")
     if not isinstance(dtype, DTYPES):
         raise TypeError(f"$binaryVector dtype must a member of bson.vector.DTYPES: {doc}")
     if not isinstance(padding, str) or len(padding) > 2:
         raise TypeError(f"$binaryVector padding must be a string at most 2 characters: {doc}")
-    if len(binary) != 3:
+    if len(doc) != 3:
         raise TypeError(
             f'$binaryVector must include only "base64", "dtype", and "padding" components: {doc}'
         )

test/bson_corpus/vector.json

@@ -0,0 +1,12 @@
+{
+  "description": "BinaryVector type [!! MORE TESTS REQUIRED. json_utils may need tweaking",
+  "bson_type": "0x05",
+  "test_key": "x",
+  "valid": [
+    {
+      "description": "DTYPES.INT8 (Zero-length)",
+      "canonical_bson": "0f0000000578000000000009030000",
+      "canonical_extjson": "{\"x\": {\"$binaryVector\": {\"base64\": \"\", \"dtype\": \"INT8\", \"padding\": \"00\"}}}"
+    }
+  ]
+}