RUST-2003 Binary vector subtype support

isabelatkinson · isabelatkinson · commit 11cd91b2f1b3 · 2025-01-23T13:15:36.000-07:00
diff --git a/Cargo.toml b/Cargo.toml
@@ -82,6 +82,7 @@ criterion = "0.3.0"
 pretty_assertions = "0.6.1"
 proptest = "1.0.0"
 serde_bytes = "0.11"
+serde_path_to_error = "0.1.16"
 chrono = { version = "0.4", features = ["serde", "clock", "std"], default-features = false }
 [target.'cfg(all(target_arch = "wasm32", target_os = "unknown"))'.dev-dependencies]
 getrandom = { version = "0.2", features = ["js"] }
diff --git a/src/binary.rs b/src/binary.rs
@@ -1,10 +1,16 @@
+#! Module containing functionality related to BSON binary values.
+
+mod vector;
+
 use crate::{spec::BinarySubtype, Document, RawBinaryRef};
 use std::{
     convert::TryFrom,
     error,
     fmt::{self, Display},
 };
 
+pub use vector::{PackedBitVector, Vector};
+
 /// Represents a BSON binary value.
 #[derive(Debug, Clone, Eq, PartialEq, Hash)]
 pub struct Binary {
@@ -98,14 +104,18 @@ impl Binary {
 pub enum Error {
     /// While trying to decode from base64, an error was returned.
     DecodingError { message: String },
+
+    /// Invalid bytes were provided to construct a [`Binary`] of subtype [`BinarySubtype::Vector`].
+    Vector { message: String },
 }
 
 impl error::Error for Error {}
 
 impl std::fmt::Display for Error {
     fn fmt(&self, fmt: &mut fmt::Formatter) -> fmt::Result {
         match self {
-            Error::DecodingError { message: m } => fmt.write_str(m),
+            Error::DecodingError { message } => fmt.write_str(message),
+            Error::Vector { message } => fmt.write_str(message),
         }
     }
 }
diff --git a/src/spec.rs b/src/spec.rs
@@ -62,6 +62,7 @@ const BINARY_SUBTYPE_MD5: u8 = 0x05;
 const BINARY_SUBTYPE_ENCRYPTED: u8 = 0x06;
 const BINARY_SUBTYPE_COLUMN: u8 = 0x07;
 const BINARY_SUBTYPE_SENSITIVE: u8 = 0x08;
+const BINARY_SUBTYPE_VECTOR: u8 = 0x09;
 const BINARY_SUBTYPE_USER_DEFINED: u8 = 0x80;
 
 /// All available BSON element types.
@@ -162,6 +163,7 @@ pub enum BinarySubtype {
     Encrypted,
     Column,
     Sensitive,
+    Vector,
     UserDefined(u8),
     Reserved(u8),
 }
@@ -179,6 +181,7 @@ impl From<BinarySubtype> for u8 {
             BinarySubtype::Encrypted => BINARY_SUBTYPE_ENCRYPTED,
             BinarySubtype::Column => BINARY_SUBTYPE_COLUMN,
             BinarySubtype::Sensitive => BINARY_SUBTYPE_SENSITIVE,
+            BinarySubtype::Vector => BINARY_SUBTYPE_VECTOR,
             BinarySubtype::UserDefined(x) => x,
             BinarySubtype::Reserved(x) => x,
         }
@@ -198,6 +201,7 @@ impl From<u8> for BinarySubtype {
             BINARY_SUBTYPE_ENCRYPTED => BinarySubtype::Encrypted,
             BINARY_SUBTYPE_COLUMN => BinarySubtype::Column,
             BINARY_SUBTYPE_SENSITIVE => BinarySubtype::Sensitive,
+            BINARY_SUBTYPE_VECTOR => BinarySubtype::Vector,
             _ if t < BINARY_SUBTYPE_USER_DEFINED => BinarySubtype::Reserved(t),
             _ => BinarySubtype::UserDefined(t),
         }
diff --git a/src/tests/spec/json/bson-corpus/binary.json b/src/tests/spec/json/bson-corpus/binary.json
@@ -74,6 +74,36 @@
             "description": "$type query operator (conflicts with legacy $binary form with $type field)",
             "canonical_bson": "180000000378001000000010247479706500020000000000",
             "canonical_extjson": "{\"x\" : { \"$type\" : {\"$numberInt\": \"2\"}}}"
+        },
+        {
+            "description": "subtype 0x09 Vector FLOAT32",
+            "canonical_bson": "170000000578000A0000000927000000FE420000E04000",
+            "canonical_extjson": "{\"x\": {\"$binary\": {\"base64\": \"JwAAAP5CAADgQA==\", \"subType\": \"09\"}}}"
+        },
+        {
+            "description": "subtype 0x09 Vector INT8",
+            "canonical_bson": "11000000057800040000000903007F0700",
+            "canonical_extjson": "{\"x\": {\"$binary\": {\"base64\": \"AwB/Bw==\", \"subType\": \"09\"}}}"
+        },
+        {
+            "description": "subtype 0x09 Vector PACKED_BIT",
+            "canonical_bson": "11000000057800040000000910007F0700",
+            "canonical_extjson": "{\"x\": {\"$binary\": {\"base64\": \"EAB/Bw==\", \"subType\": \"09\"}}}"
+        },
+        {
+            "description": "subtype 0x09 Vector (Zero-length) FLOAT32",
+            "canonical_bson": "0F0000000578000200000009270000",
+            "canonical_extjson": "{\"x\": {\"$binary\": {\"base64\": \"JwA=\", \"subType\": \"09\"}}}"
+        },
+        {
+            "description": "subtype 0x09 Vector (Zero-length) INT8",
+            "canonical_bson": "0F0000000578000200000009030000",
+            "canonical_extjson": "{\"x\": {\"$binary\": {\"base64\": \"AwA=\", \"subType\": \"09\"}}}"
+        },
+        {
+            "description": "subtype 0x09 Vector (Zero-length) PACKED_BIT",
+            "canonical_bson": "0F0000000578000200000009100000",
+            "canonical_extjson": "{\"x\": {\"$binary\": {\"base64\": \"EAA=\", \"subType\": \"09\"}}}"
         }
     ],
     "decodeErrors": [
diff --git a/src/tests/spec/json/bson-corpus/datetime.json b/src/tests/spec/json/bson-corpus/datetime.json
@@ -24,6 +24,7 @@
         {
             "description" : "Y10K",
             "canonical_bson" : "1000000009610000DC1FD277E6000000",
+            "relaxed_extjson" : "{\"a\":{\"$date\":{\"$numberLong\":\"253402300800000\"}}}",
             "canonical_extjson" : "{\"a\":{\"$date\":{\"$numberLong\":\"253402300800000\"}}}"
         },
         {
diff --git a/src/tests/spec/json/bson-corpus/decimal128-1.json b/src/tests/spec/json/bson-corpus/decimal128-1.json
@@ -312,6 +312,30 @@
            "canonical_bson": "18000000136400000000000a5bc138938d44c64d31cc3700",
            "degenerate_extjson": "{\"d\" : {\"$numberDecimal\" : \"1000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000\"}}",
            "canonical_extjson": "{\"d\" : {\"$numberDecimal\" : \"1.000000000000000000000000000000000E+999\"}}"
+        },
+        {
+            "description": "Clamped zeros with a large positive exponent",
+            "canonical_bson": "180000001364000000000000000000000000000000FE5F00",
+            "degenerate_extjson": "{\"d\" : {\"$numberDecimal\" : \"0E+2147483647\"}}",
+            "canonical_extjson": "{\"d\" : {\"$numberDecimal\" : \"0E+6111\"}}"
+        },
+        {
+            "description": "Clamped zeros with a large negative exponent",
+            "canonical_bson": "180000001364000000000000000000000000000000000000",
+            "degenerate_extjson": "{\"d\" : {\"$numberDecimal\" : \"0E-2147483647\"}}",
+            "canonical_extjson": "{\"d\" : {\"$numberDecimal\" : \"0E-6176\"}}"
+        },
+        {
+            "description": "Clamped negative zeros with a large positive exponent",
+            "canonical_bson": "180000001364000000000000000000000000000000FEDF00",
+            "degenerate_extjson": "{\"d\" : {\"$numberDecimal\" : \"-0E+2147483647\"}}",
+            "canonical_extjson": "{\"d\" : {\"$numberDecimal\" : \"-0E+6111\"}}"
+        },
+        {
+            "description": "Clamped negative zeros with a large negative exponent",
+            "canonical_bson": "180000001364000000000000000000000000000000008000",
+            "degenerate_extjson": "{\"d\" : {\"$numberDecimal\" : \"-0E-2147483647\"}}",
+            "canonical_extjson": "{\"d\" : {\"$numberDecimal\" : \"-0E-6176\"}}"
         }
     ]
 }
diff --git a/src/tests/spec/mod.rs b/src/tests/spec/mod.rs
@@ -1,4 +1,5 @@
 mod corpus;
+mod vector;
 
 use std::{
     any::type_name,
@@ -7,7 +8,7 @@ use std::{
     path::PathBuf,
 };
 
-use crate::RawDocumentBuf;
+use crate::Bson;
 use serde::de::DeserializeOwned;
 
 pub(crate) fn run_spec_test<T, F>(spec: &[&str], run_test_file: F)
@@ -31,13 +32,16 @@ where
         let file = File::open(&path)
             .unwrap_or_else(|e| panic!("Failed to open file at {:?}: {}", path, e));
 
-        let test_bson: RawDocumentBuf = serde_json::from_reader(file).unwrap_or_else(|e| {
-            panic!(
-                "Failed to deserialize test JSON to BSON in {:?}: {}",
-                path, e
-            )
-        });
-        let test: T = crate::from_slice(test_bson.as_bytes()).unwrap_or_else(|e| {
+        let mut json_deserializer = serde_json::Deserializer::from_reader(file);
+        let test_bson: Bson = serde_path_to_error::deserialize(&mut json_deserializer)
+            .unwrap_or_else(|e| {
+                panic!(
+                    "Failed to deserialize test JSON to BSON in {:?}: {}",
+                    path, e
+                )
+            });
+        let bson_deserializer = crate::Deserializer::new(test_bson);
+        let test: T = serde_path_to_error::deserialize(bson_deserializer).unwrap_or_else(|e| {
             panic!(
                 "Failed to deserialize test BSON to {} in {:?}: {}",
                 type_name::<T>(),

Original file line number	Diff line number	Diff line change
`@@ -24,6 +24,7 @@`
`24`	`24`	`{`
`25`	`25`	`"description" : "Y10K",`
`26`	`26`	`"canonical_bson" : "1000000009610000DC1FD277E6000000",`
	`27`	`+ "relaxed_extjson" : "{\"a\":{\"$date\":{\"$numberLong\":\"253402300800000\"}}}",`
`27`	`28`	`"canonical_extjson" : "{\"a\":{\"$date\":{\"$numberLong\":\"253402300800000\"}}}"`
`28`	`29`	`},`
`29`	`30`	`{`