RUST-554 Support parsing $uuid as extended JSON representation for subtype 4 binary (#213)

Author: saghm

etc/update-spec-tests.sh

@@ -19,6 +19,6 @@ fi
 tmpdir=`perl -MFile::Temp=tempdir -wle 'print tempdir(TMPDIR => 1, CLEANUP => 0)'`
 curl -sL https://github.com/mongodb/specifications/archive/master.zip -o "$tmpdir/specs.zip"
 unzip -d "$tmpdir" "$tmpdir/specs.zip" > /dev/null
-mkdir -p "tests/spec/json/$1"
-rsync -ah "$tmpdir/specifications-master/source/$1/tests/" "tests/spec/json/$1" --delete
+mkdir -p "src/tests/spec/json/$1"
+rsync -ah "$tmpdir/specifications-master/source/$1/tests/" "src/tests/spec/json/$1" --delete
 rm -rf "$tmpdir"

src/extjson/de.rs

@@ -117,6 +117,11 @@ impl TryFrom<serde_json::Map<String, serde_json::Value>> for Bson {
             return Ok(Bson::Binary(binary.parse()?));
         }
 
+        if obj.contains_key("$uuid") {
+            let uuid: models::Uuid = serde_json::from_value(obj.into())?;
+            return Ok(Bson::Binary(uuid.parse()?));
+        }
+
         if obj.contains_key("$code") {
             let code_w_scope: models::JavaScriptCodeWithScope = serde_json::from_value(obj.into())?;
             return match code_w_scope.scope {

src/extjson/models.rs

@@ -6,7 +6,7 @@ use serde::{
     Deserialize,
 };
 
-use crate::{extjson, oid, Bson};
+use crate::{extjson, oid, spec::BinarySubtype, Bson};
 
 #[derive(Deserialize)]
 #[serde(deny_unknown_fields)]
@@ -165,6 +165,50 @@ impl Binary {
     }
 }
 
+#[derive(Deserialize)]
+#[serde(deny_unknown_fields)]
+pub(crate) struct Uuid {
+    #[serde(rename = "$uuid")]
+    value: String,
+}
+
+impl Uuid {
+    pub(crate) fn parse(mut self) -> extjson::de::Result<crate::Binary> {
+        if !valid_uuid_format(&self.value) {
+            return Err(extjson::de::Error::invalid_value(
+                Unexpected::Str(&self.value),
+                &"$uuid values does not follow RFC 4122 format regarding length and hyphens",
+            ));
+        }
+
+        self.value.retain(|c| c != '-');
+
+        let bytes = hex::decode(&self.value).map_err(|_| {
+            extjson::de::Error::invalid_value(
+                Unexpected::Str(&self.value),
+                &"$uuid does not follow RFC 4122 format regarding hex bytes",
+            )
+        })?;
+
+        Ok(crate::Binary {
+            subtype: BinarySubtype::Uuid,
+            bytes,
+        })
+    }
+}
+
+fn valid_uuid_format(s: &str) -> bool {
+    // RFC 4122 defines the hyphens in a UUID as appearing in the 8th, 13th, 18th, and 23rd
+    // characters.
+    //
+    // See https://tools.ietf.org/html/rfc4122#section-3
+    s.chars().count() == 36
+        && s.chars().nth(8) == Some('-')
+        && s.chars().nth(13) == Some('-')
+        && s.chars().nth(18) == Some('-')
+        && s.chars().nth(23) == Some('-')
+}
+
 #[derive(Deserialize)]
 #[serde(deny_unknown_fields)]
 pub(crate) struct JavaScriptCodeWithScope {

src/tests/spec/json/bson-corpus/binary.json

@@ -39,6 +39,12 @@
             "canonical_bson": "1D000000057800100000000473FFD26444B34C6990E8E7D1DFC035D400",
             "canonical_extjson": "{\"x\" : { \"$binary\" : {\"base64\" : \"c//SZESzTGmQ6OfR38A11A==\", \"subType\" : \"04\"}}}"
         },
+        {
+            "description": "subtype 0x04 UUID",
+            "canonical_bson": "1D000000057800100000000473FFD26444B34C6990E8E7D1DFC035D400",
+            "canonical_extjson": "{\"x\" : { \"$binary\" : {\"base64\" : \"c//SZESzTGmQ6OfR38A11A==\", \"subType\" : \"04\"}}}",
+            "degenerate_extjson": "{\"x\" : { \"$uuid\" : \"73ffd264-44b3-4c69-90e8-e7d1dfc035d4\"}}"
+        },
         {
             "description": "subtype 0x05",
             "canonical_bson": "1D000000057800100000000573FFD26444B34C6990E8E7D1DFC035D400",
@@ -81,5 +87,15 @@
             "description": "subtype 0x02 length negative one",
             "bson": "130000000578000600000002FFFFFFFFFFFF00"
         }
+    ],
+    "parseErrors": [
+        {
+            "description": "$uuid wrong type",
+            "string": "{\"x\" : { \"$uuid\" : { \"data\" : \"73ffd264-44b3-4c69-90e8-e7d1dfc035d4\"}}}"
+        },
+        {
+            "description": "$uuid invalid value",
+            "string": "{\"x\" : { \"$uuid\" : \"73ffd264-44b3-90e8-e7d1dfc035d4\"}}"
+        }
     ]
 }

src/tests/spec/json/bson-corpus/timestamp.json

@@ -18,6 +18,11 @@
             "description": "Timestamp with high-order bit set on both seconds and increment",
             "canonical_bson": "10000000116100FFFFFFFFFFFFFFFF00",
             "canonical_extjson": "{\"a\" : {\"$timestamp\" : {\"t\" : 4294967295, \"i\" :  4294967295} } }"
+        },
+        {
+            "description": "Timestamp with high-order bit set on both seconds and increment (not UINT32_MAX)",
+            "canonical_bson": "1000000011610000286BEE00286BEE00", 
+            "canonical_extjson": "{\"a\" : {\"$timestamp\" : {\"t\" : 4000000000, \"i\" :  4000000000} } }"
         }
     ],
     "decodeErrors": [