SERVER-56492 Normalize objects within arrays in time-series metadata

(cherry picked from commit 6d66d9116bdee519bd42123a5ae30439c54dc213)

3 files changed, 157 insertions, 6 deletions

diff --git a/jstests/core/timeseries/timeseries_metadata.js b/jstests/core/timeseries/timeseries_metadata.js
index 21047da3826..9fe8de0cf0e 100644
--- a/jstests/core/timeseries/timeseries_metadata.js
+++ b/jstests/core/timeseries/timeseries_metadata.js
@@ -140,5 +140,38 @@ TimeseriesTest.run((insert) => {
         [
             {_id: 3, time: t[3], meta: {a: 1}, x: 30},
         ]);
+
+    runTest(
+        // Metadata field contains an array within an object.
+        [
+            {_id: 0, time: t[0], meta: {a: [{b: 1, c: 0}]}, x: 0},
+            {_id: 1, time: t[1], meta: {a: [{c: 0, b: 1}]}, x: 10},
+        ],
+        [
+            {_id: 2, time: t[2], meta: {a: [{b: 2, c: 0}]}, x: 20},
+            {_id: 3, time: t[3], meta: {a: [{c: 0, b: 2}]}, x: 30},
+        ]);
+
+    runTest(
+        // Metadata field contains a nested array.
+        [
+            {_id: 0, time: t[0], meta: {a: [{b: 1, c: 0}, [{e: 1, f: 0}]]}, x: 0},
+            {_id: 1, time: t[1], meta: {a: [{c: 0, b: 1}, [{f: 0, e: 1}]]}, x: 10},
+        ],
+        [
+            {_id: 2, time: t[2], meta: {a: [[{e: 1, f: 0}], {b: 1, c: 0}]}, x: 20},
+            {_id: 3, time: t[3], meta: {a: [[{f: 0, e: 1}], {c: 0, b: 1}]}, x: 30},
+        ]);
+
+    runTest(
+        // Metadata field contains an array.
+        [
+            {_id: 0, time: t[0], meta: {a: [1, 2, 3]}, x: 0},
+            {_id: 1, time: t[1], meta: {a: [1, 2, 3]}, x: 10},
+        ],
+        [
+            {_id: 2, time: t[2], meta: {a: [2, 1, 3]}, x: 20},
+            {_id: 3, time: t[3], meta: {a: [2, 1, 3]}, x: 30},
+        ]);
 });
 })();
diff --git a/src/mongo/db/timeseries/bucket_catalog.cpp b/src/mongo/db/timeseries/bucket_catalog.cpp
index 653b5583117..a0353b46449 100644
--- a/src/mongo/db/timeseries/bucket_catalog.cpp
+++ b/src/mongo/db/timeseries/bucket_catalog.cpp
@@ -46,6 +46,10 @@
 
 namespace mongo {
 namespace {
+
+void normalizeArray(BSONArrayBuilder* builder, const BSONObj& obj);
+void normalizeObject(BSONObjBuilder* builder, const BSONObj& obj);
+
 const auto getBucketCatalog = ServiceContext::declareDecoration<BucketCatalog>();
 MONGO_FAIL_POINT_DEFINE(hangTimeseriesDirectModificationBeforeWriteConflict);
 
@@ -58,6 +62,20 @@ uint8_t numDigits(uint32_t num) {
     return numDigits;
 }
 
+void normalizeArray(BSONArrayBuilder* builder, const BSONObj& obj) {
+    for (auto& arrayElem : obj) {
+        if (arrayElem.type() == BSONType::Array) {
+            BSONArrayBuilder subArray = builder->subarrayStart();
+            normalizeArray(&subArray, arrayElem.Obj());
+        } else if (arrayElem.type() == BSONType::Object) {
+            BSONObjBuilder subObject = builder->subobjStart();
+            normalizeObject(&subObject, arrayElem.Obj());
+        } else {
+            builder->append(arrayElem);
+        }
+    }
+}
+
 void normalizeObject(BSONObjBuilder* builder, const BSONObj& obj) {
     // BSONObjIteratorSorted provides an abstraction similar to what this function does. However it
     // is using a lexical comparison that is slower than just doing a binary comparison of the field
@@ -96,21 +114,27 @@ void normalizeObject(BSONObjBuilder* builder, const BSONObj& obj) {
     std::sort(it, end);
     for (; it != end; ++it) {
         auto elem = it->element();
-        if (elem.type() != BSONType::Object) {
-            builder->append(elem);
-        } else {
+        if (elem.type() == BSONType::Array) {
+            BSONArrayBuilder subArray(builder->subarrayStart(elem.fieldNameStringData()));
+            normalizeArray(&subArray, elem.Obj());
+        } else if (elem.type() == BSONType::Object) {
             BSONObjBuilder subObject(builder->subobjStart(elem.fieldNameStringData()));
             normalizeObject(&subObject, elem.Obj());
+        } else {
+            builder->append(elem);
         }
     }
 }
 
 void normalizeTopLevel(BSONObjBuilder* builder, const BSONElement& elem) {
-    if (elem.type() != BSONType::Object) {
-        builder->append(elem);
-    } else {
+    if (elem.type() == BSONType::Array) {
+        BSONArrayBuilder subArray(builder->subarrayStart(elem.fieldNameStringData()));
+        normalizeArray(&subArray, elem.Obj());
+    } else if (elem.type() == BSONType::Object) {
         BSONObjBuilder subObject(builder->subobjStart(elem.fieldNameStringData()));
         normalizeObject(&subObject, elem.Obj());
+    } else {
+        builder->append(elem);
     }
 }
 
diff --git a/src/mongo/db/timeseries/bucket_catalog_test.cpp b/src/mongo/db/timeseries/bucket_catalog_test.cpp
index b76619d9eb1..8658b6807dd 100644
--- a/src/mongo/db/timeseries/bucket_catalog_test.cpp
+++ b/src/mongo/db/timeseries/bucket_catalog_test.cpp
@@ -282,6 +282,100 @@ TEST_F(BucketCatalogTest, InsertIntoDifferentBuckets) {
     }
 }
 
+TEST_F(BucketCatalogTest, InsertIntoSameBucketArray) {
+    auto result1 = _bucketCatalog->insert(
+        _opCtx,
+        _ns1,
+        _getCollator(_ns1),
+        _getTimeseriesOptions(_ns1),
+        BSON(_timeField << Date_t::now() << _metaField << BSON_ARRAY(BSON("a" << 0 << "b" << 1))),
+        BucketCatalog::CombineWithInsertsFromOtherClients::kAllow);
+    auto result2 = _bucketCatalog->insert(
+        _opCtx,
+        _ns1,
+        _getCollator(_ns1),
+        _getTimeseriesOptions(_ns1),
+        BSON(_timeField << Date_t::now() << _metaField << BSON_ARRAY(BSON("b" << 1 << "a" << 0))),
+        BucketCatalog::CombineWithInsertsFromOtherClients::kAllow);
+
+    ASSERT_EQ(result1.getValue(), result2.getValue());
+
+    // Check metadata in buckets.
+    ASSERT_BSONOBJ_EQ(BSON(_metaField << BSON_ARRAY(BSON("a" << 0 << "b" << 1))),
+                      _bucketCatalog->getMetadata(result1.getValue()->bucket()));
+    ASSERT_BSONOBJ_EQ(BSON(_metaField << BSON_ARRAY(BSON("a" << 0 << "b" << 1))),
+                      _bucketCatalog->getMetadata(result2.getValue()->bucket()));
+}
+
+TEST_F(BucketCatalogTest, InsertIntoSameBucketObjArray) {
+    auto result1 = _bucketCatalog->insert(
+        _opCtx,
+        _ns1,
+        _getCollator(_ns1),
+        _getTimeseriesOptions(_ns1),
+        BSON(_timeField << Date_t::now() << _metaField
+                        << BSONObj(BSON("c" << BSON_ARRAY(BSON("a" << 0 << "b" << 1)
+                                                          << BSON("f" << 1 << "g" << 0))))),
+        BucketCatalog::CombineWithInsertsFromOtherClients::kAllow);
+    auto result2 = _bucketCatalog->insert(
+        _opCtx,
+        _ns1,
+        _getCollator(_ns1),
+        _getTimeseriesOptions(_ns1),
+        BSON(_timeField << Date_t::now() << _metaField
+                        << BSONObj(BSON("c" << BSON_ARRAY(BSON("b" << 1 << "a" << 0)
+                                                          << BSON("g" << 0 << "f" << 1))))),
+        BucketCatalog::CombineWithInsertsFromOtherClients::kAllow);
+
+    ASSERT_EQ(result1.getValue(), result2.getValue());
+
+    // Check metadata in buckets.
+    ASSERT_BSONOBJ_EQ(
+        BSON(_metaField << BSONObj(BSON(
+                 "c" << BSON_ARRAY(BSON("a" << 0 << "b" << 1) << BSON("f" << 1 << "g" << 0))))),
+        _bucketCatalog->getMetadata(result1.getValue()->bucket()));
+    ASSERT_BSONOBJ_EQ(
+        BSON(_metaField << BSONObj(BSON(
+                 "c" << BSON_ARRAY(BSON("a" << 0 << "b" << 1) << BSON("f" << 1 << "g" << 0))))),
+        _bucketCatalog->getMetadata(result2.getValue()->bucket()));
+}
+
+
+TEST_F(BucketCatalogTest, InsertIntoSameBucketNestedArray) {
+    auto result1 = _bucketCatalog->insert(
+        _opCtx,
+        _ns1,
+        _getCollator(_ns1),
+        _getTimeseriesOptions(_ns1),
+        BSON(_timeField << Date_t::now() << _metaField
+                        << BSONObj(BSON("c" << BSON_ARRAY(BSON("a" << 0 << "b" << 1)
+                                                          << BSON_ARRAY("123"
+                                                                        << "456"))))),
+        BucketCatalog::CombineWithInsertsFromOtherClients::kAllow);
+    auto result2 = _bucketCatalog->insert(
+        _opCtx,
+        _ns1,
+        _getCollator(_ns1),
+        _getTimeseriesOptions(_ns1),
+        BSON(_timeField << Date_t::now() << _metaField
+                        << BSONObj(BSON("c" << BSON_ARRAY(BSON("b" << 1 << "a" << 0)
+                                                          << BSON_ARRAY("123"
+                                                                        << "456"))))),
+        BucketCatalog::CombineWithInsertsFromOtherClients::kAllow);
+
+    ASSERT_EQ(result1.getValue(), result2.getValue());
+
+    // Check metadata in buckets.
+    ASSERT_BSONOBJ_EQ(BSON(_metaField << BSONObj(BSON("c" << BSON_ARRAY(BSON("a" << 0 << "b" << 1)
+                                                                        << BSON_ARRAY("123"
+                                                                                      << "456"))))),
+                      _bucketCatalog->getMetadata(result1.getValue()->bucket()));
+    ASSERT_BSONOBJ_EQ(BSON(_metaField << BSONObj(BSON("c" << BSON_ARRAY(BSON("a" << 0 << "b" << 1)
+                                                                        << BSON_ARRAY("123"
+                                                                                      << "456"))))),
+                      _bucketCatalog->getMetadata(result2.getValue()->bucket()));
+}
+
 TEST_F(BucketCatalogTest, InsertNullAndMissingMetaFieldIntoDifferentBuckets) {
     auto result1 =
         _bucketCatalog->insert(_opCtx,