SERVER-40969 No-op Document/Value to BSON conversion when Document has not been modified.

6 files changed, 36 insertions, 7 deletions

diff --git a/src/mongo/db/pipeline/document.cpp b/src/mongo/db/pipeline/document.cpp
index 64f73e9054d..29af89e0b5b 100644
--- a/src/mongo/db/pipeline/document.cpp
+++ b/src/mongo/db/pipeline/document.cpp
@@ -176,10 +176,12 @@ Position DocumentStorage::findField(StringData requested, LookupPolicy policy) c
 }
 
 Position DocumentStorage::constructInCache(const BSONElement& elem) {
+    auto savedModified = _modified;
     auto pos = getNextPosition();
     const auto fieldName = elem.fieldNameStringData();
     const unsigned hash = hashKey(fieldName);
     appendField(fieldName, hash, ValueElement::Kind::kCached) = Value(elem);
+    _modified = savedModified;
 
     return pos;
 }
@@ -296,7 +298,7 @@ void DocumentStorage::reserveFields(size_t expectedFields) {
 }
 
 intrusive_ptr<DocumentStorage> DocumentStorage::clone() const {
-    auto out = make_intrusive<DocumentStorage>(_bson, _stripMetadata);
+    auto out = make_intrusive<DocumentStorage>(_bson, _stripMetadata, _modified);
 
     if (_cache) {
         // Make a copy of the buffer with the fields.
@@ -453,6 +455,10 @@ void Document::toBson(BSONObjBuilder* builder, size_t recursionLevel) const {
 }
 
 BSONObj Document::toBson() const {
+    if (!storage().isModified() && !storage().stripMetadata()) {
+        return storage().bsonObj();
+    }
+
     BSONObjBuilder bb;
     toBson(&bb);
     return bb.obj();
diff --git a/src/mongo/db/pipeline/document.h b/src/mongo/db/pipeline/document.h
index 4f63a9fd47e..54629bc0a55 100644
--- a/src/mongo/db/pipeline/document.h
+++ b/src/mongo/db/pipeline/document.h
@@ -623,7 +623,7 @@ public:
      *  complete list is in Document::allMetadataFieldNames).
      */
     DocumentStorage& newStorageWithBson(const BSONObj& bson, bool stripMetadata) {
-        reset(make_intrusive<DocumentStorage>(bson, stripMetadata));
+        reset(make_intrusive<DocumentStorage>(bson, stripMetadata, false));
         return const_cast<DocumentStorage&>(*storagePtr());
     }
 
diff --git a/src/mongo/db/pipeline/document_internal.h b/src/mongo/db/pipeline/document_internal.h
index cdab670b5a5..843f5d4011d 100644
--- a/src/mongo/db/pipeline/document_internal.h
+++ b/src/mongo/db/pipeline/document_internal.h
@@ -390,10 +390,11 @@ public:
      * the document. If we know that the BSON does not contain any metadata fields we can set the
      * 'stripMetadata' flag to false that will speed up the field iteration.
      */
-    DocumentStorage(const BSONObj& bson, bool stripMetadata) : DocumentStorage() {
+    DocumentStorage(const BSONObj& bson, bool stripMetadata, bool modified) : DocumentStorage() {
         _bson = bson.getOwned();
         _bsonIt = BSONObjIterator(_bson);
         _stripMetadata = stripMetadata;
+        _modified = modified;
     }
 
     ~DocumentStorage();
@@ -439,10 +440,12 @@ public:
 
     // MutableDocument uses these
     ValueElement& getField(Position pos) {
+        _modified = true;
         verify(pos.found());
         return *(_firstElement->plusBytes(pos.index));
     }
     Value& getField(StringData name, LookupPolicy policy) {
+        _modified = true;
         Position pos = findField(name, policy);
         if (!pos.found())
             return appendField(name, hashKey(name), ValueElement::Kind::kMaybeInserted);
@@ -630,6 +633,13 @@ public:
 
     Position constructInCache(const BSONElement& elem);
 
+    auto isModified() const {
+        return _modified;
+    }
+    auto bsonObj() const {
+        return _bson;
+    }
+
 private:
     /// Returns the position of the named field in the cache or Position()
     Position findFieldInCache(StringData name) const;
@@ -711,6 +721,12 @@ private:
     // have any metadata we can set _stripMetadata to false that will speed up the iteration.
     bool _stripMetadata{false};
 
+    // This flag is set to true anytime the storage returns a mutable field. It is used to optimize
+    // a conversion to BSON; i.e. if there are not any modifications we can directly return _bson.
+    // Note that an empty (default) document is marked 'modified'. The reason for this is that the
+    // empty _bson is not owned but consumers expect toBson() to return owned BSON.
+    bool _modified{true};
+
     // Defined in document.cpp
     static const DocumentStorage kEmptyDoc;
 
diff --git a/src/mongo/db/pipeline/document_source_cursor.cpp b/src/mongo/db/pipeline/document_source_cursor.cpp
index f4e53f20d16..9c2d124f34d 100644
--- a/src/mongo/db/pipeline/document_source_cursor.cpp
+++ b/src/mongo/db/pipeline/document_source_cursor.cpp
@@ -75,7 +75,9 @@ DocumentSource::GetNextResult DocumentSourceCursor::getNext() {
 }
 
 Document DocumentSourceCursor::transformBSONObjToDocument(const BSONObj& obj) const {
-    return _dependencies ? _dependencies->extractFields(obj) : Document::fromBsonWithMetaData(obj);
+    return _dependencies
+        ? _dependencies->extractFields(obj)
+        : (_inputHasMetadata ? Document::fromBsonWithMetaData(obj) : Document(obj));
 }
 
 void DocumentSourceCursor::loadBatch() {
diff --git a/src/mongo/db/pipeline/document_source_cursor.h b/src/mongo/db/pipeline/document_source_cursor.h
index 215d94c3387..815b6635b14 100644
--- a/src/mongo/db/pipeline/document_source_cursor.h
+++ b/src/mongo/db/pipeline/document_source_cursor.h
@@ -119,10 +119,14 @@ public:
      *
      * @param projection The projection that has been passed down to the query system.
      * @param deps The output of DepsTracker::toParsedDeps.
+     * @param inputHasMetadata Indicates whether the input BSON object contains metadata fields.
      */
-    void setProjection(const BSONObj& projection, const boost::optional<ParsedDeps>& deps) {
+    void setProjection(const BSONObj& projection,
+                       const boost::optional<ParsedDeps>& deps,
+                       bool inputHasMetadata) {
         _projection = projection;
         _dependencies = deps;
+        _inputHasMetadata = inputHasMetadata;
     }
 
     /**
@@ -209,6 +213,7 @@ private:
     BSONObj _sort;
     BSONObj _projection;
     bool _shouldProduceEmptyDocs = false;
+    bool _inputHasMetadata = false;
     boost::optional<ParsedDeps> _dependencies;
     boost::intrusive_ptr<DocumentSourceLimit> _limit;
     long long _docsAddedToBatches;  // for _limit enforcement
diff --git a/src/mongo/db/pipeline/pipeline_d.cpp b/src/mongo/db/pipeline/pipeline_d.cpp
index dc742e0437c..3a27c6ee73f 100644
--- a/src/mongo/db/pipeline/pipeline_d.cpp
+++ b/src/mongo/db/pipeline/pipeline_d.cpp
@@ -838,7 +838,7 @@ void PipelineD::addCursorSource(Pipeline* pipeline,
     }
 
     if (!projectionObj.isEmpty()) {
-        cursor->setProjection(projectionObj, boost::none);
+        cursor->setProjection(projectionObj, boost::none, deps.getNeedsAnyMetadata());
     } else {
         // There may be fewer dependencies now if the sort was covered.
         if (!sortObj.isEmpty()) {
@@ -847,7 +847,7 @@ void PipelineD::addCursorSource(Pipeline* pipeline,
                                                  : DepsTracker::MetadataAvailable::kNoMetadata);
         }
 
-        cursor->setProjection(deps.toProjection(), deps.toParsedDeps());
+        cursor->setProjection(deps.toProjection(), deps.toParsedDeps(), deps.getNeedsAnyMetadata());
     }
 }