1 files changed, 71 insertions, 22 deletions

diff --git a/src/mongo/db/exec/document_value/document.h b/src/mongo/db/exec/document_value/document.h
index 8fcfb28dd8b..062778628a0 100644
--- a/src/mongo/db/exec/document_value/document.h
+++ b/src/mongo/db/exec/document_value/document.h
@@ -100,7 +100,9 @@ public:
     static constexpr StringData metaFieldSearchScore = "$searchScore"_sd;
     static constexpr StringData metaFieldSearchHighlights = "$searchHighlights"_sd;
     static constexpr StringData metaFieldSearchScoreDetails = "$searchScoreDetails"_sd;
+    static constexpr StringData metaFieldSearchSortValues = "$searchSortValues"_sd;
     static constexpr StringData metaFieldIndexKey = "$indexKey"_sd;
+    static constexpr StringData metaFieldVectorSearchScore = "$vectorSearchScore"_sd;
 
     static const StringDataSet allMetadataFieldNames;
 
@@ -191,7 +193,8 @@ public:
 
     /**
      * Get the approximate size of the Document, plus its underlying storage and sub-values. Returns
-     * size in bytes.
+     * size in bytes. The return value of this function is snapshotted. All subsequent calls of this
+     * method will return the same value.
      *
      * Note: Some memory may be shared with other Documents or between fields within a single
      * Document so this can overestimate usage.
@@ -202,6 +205,11 @@ public:
     size_t getApproximateSize() const;
 
     /**
+     * Same as 'getApproximateSize()', but this method re-computes the size on every call.
+     */
+    size_t getCurrentApproximateSize() const;
+
+    /**
      * Return the approximate amount of space used by metadata.
      */
     size_t getMetadataApproximateSize() const {
@@ -253,12 +261,38 @@ public:
     void hash_combine(size_t& seed, const StringData::ComparatorInterface* stringComparator) const;
 
     /**
+     * Returns true, if this document is trivially convertible to BSON, meaning the underlying
+     * storage is already in BSON format and there are no damages.
+     */
+    bool isTriviallyConvertible() const {
+        return !storage().isModified() && !storage().bsonHasMetadata();
+    }
+
+    /**
+     * Returns true, if this document is trivially convertible to BSON with metadata, meaning the
+     * underlying storage is already in BSON format and there are no damages.
+     */
+    bool isTriviallyConvertibleWithMetadata() const {
+        return !storage().isModified() && !storage().isMetadataModified();
+    }
+
+    /**
      * Serializes this document to the BSONObj under construction in 'builder'. Metadata is not
      * included. Throws a AssertionException if 'recursionLevel' exceeds the maximum allowable
      * depth.
      */
     void toBson(BSONObjBuilder* builder, size_t recursionLevel = 1) const;
-    BSONObj toBson() const;
+
+    template <typename BSONTraits = BSONObj::DefaultSizeTrait>
+    BSONObj toBson() const {
+        if (isTriviallyConvertible()) {
+            return storage().bsonObj();
+        }
+
+        BSONObjBuilder bb;
+        toBson(&bb);
+        return bb.obj<BSONTraits>();
+    }
 
     /**
      * Serializes this document iff the conversion is "trivial," meaning that the underlying storage
@@ -272,7 +306,18 @@ public:
     /**
      * Like the 'toBson()' method, but includes metadata as top-level fields.
      */
-    BSONObj toBsonWithMetaData() const;
+    void toBsonWithMetaData(BSONObjBuilder* builder) const;
+
+    template <typename BSONTraits = BSONObj::DefaultSizeTrait>
+    BSONObj toBsonWithMetaData() const {
+        if (isTriviallyConvertibleWithMetadata()) {
+            return storage().bsonObj();
+        }
+
+        BSONObjBuilder bb;
+        toBsonWithMetaData(&bb);
+        return bb.obj<BSONTraits>();
+    }
 
     /**
      * Like Document(BSONObj) but treats top-level fields with special names as metadata.
@@ -370,12 +415,6 @@ private:
     getNestedFieldNonCachingHelper(const FieldPath& dottedField, size_t level) const;
 
     boost::intrusive_ptr<const DocumentStorage> _storage;
-
-    /**
-     * Returns the approximate size of this `Document` instance without considering the size of its
-     * backing BSON object.
-     */
-    size_t getApproximateSizeWithoutBackingBSON() const;
 };
 
 //
@@ -511,13 +550,11 @@ public:
     }
 
     /**
-     * Replace the current base Document with bson.
-     *
-     * The paramater 'stripMetadata' controls whether we strip the metadata fields from the
-     * underlying bson when converting the document object back to bson.
+     * Replace the current base Document with the BSON object. Setting 'bsonHasMetadata' to true
+     * signals that the BSON object contains metadata fields.
      */
-    void reset(const BSONObj& bson, bool stripMetadata) {
-        storage().reset(bson, stripMetadata);
+    void reset(const BSONObj& bson, bool bsonHasMetadata) {
+        storage().reset(bson, bsonHasMetadata);
     }
 
     /** Add the given field to the Document.
@@ -654,6 +691,7 @@ public:
      *  TODO: there are some optimizations that may make sense at freeze time.
      */
     Document freeze() {
+        resetSnapshottedApproximateSize();
         // This essentially moves _storage into a new Document by way of temp.
         Document ret;
         boost::intrusive_ptr<const DocumentStorage> temp(storagePtr(), /*inc_ref_count=*/false);
@@ -672,8 +710,12 @@ public:
      *  Note that unlike freeze(), this indicates intention to continue
      *  modifying this document. The returned Document will not observe
      *  future changes to this MutableDocument.
+     *
+     *  Note that the computed snapshotted approximate size of the Document
+     *  is not preserved across calls.
      */
     Document peek() {
+        resetSnapshottedApproximateSize();
         return Document(storagePtr());
     }
 
@@ -695,13 +737,13 @@ public:
         storage().makeOwned();
     }
 
-    /** Create a new document storage with the BSON object.
-     *
-     *  The optional paramater 'stripMetadata' controls whether we strip the metadata fields (the
-     *  complete list is in Document::allMetadataFieldNames).
+    /**
+     * Creates a new document storage with the BSON object. Setting 'bsonHasMetadata' to true
+     * signals that the BSON object contains metadata fields (the complete list is in
+     * Document::allMetadataFieldNames).
      */
-    DocumentStorage& newStorageWithBson(const BSONObj& bson, bool stripMetadata) {
-        reset(make_intrusive<DocumentStorage>(bson, stripMetadata, false, 0));
+    DocumentStorage& newStorageWithBson(const BSONObj& bson, bool bsonHasMetadata) {
+        reset(make_intrusive<DocumentStorage>(bson, bsonHasMetadata, false, 0));
         return const_cast<DocumentStorage&>(*storagePtr());
     }
 
@@ -739,12 +781,19 @@ private:
     MutableValue getNestedFieldHelper(const FieldPath& dottedField, size_t level);
     MutableValue getNestedFieldHelper(const std::vector<Position>& positions, size_t level);
 
-    // this should only be called by storage methods and peek/freeze
+    // this should only be called by storage methods and peek/freeze/resetsnapshottedApproximateSize
     const DocumentStorage* storagePtr() const {
         dassert(!_storage || typeid(*_storage) == typeid(const DocumentStorage));
         return static_cast<const DocumentStorage*>(_storage);
     }
 
+    void resetSnapshottedApproximateSize() {
+        auto mutableStorage = const_cast<DocumentStorage*>(storagePtr());
+        if (mutableStorage) {
+            mutableStorage->resetSnapshottedApproximateSize();
+        }
+    }
+
     // These are both const to prevent modifications bypassing storage() method.
     // They always point to NULL or an object with dynamic type DocumentStorage.
     const RefCountable* _storageHolder;  // Only used in constructors and destructor