SERVER-43860 Always upsert exact source document for pipeline-insert $merge

4 files changed, 44 insertions, 18 deletions

diff --git a/src/mongo/db/ops/parsed_update.cpp b/src/mongo/db/ops/parsed_update.cpp
index 073ea2d4865..6c68a621d82 100644
--- a/src/mongo/db/ops/parsed_update.cpp
+++ b/src/mongo/db/ops/parsed_update.cpp
@@ -32,6 +32,7 @@
 #include "mongo/db/ops/parsed_update.h"
 
 #include "mongo/db/ops/update_request.h"
+#include "mongo/db/ops/write_ops_gen.h"
 #include "mongo/db/query/canonical_query.h"
 #include "mongo/db/query/collation/collator_factory_interface.h"
 #include "mongo/db/query/query_planner_common.h"
@@ -53,6 +54,22 @@ Status ParsedUpdate::parseRequest() {
     // of a document during a multi-update.
     invariant(!(_request->shouldReturnAnyDocs() && _request->isMulti()));
 
+    // It is invalid to specify 'upsertSupplied:true' for a non-upsert operation, or if no upsert
+    // document was supplied with the request.
+    if (_request->shouldUpsertSuppliedDocument()) {
+        uassert(ErrorCodes::FailedToParse,
+                str::stream() << "cannot specify '"
+                              << write_ops::UpdateOpEntry::kUpsertSuppliedFieldName
+                              << ": true' for a non-upsert operation",
+                _request->isUpsert());
+        const auto& constants = _request->getUpdateConstants();
+        uassert(ErrorCodes::FailedToParse,
+                str::stream() << "the parameter '"
+                              << write_ops::UpdateOpEntry::kUpsertSuppliedFieldName
+                              << "' is set to 'true', but no document was supplied",
+                constants && (*constants)["new"_sd].type() == BSONType::Object);
+    }
+
     // It is invalid to request that a ProjectionStage be applied to the UpdateStage if the
     // UpdateStage would not return any document.
     invariant(_request->getProj().isEmpty() || _request->shouldReturnAnyDocs());
diff --git a/src/mongo/db/ops/update_request.h b/src/mongo/db/ops/update_request.h
index f946302c6b9..5fa0a561b7e 100644
--- a/src/mongo/db/ops/update_request.h
+++ b/src/mongo/db/ops/update_request.h
@@ -56,16 +56,7 @@ public:
         RETURN_NEW
     };
 
-    inline UpdateRequest(const NamespaceString& nsString)
-        : _nsString(nsString),
-          _god(false),
-          _upsert(false),
-          _multi(false),
-          _fromMigration(false),
-          _fromOplogApplication(false),
-          _isExplain(false),
-          _returnDocs(ReturnDocOption::RETURN_NONE),
-          _yieldPolicy(PlanExecutor::NO_YIELD) {}
+    inline UpdateRequest(const NamespaceString& nsString) : _nsString(nsString) {}
 
     const NamespaceString& getNamespaceString() const {
         return _nsString;
@@ -154,6 +145,14 @@ public:
         return _upsert;
     }
 
+    inline void setUpsertSuppliedDocument(bool value = true) {
+        _upsertSuppliedDocument = value;
+    }
+
+    bool shouldUpsertSuppliedDocument() const {
+        return _upsertSuppliedDocument;
+    }
+
     inline void setMulti(bool value = true) {
         _multi = value;
     }
@@ -306,22 +305,26 @@ private:
 
     // God bypasses _id checking and index generation. It is only used on behalf of system
     // updates, never user updates.
-    bool _god;
+    bool _god = false;
 
     // True if this should insert if no matching document is found.
-    bool _upsert;
+    bool _upsert = false;
+
+    // True if this upsert operation should insert the document supplied as 'c.new' if the query
+    // does not match any documents.
+    bool _upsertSuppliedDocument = false;
 
     // True if this update is allowed to affect more than one document.
-    bool _multi;
+    bool _multi = false;
 
     // True if this update is on behalf of a chunk migration.
-    bool _fromMigration;
+    bool _fromMigration = false;
 
     // True if this update was triggered by the application of an oplog entry.
-    bool _fromOplogApplication;
+    bool _fromOplogApplication = false;
 
     // Whether or not we are requesting an explained update. Explained updates are read-only.
-    bool _isExplain;
+    bool _isExplain = false;
 
     // Specifies which version of the documents to return, if any.
     //
@@ -335,10 +338,10 @@ private:
     //
     // This allows findAndModify to execute an update and retrieve the resulting document
     // without another query before or after the update.
-    ReturnDocOption _returnDocs;
+    ReturnDocOption _returnDocs = ReturnDocOption::RETURN_NONE;
 
     // Whether or not the update should yield. Defaults to NO_YIELD.
-    PlanExecutor::YieldPolicy _yieldPolicy;
+    PlanExecutor::YieldPolicy _yieldPolicy = PlanExecutor::NO_YIELD;
 };
 
 }  // namespace mongo
diff --git a/src/mongo/db/ops/write_ops.idl b/src/mongo/db/ops/write_ops.idl
index 9409aa73041..42f86b1f481 100644
--- a/src/mongo/db/ops/write_ops.idl
+++ b/src/mongo/db/ops/write_ops.idl
@@ -127,6 +127,11 @@ structs:
                               operation inserts only a single document."
                 type: bool
                 default: false
+            upsertSupplied:
+                description: "Only applicable when upsert is true. If set, and if no documents match
+                              the query, the update subsystem will insert the document supplied as
+                              'c.new' rather than generating a new document from the update spec."
+                type: optionalBool
             collation:
                 description: "Specifies the collation to use for the operation."
                 type: object
diff --git a/src/mongo/db/ops/write_ops_exec.cpp b/src/mongo/db/ops/write_ops_exec.cpp
index 38da11a7d0e..3b1c09b6a62 100644
--- a/src/mongo/db/ops/write_ops_exec.cpp
+++ b/src/mongo/db/ops/write_ops_exec.cpp
@@ -724,6 +724,7 @@ static SingleWriteResult performSingleUpdateOpWithDupKeyRetry(OperationContext*
     request.setArrayFilters(write_ops::arrayFiltersOf(op));
     request.setMulti(op.getMulti());
     request.setUpsert(op.getUpsert());
+    request.setUpsertSuppliedDocument(op.getUpsertSupplied());
     request.setHint(op.getHint());
 
     request.setYieldPolicy(opCtx->inMultiDocumentTransaction() ? PlanExecutor::INTERRUPT_ONLY