CBG-4411: Improvements to Diagnostic Sync Function dry run endpoint (#7932)

Author: RIT3shSapata

base/error.go

@@ -361,3 +361,28 @@ func (me *MultiError) ErrorOrNil() error {
 	}
 	return me
 }
+
+const syncFnDryRunErrorPrefix = "Error returned from Sync Function"
+
+// SyncFnDryRunError is returned when the sync function dry run returns an error.
+// It wraps the original error for errors.Is and the type supports errors.As
+type SyncFnDryRunError struct {
+	Err error
+}
+
+func (e *SyncFnDryRunError) Error() string {
+	if e == nil {
+		return syncFnDryRunErrorPrefix
+	}
+	if e.Err == nil {
+		return syncFnDryRunErrorPrefix
+	}
+	return syncFnDryRunErrorPrefix + ": " + e.Err.Error()
+}
+
+func (e *SyncFnDryRunError) Unwrap() error {
+	if e == nil {
+		return nil
+	}
+	return e.Err
+}

db/crud.go

@@ -1694,90 +1694,43 @@ func (db *DatabaseCollectionWithUser) PutExistingRevWithBody(ctx context.Context
 
 }
 
-// SyncFnDryrun Runs a document through the sync function and returns expiry, channels doc was placed in, access map for users, roles, handler errors and sync fn exceptions
-func (db *DatabaseCollectionWithUser) SyncFnDryrun(ctx context.Context, body Body, docID string) (*channels.ChannelMapperOutput, error, error) {
-	doc := &Document{
-		ID:    docID,
-		_body: body,
-	}
-	oldDoc := doc
-	if docID != "" {
-		if docInBucket, err := db.GetDocument(ctx, docID, DocUnmarshalAll); err == nil {
-			oldDoc = docInBucket
-			if doc._body == nil {
-				body = oldDoc.Body(ctx)
-				doc._body = body
-				// If no body is given, use doc in bucket as doc with no old doc
-				oldDoc._body = nil
-			}
-			doc._body[BodyRev] = oldDoc.SyncData.GetRevTreeID()
-		} else {
-			return nil, err, nil
-		}
-	} else {
-		oldDoc._body = nil
-	}
-
-	delete(body, BodyId)
-
-	// Get the revision ID to match, and the new generation number:
-	matchRev, _ := body[BodyRev].(string)
-	generation, _ := ParseRevID(ctx, matchRev)
-	if generation < 0 {
-		return nil, base.HTTPErrorf(http.StatusBadRequest, "Invalid revision ID"), nil
-	}
-	generation++
-
-	// Create newDoc which will be used to pass around Body
-	newDoc := &Document{
-		ID: docID,
-	}
-	// Pull out attachments
-	newDoc.SetAttachments(GetBodyAttachments(body))
-	delete(body, BodyAttachments)
-
-	delete(body, BodyRevisions)
+// SyncFnDryrun Runs the given document body through a sync function and returns expiry, channels doc was placed in,
+// access map for users, roles, handler errors and sync fn exceptions.
+// If syncFn is provided, it will be used instead of the one configured on the database.
+func (db *DatabaseCollectionWithUser) SyncFnDryrun(ctx context.Context, newDoc, oldDoc *Document, syncFn string) (*channels.ChannelMapperOutput, error) {
 
-	err := validateAPIDocUpdate(body)
-	if err != nil {
-		return nil, err, nil
-	}
-	bodyWithoutInternalProps, wasStripped := StripInternalProperties(body)
-	canonicalBytesForRevID, err := base.JSONMarshalCanonical(bodyWithoutInternalProps)
-	if err != nil {
-		return nil, err, nil
-	}
-
-	// We needed to keep _deleted around in the body until we generated a rev ID, but now we can ditch it.
-	_, isDeleted := body[BodyDeleted]
-	if isDeleted {
-		delete(body, BodyDeleted)
-	}
-
-	// and now we can finally update the newDoc body to be without any special properties
-	newDoc.UpdateBody(body)
-
-	// If no special properties were stripped and document wasn't deleted, the canonical bytes represent the current
-	// body.  In this scenario, store canonical bytes as newDoc._rawBody
-	if !wasStripped && !isDeleted {
-		newDoc._rawBody = canonicalBytesForRevID
-	}
-
-	newRev := CreateRevIDWithBytes(generation, matchRev, canonicalBytesForRevID)
-	newDoc.RevID = newRev
 	mutableBody, metaMap, _, err := db.prepareSyncFn(oldDoc, newDoc)
 	if err != nil {
 		base.InfofCtx(ctx, base.KeyDiagnostic, "Failed to prepare to run sync function: %v", err)
-		return nil, err, nil
+		return nil, err
 	}
 
 	syncOptions, err := MakeUserCtx(db.user, db.ScopeName, db.Name)
 	if err != nil {
-		return nil, err, nil
+		return nil, err
+	}
+	var output *channels.ChannelMapperOutput
+	var syncErr error
+	if syncFn == "" {
+		output, err = db.ChannelMapper.MapToChannelsAndAccess(ctx, mutableBody, string(oldDoc._rawBody), metaMap, syncOptions)
+		if err != nil {
+			return nil, &base.SyncFnDryRunError{Err: err}
+		}
+	} else {
+		jsTimeout := time.Duration(base.DefaultJavascriptTimeoutSecs) * time.Second
+		syncRunner, err := channels.NewSyncRunner(ctx, syncFn, jsTimeout)
+		if err != nil {
+			return nil, fmt.Errorf("failed to create sync runner: %v", err)
+		}
+		jsOutput, err := syncRunner.Call(ctx, mutableBody, string(oldDoc._rawBody), metaMap, syncOptions)
+		if err != nil {
+
+			return nil, &base.SyncFnDryRunError{Err: err}
+		}
+		output = jsOutput.(*channels.ChannelMapperOutput)
 	}
-	output, err := db.ChannelMapper.MapToChannelsAndAccess(ctx, mutableBody, string(oldDoc._rawBody), metaMap, syncOptions)
 
-	return output, nil, err
+	return output, syncErr
 }
 
 // revTreeConflictCheck checks for conflicts in the rev tree history and returns the parent revid, currentRevIndex

docs/api/paths/diagnostic/keyspace-sync.yaml

@@ -7,16 +7,40 @@
 # the file licenses/APL2.txt.
 parameters:
   - $ref: ../../components/parameters.yaml#/keyspace
-get:
+  - $ref: ../../components/parameters.yaml#/doc_id
+post:
   summary: Run a doc body through the sync function and return sync data.
   description: |-
-    Run a document body through the sync function and return document sync data.
+    Runs a document body through the sync function and returns document sync
+    data. If no custom sync function is provided in the request body, the
+    default or user-defined sync function for the collection is used.
+    | Document | DocID | Behaviour                                                                                                                                                                            |
+    | -------- | ----- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
+    | Yes      | No    | The document passed will be considered as newDoc and oldDoc will be empty                                                                                                            |
+    | Yes      | Yes   | The document passed in the body will be newDoc and DocID will be read from the bucket/collection and will be passed as the oldDoc. If DocID doesn't exist, then oldDoc will be empty |
+    | No       | No    | Will throw an error                                                                                                                                                                  |
+    | No       | Yes   | The docID will be passed in as the newDoc and oldDoc will be empty. If the document is not found, an error will be returned                                                          |
+
     * Sync Gateway Application Read Only
   requestBody:
     content:
       application/json:
         schema:
-          $ref: ../../components/schemas.yaml#/Document
+          type: object
+          properties:
+            sync_function:
+              description: |-
+                A JavaScript function that defines custom access, channel, and
+                validation logic for documents. This function will be evaluated
+                by the Sync Gateway to determine document routing, access
+                grants, and validation outcomes during synchronization.
+              type: string
+              example: |-
+                function (doc, oldDoc) {
+                  channel(doc.channels);
+                }
+            doc:
+              $ref: ../../components/schemas.yaml#/Document
   responses:
     '200':
       description: Document Processed by sync function successfully

rest/diagnostic_doc_api.go

@@ -11,6 +11,7 @@ licenses/APL2.txt.
 package rest
 
 import (
+	"errors"
 	"fmt"
 	"net/http"
 
@@ -33,6 +34,11 @@ type ImportFilterDryRun struct {
 	Error        string `json:"error"`
 }
 
+type SyncFnDryRunPayload struct {
+	Function string  `json:"sync_function"`
+	Doc      db.Body `json:"doc,omitempty"`
+}
+
 func populateDocChannelInfo(doc db.Document) map[string][]auth.GrantHistorySequencePair {
 	resp := make(map[string][]auth.GrantHistorySequencePair, len(doc.Channels))
 
@@ -70,20 +76,84 @@ func (h *handler) handleGetDocChannels() error {
 func (h *handler) handleSyncFnDryRun() error {
 	docid := h.getQuery("doc_id")
 
-	body, err := h.readDocument()
+	var syncDryRunPayload SyncFnDryRunPayload
+	err := h.readJSONInto(&syncDryRunPayload)
 	if err != nil {
-		if docid == "" {
-			return fmt.Errorf("no doc id provided for dry run and error reading body: %s", err)
+		return base.HTTPErrorf(http.StatusBadRequest, "Error reading sync function payload: %v", err)
+	}
+
+	if syncDryRunPayload.Doc == nil && docid == "" {
+		return base.HTTPErrorf(http.StatusBadRequest, "no docid or document provided")
+	}
+
+	oldDoc := &db.Document{ID: docid}
+	oldDoc.UpdateBody(syncDryRunPayload.Doc)
+	if docid != "" {
+		if docInbucket, err := h.collection.GetDocument(h.ctx(), docid, db.DocUnmarshalAll); err == nil {
+			oldDoc = docInbucket
+			if len(syncDryRunPayload.Doc) == 0 {
+				syncDryRunPayload.Doc = oldDoc.Body(h.ctx())
+				oldDoc.UpdateBody(nil)
+			}
+		} else {
+			return base.HTTPErrorf(http.StatusNotFound, "Error reading document: %v", err)
 		}
+	} else {
+		oldDoc.UpdateBody(nil)
+	}
+
+	delete(syncDryRunPayload.Doc, db.BodyId)
+
+	// Get the revision ID to match, and the new generation number:
+	matchRev, _ := syncDryRunPayload.Doc[db.BodyRev].(string)
+	generation, _ := db.ParseRevID(h.ctx(), matchRev)
+	if generation < 0 {
+		return base.HTTPErrorf(http.StatusBadRequest, "Invalid revision ID")
+	}
+	generation++
+
+	// Create newDoc which will be used to pass around Body
+	newDoc := &db.Document{
+		ID: docid,
 	}
+	// Pull attachments
+	newDoc.SetAttachments(db.GetBodyAttachments(syncDryRunPayload.Doc))
+	delete(syncDryRunPayload.Doc, db.BodyAttachments)
+	delete(syncDryRunPayload.Doc, db.BodyRevisions)
 
-	output, err, syncFnErr := h.collection.SyncFnDryrun(h.ctx(), body, docid)
+	if _, ok := syncDryRunPayload.Doc[base.SyncPropertyName]; ok {
+		return base.HTTPErrorf(http.StatusBadRequest, "document-top level property '_sync' is a reserved internal property")
+	}
+
+	db.StripInternalProperties(syncDryRunPayload.Doc)
+
+	// We needed to keep _deleted around in the body until we generate rev ID, but now it can be removed
+	_, isDeleted := syncDryRunPayload.Doc[db.BodyDeleted]
+	if isDeleted {
+		delete(syncDryRunPayload.Doc, db.BodyDeleted)
+	}
+
+	//update the newDoc body to be without any special properties
+	newDoc.UpdateBody(syncDryRunPayload.Doc)
+
+	rawDocBytes, err := newDoc.BodyBytes(h.ctx())
 	if err != nil {
-		return err
+		return base.HTTPErrorf(http.StatusBadRequest, "Error marshalling document: %v", err)
 	}
-	if syncFnErr != nil {
+
+	newRev := db.CreateRevIDWithBytes(generation, matchRev, rawDocBytes)
+	newDoc.RevID = newRev
+
+	output, err := h.collection.SyncFnDryrun(h.ctx(), newDoc, oldDoc, syncDryRunPayload.Function)
+	if err != nil {
+		var syncFnDryRunErr *base.SyncFnDryRunError
+		if !errors.As(err, &syncFnDryRunErr) {
+			return err
+		}
+
+		errMsg := syncFnDryRunErr.Error()
 		resp := SyncFnDryRun{
-			Exception: syncFnErr.Error(),
+			Exception: errMsg,
 		}
 		h.writeJSON(resp)
 		return nil