fix(NODE-6362): cache cursor deserialization options across deserialize calls (#4221)

Author: baileympearson

src/bson.ts

@@ -132,3 +132,14 @@ export function resolveBSONOptions(
       options?.enableUtf8Validation ?? parentOptions?.enableUtf8Validation ?? true
   };
 }
+
+/** @internal */
+export function parseUtf8ValidationOption(options?: { enableUtf8Validation?: boolean }): {
+  utf8: { writeErrors: false } | false;
+} {
+  const enableUtf8Validation = options?.enableUtf8Validation;
+  if (enableUtf8Validation === false) {
+    return { utf8: false };
+  }
+  return { utf8: { writeErrors: false } };
+}

src/cmap/connection.ts

@@ -1,3 +1,4 @@
+import { type DeserializeOptions } from 'bson';
 import { type Readable, Transform, type TransformCallback } from 'stream';
 import { clearTimeout, setTimeout } from 'timers';
 
@@ -487,7 +488,7 @@ export class Connection extends TypedEventEmitter<ConnectionEvents> {
 
     // If `documentsReturnedIn` not set or raw is not enabled, use input bson options
     // Otherwise, support raw flag. Raw only works for cursors that hardcode firstBatch/nextBatch fields
-    const bsonOptions =
+    const bsonOptions: DeserializeOptions =
       options.documentsReturnedIn == null || !options.raw
         ? options
         : {

src/cmap/wire_protocol/on_demand/document.ts

@@ -1,16 +1,16 @@
+import { type DeserializeOptions } from 'bson';
+
 import {
   Binary,
   type BSONElement,
   BSONError,
-  type BSONSerializeOptions,
   BSONType,
   deserialize,
   getBigInt64LE,
   getFloat64LE,
   getInt32LE,
   ObjectId,
   parseToElementsToArray,
-  pluckBSONSerializeOptions,
   Timestamp,
   toUTF8
 } from '../../../bson';
@@ -44,6 +44,14 @@ export type JSTypeOf = {
 /** @internal */
 type CachedBSONElement = { element: BSONElement; value: any | undefined };
 
+/**
+ * @internal
+ *
+ * Options for `OnDemandDocument.toObject()`. Validation is required to ensure
+ * that callers provide utf8 validation options. */
+export type OnDemandDocumentDeserializeOptions = Omit<DeserializeOptions, 'validation'> &
+  Required<Pick<DeserializeOptions, 'validation'>>;
+
 /** @internal */
 export class OnDemandDocument {
   /**
@@ -330,24 +338,12 @@ export class OnDemandDocument {
    * Deserialize this object, DOES NOT cache result so avoid multiple invocations
    * @param options - BSON deserialization options
    */
-  public toObject(options?: BSONSerializeOptions): Record<string, any> {
-    const exactBSONOptions = {
-      ...pluckBSONSerializeOptions(options ?? {}),
-      validation: this.parseBsonSerializationOptions(options),
+  public toObject(options?: OnDemandDocumentDeserializeOptions): Record<string, any> {
+    return deserialize(this.bson, {
+      ...options,
       index: this.offset,
       allowObjectSmallerThanBufferSize: true
-    };
-    return deserialize(this.bson, exactBSONOptions);
-  }
-
-  private parseBsonSerializationOptions(options?: { enableUtf8Validation?: boolean }): {
-    utf8: { writeErrors: false } | false;
-  } {
-    const enableUtf8Validation = options?.enableUtf8Validation;
-    if (enableUtf8Validation === false) {
-      return { utf8: false };
-    }
-    return { utf8: { writeErrors: false } };
+    });
   }
 
   /** Returns this document's bytes only */

src/cmap/wire_protocol/responses.ts

@@ -1,16 +1,24 @@
+import { type DeserializeOptions } from 'bson';
+
 import {
   type BSONElement,
   type BSONSerializeOptions,
   BSONType,
   type Document,
   Long,
   parseToElementsToArray,
+  parseUtf8ValidationOption,
+  pluckBSONSerializeOptions,
   type Timestamp
 } from '../../bson';
 import { MongoUnexpectedServerResponseError } from '../../error';
 import { type ClusterTime } from '../../sdam/common';
 import { decorateDecryptionResult, ns } from '../../utils';
-import { type JSTypeOf, OnDemandDocument } from './on_demand/document';
+import {
+  type JSTypeOf,
+  OnDemandDocument,
+  type OnDemandDocumentDeserializeOptions
+} from './on_demand/document';
 
 // eslint-disable-next-line no-restricted-syntax
 const enum BSONElementOffset {
@@ -112,7 +120,8 @@ export class MongoDBResponse extends OnDemandDocument {
       this.get('recoveryToken', BSONType.object)?.toObject({
         promoteValues: false,
         promoteLongs: false,
-        promoteBuffers: false
+        promoteBuffers: false,
+        validation: { utf8: true }
       }) ?? null
     );
   }
@@ -165,6 +174,14 @@ export class MongoDBResponse extends OnDemandDocument {
     }
     return this.clusterTime ?? null;
   }
+
+  public override toObject(options?: BSONSerializeOptions): Record<string, any> {
+    const exactBSONOptions = {
+      ...pluckBSONSerializeOptions(options ?? {}),
+      validation: parseUtf8ValidationOption(options)
+    };
+    return super.toObject(exactBSONOptions);
+  }
 }
 
 /** @internal */
@@ -248,12 +265,13 @@ export class CursorResponse extends MongoDBResponse {
       this.cursor.get('postBatchResumeToken', BSONType.object)?.toObject({
         promoteValues: false,
         promoteLongs: false,
-        promoteBuffers: false
+        promoteBuffers: false,
+        validation: { utf8: true }
       }) ?? null
     );
   }
 
-  public shift(options?: BSONSerializeOptions): any {
+  public shift(options: OnDemandDocumentDeserializeOptions): any {
     if (this.iterated >= this.batchSize) {
       return null;
     }
@@ -305,7 +323,7 @@ export class ExplainedCursorResponse extends CursorResponse {
     return this._length;
   }
 
-  override shift(options?: BSONSerializeOptions | undefined) {
+  override shift(options?: DeserializeOptions) {
     if (this._length === 0) return null;
     this._length -= 1;
     return this.toObject(options);

src/cursor/abstract_cursor.ts

@@ -1,6 +1,7 @@
 import { Readable, Transform } from 'stream';
 
 import { type BSONSerializeOptions, type Document, Long, pluckBSONSerializeOptions } from '../bson';
+import { type OnDemandDocumentDeserializeOptions } from '../cmap/wire_protocol/on_demand/document';
 import { type CursorResponse } from '../cmap/wire_protocol/responses';
 import {
   MongoAPIError,
@@ -157,6 +158,9 @@ export abstract class AbstractCursor<
   /** @event */
   static readonly CLOSE = 'close' as const;
 
+  /** @internal */
+  protected deserializationOptions: OnDemandDocumentDeserializeOptions;
+
   /** @internal */
   protected constructor(
     client: MongoClient,
@@ -211,6 +215,13 @@ export abstract class AbstractCursor<
     } else {
       this.cursorSession = this.cursorClient.startSession({ owner: this, explicit: false });
     }
+
+    this.deserializationOptions = {
+      ...this.cursorOptions,
+      validation: {
+        utf8: options?.enableUtf8Validation === false ? false : true
+      }
+    };
   }
 
   /**
@@ -304,7 +315,7 @@ export abstract class AbstractCursor<
     );
 
     for (let count = 0; count < documentsToRead; count++) {
-      const document = this.documents?.shift(this.cursorOptions);
+      const document = this.documents?.shift(this.deserializationOptions);
       if (document != null) {
         bufferedDocs.push(document);
       }
@@ -406,7 +417,7 @@ export abstract class AbstractCursor<
     }
 
     do {
-      const doc = this.documents?.shift(this.cursorOptions);
+      const doc = this.documents?.shift(this.deserializationOptions);
       if (doc != null) {
         if (this.transform != null) return await this.transformDocument(doc);
         return doc;
@@ -425,15 +436,15 @@ export abstract class AbstractCursor<
       throw new MongoCursorExhaustedError();
     }
 
-    let doc = this.documents?.shift(this.cursorOptions);
+    let doc = this.documents?.shift(this.deserializationOptions);
     if (doc != null) {
       if (this.transform != null) return await this.transformDocument(doc);
       return doc;
     }
 
     await this.fetchBatch();
 
-    doc = this.documents?.shift(this.cursorOptions);
+    doc = this.documents?.shift(this.deserializationOptions);
     if (doc != null) {
       if (this.transform != null) return await this.transformDocument(doc);
       return doc;

src/cursor/aggregation_cursor.ts

@@ -76,7 +76,7 @@ export class AggregationCursor<TSchema = any> extends AbstractCursor<TSchema> {
           explain: verbosity ?? true
         })
       )
-    ).shift(this.aggregateOptions);
+    ).shift(this.deserializationOptions);
   }
 
   /** Add a stage to the aggregation pipeline

src/cursor/find_cursor.ts

@@ -143,7 +143,7 @@ export class FindCursor<TSchema = any> extends AbstractCursor<TSchema> {
           explain: verbosity ?? true
         })
       )
-    ).shift(this.findOptions);
+    ).shift(this.deserializationOptions);
   }
 
   /** Set the cursor query */

src/index.ts

@@ -302,7 +302,11 @@ export type { ClientMetadata, ClientMetadataOptions } from './cmap/handshake/cli
 export type { ConnectionPoolMetrics } from './cmap/metrics';
 export type { StreamDescription, StreamDescriptionOptions } from './cmap/stream_description';
 export type { CompressorName } from './cmap/wire_protocol/compression';
-export type { JSTypeOf, OnDemandDocument } from './cmap/wire_protocol/on_demand/document';
+export type {
+  JSTypeOf,
+  OnDemandDocument,
+  OnDemandDocumentDeserializeOptions
+} from './cmap/wire_protocol/on_demand/document';
 export type {
   CursorResponse,
   MongoDBResponse,