refactor(NODE-6325): implement document sequence support (#4201)

src/cmap/commands.ts

@@ -30,6 +30,8 @@ const QUERY_FAILURE = 2;
 const SHARD_CONFIG_STALE = 4;
 const AWAIT_CAPABLE = 8;
 
+const encodeUTF8Into = BSON.BSON.onDemand.ByteUtils.encodeUTF8Into;
+
 /** @internal */
 export type WriteProtocolMessageType = OpQueryRequest | OpMsgRequest;
 
@@ -411,6 +413,15 @@ export interface OpMsgOptions {
   readPreference: ReadPreference;
 }
 
+/** @internal */
+export class DocumentSequence {
+  documents: Document[];
+
+  constructor(documents: Document[]) {
+    this.documents = documents;
+  }
+}
+
 /** @internal */
 export class OpMsgRequest {
   requestId: number;
@@ -480,7 +491,7 @@ export class OpMsgRequest {
 
     let totalLength = header.length;
     const command = this.command;
-    totalLength += this.makeDocumentSegment(buffers, command);
+    totalLength += this.makeSections(buffers, command);
 
     header.writeInt32LE(totalLength, 0); // messageLength
     header.writeInt32LE(this.requestId, 4); // requestID
@@ -490,15 +501,65 @@ export class OpMsgRequest {
     return buffers;
   }
 
-  makeDocumentSegment(buffers: Uint8Array[], document: Document): number {
-    const payloadTypeBuffer = Buffer.alloc(1);
+  /**
+   * Add the sections to the OP_MSG request's buffers and returns the length.
+   */
+  makeSections(buffers: Uint8Array[], document: Document): number {
+    const sequencesBuffer = this.extractDocumentSequences(document);
+    const payloadTypeBuffer = Buffer.allocUnsafe(1);
     payloadTypeBuffer[0] = 0;
 
     const documentBuffer = this.serializeBson(document);
+    // First section, type 0
     buffers.push(payloadTypeBuffer);
     buffers.push(documentBuffer);
+    // Subsequent sections, type 1
+    buffers.push(sequencesBuffer);
 
-    return payloadTypeBuffer.length + documentBuffer.length;
+    return payloadTypeBuffer.length + documentBuffer.length + sequencesBuffer.length;
+  }
+
+  /**
+   * Extracts the document sequences from the command document and returns
+   * a buffer to be added as multiple sections after the initial type 0
+   * section in the message.
+   */
+  extractDocumentSequences(document: Document): Uint8Array {
+    // Pull out any field in the command document that's value is a document sequence.
+    const chunks = [];
+    for (const [key, value] of Object.entries(document)) {
+      if (value instanceof DocumentSequence) {
+        // Document sequences starts with type 1 at the first byte.
+        const buffer = Buffer.allocUnsafe(1 + 4 + key.length);
+        buffer[0] = 1;
+        // Third part is the field name at offset 5.
+        encodeUTF8Into(buffer, key, 5);
+        chunks.push(buffer);
+        // Fourth part are the documents' bytes.
+        let docsLength = 0;
+        for (const doc of value.documents) {
+          const docBson = this.serializeBson(doc);
+          docsLength += docBson.length;
+          chunks.push(docBson);
+        }
+        // Second part of the sequence is the length at offset 1;
+        buffer.writeInt32LE(key.length + docsLength, 1);
+        // Why are we removing the field from the command? This is because it needs to be
+        // removed in the OP_MSG request first section, and DocumentSequence is not a
+        // BSON type and is specific to the MongoDB wire protocol so there's nothing
+        // our BSON serializer can do about this. Since DocumentSequence is not exposed
+        // in the public API and only used internally, we are never mutating an original
+        // command provided by the user, just our own, and it's cheaper to delete from
+        // our own command than copying it.
+        delete document[key];
+      }
+    }
+    if (chunks.length > 0) {
+      return Buffer.concat(chunks);
+    }
+    // If we have no document sequences we return an empty buffer for nothing to add
+    // to the payload.
+    return Buffer.alloc(0);
   }
 
   serializeBson(document: Document): Uint8Array {