feat(document): add schemaFieldsOnly option to toObject() and toJSON()

Fix Automattic#15218

Author: vkarpov15

lib/document.js

@@ -3834,15 +3834,39 @@ Document.prototype.$toObject = function(options, json) {
   // Parent options should only bubble down for subdocuments, not populated docs
   options._parentOptions = this.$isSubdocument ? options : null;
 
-  // remember the root transform function
-  // to save it from being overwritten by sub-transform functions
-  // const originalTransform = options.transform;
+  const schemaFieldsOnly = options._calledWithOptions.schemaFieldsOnly
+    ?? options.schemaFieldsOnly
+    ?? defaultOptions.schemaFieldsOnly
+    ?? false;
 
   let ret;
   if (hasOnlyPrimitiveValues && !options.flattenObjectIds) {
     // Fast path: if we don't have any nested objects or arrays, we only need a
     // shallow clone.
-    ret = this.$__toObjectShallow();
+    ret = this.$__toObjectShallow(schemaFieldsOnly);
+  } else if (schemaFieldsOnly) {
+    ret = {};
+    for (const path of Object.keys(this.$__schema.paths)) {
+      const value = this.$__getValue(path);
+      if (value === undefined) {
+        continue;
+      }
+      let pathToSet = path;
+      let objToSet = ret;
+      if (path.indexOf('.') !== -1) {
+        const segments = path.split('.');
+        pathToSet = segments[segments.length - 1];
+        for (let i = 0; i < segments.length - 1; ++i) {
+          objToSet[segments[i]] = objToSet[segments[i]] ?? {};
+          objToSet = objToSet[segments[i]];
+        }
+      }
+      if (value === null) {
+        objToSet[pathToSet] = null;
+        continue;
+      }
+      objToSet[pathToSet] = clone(value, options);
+    }
   } else {
     ret = clone(this._doc, options) || {};
   }
@@ -3908,10 +3932,12 @@ Document.prototype.$toObject = function(options, json) {
  * Internal shallow clone alternative to `$toObject()`: much faster, no options processing
  */
 
-Document.prototype.$__toObjectShallow = function $__toObjectShallow() {
+Document.prototype.$__toObjectShallow = function $__toObjectShallow(schemaFieldsOnly) {
   const ret = {};
   if (this._doc != null) {
-    for (const key of Object.keys(this._doc)) {
+    const keys = schemaFieldsOnly ? Object.keys(this.$__schema.paths) : Object.keys(this._doc);
+    for (const key of keys) {
+      // Safe to do this even in the schemaFieldsOnly case because we assume there's no nested paths
       const value = this._doc[key];
       if (value instanceof Date) {
         ret[key] = new Date(value);
@@ -4064,6 +4090,7 @@ Document.prototype.$__toObjectShallow = function $__toObjectShallow() {
  * @param {Boolean} [options.flattenMaps=false] if true, convert Maps to POJOs. Useful if you want to `JSON.stringify()` the result of `toObject()`.
  * @param {Boolean} [options.flattenObjectIds=false] if true, convert any ObjectIds in the result to 24 character hex strings.
  * @param {Boolean} [options.useProjection=false] - If true, omits fields that are excluded in this document's projection. Unless you specified a projection, this will omit any field that has `select: false` in the schema.
+ * @param {Boolean} [options.schemaFieldsOnly=false] - If true, the resulting object will only have fields that are defined in the document's schema. By default, `toObject()` returns all fields in the underlying document from MongoDB, including ones that are not listed in the schema.
  * @return {Object} document as a plain old JavaScript object (POJO). This object may contain ObjectIds, Maps, Dates, mongodb.Binary, Buffers, and other non-POJO values.
  * @see mongodb.Binary https://mongodb.github.io/node-mongodb-native/4.9/classes/Binary.html
  * @api public
@@ -4334,6 +4361,7 @@ function omitDeselectedFields(self, json) {
  * @param {Object} options
  * @param {Boolean} [options.flattenMaps=true] if true, convert Maps to [POJOs](https://masteringjs.io/tutorials/fundamentals/pojo). Useful if you want to `JSON.stringify()` the result.
  * @param {Boolean} [options.flattenObjectIds=false] if true, convert any ObjectIds in the result to 24 character hex strings.
+ * @param {Boolean} [options.schemaFieldsOnly=false] - If true, the resulting object will only have fields that are defined in the document's schema. By default, `toJSON()` returns all fields in the underlying document from MongoDB, including ones that are not listed in the schema.
  * @return {Object}
  * @see Document#toObject https://mongoosejs.com/docs/api/document.html#Document.prototype.toObject()
  * @see JSON.stringify() in JavaScript https://thecodebarbarian.com/the-80-20-guide-to-json-stringify-in-javascript.html

test/document.test.js

@@ -14296,6 +14296,93 @@ describe('document', function() {
 
     delete mongoose.Schema.Types.CustomType;
   });
+
+  it('supports schemaFieldsOnly option for toObject() (gh-15258)', async function() {
+    const schema = new Schema({ key: String }, { discriminatorKey: 'key' });
+    const subschema1 = new Schema({ field1: String });
+    const subschema2 = new Schema({ field2: String });
+
+    const Discriminator = db.model('Test', schema);
+    Discriminator.discriminator('type1', subschema1);
+    Discriminator.discriminator('type2', subschema2);
+
+    const type1Key = 'type1';
+    const type2Key = 'type2';
+
+    const doc = await Discriminator.create({
+      key: type1Key,
+      field1: 'test value'
+    });
+
+    await Discriminator.updateOne(
+      { _id: doc._id },
+      {
+        key: type2Key,
+        field2: 'test2'
+      },
+      { overwriteDiscriminatorKey: true }
+    );
+
+    const doc2 = await Discriminator.findById(doc).orFail();
+    assert.strictEqual(doc2.field1, undefined);
+    assert.strictEqual(doc2.field2, 'test2');
+
+    const obj = doc2.toObject();
+    assert.strictEqual(obj.field2, 'test2');
+    assert.strictEqual(obj.field1, 'test value');
+
+    const obj2 = doc2.toObject({ schemaFieldsOnly: true });
+    assert.strictEqual(obj.field2, 'test2');
+    assert.strictEqual(obj2.field1, undefined);
+  });
+
+  it('supports schemaFieldsOnly on nested paths, subdocuments, and arrays (gh-15258)', async function() {
+    const subSchema = new Schema({
+      title: String,
+      description: String
+    }, { _id: false });
+    const taskSchema = new Schema({
+      name: String,
+      details: {
+        dueDate: Date,
+        priority: Number
+      },
+      subtask: subSchema,
+      tasks: [subSchema]
+    });
+    const Task = db.model('Test', taskSchema);
+
+    const doc = await Task.create({
+      _id: '0'.repeat(24),
+      name: 'Test Task',
+      details: {
+        dueDate: new Date('2024-01-01'),
+        priority: 1
+      },
+      subtask: {
+        title: 'Subtask 1',
+        description: 'Test Description'
+      },
+      tasks: [{
+        title: 'Array Task 1',
+        description: 'Array Description 1'
+      }]
+    });
+
+    doc._doc.details.extraField = 'extra';
+    doc._doc.subtask.extraField = 'extra';
+    doc._doc.tasks[0].extraField = 'extra';
+
+    const obj = doc.toObject({ schemaFieldsOnly: true });
+    assert.deepStrictEqual(obj, {
+      name: 'Test Task',
+      details: { dueDate: new Date('2024-01-01T00:00:00.000Z'), priority: 1 },
+      subtask: { title: 'Subtask 1', description: 'Test Description' },
+      tasks: [{ title: 'Array Task 1', description: 'Array Description 1' }],
+      _id: new mongoose.Types.ObjectId('0'.repeat(24)),
+      __v: 0
+    });
+  });
 });
 
 describe('Check if instance function that is supplied in schema option is available', function() {

types/index.d.ts

@@ -204,30 +204,32 @@ declare module 'mongoose' {
   }
 
   export interface ToObjectOptions<THydratedDocumentType = HydratedDocument<unknown>> {
-    /** apply all getters (path and virtual getters) */
-    getters?: boolean;
-    /** apply virtual getters (can override getters option) */
-    virtuals?: boolean | string[];
     /** if `options.virtuals = true`, you can set `options.aliases = false` to skip applying aliases. This option is a no-op if `options.virtuals = false`. */
     aliases?: boolean;
+    /** if true, replace any conventionally populated paths with the original id in the output. Has no affect on virtual populated paths. */
+    depopulate?: boolean;
+    /** if true, convert Maps to POJOs. Useful if you want to `JSON.stringify()` the result of `toObject()`. */
+    flattenMaps?: boolean;
+    /** if true, convert any ObjectIds in the result to 24 character hex strings. */
+    flattenObjectIds?: boolean;
+    /** apply all getters (path and virtual getters) */
+    getters?: boolean;
     /** remove empty objects (defaults to true) */
     minimize?: boolean;
+    /** If true, the resulting object will only have fields that are defined in the document's schema. By default, `toJSON()` returns all fields in the underlying document from MongoDB, including ones that are not listed in the schema. */
+    schemaFieldsOnly?: boolean;
     /** if set, mongoose will call this function to allow you to transform the returned object */
     transform?: boolean | ((
       doc: THydratedDocumentType,
       ret: Record<string, any>,
       options: ToObjectOptions<THydratedDocumentType>
     ) => any);
-    /** if true, replace any conventionally populated paths with the original id in the output. Has no affect on virtual populated paths. */
-    depopulate?: boolean;
-    /** if false, exclude the version key (`__v` by default) from the output */
-    versionKey?: boolean;
-    /** if true, convert Maps to POJOs. Useful if you want to `JSON.stringify()` the result of `toObject()`. */
-    flattenMaps?: boolean;
-    /** if true, convert any ObjectIds in the result to 24 character hex strings. */
-    flattenObjectIds?: boolean;
     /** If true, omits fields that are excluded in this document's projection. Unless you specified a projection, this will omit any field that has `select: false` in the schema. */
     useProjection?: boolean;
+    /** if false, exclude the version key (`__v` by default) from the output */
+    versionKey?: boolean;
+    /** apply virtual getters (can override getters option) */
+    virtuals?: boolean | string[];
   }
 
   export type DiscriminatorModel<M, T> = T extends Model<infer T, infer TQueryHelpers, infer TInstanceMethods, infer TVirtuals>