feat(NODE-3784): Add enableUtf8Validation option (#3074)

durran · baileympearson · web-flow · commit 4f56409e5dde · 2021-12-13T16:07:10.000-05:00
Co-authored-by: Bailey Pearson &lt;bailey.pearson@mongodb.com&gt;
diff --git a/src/bson.ts b/src/bson.ts
@@ -55,6 +55,9 @@ export interface BSONSerializeOptions
     > {
   /** Return BSON filled buffers from operations */
   raw?: boolean;
+
+  /** Enable utf8 validation when deserializing BSON documents.  Defaults to true. */
+  enableUtf8Validation?: boolean;
 }
 
 export function pluckBSONSerializeOptions(options: BSONSerializeOptions): BSONSerializeOptions {
@@ -66,7 +69,8 @@ export function pluckBSONSerializeOptions(options: BSONSerializeOptions): BSONSe
     serializeFunctions,
     ignoreUndefined,
     bsonRegExp,
-    raw
+    raw,
+    enableUtf8Validation
   } = options;
   return {
     fieldsAsRaw,
@@ -76,7 +80,8 @@ export function pluckBSONSerializeOptions(options: BSONSerializeOptions): BSONSe
     serializeFunctions,
     ignoreUndefined,
     bsonRegExp,
-    raw
+    raw,
+    enableUtf8Validation
   };
 }
 
@@ -99,6 +104,8 @@ export function resolveBSONOptions(
     ignoreUndefined: options?.ignoreUndefined ?? parentOptions?.ignoreUndefined ?? false,
     bsonRegExp: options?.bsonRegExp ?? parentOptions?.bsonRegExp ?? false,
     serializeFunctions: options?.serializeFunctions ?? parentOptions?.serializeFunctions ?? false,
-    fieldsAsRaw: options?.fieldsAsRaw ?? parentOptions?.fieldsAsRaw ?? {}
+    fieldsAsRaw: options?.fieldsAsRaw ?? parentOptions?.fieldsAsRaw ?? {},
+    enableUtf8Validation:
+      options?.enableUtf8Validation ?? parentOptions?.enableUtf8Validation ?? true
   };
 }
diff --git a/src/cmap/commands.ts b/src/cmap/commands.ts
@@ -469,8 +469,6 @@ export interface MessageHeader {
 export interface OpResponseOptions extends BSONSerializeOptions {
   raw?: boolean;
   documentsReturnedIn?: string | null;
-  // For now we use this internally to only prevent writeErrors from crashing the driver
-  validation?: { utf8: { writeErrors: boolean } };
 }
 
 /** @internal */
@@ -839,7 +837,7 @@ export class BinMsg {
     const promoteValues = options.promoteValues ?? this.opts.promoteValues;
     const promoteBuffers = options.promoteBuffers ?? this.opts.promoteBuffers;
     const bsonRegExp = options.bsonRegExp ?? this.opts.bsonRegExp;
-    const validation = options.validation ?? { utf8: { writeErrors: false } };
+    const validation = this.parseBsonSerializationOptions(options);
 
     // Set up the options
     const bsonOptions: BSONSerializeOptions = {
@@ -876,4 +874,14 @@ export class BinMsg {
 
     this.parsed = true;
   }
+
+  parseBsonSerializationOptions({ enableUtf8Validation }: BSONSerializeOptions): {
+    utf8: { writeErrors: false } | false;
+  } {
+    if (enableUtf8Validation === false) {
+      return { utf8: false };
+    }
+
+    return { utf8: { writeErrors: false } };
+  }
 }
diff --git a/src/cmap/connection.ts b/src/cmap/connection.ts
@@ -787,6 +787,8 @@ function write(
     promoteValues: typeof options.promoteValues === 'boolean' ? options.promoteValues : true,
     promoteBuffers: typeof options.promoteBuffers === 'boolean' ? options.promoteBuffers : false,
     bsonRegExp: typeof options.bsonRegExp === 'boolean' ? options.bsonRegExp : false,
+    enableUtf8Validation:
+      typeof options.enableUtf8Validation === 'boolean' ? options.enableUtf8Validation : true,
     raw: typeof options.raw === 'boolean' ? options.raw : false,
     started: 0
   };
diff --git a/src/cmap/wire_protocol/shared.ts b/src/cmap/wire_protocol/shared.ts
@@ -45,7 +45,9 @@ export function applyCommonQueryOptions(
     promoteLongs: typeof options.promoteLongs === 'boolean' ? options.promoteLongs : true,
     promoteValues: typeof options.promoteValues === 'boolean' ? options.promoteValues : true,
     promoteBuffers: typeof options.promoteBuffers === 'boolean' ? options.promoteBuffers : false,
-    bsonRegExp: typeof options.bsonRegExp === 'boolean' ? options.bsonRegExp : false
+    bsonRegExp: typeof options.bsonRegExp === 'boolean' ? options.bsonRegExp : false,
+    enableUtf8Validation:
+      typeof options.enableUtf8Validation === 'boolean' ? options.enableUtf8Validation : true
   });
 
   if (options.session) {
diff --git a/src/connection_string.ts b/src/connection_string.ts
@@ -688,6 +688,7 @@ export const OPTIONS = {
       });
     }
   },
+  enableUtf8Validation: { type: 'boolean', default: true },
   family: {
     transform({ name, values: [value] }): 4 | 6 {
       const transformValue = getInt(name, value);
diff --git a/src/db.ts b/src/db.ts
@@ -73,6 +73,7 @@ const DB_OPTIONS_ALLOW_LIST = [
   'promoteBuffers',
   'promoteLongs',
   'bsonRegExp',
+  'enableUtf8Validation',
   'promoteValues',
   'compression',
   'retryWrites'
diff --git a/src/operations/create_collection.ts b/src/operations/create_collection.ts
@@ -27,7 +27,8 @@ const ILLEGAL_COMMAND_FIELDS = new Set([
   'promoteBuffers',
   'bsonRegExp',
   'serializeFunctions',
-  'ignoreUndefined'
+  'ignoreUndefined',
+  'enableUtf8Validation'
 ]);
 
 /** @public
diff --git a/src/operations/map_reduce.ts b/src/operations/map_reduce.ts
@@ -33,6 +33,7 @@ const exclusionList = [
   'bsonRegExp',
   'serializeFunctions',
   'ignoreUndefined',
+  'enableUtf8Validation',
   'scope' // this option is reformatted thus exclude the original
 ];
 
diff --git a/test/functional/cmap/commands.test.ts b/test/functional/cmap/commands.test.ts
@@ -0,0 +1,122 @@
+import { spy } from 'sinon';
+import { expect } from 'chai';
+import * as BSON from '../../../src/bson';
+
+const deserializeSpy = spy(BSON, 'deserialize');
+
+const EXPECTED_VALIDATION_DISABLED_ARGUMENT = {
+  utf8: false
+};
+
+const EXPECTED_VALIDATION_ENABLED_ARGUMENT = {
+  utf8: {
+    writeErrors: false
+  }
+};
+
+describe('class BinMsg', () => {
+  beforeEach(() => {
+    deserializeSpy.resetHistory();
+  });
+
+  describe('enableUtf8Validation option set to false', () => {
+    let client;
+    const option = { enableUtf8Validation: false };
+
+    for (const passOptionTo of ['client', 'db', 'collection', 'operation']) {
+      it(`should disable validation with option passed to ${passOptionTo}`, async function () {
+        try {
+          client = this.configuration.newClient(passOptionTo === 'client' ? option : undefined);
+          await client.connect();
+
+          const db = client.db(
+            'bson_utf8Validation_db',
+            passOptionTo === 'db' ? option : undefined
+          );
+          const collection = db.collection(
+            'bson_utf8Validation_coll',
+            passOptionTo === 'collection' ? option : undefined
+          );
+
+          await collection.insertOne(
+            { name: 'John Doe' },
+            passOptionTo === 'operation' ? option : {}
+          );
+
+          expect(deserializeSpy.called).to.be.true;
+          const validationArgument = deserializeSpy.lastCall.lastArg.validation;
+          expect(validationArgument).to.deep.equal(EXPECTED_VALIDATION_DISABLED_ARGUMENT);
+        } finally {
+          await client.close();
+        }
+      });
+    }
+  });
+
+  describe('enableUtf8Validation option set to true', () => {
+    // define client and option for tests to use
+    let client;
+    const option = { enableUtf8Validation: true };
+    for (const passOptionTo of ['client', 'db', 'collection', 'operation']) {
+      it(`should enable validation with option passed to ${passOptionTo}`, async function () {
+        try {
+          client = this.configuration.newClient(passOptionTo === 'client' ? option : undefined);
+          await client.connect();
+
+          const db = client.db(
+            'bson_utf8Validation_db',
+            passOptionTo === 'db' ? option : undefined
+          );
+          const collection = db.collection(
+            'bson_utf8Validation_coll',
+            passOptionTo === 'collection' ? option : undefined
+          );
+
+          await collection.insertOne(
+            { name: 'John Doe' },
+            passOptionTo === 'operation' ? option : {}
+          );
+
+          expect(deserializeSpy.called).to.be.true;
+          const validationArgument = deserializeSpy.lastCall.lastArg.validation;
+          expect(validationArgument).to.deep.equal(EXPECTED_VALIDATION_ENABLED_ARGUMENT);
+        } finally {
+          await client.close();
+        }
+      });
+    }
+  });
+
+  describe('enableUtf8Validation option not set', () => {
+    let client;
+    const option = { enableUtf8Validation: true };
+    for (const passOptionTo of ['client', 'db', 'collection', 'operation']) {
+      it(`should default to enabled with option passed to ${passOptionTo}`, async function () {
+        try {
+          client = this.configuration.newClient(passOptionTo === 'client' ? option : undefined);
+          await client.connect();
+
+          const db = client.db(
+            'bson_utf8Validation_db',
+            passOptionTo === 'db' ? option : undefined
+          );
+          const collection = db.collection(
+            'bson_utf8Validation_coll',
+            passOptionTo === 'collection' ? option : undefined
+          );
+
+          await collection.insertOne(
+            { name: 'John Doe' },
+            passOptionTo === 'operation' ? option : {}
+          );
+
+          expect(deserializeSpy.called).to.be.true;
+          const validationArgument = deserializeSpy.lastCall.lastArg.validation;
+          expect(validationArgument).to.deep.equal(EXPECTED_VALIDATION_ENABLED_ARGUMENT);
+        } finally {
+          await client.close();
+        }
+      });
+    }
+  });
+});
diff --git a/test/types/bson.test-d.ts b/test/types/bson.test-d.ts
@@ -21,6 +21,7 @@ type PermittedBSONOptionKeys =
   | 'promoteValues'
   | 'bsonRegExp'
   | 'fieldsAsRaw'
+  | 'enableUtf8Validation'
   | 'raw';
 
 const keys = null as unknown as PermittedBSONOptionKeys;
diff --git a/test/unit/commands.test.ts b/test/unit/commands.test.ts

Original file line number	Diff line number	Diff line change
`@@ -688,6 +688,7 @@ export const OPTIONS = {`
`688`	`688`	`});`
`689`	`689`	`}`
`690`	`690`	`},`
	`691`	`+ enableUtf8Validation: { type: 'boolean', default: true },`
`691`	`692`	`family: {`
`692`	`693`	`transform({ name, values: [value] }): 4 \| 6 {`
`693`	`694`	`const transformValue = getInt(name, value);`