buffer: add isAscii method

anonrig · ruyadorno · commit 5ad6c2088e2f · 2023-01-31T18:16:05.000-05:00
PR-URL: #46046 Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com> Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
diff --git a/doc/api/buffer.md b/doc/api/buffer.md
@@ -5130,6 +5130,20 @@ For code running using Node.js APIs, converting between base64-encoded strings
 and binary data should be performed using `Buffer.from(str, 'base64')` and
 `buf.toString('base64')`.**
 
+### `buffer.isAscii(input)`
+
+<!-- YAML
+added: REPLACEME
+-->
+
+* input {Buffer | ArrayBuffer | TypedArray} The input to validate.
+* Returns: {boolean}
+
+This function returns `true` if `input` contains only valid ASCII-encoded data,
+including the case in which `input` is empty.
+
+Throws if the `input` is a detached array buffer.
+
 ### `buffer.isUtf8(input)`
 
 <!-- YAML
diff --git a/lib/buffer.js b/lib/buffer.js
@@ -56,6 +56,7 @@ const {
   compareOffset,
   createFromString,
   fill: bindingFill,
+  isAscii: bindingIsAscii,
   isUtf8: bindingIsUtf8,
   indexOfBuffer,
   indexOfNumber,
@@ -1323,11 +1324,20 @@ function isUtf8(input) {
   throw new ERR_INVALID_ARG_TYPE('input', ['TypedArray', 'Buffer'], input);
 }
 
+function isAscii(input) {
+  if (isTypedArray(input) || isAnyArrayBuffer(input)) {
+    return bindingIsAscii(input);
+  }
+
+  throw new ERR_INVALID_ARG_TYPE('input', ['ArrayBuffer', 'Buffer', 'TypedArray'], input);
+}
+
 module.exports = {
   Buffer,
   SlowBuffer,
   transcode,
   isUtf8,
+  isAscii,
 
   // Legacy
   kMaxLength,
diff --git a/src/node_buffer.cc b/src/node_buffer.cc
@@ -1238,6 +1238,21 @@ static void IsUtf8(const FunctionCallbackInfo<Value>& args) {
   args.GetReturnValue().Set(simdutf::validate_utf8(abv.data(), abv.length()));
 }
 
+static void IsAscii(const FunctionCallbackInfo<Value>& args) {
+  Environment* env = Environment::GetCurrent(args);
+  CHECK_EQ(args.Length(), 1);
+  CHECK(args[0]->IsTypedArray() || args[0]->IsArrayBuffer() ||
+        args[0]->IsSharedArrayBuffer());
+  ArrayBufferViewContents<char> abv(args[0]);
+
+  if (abv.WasDetached()) {
+    return node::THROW_ERR_INVALID_STATE(
+        env, "Cannot validate on a detached buffer");
+  }
+
+  args.GetReturnValue().Set(simdutf::validate_ascii(abv.data(), abv.length()));
+}
+
 void SetBufferPrototype(const FunctionCallbackInfo<Value>& args) {
   Environment* env = Environment::GetCurrent(args);
 
@@ -1373,6 +1388,7 @@ void Initialize(Local<Object> target,
   SetMethodNoSideEffect(context, target, "encodeUtf8String", EncodeUtf8String);
 
   SetMethodNoSideEffect(context, target, "isUtf8", IsUtf8);
+  SetMethodNoSideEffect(context, target, "isAscii", IsAscii);
 
   target
       ->Set(context,
@@ -1430,6 +1446,7 @@ void RegisterExternalReferences(ExternalReferenceRegistry* registry) {
   registry->Register(EncodeUtf8String);
 
   registry->Register(IsUtf8);
+  registry->Register(IsAscii);
 
   registry->Register(StringSlice<ASCII>);
   registry->Register(StringSlice<BASE64>);
diff --git a/test/parallel/test-buffer-isascii.js b/test/parallel/test-buffer-isascii.js
@@ -0,0 +1,42 @@
+'use strict';
+
+require('../common');
+const assert = require('assert');
+const { isAscii, Buffer } = require('buffer');
+const { TextEncoder } = require('util');
+
+const encoder = new TextEncoder();
+
+assert.strictEqual(isAscii(encoder.encode('hello')), true);
+assert.strictEqual(isAscii(encoder.encode('ğ')), false);
+assert.strictEqual(isAscii(Buffer.from([])), true);
+
+[
+  undefined,
+  '', 'hello',
+  false, true,
+  0, 1,
+  0n, 1n,
+  Symbol(),
+  () => {},
+  {}, [], null,
+].forEach((input) => {
+  assert.throws(
+    () => { isAscii(input); },
+    {
+      code: 'ERR_INVALID_ARG_TYPE',
+    },
+  );
+});
+
+{
+  // Test with detached array buffers
+  const arrayBuffer = new ArrayBuffer(1024);
+  structuredClone(arrayBuffer, { transfer: [arrayBuffer] });
+  assert.throws(
+    () => { isAscii(arrayBuffer); },
+    {
+      code: 'ERR_INVALID_STATE'
+    }
+  );
+}
