util: add getStringWidth()

Expose the string width helper that readline, the REPL and the test
runner already use. It returns the number of terminal columns a string
occupies: full-width characters count as two, zero-width characters as
zero, and ANSI escape sequences are ignored. The API is experimental.

Refs: #40214
Signed-off-by: marcopiraccini <marco.piraccini@gmail.com>

Author: marcopiraccini

doc/api/util.md

@@ -946,6 +946,41 @@ const callSites = getCallSites({ sourceMap: true });
 // Column Number: 26
 ```
 
+## `util.getStringWidth(str)`
+
+<!-- YAML
+added: REPLACEME
+-->
+
+> Stability: 1 - Experimental
+
+* `str` {string}
+* Returns: {integer} An estimate of the number of columns needed to display
+  `str`.
+
+Returns an estimate of the width of `str` as displayed in a terminal, counted
+in columns. Full-width characters, such as CJK ideographs and most emoji, count
+as two columns. Zero-width characters, such as combining marks, control
+characters and zero-width joiners, count as zero. ANSI escape sequences, as
+produced by [`util.styleText()`][], are ignored.
+
+```js
+console.log(util.getStringWidth('hello'));
+// Prints: 5
+console.log(util.getStringWidth('你好'));
+// Prints: 4
+console.log(util.getStringWidth(util.styleText('red', 'hi')));
+// Prints: 2
+```
+
+The result is an estimate because terminals differ in how they render some
+characters. In particular, the emoji of a joined sequence are counted
+separately, so a sequence that a terminal renders as a single glyph can be
+overcounted, with or without ICU: `'\u{1F469}\u200D\u{1F469}\u200D\u{1F467}'`
+is counted as 6 columns. When Node.js is built without ICU, a simpler table of
+full-width and zero-width code points is used, which is less accurate for less
+common scripts.
+
 ## `util.getSystemErrorName(err)`
 
 <!-- YAML
@@ -4176,6 +4211,7 @@ npx codemod@latest @nodejs/util-is
 [`util.format()`]: #utilformatformat-args
 [`util.inspect()`]: #utilinspectobject-options
 [`util.promisify()`]: #utilpromisifyoriginal
+[`util.styleText()`]: #utilstyletextformat-text-options
 [`util.types.isAnyArrayBuffer()`]: #utiltypesisanyarraybuffervalue
 [`util.types.isArrayBuffer()`]: #utiltypesisarraybuffervalue
 [`util.types.isSharedArrayBuffer()`]: #utiltypesissharedarraybuffervalue

lib/util.js

@@ -58,6 +58,7 @@ const { Buffer } = require('buffer');
 const {
   format,
   formatWithOptions,
+  getStringWidth: internalGetStringWidth,
   inspect,
   stripVTControlCharacters,
 } = require('internal/util/inspect');
@@ -236,6 +237,16 @@ function rgbToAnsi24Bit(r, g, b) {
   return `38;2;${r};${g};${b}`;
 }
 
+/**
+ * Returns the number of columns needed to display `str` in a terminal.
+ * @param {string} str
+ * @returns {number}
+ */
+function getStringWidth(str) {
+  validateString(str, 'str');
+  return internalGetStringWidth(str);
+}
+
 /**
  * @param {string | string[]} format
  * @param {string} text
@@ -612,6 +623,7 @@ module.exports = {
   styleText,
   formatWithOptions,
   getCallSites,
+  getStringWidth,
   getSystemErrorMap,
   getSystemErrorName,
   getSystemErrorMessage,

test/parallel/test-util-get-string-width.js

@@ -0,0 +1,61 @@
+'use strict';
+
+const common = require('../common');
+const assert = require('assert');
+const util = require('util');
+
+const { getStringWidth } = util;
+
+assert.strictEqual(getStringWidth(''), 0);
+assert.strictEqual(getStringWidth('a'), 1);
+assert.strictEqual(getStringWidth('hello'), 5);
+assert.strictEqual(getStringWidth('a你b'), 4);
+
+// Full-width characters count as two columns.
+assert.strictEqual(getStringWidth('丁'), 2);
+assert.strictEqual(getStringWidth('你好'), 4);
+assert.strictEqual(getStringWidth('가'), 2);
+assert.strictEqual(getStringWidth('👅'), 2);
+assert.strictEqual(getStringWidth('Ｆ'), 2);  // Fullwidth Latin letter.
+assert.strictEqual(getStringWidth('ｱ'), 1);  // Halfwidth katakana.
+assert.strictEqual(getStringWidth('\u{1D49C}'), 1);  // Astral, not wide.
+assert.strictEqual(getStringWidth('\u{20000}'), 2);  // CJK Extension B.
+
+// Zero-width characters count as zero.
+assert.strictEqual(getStringWidth('\u0000'), 0);
+assert.strictEqual(getStringWidth('\u0007'), 0);
+assert.strictEqual(getStringWidth('\n'), 0);
+assert.strictEqual(getStringWidth('\r\n'), 0);
+assert.strictEqual(getStringWidth('\t'), 0);
+assert.strictEqual(getStringWidth('́'), 0);
+assert.strictEqual(getStringWidth('é'), 1);
+assert.strictEqual(getStringWidth('​'), 0);
+assert.strictEqual(getStringWidth('a‏b'), 2);  // Right-to-left mark.
+
+// ANSI escape sequences are ignored.
+assert.strictEqual(getStringWidth('\u001B[31mred\u001B[39m'), 3);
+assert.strictEqual(getStringWidth(util.styleText('bold', 'hi')), 2);
+assert.strictEqual(getStringWidth(util.styleText(['bold', 'red'], 'ok')), 2);
+assert.strictEqual(
+  getStringWidth('\u001B]8;;https://nodejs.org\u0007text\u001B]8;;\u0007'), 4);
+
+// Lone surrogates take one column.
+assert.strictEqual(getStringWidth('\uD83D'), 1);
+
+if (common.hasIntl) {
+  assert.strictEqual(getStringWidth(' '), 1);
+  assert.strictEqual(getStringWidth('กิ'), 1);  // Thai with a vowel mark.
+  // Joined sequences count each emoji; a terminal may render them narrower.
+  assert.strictEqual(getStringWidth('\u{1F469}‍\u{1F469}‍\u{1F467}‍\u{1F467}'), 8);
+  assert.strictEqual(getStringWidth('\u{1F44D}\u{1F3FD}'), 4);  // Skin tone modifier.
+  assert.strictEqual(getStringWidth('\u{1F1EE}\u{1F1F9}'), 4);  // Flag.
+  assert.strictEqual(getStringWidth('❤️'), 1);
+  assert.strictEqual(getStringWidth('1️⃣'), 1);  // Keycap.
+}
+
+for (const value of [undefined, null, 1, true, {}, [], Symbol('s')]) {
+  assert.throws(() => getStringWidth(value), {
+    code: 'ERR_INVALID_ARG_TYPE',
+    name: 'TypeError',
+  });
+}