perf: don't use grapheme-splitter on ASCII strings in key-spacing r…

…ule (#17122)

* perf: don't use `grapheme-splitter` on ASCII strings in key-spacing rule

* extract to string-utils

* remove extra blank line

lib/rules/id-length.js

@@ -9,41 +9,8 @@
 //------------------------------------------------------------------------------
 // Requirements
 //------------------------------------------------------------------------------
-const GraphemeSplitter = require("grapheme-splitter");
 
-//------------------------------------------------------------------------------
-// Helpers
-//------------------------------------------------------------------------------
-
-/**
- * Checks if the string given as argument is ASCII or not.
- * @param {string} value A string that you want to know if it is ASCII or not.
- * @returns {boolean} `true` if `value` is ASCII string.
- */
-function isASCII(value) {
-    if (typeof value !== "string") {
-        return false;
-    }
-    return /^[\u0020-\u007f]*$/u.test(value);
-}
-
-/** @type {GraphemeSplitter | undefined} */
-let splitter;
-
-/**
- * Gets the length of the string. If the string is not in ASCII, counts graphemes.
- * @param {string} value A string that you want to get the length.
- * @returns {number} The length of `value`.
- */
-function getStringLength(value) {
-    if (isASCII(value)) {
-        return value.length;
-    }
-    if (!splitter) {
-        splitter = new GraphemeSplitter();
-    }
-    return splitter.countGraphemes(value);
-}
+const { getGraphemeCount } = require("../shared/string-utils");
 
 //------------------------------------------------------------------------------
 // Rule Definition
@@ -169,7 +136,7 @@ module.exports = {
                 const name = node.name;
                 const parent = node.parent;
 
-                const nameLength = getStringLength(name);
+                const nameLength = getGraphemeCount(name);
 
                 const isShort = nameLength < minLength;
                 const isLong = nameLength > maxLength;

lib/rules/key-spacing.js

@@ -9,13 +9,7 @@
 //------------------------------------------------------------------------------
 
 const astUtils = require("./utils/ast-utils");
-const GraphemeSplitter = require("grapheme-splitter");
-
-const splitter = new GraphemeSplitter();
-
-//------------------------------------------------------------------------------
-// Helpers
-//------------------------------------------------------------------------------
+const { getGraphemeCount } = require("../shared/string-utils");
 
 /**
  * Checks whether a string contains a line terminator as defined in
@@ -523,7 +517,7 @@ module.exports = {
             const startToken = sourceCode.getFirstToken(property);
             const endToken = getLastTokenBeforeColon(property.key);
 
-            return splitter.countGraphemes(sourceCode.getText().slice(startToken.range[0], endToken.range[1]));
+            return getGraphemeCount(sourceCode.getText().slice(startToken.range[0], endToken.range[1]));
         }
 
         /**

lib/shared/string-utils.js

@@ -5,6 +5,26 @@
 
 "use strict";
 
+//------------------------------------------------------------------------------
+// Requirements
+//------------------------------------------------------------------------------
+
+const GraphemeSplitter = require("grapheme-splitter");
+
+//------------------------------------------------------------------------------
+// Helpers
+//------------------------------------------------------------------------------
+
+// eslint-disable-next-line no-control-regex -- intentionally including control characters
+const ASCII_REGEX = /^[\u0000-\u007f]*$/u;
+
+/** @type {GraphemeSplitter | undefined} */
+let splitter;
+
+//------------------------------------------------------------------------------
+// Public Interface
+//------------------------------------------------------------------------------
+
 /**
  * Converts the first letter of a string to uppercase.
  * @param {string} string The string to operate on
@@ -17,6 +37,24 @@ function upperCaseFirst(string) {
     return string[0].toUpperCase() + string.slice(1);
 }
 
+/**
+ * Counts graphemes in a given string.
+ * @param {string} value A string to count graphemes.
+ * @returns {number} The number of graphemes in `value`.
+ */
+function getGraphemeCount(value) {
+    if (ASCII_REGEX.test(value)) {
+        return value.length;
+    }
+
+    if (!splitter) {
+        splitter = new GraphemeSplitter();
+    }
+
+    return splitter.countGraphemes(value);
+}
+
 module.exports = {
-    upperCaseFirst
+    upperCaseFirst,
+    getGraphemeCount
 };

tests/lib/shared/string-utils.js

@@ -11,7 +11,23 @@
 
 const assert = require("chai").assert;
 
-const { upperCaseFirst } = require("../../../lib/shared/string-utils");
+const { upperCaseFirst, getGraphemeCount } = require("../../../lib/shared/string-utils");
+
+//------------------------------------------------------------------------------
+// Helpers
+//------------------------------------------------------------------------------
+
+/**
+ * Replaces raw control characters with the `\xXX` form.
+ * @param {string} text The text to process.
+ * @returns {string} `text` with escaped control characters.
+ */
+function escapeControlCharacters(text) {
+    return text.replace(
+        /[\u0000-\u001F\u007F-\u009F]/gu, // eslint-disable-line no-control-regex -- intentionally including control characters
+        c => `\\x${c.codePointAt(0).toString(16).padStart(2, "0")}`
+    );
+}
 
 //------------------------------------------------------------------------------
 // Tests
@@ -39,3 +55,34 @@ describe("upperCaseFirst", () => {
         assert(upperCaseFirst("") === "");
     });
 });
+
+describe("getGraphemeCount", () => {
+    /* eslint-disable quote-props -- Make consistent here for readability */
+    const expectedResults = {
+        "": 0,
+        "a": 1,
+        "ab": 2,
+        "aa": 2,
+        "123": 3,
+        "cccc": 4,
+        [Array.from({ length: 128 }, (_, i) => String.fromCharCode(i)).join("")]: 128, // all ASCII characters
+        "👍": 1, // 1 grapheme, 1 code point, 2 code units
+        "👍👍": 2,
+        "👍9👍": 3,
+        "a👍b": 3,
+        "👶🏽": 1, // 1 grapheme, 2 code points, 4 code units
+        "👨‍👩‍👦": 1, // 1 grapheme, 5 code points, 8 code units
+        "👨‍👩‍👦👨‍👩‍👦": 2,
+        "👨‍👩‍👦a👨‍👩‍👦": 3,
+        "a👨‍👩‍👦b👨‍👩‍👦c": 5,
+        "👨‍👩‍👦👍": 2,
+        "👶🏽👨‍👩‍👦": 2
+    };
+    /* eslint-enable quote-props -- Make consistent here for readability */
+
+    Object.entries(expectedResults).forEach(([key, value]) => {
+        it(`should return ${value} for ${escapeControlCharacters(key)}`, () => {
+            assert.strictEqual(getGraphemeCount(key), value);
+        });
+    });
+});