Faster readRegexp (#13453)

packages/babel-parser/benchmark/large-regexp/bench.mjs

@@ -0,0 +1,22 @@
+import Benchmark from "benchmark";
+import baseline from "@babel-baseline/parser";
+import current from "../../lib/index.js";
+import { report } from "../util.mjs";
+
+const suite = new Benchmark.Suite();
+function createInput(length) {
+  return "const a = /" + "[/\\\\]".repeat(length / 4) + "/igsudm";
+}
+function benchCases(name, implementation, options) {
+  for (const length of [256, 512, 1024, 2048]) {
+    const input = createInput(length);
+    suite.add(`${name} ${length}-size RegExp literal `, () => {
+      implementation.parse(input, options);
+    });
+  }
+}
+
+benchCases("baseline", baseline);
+benchCases("current", current);
+
+suite.on("cycle", report).run();

packages/babel-parser/benchmark/many-small-all-flags-regexp/bench.mjs

@@ -0,0 +1,22 @@
+import Benchmark from "benchmark";
+import baseline from "../../lib/index-v2.js";
+import current from "../../lib/index.js";
+import { report } from "../util.mjs";
+
+const suite = new Benchmark.Suite();
+function createInput(length) {
+  return "/x/dgimsuy;".repeat(length);
+}
+function benchCases(name, implementation, options) {
+  for (const length of [256, 512, 1024, 2048]) {
+    const input = createInput(length);
+    suite.add(`${name} ${length} small regexp literal with all flags`, () => {
+      implementation.parse(input, options);
+    });
+  }
+}
+
+benchCases("baseline", baseline);
+benchCases("current", current);
+
+suite.on("cycle", report).run();

packages/babel-parser/package.json

@@ -33,7 +33,7 @@
     "node": ">=6.0.0"
   },
   "devDependencies": {
-    "@babel-baseline/parser": "npm:@babel/parser@^7.14.4",
+    "@babel-baseline/parser": "npm:@babel/parser@^7.14.5",
     "@babel/code-frame": "workspace:*",
     "@babel/helper-fixtures": "workspace:*",
     "@babel/helper-validator-identifier": "workspace:*",

packages/babel-parser/src/tokenizer/index.js

@@ -12,7 +12,6 @@ import { type TokContext, types as ct } from "./context";
 import ParserErrors, { Errors, type ErrorTemplate } from "../parser/error";
 import { SourceLocation } from "../util/location";
 import {
-  lineBreak,
   lineBreakG,
   isNewLine,
   isWhitespace,
@@ -21,7 +20,15 @@ import {
 import State from "./state";
 import type { LookaheadState } from "./state";
 
-const VALID_REGEX_FLAGS = new Set(["g", "m", "s", "i", "y", "u", "d"]);
+const VALID_REGEX_FLAGS = new Set([
+  charCodes.lowercaseG,
+  charCodes.lowercaseM,
+  charCodes.lowercaseS,
+  charCodes.lowercaseI,
+  charCodes.lowercaseY,
+  charCodes.lowercaseU,
+  charCodes.lowercaseD,
+]);
 
 // The following character codes are forbidden from being
 // an immediate sibling of NumericLiteralSeparator _
@@ -976,53 +983,52 @@ export default class Tokenizer extends ParserErrors {
   readRegexp(): void {
     const start = this.state.start + 1;
     let escaped, inClass;
-    for (;;) {
+    let { pos } = this.state;
-      if (this.state.pos >= this.length) {
+    for (; ; ++pos) {
+      if (pos >= this.length) {
         throw this.raise(start, Errors.UnterminatedRegExp);
       }
-      const ch = this.input.charAt(this.state.pos);
+      const ch = this.input.charCodeAt(pos);
-      if (lineBreak.test(ch)) {
+      if (isNewLine(ch)) {
         throw this.raise(start, Errors.UnterminatedRegExp);
       }
       if (escaped) {
         escaped = false;
       } else {
-        if (ch === "[") {
+        if (ch === charCodes.leftSquareBracket) {
           inClass = true;
-        } else if (ch === "]" && inClass) {
+        } else if (ch === charCodes.rightSquareBracket && inClass) {
           inClass = false;
-        } else if (ch === "/" && !inClass) {
+        } else if (ch === charCodes.slash && !inClass) {
           break;
         }
-        escaped = ch === "\\";
+        escaped = ch === charCodes.backslash;
       }
-      ++this.state.pos;
     }
-    const content = this.input.slice(start, this.state.pos);
+    const content = this.input.slice(start, pos);
-    ++this.state.pos;
+    ++pos;
 
     let mods = "";
 
-    while (this.state.pos < this.length) {
+    while (pos < this.length) {
-      const char = this.input[this.state.pos];
+      const cp = this.codePointAtPos(pos);
-      const charCode = this.codePointAtPos(this.state.pos);
+      // It doesn't matter if cp > 0xffff, the loop will either throw or break because we check on cp
+      const char = String.fromCharCode(cp);
 
-      if (VALID_REGEX_FLAGS.has(char)) {
+      if (VALID_REGEX_FLAGS.has(cp)) {
-        if (mods.indexOf(char) > -1) {
+        if (mods.includes(char)) {
-          this.raise(this.state.pos + 1, Errors.DuplicateRegExpFlags);
+          this.raise(pos + 1, Errors.DuplicateRegExpFlags);
         }
-      } else if (
+      } else if (isIdentifierChar(cp) || cp === charCodes.backslash) {
-        isIdentifierChar(charCode) ||
+        this.raise(pos + 1, Errors.MalformedRegExpFlags);
-        charCode === charCodes.backslash
-      ) {
-        this.raise(this.state.pos + 1, Errors.MalformedRegExpFlags);
       } else {
         break;
       }
 
-      ++this.state.pos;
+      ++pos;
       mods += char;
     }
+    this.state.pos = pos;
 
     this.finishToken(tt.regexp, {
       pattern: content,

yarn.lock

@@ -5,12 +5,12 @@ __metadata:
   version: 4
   cacheKey: 7
 
-"@babel-baseline/parser@npm:@babel/parser@^7.14.4":
+"@babel-baseline/parser@npm:@babel/parser@^7.14.5":
-  version: 7.14.4
+  version: 7.14.5
-  resolution: "@babel/parser@npm:7.14.4"
+  resolution: "@babel/parser@npm:7.14.5"
   bin:
     parser: ./bin/babel-parser.js
-  checksum: 3bc067c1ee0e0178d365e1b2988ea1a0d6d37af37870ea1a7e80729b3bdc40acda083cac44ce72f63a5b31a489e35120f617bd41f312dec4c86cf814cff8e64a
+  checksum: 55c14793888cb7d54275811e7f13136875df1ee4fc368f3f10cff46ebdf95b6a072e706a0486be0ac5686a597cbfb82f33b5f66aa6ba80ff50b73bca945035c6
   languageName: node
   linkType: hard
 
@@ -658,6 +658,7 @@ __metadata:
   resolution: "@babel/helper-module-transforms@condition:BABEL_8_BREAKING?:workspace:^7.14.5#2510a1"
   dependencies:
     "@babel/helper-module-transforms-BABEL_8_BREAKING-false": "npm:@babel/helper-module-transforms@workspace:^7.14.5"
+  checksum: eb4895913562bf398b8bf7e6c68a0380f153f52f2715b3685f9d07e376725227678c2f920dfe0772012dfed655e037534619de86bb9f3284b92555f8bf9d0f42
   languageName: node
   linkType: hard
 
@@ -972,7 +973,7 @@ __metadata:
   version: 0.0.0-use.local
   resolution: "@babel/parser@workspace:packages/babel-parser"
   dependencies:
-    "@babel-baseline/parser": "npm:@babel/parser@^7.14.4"
+    "@babel-baseline/parser": "npm:@babel/parser@^7.14.5"
     "@babel/code-frame": "workspace:*"
     "@babel/helper-fixtures": "workspace:*"
     "@babel/helper-validator-identifier": "workspace:*"