Add transform support for the "regexp unicode sets" proposal (#14125)

packages/babel-core/src/parser/util/missing-plugin-helper.ts

@@ -201,6 +201,16 @@ const pluginNameMap = {
       url: "https://git.io/JvKp3",
     },
   },
+  regexpUnicodeSets: {
+    syntax: {
+      name: "@babel/plugin-syntax-unicode-sets-regex",
+      url: "https://git.io/J9GTd",
+    },
+    transform: {
+      name: "@babel/plugin-proposal-unicode-sets-regex",
+      url: "https://git.io/J9GTQ",
+    },
+  },
   throwExpressions: {
     syntax: {
       name: "@babel/plugin-syntax-throw-expressions",

packages/babel-helper-create-regexp-features-plugin/package.json

@@ -19,7 +19,7 @@
   ],
   "dependencies": {
     "@babel/helper-annotate-as-pure": "workspace:^",
-    "regexpu-core": "^4.7.1"
+    "regexpu-core": "^5.0.1"
   },
   "peerDependencies": {
     "@babel/core": "^7.0.0"

packages/babel-helper-create-regexp-features-plugin/src/features.ts

@@ -3,6 +3,8 @@ export const FEATURES = Object.freeze({
   dotAllFlag: 1 << 1,
   unicodePropertyEscape: 1 << 2,
   namedCaptureGroups: 1 << 3,
+  unicodeSetsFlag_syntax: 1 << 4,
+  unicodeSetsFlag: 1 << 5,
 });
 
 // We can't use a symbol because this needs to always be the same, even if

packages/babel-helper-create-regexp-features-plugin/src/index.ts

@@ -1,29 +1,10 @@
 import rewritePattern from "regexpu-core";
-import {
-  featuresKey,
-  FEATURES,
-  enableFeature,
-  runtimeKey,
-  hasFeature,
-} from "./features";
-import { generateRegexpuOptions } from "./util";
+import { featuresKey, FEATURES, enableFeature, runtimeKey } from "./features";
+import { generateRegexpuOptions, canSkipRegexpu, transformFlags } from "./util";
 
 import { types as t } from "@babel/core";
 import annotateAsPure from "@babel/helper-annotate-as-pure";
 
-type RegExpFlags = "i" | "g" | "m" | "s" | "u" | "y";
-
-/**
- * Remove given flag from given RegExpLiteral node
- *
- * @param {RegExpLiteral} node
- * @param {RegExpFlags} flag
- * @returns {void}
- */
-function pullFlag(node, flag: RegExpFlags): void {
-  node.flags = node.flags.replace(flag, "");
-}
-
 declare const PACKAGE_JSON: { name: string; version: string };
 
 // Note: Versions are represented as an integer. e.g. 7.1.5 is represented
@@ -39,9 +20,13 @@ export function createRegExpFeaturePlugin({
   name,
   feature,
   options = {} as any,
+  manipulateOptions = (() => {}) as (opts: any, parserOpts: any) => void,
 }) {
   return {
     name,
+
+    manipulateOptions,
+
     pre() {
       const { file } = this;
       const features = file.get(featuresKey) ?? 0;
@@ -70,20 +55,21 @@ export function createRegExpFeaturePlugin({
         const { file } = this;
         const features = file.get(featuresKey);
         const runtime = file.get(runtimeKey) ?? true;
-        const regexpuOptions = generateRegexpuOptions(node, features);
-        if (regexpuOptions === null) {
-          return;
-        }
+
+        const regexpuOptions = generateRegexpuOptions(features);
+        if (canSkipRegexpu(node, regexpuOptions)) return;
+
         const namedCaptureGroups = {};
-        if (regexpuOptions.namedGroup) {
+        if (regexpuOptions.namedGroups === "transform") {
           regexpuOptions.onNamedGroup = (name, index) => {
             namedCaptureGroups[name] = index;
           };
         }
+
         node.pattern = rewritePattern(node.pattern, node.flags, regexpuOptions);
 
         if (
-          regexpuOptions.namedGroup &&
+          regexpuOptions.namedGroups === "transform" &&
           Object.keys(namedCaptureGroups).length > 0 &&
           runtime &&
           !isRegExpTest(path)
@@ -96,12 +82,8 @@ export function createRegExpFeaturePlugin({
 
           path.replaceWith(call);
         }
-        if (hasFeature(features, FEATURES.unicodeFlag)) {
-          pullFlag(node, "u");
-        }
-        if (hasFeature(features, FEATURES.dotAllFlag)) {
-          pullFlag(node, "s");
-        }
+
+        node.flags = transformFlags(regexpuOptions, node.flags);
       },
     },
   };

packages/babel-helper-create-regexp-features-plugin/src/util.ts

@@ -1,65 +1,77 @@
+import type { types as t } from "@babel/core";
 import { FEATURES, hasFeature } from "./features";
 
 type RegexpuOptions = {
-  useUnicodeFlag: boolean;
+  unicodeFlag: "transform" | false;
+  unicodeSetsFlag: "transform" | "parse" | false;
+  dotAllFlag: "transform" | false;
+  unicodePropertyEscapes: "transform" | false;
+  namedGroups: "transform" | false;
   onNamedGroup: (name: string, index: number) => void;
-  namedGroup: boolean;
-  unicodePropertyEscape: boolean;
-  dotAllFlag: boolean;
-  lookbehind: boolean;
 };
 
-export function generateRegexpuOptions(node, features): RegexpuOptions | null {
-  let useUnicodeFlag = false,
-    dotAllFlag = false,
-    unicodePropertyEscape = false,
-    namedGroup = false;
+export function generateRegexpuOptions(toTransform: number): RegexpuOptions {
+  type Experimental = 1;
+
+  const feat = <Stability extends 0 | 1 = 0>(
+    name: keyof typeof FEATURES,
+    ok: "transform" | (Stability extends 0 ? never : "parse") = "transform",
+  ) => {
+    return hasFeature(toTransform, FEATURES[name]) ? ok : false;
+  };
+
+  return {
+    unicodeFlag: feat("unicodeFlag"),
+    unicodeSetsFlag:
+      feat<Experimental>("unicodeSetsFlag") ||
+      feat<Experimental>("unicodeSetsFlag_syntax", "parse"),
+    dotAllFlag: feat("dotAllFlag"),
+    unicodePropertyEscapes: feat("unicodePropertyEscape"),
+    namedGroups: feat("namedCaptureGroups"),
+    onNamedGroup: () => {},
+  };
+}
+
+export function canSkipRegexpu(
+  node: t.RegExpLiteral,
+  options: RegexpuOptions,
+): boolean {
   const { flags, pattern } = node;
-  const flagsIncludesU = flags.includes("u");
 
-  if (flagsIncludesU) {
-    if (!hasFeature(features, FEATURES.unicodeFlag)) {
-      useUnicodeFlag = true;
-    }
+  if (flags.includes("v")) {
+    if (options.unicodeSetsFlag === "transform") return false;
+  }
+
+  if (flags.includes("u")) {
+    if (options.unicodeFlag === "transform") return false;
     if (
-      hasFeature(features, FEATURES.unicodePropertyEscape) &&
+      options.unicodePropertyEscapes === "transform" &&
       /\\[pP]{/.test(pattern)
     ) {
-      unicodePropertyEscape = true;
+      return false;
     }
   }
 
-  if (hasFeature(features, FEATURES.dotAllFlag) && flags.indexOf("s") >= 0) {
-    dotAllFlag = true;
+  if (flags.includes("s")) {
+    if (options.dotAllFlag === "transform") return false;
   }
-  if (
-    hasFeature(features, FEATURES.namedCaptureGroups) &&
-    /\(\?<(?![=!])/.test(pattern)
-  ) {
-    namedGroup = true;
+
+  if (options.namedGroups === "transform" && /\(\?<(?![=!])/.test(pattern)) {
+    return false;
   }
-  if (
-    !namedGroup &&
-    !unicodePropertyEscape &&
-    !dotAllFlag &&
-    (!flagsIncludesU || useUnicodeFlag)
-  ) {
-    return null;
+
+  return true;
+}
+
+export function transformFlags(regexpuOptions: RegexpuOptions, flags: string) {
+  if (regexpuOptions.unicodeSetsFlag === "transform") {
+    flags = flags.replace("v", "u");
   }
-  // Now we have to feed regexpu-core the regex
-  if (flagsIncludesU && flags.indexOf("s") >= 0) {
-    // When flags includes u, `config.unicode` will be enabled even if `u` is supported natively.
-    // In this case we have to enable dotAllFlag, otherwise `rewritePattern(/./su)` will return
-    // incorrect result
-    // https://github.com/mathiasbynens/regexpu-core/blob/v4.6.0/rewrite-pattern.js#L191
-    dotAllFlag = true;
+  if (regexpuOptions.unicodeFlag === "transform") {
+    flags = flags.replace("u", "");
   }
-  return {
-    useUnicodeFlag,
-    onNamedGroup: () => {},
-    namedGroup,
-    unicodePropertyEscape,
-    dotAllFlag,
-    lookbehind: true,
-  };
+  if (regexpuOptions.dotAllFlag === "transform") {
+    flags = flags.replace("s", "");
+  }
+  return flags;
 }

packages/babel-helper-create-regexp-features-plugin/test/fixtures/plugin-transform-named-capturing-groups-regex/runtime-false-not-overwritten/output.mjs

@@ -1 +1 @@
-/([0-9]{4})/;
+/(\d{4})/;

packages/babel-plugin-proposal-unicode-sets-regex/.npmignore

@@ -0,0 +1,3 @@
+src
+test
+*.log

packages/babel-plugin-proposal-unicode-sets-regex/README.md

@@ -0,0 +1,19 @@
+# @babel/plugin-proposal-unicode-sets-regex
+
+> Compile regular expressions' unicodeSets (v) flag.
+
+See our website [@babel/plugin-proposal-unicode-sets-regex](https://babeljs.io/docs/en/babel-plugin-proposal-unicode-sets-regex) for more information.
+
+## Install
+
+Using npm:
+
+```sh
+npm install --save-dev @babel/plugin-proposal-unicode-sets-regex
+```
+
+or using yarn:
+
+```sh
+yarn add @babel/plugin-proposal-unicode-sets-regex --dev
+```

packages/babel-plugin-proposal-unicode-sets-regex/package.json

@@ -0,0 +1,48 @@
+{
+  "name": "@babel/plugin-proposal-unicode-sets-regex",
+  "version": "7.16.7",
+  "description": "Compile regular expressions' unicodeSets (v) flag.",
+  "homepage": "https://babel.dev/docs/en/next/babel-plugin-proposal-unicode-sets-regex",
+  "license": "MIT",
+  "publishConfig": {
+    "access": "public"
+  },
+  "main": "./lib/index.js",
+  "keywords": [
+    "babel-plugin",
+    "regex",
+    "regexp",
+    "unicode",
+    "sets",
+    "properties",
+    "property",
+    "string",
+    "strings",
+    "regular expressions"
+  ],
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/babel/babel.git",
+    "directory": "packages/babel-plugin-proposal-unicode-sets-regex"
+  },
+  "bugs": "https://github.com/babel/babel/issues",
+  "dependencies": {
+    "@babel/helper-create-regexp-features-plugin": "workspace:^",
+    "@babel/helper-plugin-utils": "workspace:^"
+  },
+  "peerDependencies": {
+    "@babel/core": "^7.0.0"
+  },
+  "devDependencies": {
+    "@babel/core": "workspace:^",
+    "@babel/helper-plugin-test-runner": "workspace:^"
+  },
+  "author": "The Babel Team (https://babel.dev/team)",
+  "exports": {
+    ".": "./lib/index.js",
+    "./package.json": "./package.json"
+  },
+  "engines": {
+    "node": ">=6.9.0"
+  }
+}

packages/babel-plugin-proposal-unicode-sets-regex/src/index.ts

@@ -0,0 +1,15 @@
+/* eslint-disable @babel/development/plugin-name */
+import { createRegExpFeaturePlugin } from "@babel/helper-create-regexp-features-plugin";
+import { declare } from "@babel/helper-plugin-utils";
+
+export default declare(api => {
+  api.assertVersion(7);
+
+  return createRegExpFeaturePlugin({
+    name: "transform-unicode-sets-regex",
+    feature: "unicodeSetsFlag",
+    manipulateOptions(opts, parserOpts) {
+      parserOpts.plugins.push("regexpUnicodeSets");
+    },
+  });
+});

packages/babel-plugin-proposal-unicode-sets-regex/test/fixtures/basic/basic/input.js

@@ -0,0 +1 @@
+/[[0-7]&&[5-9]]\u{10FFFF}/v;

packages/babel-plugin-proposal-unicode-sets-regex/test/fixtures/basic/basic/options.json

@@ -0,0 +1,3 @@
+{
+  "plugins": ["proposal-unicode-sets-regex"]
+}

packages/babel-plugin-proposal-unicode-sets-regex/test/fixtures/basic/basic/output.js

@@ -0,0 +1 @@
+/[5-7]\u{10FFFF}/u;

packages/babel-plugin-proposal-unicode-sets-regex/test/fixtures/basic/transform-u/input.js

@@ -0,0 +1 @@
+/[[0-7]&&[5-9]]\u{10FFFF}/v;

packages/babel-plugin-proposal-unicode-sets-regex/test/fixtures/basic/transform-u/options.json

@@ -0,0 +1,3 @@
+{
+  "plugins": ["proposal-unicode-sets-regex", "transform-unicode-regex"]
+}

packages/babel-plugin-proposal-unicode-sets-regex/test/fixtures/basic/transform-u/output.js

@@ -0,0 +1 @@
+/[5-7](?:\uDBFF\uDFFF)/;

packages/babel-plugin-proposal-unicode-sets-regex/test/index.js

@@ -0,0 +1,3 @@
+import runner from "@babel/helper-plugin-test-runner";
+
+runner(import.meta.url);

packages/babel-plugin-proposal-unicode-sets-regex/test/package.json

@@ -0,0 +1 @@
+{ "type": "module" }

packages/babel-plugin-syntax-unicode-sets-regex/.npmignore

@@ -0,0 +1,3 @@
+src
+test
+*.log

packages/babel-plugin-syntax-unicode-sets-regex/README.md

@@ -0,0 +1,19 @@
+# @babel/plugin-syntax-unicode-sets-regex
+
+> Parse regular expressions' unicodeSets (v) flag.
+
+See our website [@babel/plugin-syntax-unicode-sets-regex](https://babeljs.io/docs/en/babel-plugin-syntax-unicode-sets-regex) for more information.
+
+## Install
+
+Using npm:
+
+```sh
+npm install --save-dev @babel/plugin-syntax-unicode-sets-regex
+```
+
+or using yarn:
+
+```sh
+yarn add @babel/plugin-syntax-unicode-sets-regex --dev
+```