Add remaining Unicode properties

Author: ota-meshi

.eslintrc.js

@@ -13,7 +13,7 @@ module.exports = {
   },
   overrides: [
     {
-      files: ["acorn/src/bin/*.js", "bin/generate-identifier-regex.js"],
+      files: ["acorn/src/bin/*.js", "bin/generate-identifier-regex.js", "bin/generate-unicode-script-values.js"],
       rules: {
         "no-console": "off"
       }

acorn/src/generated/scriptValuesAddedInUnicode.js

@@ -0,0 +1,2 @@
+// This file was generated by "bin/generate-unicode-script-values.js". Do not modify manually!
+export default "Gara Garay Gukh Gurung_Khema Hrkt Katakana_Or_Hiragana Kawi Kirat_Rai Krai Nag_Mundari Nagm Ol_Onal Onao Sunu Sunuwar Todhri Todr Tulu_Tigalari Tutg Unknown Zzzz"

acorn/src/unicode-property-data.js

@@ -1,4 +1,5 @@
 import {wordsRegexp} from "./util.js"
+import scriptValuesAddedInUnicode from "./generated/scriptValuesAddedInUnicode.js"
 
 // This file contains Unicode properties extracted from the ECMAScript specification.
 // The lists are extracted like so:
@@ -42,7 +43,7 @@ const ecma10ScriptValues = ecma9ScriptValues + " Dogra Dogr Gunjala_Gondi Gong H
 const ecma11ScriptValues = ecma10ScriptValues + " Elymaic Elym Nandinagari Nand Nyiakeng_Puachue_Hmong Hmnp Wancho Wcho"
 const ecma12ScriptValues = ecma11ScriptValues + " Chorasmian Chrs Diak Dives_Akuru Khitan_Small_Script Kits Yezi Yezidi"
 const ecma13ScriptValues = ecma12ScriptValues + " Cypro_Minoan Cpmn Old_Uyghur Ougr Tangsa Tnsa Toto Vithkuqi Vith"
-const ecma14ScriptValues = ecma13ScriptValues + " Hrkt Katakana_Or_Hiragana Kawi Nag_Mundari Nagm Unknown Zzzz"
+const ecma14ScriptValues = ecma13ScriptValues + " " + scriptValuesAddedInUnicode
 
 const unicodeScriptValues = {
   9: ecma9ScriptValues,

bin/generate-unicode-script-values.js

@@ -0,0 +1,59 @@
+"use strict"
+
+const fs = require("fs")
+const path = require("path")
+
+import("../acorn/src/unicode-property-data.js")
+  .then(m => {
+    return m.default[13].nonBinary.Script
+  })
+  .then(async(reScriptValuesAddedInES) => {
+    const scriptValues = new Set()
+    for await (const value of getLatestUnicodeScriptValues()) {
+      scriptValues.add(value)
+    }
+    const scriptValuesAddedInUnicode = "export default " +
+    JSON.stringify(
+      [...scriptValues]
+        // The unicode script values now follow the Unicode spec as of ES2023,
+        // but prior to ES2022 they were listed in the ES2022 spec.
+        // The generated file lists all the unicode script values except those listed before ES2022.
+        .filter(value => !reScriptValuesAddedInES.test(value))
+        .sort()
+        .join(" ")
+    )
+
+    writeGeneratedFile("scriptValuesAddedInUnicode", scriptValuesAddedInUnicode)
+
+    console.log("Done. The generated files must be committed.")
+  })
+
+function writeGeneratedFile(filename, content) {
+  const comment = "// This file was generated by \"bin/" + path.basename(__filename) + "\". Do not modify manually!"
+  fs.writeFileSync(path.resolve("./acorn/src/generated", filename + ".js"), comment + "\n" + content + "\n", "utf8")
+}
+
+/**
+ * Gets the all unicode script values from the latest PropertyValueAliases.
+ */
+async function * getLatestUnicodeScriptValues() {
+  const response = await fetch("https://unicode.org/Public/UCD/latest/ucd/PropertyValueAliases.txt")
+  const lines = (await response.text()).split("\n")
+  for (const line of lines) {
+    if (!line || line.startsWith("#")) {
+      continue
+    }
+    const [propertyAlias, alias, canonical, ...remaining] = line
+      .split("#")[0] // strip comments
+      .split(";") // split by semicolon
+      .map((x) => x.trim()) // trim
+
+    if (propertyAlias !== "sc") {
+      continue
+    }
+
+    yield canonical
+    yield alias
+    yield * remaining
+  }
+}

package.json

@@ -25,7 +25,9 @@
     "build:loose": "rollup -c acorn-loose/rollup.config.mjs",
     "build:main": "rollup -c acorn/rollup.config.mjs",
     "build:walk": "rollup -c acorn-walk/rollup.config.mjs",
-    "generate": "node bin/generate-identifier-regex.js",
+    "generate": "npm run generate:identifier-regex && npm run generate:unicode-script-values",
+    "generate:identifier-regex": "node bin/generate-identifier-regex.js",
+    "generate:unicode-script-values": "node bin/generate-unicode-script-values.js",
     "lint": "eslint .",
     "prepare": "npm run test",
     "pretest": "npm run build:main && npm run build:loose",