Generate char-utils.ts and known-tlds.ts. Remove punycode support for known tlds

Author: gregjacobs

BUILD.bazel

@@ -1,5 +1,5 @@
 load("@npm//:defs.bzl", "npm_link_all_packages")
-load("@aspect_rules_js//js:defs.bzl", "js_library")
+load("@aspect_rules_js//js:defs.bzl", "js_library", "js_run_binary")
 load("@aspect_rules_js//npm:defs.bzl", "npm_package")
 load("//build-tools:ts-bundle.bzl", "ts_bundle")
 
@@ -9,7 +9,14 @@ load("@bazel_skylib//rules:write_file.bzl", "write_file")
 
 npm_link_all_packages(name = "node_modules")
 
-src_files = glob(["src/**/*.ts"])
+src_files = glob(["src/**/*.ts"]) + [
+    "src/char-utils.ts", # Generated file
+    "src/parser/known-tlds.ts", # Generated file
+]
+
+test_files = glob(["tests/**/*.ts"]) + [
+    "tests/char-utils.spec.ts", # Generated file
+]
 
 npm_package(
     name = "package",
@@ -21,6 +28,7 @@ npm_package(
 
 js_library(
     name = "library",
+    visibility = ["//visiblity:public"],
     deps = [
         ":src_esm",
         ":src_cjs",
@@ -30,6 +38,7 @@ js_library(
 # Create ESM bundle
 ts_bundle(
     name = "src_esm",
+    visibility = ["//visibility:private"],
     entry_point = "src/index.ts",
     root_dir = "src",
     srcs = src_files,
@@ -46,6 +55,7 @@ ts_bundle(
 # Create CommonJS bundle
 ts_bundle(
     name = "src_cjs",
+    visibility = ["//visibility:private"],
     entry_point = "src/index.ts",
     root_dir = "src",
     srcs = src_files,
@@ -61,6 +71,25 @@ ts_bundle(
 
 ts_config(
     name = "tsconfig",
-    src = "tsconfig.json",
     visibility = [":__subpackages__"],
+    src = "tsconfig.json",
+)
+
+# Generate the 'src/char-utils.ts' and 'tests/char-utils.spec.ts' files
+js_run_binary(
+    name = "generate_char_utils",
+    tool = "//scripts:generate_char_utils",
+    outs = [
+        "src/char-utils.ts", 
+        "tests/char-utils.spec.ts"
+    ],
+)
+
+# Generate the 'parser/known-tlds.ts' file
+js_run_binary(
+    name = "generate_known_tlds",
+    tool = "//scripts:generate_known_tlds",
+    outs = [
+        "src/parser/known-tlds.ts",
+    ],
 )

MODULE.bazel.lock

@@ -17,6 +17,7 @@ def ts_bundle(
     deps = [],
     target = "esnext",
     platform = "node",
+    visibility = None,
 ):
     """
     Compiles TypeScript and bundles the output into a single .js / .d.ts files.
@@ -43,6 +44,7 @@ def ts_bundle(
         See https://esbuild.github.io/api/#target for reference.
       platform: The platform ('browser', 'node', or 'neutral') to build for. 
         See: https://esbuild.github.io/api/#platform
+      visibility: The visibility for the generated target.
     """
     js_bundle_target = name + "_bundle"
     declarations_bundle_target = name + "_declarations_bundle"
@@ -51,6 +53,7 @@ def ts_bundle(
 
     js_library(
         name = name,
+        visibility = visibility,
         srcs = [
             ":" + js_bundle_target,
             ":" + declarations_bundle_target

build-tools/ts-bundle.bzl

@@ -67,7 +67,6 @@ def ts_declarations(
                     "sourceMap": true,
                     "declarationMap": true,
                     "emitDeclarationOnly": true,
-                    "noEmit": false
                 }
             }
         """ % (tsconfig_file_path, target, module)]

build-tools/ts-declarations.bzl

@@ -94,14 +94,14 @@
     "npm-run-all": "^4.1.5",
     "nyc": "^17.1.0",
     "prettier": "3.4.2",
-    "punycode": "^2.3.1",
     "puppeteer": "^24.7.2",
     "rimraf": "^6.0.1",
     "rollup": "^4.29.1",
     "rollup-plugin-dts": "^6.2.1",
     "sinon": "^20.0.0",
     "style-loader": "^4.0.0",
     "terser": "^5.37.0",
+    "tlds": "^1.258.0",
     "ts-loader": "^9.5.1",
     "ts-node": "^10.9.2",
     "typescript": "5.8.3",

package.json

@@ -0,0 +1,48 @@
+load("@aspect_rules_js//js:defs.bzl", "js_binary")
+load("@aspect_rules_ts//ts:defs.bzl", "ts_config", "ts_project")
+
+# Script to generate the 'src/char-utils.ts' and 'tests/char-utils.spec.ts' files
+js_binary(
+    name = "generate_char_utils",
+    visibility = ["//visibility:public"],
+    entry_point = "generate-char-utils.mjs",
+    data = [
+        ":tsc",
+    ],
+)
+
+# Script to generate the 'parser/known-tlds.ts' file
+js_binary(
+    name = "generate_known_tlds",
+    visibility = ["//visibility:public"],
+    entry_point = "generate-known-tlds.mjs",
+    data = [
+        ":tsc",
+    ],
+)
+
+# Compile the scripts/ folder's TypeScript into JavaScript for execution
+ts_project(
+    name = "tsc",
+    tsconfig = ":tsconfig",
+    srcs = glob(["**/*.ts", "**/*.mts"]),
+    deps = [
+        "//:node_modules/@types/fs-extra",
+        "//:node_modules/@types/node",
+        "//:node_modules/dedent",
+        "//:node_modules/fast-glob",
+        "//:node_modules/fs-extra",
+        "//:node_modules/html-webpack-plugin",
+        "//:node_modules/tlds",
+        "//:node_modules/webpack",
+    ],
+    source_map = True,
+    declaration = True,
+    declaration_map = True,
+)
+
+ts_config(
+    name = "tsconfig",
+    src = "tsconfig.json",
+    deps = ["//:tsconfig"]
+)

pnpm-lock.yaml

@@ -1,7 +1,7 @@
 import dedent from 'dedent';
-import fs from 'fs';
+import fse from 'fs-extra';
 import path from 'path';
-import { alphaNumericAndMarksRe } from './alpha-numeric-and-marks-re';
+import { alphaNumericAndMarksRe } from './alpha-numeric-and-marks-re.mjs';
 
 /*
  * This script generates functions which check that a single character matches
@@ -15,7 +15,7 @@ import { alphaNumericAndMarksRe } from './alpha-numeric-and-marks-re';
  * as fast as possible.
  */
 
-const rootPath = path.normalize(`${__dirname}/..`);
+const rootPath = path.normalize(`${import.meta.dirname}/..`);
 const generateScriptName = 'generate-char-utils';
 
 const { srcFileContents, specFileContents } = generateCharUtils([
@@ -32,15 +32,15 @@ const { srcFileContents, specFileContents } = generateCharUtils([
     ['isCloseBraceChar', /[)}\]]/],
 ]);
 
-// console.log(srcFileContents);
-// console.log(specFileContents);
+console.log(srcFileContents);
+console.log(specFileContents);
 
 const srcFilePath = `src/char-utils.ts`;
-fs.writeFileSync(`${rootPath}/${srcFilePath}`, srcFileContents, 'utf-8');
+fse.outputFileSync(`${rootPath}/${srcFilePath}`, srcFileContents, 'utf-8');
 console.log(`Wrote ${srcFilePath}`);
 
 const specFilePath = `tests/char-utils.spec.ts`;
-fs.writeFileSync(`${rootPath}/${specFilePath}`, specFileContents, 'utf-8');
+fse.outputFileSync(`${rootPath}/${specFilePath}`, specFileContents, 'utf-8');
 console.log(`Wrote ${specFilePath}`);
 
 // -------------------------------------------------------------

scripts/BUILD.bazel

@@ -0,0 +1,32 @@
+
+import dedent from 'dedent';
+import fse from 'fs-extra';
+import tlds from 'tlds' with { type: "json" };
+
+generateKnownTlds().catch(error => {
+    console.error(error);
+    process.exit(1);
+});
+
+/**
+ * Updates the set of known top-level domains (TLDs). Ex: '.com', '.net', etc.
+ */
+export async function generateKnownTlds() {
+    const tldRegexStr = tlds.toSorted(compareLengthLongestFirst).join('|');
+
+    const outputFile =
+        dedent`
+        // NOTE: THIS IS A GENERATED FILE\n// To update with the latest TLD list, update the 'tlds' dependency version and then run a build
+        
+        export const tldRegex = /^(?:${tldRegexStr})$/;
+    ` + '\n';
+    await fse.outputFile('./src/parser/known-tlds.ts', outputFile);
+}
+
+function compareLengthLongestFirst(a: string, b: string): number {
+    let result = b.length - a.length;
+    if (result === 0) {
+        result = a.localeCompare(b);
+    }
+    return result;
+}