SyntaxError: Invalid regular expression

[inherithandle]

Version

2.6.3

Reproduction link

https://github.com/inherithandle/vue-webpack-simple-project

Steps to reproduce

1. git clone https://github.com/inherithandle/vue-webpack-simple-project
2. run "npm install"
3. run "npm run build"
4. open ./dist/index.html

What is expected?

I expected index.html to be rendered successfully.

What is actually happening?

I get the error: "SyntaxError: invalid range in character class"

---

I think it is similar to the issue #2291.

[aizquier]

I have the same problem. The bug seems to have been introduced since the version 2.6.0.

[Justineo]

It seems to be caused by Terser.

You can workaround this as follows for now:

const path = require('path');
const TerserPlugin = require('terser-webpack-plugin');

module.exports = {
  entry: './src/js/vue-first.js',
  output: {
    filename: 'todo.js',
    path: path.resolve(__dirname, 'dist')
  },
  optimization: {
    minimizer: [
      new TerserPlugin({
        cache: true,
        parallel: true,
        sourceMap: true,
        terserOptions: {
          output: {
            ascii_only: true // <- Add this to keep the regex valid
          }
        }
      }),
    ],
  }
};

[posva]

I'm taking a look at this

[posva]

I couldn't spend more time on this for now but it's weird since using raw terser with the regex at https://github.com/vuejs/vue/blob/dev/src/core/util/lang.js#L8, and more specifically new RegExp('[^\uF900-\uFDCF]') yields a valid result (the same we have in project with the CLI). It's as if the problem was with webpack. We cannot really expand the regex as it would be too much either

[yyx990803]

/cc @sodatea not sure if this is affecting the CLI which also uses Terser. Maybe it has to do with a specific version of Terser?

[haoqunjiang]

It works with http, but not file:///

[haoqunjiang]

Or with charset="utf8" attribute on the script tag.
So the issue is browser's auto charset detection.
Not sure how to fix it.

[haoqunjiang]

So, as I tested out on https://xem.github.io/terser-online/, unicode sequences in strings (e.g. '\uF900-\uFDCF') are compressed into single characters, while those in regexps aren't (e.g. /\uF900-\uFDCF/). It'll be an easy fix.