RegExp._escape

[coolaj86]

Implementation

RegExp._escape = function escapeRegExp(str) {
  return str.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
};

Good Usage:

new RegExp(escapeRegExp(str), 'g');

Bad Usage:

new RegExp('\\' + escapeRegExp(str), 'g');

Unlikely to ever be accepted into ECMAScript. See:

• https://esdiscuss.org/topic/regexp-escape
• Results of TC39 presentation tc39/proposal-regex-escaping#37

[DBBrowne]

What would someone be trying to achieve with the "Bad Usage" pattern?

[coolaj86]

That's the question of the decade isn't it... tc39/proposal-regex-escaping#37 (comment)

[DBBrowne]

So someone might use the '\\' + escapeRegExp(...) pattern if they were trying to escape a \ at the beginning of their regex?
Which would be unnecessary here, because escaping that initial \ should be handled by the escapeRegEx function?

[coolaj86]

It doesn't make any sense. I couldn't find any explanation of "the even/odd problem", just a reference to it.

I'll summarize it here as well:

You'd have to be stupid to do that. Obviously if you only use an escape on part of your string, only part of your string is escaped. If you use \ then it may escape whatever comes next. If you use \\ then the \ is escaped. If it's a string then \\\ means nothing and \\\\ would mean \\ which, as regex would be a single literal \.

I can't explain it better because it's literally nonsense, which is what everyone is so upset about.

[coolaj86]

(seriously, don't dig too deep trying to understand it, it's lost cause - someone just made up a bad use case and I'm quoting it here not because it would ever make sense, but just to document exactly what was shown to be an invalid use)