gistfile1.md

RegExp.escape(string)

Computes a new version of a String value in which certain characters have been escaped, so that the regular expression engine will interpret any metacharacters that it may contain as character literals.

When the escape function is called with one argument string, the following steps are taken:

1. Let string be ToString(string).
2. ReturnIfAbrupt(string).
3. Let length be the number of characters in string.
4. Let R be the empty string.
5. Let k be 0.
6. Repeat, while k < length, 1. Let C be the character at position k within string. 1. If C is one of the 16 nonblank characters "-[]{}()*+?.,\^$|" then,
   1. Let S be a String containing two characters "\x" where x is a C character. 1. Else,
   2. Let S be a String containing the single C character. 1. Let R be a new String value computed by concatenating the previous value of R and S. 1. Increase k by 1.
7. Return R.

There's also a precendent in the EscapeRegExpPattern production, which is careful to specify the results without constraining the exact implementation too closely. On the other hand, specifying an exact result might be better...

What about the backslash character?
Upd: It's lost in Markdown. Fix, please

@inexorabletash

Rather than the explicit list in 6.2 you should define it in terms of SyntaxCharacter production in the RegExp grammar

Good point. Only SyntaxCharacter doesn't have - and ,.

Also, best to avoid "character" when discussing strings. Use "code units" like B.2.1.1 escape does.

I thought about this as well, but saw that escape gives a list of characters as well, so left it as is. Or do you mean to rephrase "If char is the code point of one of the 16 nonblank characters"?

@zloirock

What about the backslash character?

Thanks, fixed :)

@kangax re code unit, i think @inexorabletash was referring to step 3 and 6i. In particular, since regular expressions now have a "unicode" mode, you probably need to decide how you want to represent astral characters -- is that single "code point" encoded as two UTF-16 "code units" in an ordinary regexp, or as a single code point in a unicode mode regexp?

@arv can you take a look at this?

For reference here is lodash's implementation of lodash.escaperegexp.
The primary difference is it doesn't escape - or , because they fall out of escaping {} and [].
It also escapes / though I need to dig as to why it was added.

@kangax any news about state of this proposal? I don't see it in this list, but it would be great to see it in ES7.

I made a repo here: https://github.com/benjamingr/RexExp.escape/blob/master/README.md any help would be appreciated.

https://github.com/benjamingr/RexExp.escape/blob/master/README.md#regexpescapes