javascript正则表达式，以匹配单引号，双引号和正则斜杠之间的任何内容

Question

I'm trying to match anything between either double quotes, single quotes, or regex slashes, basically anything that isn't tokenized by javascript as a string or regex. So far what I came up with is:

/"[^\\"\n]*(\\"[^\\"\n]*)*"|'[^\\'\n]*(\\'[^\\'\n]*)*'|\/[^\\\/\n]*(\\\/[^\\\/\n]*)*\//

But there are a couple of problems with this as you can see here

http://goo.gl/4Yn9pR

Basically this shouldn't match 1+2/3+4/5 since it isn't a regex. Also
Dont match "Match here\\\\" Dont match" should match the first part and not the second (thats true for single quotes and regexes too)

How should this be written?

Edit: If it's not possible differentiate between 1+2/3+4/5 , /*comment*/ and /regex/ using regular expressions, how would I just solve the Dont match "Match here\\\\" Dont match" problem

Answer 1

The trick to match c-alike escaped strings is like this:

" (\\. | [^"]) * "

That is,

 - quote
 - repeat (
    - one escaped char
    - or not a quote
   )
  - quote

Similarly with single quotes. Illustration in python since JS regexes are ugly:

import re

test = r"""
    foo "bar" and "bar\"bar" and "bar\\bar" and "bar \\"
    foo 'bar' and 'bar\'bar' and 'bar\\bar' and 'bar \\'
"""

rr = r"""(?x)
    " (\\. | [^"]) * "
    |
    ' (\\. | [^']) * '
"""

print re.sub(rr, '@@', test)

> foo @@ and @@ and @@ and @@
> foo @@ and @@ and @@ and @@

It might be necessary to add newlines to the [^"] group.

Do note that this expression is quite forgiving and allows many constructs that aren't valid javascript. See https://stackoverflow.com/a/13800082/989121 for the complete and accurate implementation.

Answer 2

Just figured it out. I was very close. Here's the solution:

/"[^\\"\n]*(\\["\\][^\\"\n]*)*"|'[^\\'\n]*(\\['\\][^\\'\n]*)*'|\/[^\\\/\n]*(\\[\/\\][^\\\/\n]*)*\//

DEMO

It's very similar to thg435 answer but I think it's a little more performent because it doesn't backtrack as much

What I was missing was when looking for an escaped quote, I should have also been looking for an escaped backslash too, so i changed \\\\" to \\\\["\\\\] As opposed to thg435's answer which looks at anything after a backslash which while valid can use up more states in the regex engine