Regular expression literal in JS code can be mis-parsed as unterminated strings

[schlessera]

The JavaScript parsing code contains a bug in the upstream mck89/peast library.

If the parsed code has regular expression literals containing quotes, these quotes are erroneously being parsed as opening or closing a string literal.

This will in turn cause the parser to throw an Unterminated string error once it hits the end of the line.

As an example, the WordPress Core code contains regular expression literals like this:

function escapeHTML(s) {
   var n = s;
   n = n.replace(/&/g, '&');
   n = n.replace(/</g, '&lt;');
   n = n.replace(/>/g, '&gt;');
   n = n.replace(/"/g, '&quot;');

   return n;
}

In this example, the regular expression literal /"/g will be parsed as opening a double-quoted string and throw an error once it hits the next line.

[schlessera]

After some discussion on Slack, we opted to use a work-around in wp i18n makepot while waiting on the upstream bug to get a fix (which we expect is more involved).

Planned work-around is to string-replace away the offending regular expression literals in the in-memory file representation that the peast library works off of.

[schlessera]

Maybe we should keep this open until the upstream bug is fixed.

[schlessera]

We should look into this again, as the upstream issue seems to have been fixed: mck89/peast#3