Greedy matching bug

[RunDevelopment]

Consider the following code:

{
	'template-string-pattern': /`(?:\\[\s\S]|\${[^}]+}|[^\\`])*`/,
	'a': 0,
	'b': 1
};
{
	'string' : /(`|')(?:\\.|(?!\1)[^\\\r\n])*\1/m,
	'c': 0
};

Highlighted using Test Drive:

What's happening is quite simple:

1. template-string matches (?:\\[\s\S]|\${[^}]+}|[^\\ and /, 'a': 0 ....... 'string': /( (plus the grave accents).
2. string matches 'template-string-pattern' and 'c'. The rest is blocked by template-string.
3. regex breaks the matches of template-string and highlights the regexes.
4. string gets a chance from regex to match and does so matching 'a'. But that's it, because oneshot=true.

Btw: You can see this bug all over the place when you look at the Prism code with lots of language definitions.

[RunDevelopment]

Here is a minimal example showing the bug in action:

Language definition

I used aliases for styling.

Prism.languages['greedy-test'] = {
	'a': {
		pattern: /'[^'\r\n]*'/,
		alias: 'comment'
	},
	'b': {
		pattern: /"[^"\r\n]*"/,
		greedy: true,
		alias: 'string'
	},
	'c': {
		pattern: /<[^>\r\n]*>/,
		greedy: true,
		alias: 'keyword'
	}
}

Code:

<'> '' ''
<"> "" ""

(Each line is one example.)

Result:

Notes:

1. The order of a and b is not significant. They just have to be before c.
2. Both example lines of code are independent of each other.