RegExp Parser for Tag-Like String

Question

In an application I'm writing, at some point, I want to be able to parse a tag-like string with RegExp so that it is modified, such as:

"{b}This is BOLD{/b}".replace(/\{b\}(.*?)\{\/b\}/gi, "00b3f[ $1 00b3d]");

// Returns "00b3f[ This is BOLD 00b3d]"

I am able to do this easily, but it gets complicated when a more complex string is passed to the function, for example:

"{red} This is RED {red} This also should be red {/red} and this {/red}"
.replace(/\{red\}(.*?)\{\/red\}/gi, "00b4f[ $1 00b4d]");

// Returns:
// "00b4f[  This is RED {red} This also should be red  00b4d] and this {/red}"

// Where the output should be:
// "00b4f[  This is RED 00b4f[ This also should be red 00b4d] and this 00b4d]"

I would like to solve this problem with a simple RegExp but I can't find a way to do it! I think I could do this with a while-loop but it would get too messy. Any suggestions?

You can't solve this problem with a regexp because the language you are describing is context-sensitive but not regular - in simple terms it means "you need memory to parse it" which is what you've observed. I recommend a simple recursive descent parser. — Benjamin Gruenbaum, Dec 30 '14 at 20:38

score 2 · Accepted Answer · edited Dec 30 '14 at 21:08

2

Regex can't deal with nested expressions (unless you have access to a powerful implementation of regex, which javascript doesn't), so a pure-regex solution is out of the question. But there's still an easy way to do this:

replace all occurences of \{(\w+)\}((?:(?!\{\w+\}).)*)\{\/\1\} (this matches a {tag}...{/tag} pair, but only if it does not contain another {tag} ) with 00b4f[ $2 00b4d] .
repeat until there are no more matches.

To make this dynamic, use a callback function for the replacement:

var tagPattern = /\{(\w+)\}((?:(?!\{\w+\}).)*)\{\/\1\}/g,
    tagReplacer = function ($0, $1, $2) {
        switch ($1) {
            case "b": return "00b3f[" + $2 + " 00b3d]";
            case "red": return "00b4f[" + $2 + " 00b4d]";
            default: return $2;
        }
    };

while (tagPattern.test(sourceString)) {
    sourceString = sourceString.replace(tagPattern, tagReplacer);
}

edited Dec 30 '14 at 21:08

Tomalak

332,285
67
532
628

answered Dec 30 '14 at 20:48

Aran-Fey

39,665
11
104
149

1

Very nice solution. You could extend that by mentioning replace callbacks, I'm pretty sure the OP will need them. – Tomalak Dec 30 '14 at 20:50
1

+ 1 Very very cool. A small while loop and a callback will make this very dynamic and easy to implement! – Afonso Matos Dec 30 '14 at 20:53
@Tomalak: I've never really used JavaScript, so chances are OP knows more about replace callbacks than I do. – Aran-Fey Dec 30 '14 at 20:55
1

@Rawing I took the liberty to add that. – Tomalak Dec 30 '14 at 21:03
@Tomalak: In that case, I will take the liberty to upvote your comment, at the very least. Thank you. – Aran-Fey Dec 30 '14 at 21:05
You're welcome. That's what this whole "everything can be edited" thing is for, after all. – Tomalak Dec 30 '14 at 21:07

RegExp Parser for Tag-Like String

1 Answers1