Converting Straight Quotes to Curly Quotes

Question

I have an application which uses a javascript based rules engine. I need a way to convert regular straight quotes into curl (or smart) quotes. It'd be easy to just do a string.replace for ["], only this will only insert one case of the curly quote.

The best way I could think of was replace the first occurrence of a quote with a left curly quote and every other one following with a left, and the rest right curly.

Is there a way to accomplish this using javascript?

Answer 1

'foo "foo bar" "bar"'.replace(/"([-a-zA-Z0-9 ]+)"/g, function(wholeMatch, m1){
    return "“" + m1 + "”";
});

Answer 2

You could replace all that preceed a word character with the left quote, and all that follow a word character with a right quote.

str = str.replace(/"(?=\w|$)/g, "“");
str = str.replace(/(?<=\w|^)"/g, "&#8221;"); // IF the language supports look-
                                             // behind. Otherwise, see below.

As pointed out in the comments below, this doesn't take punctuation into account, but easily can:

/(?<=[\w,.?!\)]|^)"/g

[Edit:] For languages that don't support look-behind, like Javascript, as long as you replace all the front-facing ones first, you have two options:

str = str.replace(/"/g, "”"); // Replace the rest with right curly quotes
// or...
str = str.replace(/\b"/g, "”"); // Replace any quotes after a word
                                      // boundary with right curly quotes

(I've left the original solution above in case this is helpful to someone using a language that does support look-behind)

Answer 3

I don't think something like that in general is easy at all, because you'd have to interpret exactly what each double-quote character in your content means. That said, what I'd do is collect all the text nodes I was interested in, and then go through and keep track of the "on/off" (or "odd/even"; whatever) nature of each double quote instance. Then you can know which replacement entity to use.

Answer 4

You might want to look at what Pandoc does—apparently with the --smart option, it handles quotes properly in all cases (including e.g. ’tis and ’twere).

I recently wrote a Javascript typography prettification engine that does, among other things, quote replacement; I wound up using basically the algorithm suggested by Renesis, but there’s currently a failing test up waiting for a smarter solution.

If you’re interested in cribbing my code (and/or submitting a patch based on work you’ve done), check it out: jsPrettify. jsprettify.prettifyStr does what you’re looking for. If you don’t want to deal with the Closure dependency, there’s an older version that runs on its own—it even works in Rhino.