finds a url inside a base64 encoded image string

[ralyodio]

I don't know if this is intended or not, but its causing problems for me.

a base64 encoded image data url is being recognized as a url, so there ends up being a n anchor tag wrapped inside the src attribute of the image when using

If my text contains something like this:

<img src="data:image/png;base64,iVBORw0K...">

Using this method

         text = URI.withinString(text, function(url) {
                return '<a href="'+url+'" class="ext">'+url+'</a>
         });

I end up with this:

<img src="<a href="data:image/png;base64,iVBORw0K...">data:image/png;base64,iVBORw0K...</a>">

...which obviously is invalid html.

[rodneyrehm]

.withinString() can't handle HTML. It never could, it never will. If you want to convert URLs to links on HTML content or the DOM, have a look at this example. Note that the example is using TreeWalker which is not available/fully supported in older browsers.

[ralyodio]

its actually a string with HTML embedded in it. Before it gets added to the dom.

I wonder if I can add it to the dom first, then filter on text nodes, and apy withinString to those nodes instead of the html string.

[rodneyrehm]

that's exactly what the jsbin demo is supposed to show ;)

let me know what you made of it. maybe we should add this method to URI.js' jQuery plugin?

[ralyodio]

this is how i'm doing it now:

https://gist.github.com/chovy/1e3f3df674e1557c230e

pretty easy to filter by textnodes first.

but my $message contains text which has <img src="data"> string, not a dom node...because its escaped...I think I need to unescape it first, then filter. but then I'm not sure how do add the non-text nodes back into the dom.

[rodneyrehm]

well, you can establish context within the callback function like so:

var text = "hello {{hello}} ((hello)) {{hello}} world";
text = text.replace(/(hello)/g, function(url) { 
    var original = arguments[arguments.length -1];
    var start = arguments[arguments.length -2];
    var length = url.length;

    if (original.substring(start -2, start) !== '{{') {
        return url;
    }

    if (original.substring(start + length, start + length + 2) !== '}}') {
        return url;
    }

    return "within double curly braces";
});

so this could be used to see if your match is preceded by /<[^<]+=("')?/

[ralyodio]

is there some kind of limit to the length of the url passed to the callback function for URI.withinString?

I noticed that on larger files, they all seem to be truncated. around 65k characters.

[rodneyrehm]

URI.js does not impose any limits. if there is something fishy, it may be RegExp itself.

[rodneyrehm]

The issue is resolved in v1.12.0 see the docs for an enhanced API.