[Feature] linkifying URLs

Hi! I was just checking out the roadmap [hp.jpsband.org] and it said you were planning in the 2.0 release for linkifying URLs. I just want to say that would be a EXTREMELY helpful feature because that is one of the biggest nightmares I have had to deal with. I spent over a week on it and I still couldn't figure it out. Its easy to try and linkify text with a simple regex but tons of problems arise if the text it detects already happens to be a link and the fix for it to be done right is complicated.

I don't know if these will come in handy for because you have a better way of parsing it, but when you do it here are some links for code done for it:

(one of the better ones) [code.iamcal.com]

(other ones) [www.zend.com] [www.truerwords.net] [www.coffee2code.com]

Oh yeah, these options would be nice for it: If you set it to detect Everything it will pick up: google.com, coolsite.us, blah.net (major domain extensions) If you set it to detect Strict it will pick up only: http ://www.google.com, http ://coolsite.us, or http: //blah.net

And options that allow you to truncate the displayed URL, or just show domain, or no mangling of the displayed URL.

Edited 3 time(s). Last edit at 04/04/2007 12:05PM by Ambush Commander.

Yeah, it's tough (there's a reason why it's marked COMPLEX). What I was planning on doing was hooking in a token filter, which would scan Text tokens for URI-like constructs, and linkify them. It's simpler than auto-paragraphing, to be sure, but still a bear.

I suppose this could be implemented with regexps. What you'll need to do is create a regular expression that globs up things that look like URIs, as well as things around them that might indicate that they are inside a tag (quotes and gt/lt signs come to mind). Then, do a preg_replace_callback(), where the callback function analyzes the matches and determines whether or not to do the replacement, or simply return $matches[0]. Complicated, to be sure. I'll bump up the legit approach on my priority list.

HTML Purifier, Standards Compliant HTML Filtering

I wonder if 'linkifying' of URLs, when the feature arrives, could be optionally turned off (or on).

Some may say that HTMLPurifier should do only what its name suggests - purify HTML, correcting input data only to the extent that is needed, and that parsing text to linkify, which ? might interfere with the functionality of the larger software that HTMLPurifier is embedded in, should not be its goal.

Ah yes, it wouldn't be enabled by default. Definitely no. This is part of the reason why it's so late in the changelog: it has nothing to do with filtering per-say. "Beyond HTML" sums it up nicely: there are certain features not part of HTML that are very convenient, and are easier to implement within HTML Purifier.

HTML Purifier, Standards Compliant HTML Filtering