Really deep linking: Url + regex

Here's today's ~~boil the ocean scheme~~ idea — and it's not a new idea by any means. I'm sure it has occurred to many people at many times, I've just never seen it written down. (links to further reading welcome)

Just say you want to bookmark a particular paragraph on a particular web document.

Perhaps you want to perform the electronic equivalent of using a highlighter pen to point out a particular fragment of a document in its original context.

You can give the uri of the page, but you can't give a specific link to the actual paragraph you are interested in.

(Sure, if the author of the document provided a named anchor tag for the paragraph then you are in luck. But only semantic web fanboys and egomaniacs go to this kind of extreme)

What I'm thinking is that, instead, you could provide a little extra bit of regex-goodness in the url. Say for example you said:

href='http://secretGeek.net/regexUri.asp --highlight "Say for example you said:"'

...then this could act as a pair of instructions to the browser: the first one is "get from 'http://secretGeek.net/regexUri.asp' and the second instruction is "highlight the text that matches this regex: "Say for example you said:" (excluding the quotes). The instructions are separated by spaces, qualified by quotes and so on. Click on the link and the page is shown, with the relevant text highlighted.

Okay — i'm thinking four different things at once here:

Why stop at highlight? what other commands could the new commandline accept?
"Major security flaws waiting to happen"
Some people, when confronted with a problem, think "I know, I'll use regular expressions." (etc...)
Subtle differences in implementations between browsers.... NO!!!1!

And of course, as always, I'm hearing the voice of some strange ever-present character in my head (who sounds a lot like comic-book-store-guy from the simpsons) and he alternates between:

You can already do that in firefox with addins/greasemonkey, idiot. And
'course, there'a an emacs command to do that

Thoughts? Additions? Subtractions? Jakob?

You could, of course implement it unilaterally on your site in javascript. When it saw that kind of URL it could run a regex over the content of the page and do the highlighting of any matching elements.

Of course creating the URL in the first place would rely on end users learning regular expressions. While Jeff Friedl would probably love this to come to pass it doesn't seem likely to happen in the short to medium term.

A few random comments:

Since the space character has to be Url-encoded in links, you're relying on all browsers correctly splitting the Url and the "highlight" before making any requests - which is not backwards compatible with older browsers. And would result in a bunch of "href='http://secretGeek.net/regexUri.asp+--highlight+%22Say+for+example+you+said%3A%22%27"" requests and related server 404 errors.

I would propose re-using the existing fragment separator ('#') and making an additional constraint that the "highlight" is Url-encoded. That way, older browsers will just ignore it, while newer browsers can add the highlighting logic.

Of course, then we would have to add some way for those newer browsers to know what's a regular fragment and what is the new "highlight" fragment. Maybe it's treated as a regular fragment if it exists in the document, and as a highlight if the string doesn't match any element id's?

while it's always nice to puzzle over a regex now and then, perhaps XPath would be good here? However, with XPath you would be specifying the markup rather than the content which may be less enduring.

I saw a related idea for permanent url linking on the horror a while ago http://www.codinghorror.com/blog/archives/000068.html
Not quite the same kind of thing your proposing...but

Adobe has actually had this built into their browser plugin for a little bit: http://partners.adobe.com/public/developer/en/acrobat/PDFOpenParameters.pdf

Of course, it's not always reliable and gets tripped up sometimes.

There was/is an attempt to create a standard for this based on XPath:

http://www.w3.org/TR/WD-xptr

I've thought of this idea for a long time and think that properly executed it can be a hit. Myself though, I have still to figure out a good solution for it. But I often find I want to e-mail/message snippets to friends and colleagues with a highlight on the page.

@TextMarker
interesting stuff.
It worked ok for me, though it didn't load any style sheets (which use relative urls) and relative links were also broke.
Plus, not seeing the original page's url anywhere was a bit disconcerting.
see http://beta.marker.to/25463
(I'd like it if there was an 'about' link at beta.marker.to ;-) I want to see who the geniuses were behind it)

Really deep linking: Url + regex

NimbleText is a Powerful FREE Tool

Articles

Twitter, baby

Recommended Reading

Recommended blogs

Aggregated Links

Human Link Machines