Luke's uberFeed

04/04/09 @ 10:47:58

with this itchy rash of web 2.0 sites out thee i've found i have a host of sites that im updating (twitter being the most regular, i really do like micro-bloggin') and it's all been a bit disjointed. now for the the more techi amongst you, why would you want multiple feeds coming in from one person? just fills your newsreader with junk. it's also kinda annoying having to point people in the right direction whenever something get's updated, eg:

a new crop of photos have gone online and i give my old man a call: "hey dad, want to see my new photos?", "sure son, are they on that twithead thing you gave me?", "uhhh, no - they're on picasa, you know the photo-sharing place", "nope, don't recall - do you have a link?".

something had to be done.

RSS to the rescue

you might have seen the little orange rss buttons floating around the web?! basically they provide a link to the raw data of a webpage and are usually consumed by another electronic service somewhere to re-produce the headline, link and brief description of the original article. so, in a nutshell, im able to grab these rss feeds from the sites i post to (blog, twitter, youtube, picasa etc) mix them all together and chuck them back out in date order and in the format i decide.

technical challenges (management speak for "pains in the arse")

only bother reading on if you're a web techie... (if not just go to Luke's uberFeed you'll get the idea.

feeds can be slightly different to one another which means some intervention would be required to make them all play nice. for instance, an atom feed is different to an rss feed - to be honest im not sure of all the differences - something about harvard doing one and someone doing something else, and.. well i don't really care, all i know is if it's structured - i can get at it :-) which is what i did. coldfusion offers some excellent xml functions and with a little bit of messing you can soon pull the fields you need out.

simply put if you create an object by calling a url with cfhttp, then parse it with xmlParse you can then get at the data fairly easily:

something like:

<cfhttp url="http://lukenukum.blogspot.com/feeds/posts/default?alt=rss" method="get">
<cfset doc = xmlParse(cfhttp.fileContent) />
<cfoutput>#doc.rss.channel.item[1].title.xmlText#</cfoutput>

will bring you back my latest blog post title (swap the 1 for 2 to get the second title etc). you can then adapt this to get the url and description, it's not rocket science but it's a handy thing to know. all well and good, this enabled me to produce a pretty list and then repeat the process for my other feeds - all similarly tagged (albeit slightly differently) in the XML.

but how to get them mingled with each other so that they form a nice time line? a faux query of course. creating a query object from scratch means that once you have it you can query it just like you would say an SQL database, so using something like the following (ideally in a loop over your XML object):

<cfset thisCell = QuerySetCell(thisRss, "title", doc.rss.channel.item[1].title.XmlText) />
<cfset thisCell = QuerySetCell(thisRss, "published", doc.rss.channel.item[1].pubDate.XmlText) />

it populate that query object (thisRss) which you can then treat just as you would a database query (ie - select title from thisRss order by published) - magic. theres a whole bunch of other stuff going behind the scenes including date formatting and a fix for twitter which clears @[twittername] replies from tweets - they now get filtered out along with other bits of ugly HTML via some regular expressions.

tbh, im getting a bit bored of typing now, suffice to say im fairly happy with Luke's uberFeed (oh, and there's an rss spawn of it here). i know friendfeed is a very similar and far superior option *but* it's all a bit too much... i just don't need more than a link and brief description of something i posted.

i guess the only question that remains is why you'd want to efficiently bring together a bunch of prattle... good question, good question :-/

keep it real campers.

blog posts

Coldfusion
- a much faster homepage
- pop3 picture extraction and API fun
- lukes uberfeed
Random
- download festival 2009
- hashtag invented
- learn some new words
- xmas messages