First off, let’s define aggregation as taking a bunch of things and putting them all in one place.
Let’s try something. I’m going to read off a bunch of web services. Which ones are you a member of?
- Digg
- Google Reader
- Google Talk
- Gmail
- Mixx
- Reddit
- del.icio.us
- Furl
- Ma.gnolia
- StumbleUpon
- Jaiku
- Pownce
- Twitter
- Seesmic
- Vimeo
- YouTube
- Flickr
- Picasa
- SmugMug
- Zooomr
- Tumblr
- iLike
- Last.fm
- Pandora
- Goodreads
- LibraryThing
- Amazon
- Disqus
- LinkedIn
- Netflix Queue
- Netvibes
- SlideShare
- Upcoming
- Yelp
- or any blog
- or anything else putting off an RSS feed
Each of these can be pulled into one feed by FriendFeed.com .
How many of those services are you a member of? What do you do when you want visitors to see all your posts in one place? Think about using FriendFeed and harnessing it’s API . We’ll use it to create a single central feed out of the chaos.
So how does FriendFeed work? Each member loads up their various accounts from around the web – Google Reader shared articles, Twitter posts, blog entries, YouTube videos, Flickr photos, del.icio.us bookmarks, all that stuff I listed earlier – and members can have discussions around each entry. It’s a furious stream of information and comments coming from all directions. blogs, news, pictures, etc – all flying by, with comments. It’s the daily web posting activity of thousands of people in one place.
Facebook already has a nice FriendFeed app so your friends can see your FriendFeed-aggregated posts.
Otherwise showing it on another web page (like a very cool business’ web page) means we’re off to the FriendFeed API.
First thing we need is a FriendFeed account. They’re free. Go to friendfeed.com and follow the instructions.
Then load all of your web memberships into FriendFeed so you’ll have stuff in your feed.
Then we need to download the API from http://code.google.com/p/friendfeed-api/ . Documentation and forums are available at http://friendfeed.com/api/ .
Then, in order to authenticate with the API, we get our new account’s Remote Key from http://friendfeed.com/remotekey .
That’s all the official FriendFeed stuff we need. Now on to our application. I called it FriendFeedFormatter.
It is not affiliated with FriendFeed. Use it all at your own risk. FriendFeed has stated that the API is young and could change at any time. That disclaimer out of the way…
My code does all the work of taking our nickname and remote key and producing an HTML snippet from our FriendFeed feed that we can then have included in a PHP page. Download example.zip from the show notes. It’s got all the code that I’ve written using the API.
http://www.davidvanvickle.com/friendfeed/example.zip
In the show notes is a link to see sample output. That will give you an idea of what to expect, and you can begin to think of how to embellish it.
http://www.davidvanvickle.com/friendfeed/build.php
There are only two files in example.zip. One is my FriendFeedFormatter PHP class. The other is build.php, which is the file you’ll customize and run.
Now find the FriendFeed API you downloaded. In order for my custom class to work, you’ll need to include the friendfeed.php file from the API in the same folder.
So when you’re done there will be 3 files uploaded to a folder on your PHP enabled web server. friendfeed.php from the API, plus FriendFeedFormatter.php and build.php.
Before we try to run anything we need to change a couple things inside of build.php. So open that file in a text editor.
Enter your FriendFeed nickname where it says “MY_NICKNAME”. Then enter your remote key where it says “MY_REMOTEKEY”. That’s it. There are more things we could toggle but let’s start here.
So save and close build.php and upload the 3 files to the web server. Now you should be able to hit your build.php with a web browser and see what happens.
If all is well then you’ll see the HTML table that the FriendFeedFormatter produced using the API. Now look in the webserver folder where you uploaded the 3 files. You should now see a 4th file. That file is what you’ll be including in your normal web pages. It is JUST the HTML table displaying your feed. Just a little piece of a web page, not the whole thing. Perfect for inserting into a page that already exists.
Why don’t I just write a Javascript snippet like everyone else? Well because I don’t want to slow my web pages with a call to a remote site. The FriendFeed servers are pretty fast, but I still don’t want that dependancy if I can help it. As long as I can run that build.php file on a regular basis, I will have a new enough include file for my site. And it will load right away.
So how do I run build.php on a regular basis? Good question. If you want to automate that, I have half an answer. If you’re familiar with running cron jobs, or if you want to learn how, cron would be a good way. Cron allows your server to perform a task – such as calling build.php – according to a schedule, like every few minutes, instead of waiting for a web user to come along and trigger it. Cron is usually running already. You don’t have to install it. Just add a crontab to call lynx, wget, or on older servers, the php cgi executable and pass it build.php and how often it should run.
Truth be told, I don’t have much experience with cron, but it seems to be where people go for scheduling scripts. I included some examples in the show notes that I scraped from the web to get you started. I’m sorry that I don’t have more tested examples.
Run every Monday morning at 4:41AM. (minute hour day month weekday command)
41 04 * * 1 /path/to/lynx http://www.mydomain.com/path/to/your/cron.php
or
41 04 * * 1 /path/to/wget http://www.mydomain.com/path/to/your/cron.php
or
41 04 * * 1 /path/to/php /path/to/your/cron.php
Samples from:
http://www.hackernotcracker.com/2007-04/run-php-scripts-with-crond-and-crontab.html
http://www.modwest.com/help/kb.phtml?cat=2&qid=105
More at
http://www.webmasterworld.com/forum40/1258.htm
Otherwise just keep build.php bookmarked and hit it every time you do something that updates your feed.
Fortunately for server performance, but unfortunate for us, is that FriendFeed is on it’s own schedule. Just because I update my blog, doesn’t mean FriendFeed is immediately aware of that post. It may take a few minutes for it to show up in the FriendFeed feed. (That’s a great reason to figure out how to make the cron way work)
The code to pull in the include file in your PHP page is the same code in build.php that displays the output in the browser. The FriendFeedFormatter->get_output() method.
Once you see the output, I’m sure you’ll want to customize it more. Go for it.
I built in a few enhancements.
One enhancement is date formatting. The date may come to you a couple different ways. One is in ISO8601 format. That looks like this. “2008-05-21T15:02:22Z”
Another format would be as a UNIX timestamp, or a long number that represents the number of seconds since January 1, 1970.
You should be able to look in the date column, see which kind of date you have, and change the value of the date format variable in build.php.
$ff_date_format = ‘ ISO8601’; // or ‘unix’
On one server I got one format, on another I got the other format. I think that’s based on the PHP version I had running but I’m not sure. Anyway, that’s why my code has two ways to format dates.Another customization is the title of the table. I recommend making that a link to your
Friendfeed page. The current set up has a link to your FF page, plus a link to subscribe to your feed.
The formatter has a couple features users may appreciate. One is inline Youtube videos. First the user sees a JPG image of the video, but when it is clicked, it turns into a playable video.
Another feature is link detection, which is programmed to happen with Twitter entries. The static URL’s become clickable.
The last feature is, when Flickr is detected, it lists the thumbnails of the images.
These are things I expected to happen from the API, but they weren’t there, so I added them to my formatter.
Open the formatter and notice how I can treat each service’s entries differently. I can detect Twitter, for instance, and scan the entry for URL’s to convert into links. Or I can detect Youtube and make the inline play feature happen.
The important thing is just to hack away to get going using FriendFeed to aggregate your public web activity.
