I went to Codeworks Austin on Saturday, and it was kind of eye-opening. Lots of good info about how to better approach PHP development.  Here’s what I learned:

1. First and foremost, even after developing for over 10 years, I have a LOT to learn. In fact, there’s an alarming amount of stuff I DON’T know.My biggest issue, is that I originally learned PHP4 10 years ago, and it was very procedural. You wrote code from top to bottom, and it ran from top to bottom. After I started getting into it, I got my first programming job doing mostly Flash, ActionScript 2, and classic ASP using VBscript. I basically set aside learning PHP, since there wasn’t really a need in my day-to-day life.  I picked it back up a couple of years ago, mostly for personal sites, but PHP5 had many new features, and I never had need to learn them all.
2. I really need to master OOP. Again, my previous programming experience was ActionScript 2, classic ASP, and PHP4… which are all mostly procedural or functional.  I’ve jumped ahead quite a bit in my OOP experience with getting into some of PHP Frameworks like CodeIgniter and CakePHP… but I need to get more comfortable writing my own classes outside of the frameworks.
3. Speaking of frameworks, I might need to check out Zend. It seems a lot of people there were using Zend Framework, and the speakers seemed pretty happy with it.  Also, I think I was the ONLY person there that isn’t using PEAR.  So I should probably get on that as well. Though, working on a Windows machine, it doesn’t look all that straight forward.
4. Speaking of Windows not being easy… it seems most people there were on Macs, and many seemed to be Terminal/Command Prompt ninjas. Since current Macs have Unix at their base, there is a lot of built in functionality that makes setting up a PHP dev environment a lot easier.  However, I think I may go the Linux route, because… why not just go as hardcore as possible?
5. Subversion/ Version Control. I’ve submitted one bit of code to Github, and that’s the ONLY time I’ve ever checked anything in or out, ever.  In a multiple dev environment having version control is a necessity,  but since I’ve been a “lone ranger” for all of programming career, I’ve never bothered to learn or implement it. Moving forward, with both my personal advancement and expanding my current job, I really need to get this up and running.

While the actual talks themselves were okay, they were mostly “overview” type stuff. Very little detail.  But just seeing how other devs worked was the most education thing I took away. Lastly, I really want to get more involved with other events like that… and take my coding to the next level.

I like to play around with PHP and APIs and create silly, stupid things every now and then.  After recently watching a number of zombie-themed shows, namely The Walking Dead and Dead Set, I thought it would be funny to have an app that would “eat” the brains of people’s Twitter avatars. And for whatever reason, I thought having a Zombie Jesus doing the eating would be even funnier.

I worked out the steps that needed to happen: 1) get a submitted user’s avatar from Twitter, 2) process the image so it looks like Jesus is eating the “brains”, 3) display/save the image, 4) send out a Tweet with a link to the picture.

Getting the avatar from Twitter is a fairly simple function:

function getAvatar($screen_name){
     $url = "http://api.twitter.com/1/users/show.json?screen_name=".$screen_name;
     $ch = curl_init();
     curl_setopt($ch,CURLOPT_URL,$url);
     curl_setopt($ch,CURLOPT_RETURNTRANSFER,1);
     $return = curl_exec($ch);
     curl_close($ch);
     $return = json_decode($return);
     return $return->profile_image_url;
}

Since you don’t need authentication, you just “curl” to get the user’s info, parse the json, and return the avatar’s url.  Processing the image isn’t so easy. I’ve used the GD2 library in the past… for another simple app that put a custom message on Charlie from “Lost”‘s hand, in the episode where he wrote “Not Pennys Boat” on his hand, which I can’t seem to find now.  GD2 is powerful, but not user-friendly at all. So I searched for a better solution, and found the amazing WideImage library. It does everything I needed, and I was able to map out the process I needed in a few minutes.

I grabbed the image from Twitter, resize it to 60px by 60px, then rotate it -45 degrees. Then I have a Jesus png that I use as a mask, with the section where the avatar goes being cut out… and some blood coming down for effect. To do the masking effectively, I had to then resize the avatar’s canvas to the size of the main Jesus pic, and position the avatar on the canvas to the area where the open area of the Jesus pic is. Then you merge the two pictures to make the final image.

After that, I just used “writeText” to put the caption on. I had do a second “writeText” in black for the bottom part, to give it a shadow, since just the white was being washed out. Once all that is done, I save the file. For the filename, I was originally using the timecode… but if by chance multiple people hit the site simultaneously, it might be hosed. So instead, I took the two submitted usernames and the timecode… and ran md5 on it. I’m assuming that will give totally unique names.

I didn’t want to be serving up the images solely on my machine, so I decided to use the TwitPic API to save the images. This also allowed me to skip using the Twitter API and do the actual post from TwitPic as well. Since this is a simple, fun project, I decided to use the legacy v1 of the API, because I didn’t want to mess with oAuth.  Here’s that function:

function sendPic($file,$to,$from) {
     $key = ""; // TwitPic API key
     $consumer_token = ""; // Twitter application consumer key
     $consumer_secret = ""; // Twitter application consumer secret
     $oauth_token = ""; // the user's OAuth token
     $oauth_secret = ""; // the user's OAuth secret
     $ar = array (
         "consumer_token" => $consumer_token,
         "consumer_secret" => $consumer_secret,
         "oauth_token" => $oauth_token,
         "oauth_secret" => $oauth_secret,
         "key" => $key,
         "message" => "I just ate @" . $to . "'s brains at the request of @". $from ,
         "media" => "@$file"
     );
     $url = "http://api.twitpic.com/1/uploadAndPost.json";
     $ch = curl_init();
     curl_setopt($ch, CURLOPT_URL, $url);
     curl_setopt($ch, CURLOPT_POST, 1);
     curl_setopt($ch, CURLOPT_POSTFIELDS, $ar);
     curl_setopt($ch, CURLOPT_RETURNTRANSFER,1);
     curl_exec($ch);
     curl_close($ch);
     return true;
}

And that’s basically it. The only other bit of code-fu I used was to check the Twitter names people might put in… since sometimes people will use the “@” in front of the name, and other times they won’t. So I just ran a “strpos” checking for “@”, and if it’s at the 0 position, I “substr” to position 1.

As far as security, there really isn’t any.  Since there’s no mySQL, I wasn’t too worried about it… as there’s not much to exploit with the simple code I used. I did set a session with the timecode, and if you try to submit the form more than once a minute, it sends an error… though that would be trivial to overcome.

Other than toying with it some more, I’m pretty much finished. Being wide open as it is, it’s very very VERY susceptible to spam… so I may implement a bit of oAuth eventually. Other than that, it’s really only amusing for a few times… after that, it’s just a bit annoying. So we’ll see if anyone else finds it humorous.