I don't do much PHP, so its probably pretty unoptimal. What it does is:
- Opens the index of all journals
- Finds the links to all journals (handily in most-recently-updated order)
- Opens each journal, finds and scans the contents of each journal entry
- Pulls out the first image link that looks suitable. Trys to ignore gamedev smilies and post icons.
- Splurges the results to your browser.
A brute-force takes quite a while, so its now restricted to only scanning the first few journals, and the first few entries in each. This cuts the time considerable, and as most people don't post images much it doesn't miss anything if you take a look every couple of days or so.
In other stuff, I've been making my little shooty thing more data-driven by pulling out config etc. info from xml files. Seems to be working nicely and dead easy to use as well. :)