This is pretty fucking cool – essentially an extension to Python (a pretty accessible programming language) that can take data from screenshots. So, if your bus transit company only has a webpage that outputs a bus' location using HTML but doesn't offer any sort of XML/formatted document/RSS feed/whatever, you can use the webpage's screenshot as an input. It brings to mind Apple's not-so-successful Web Clips on Dashboard that let's you isolate a part of a web page to display as a widget. This takes that one step further and lets you use the clip as an input, likely turning it into data a little more refined than just a .jpg (perhaps using OCR technology to convert it into text). With straight HTML this is a bit convoluted, as you could just take the text straight off the .html (or .php or whatever) file, but this system seems easier for the end users, as you don't have to use regexes or whatever to isolate the text you want. Sorry if this post was a bit geeky for most readers, but the whole idea seems revolutionary to me.