At work, I'm hacking on a courses database. We use AxKit to generate course information pages, searches, bookings, etc., all back-ended by a simple OO library I've written and ultimately an SQL database. Very cool, all perfectly buzzword-friendly.
Unfortunately, the data is currently in all sorts of places, and the best way to get it to me as in several flat text files. So once I think I've got everything sorted, someone comes along and says "Can you add this list of descriptions to the database?" and provides me a text file linking course codes to descriptions. I had been writing a series of little hacks to parse the file and insert the data into the database, and then I thought "Why not solve the general case?" The result is squalk. ("SQL awk") It's not very awk-like, but it lets you do things like this:
squalk -F: -d courses -e 'update courses set category=$2 where code=$1'
That's to say, connect to the courses database, then rip through the text file splitting on :, setting the category to be the second field where you match course codes in the first.
Neat, eh? It's on CPAN; the version there doesn't support transactions but the one on my hard disk does, so there.
(Of course, then my cow-orkers pointed out that this doesn't buy you much more than awk -F: '{print "update courses set ..."}' | psql courses. Ah well. It was fun while it lasted.)
