Niall’s Weblog

One of the most annoying things about this whole weblog business is having to write your posts in the browser. I can’t stand entering any more than a few sentences in a web browser text area. Not only because web browsers are so unstable, prone to crashing, or that its so easy to press the wrong key and kill the window or go back in the history and lose what you’ve written - but that you are forced to use the crummy browser text field component. I use Vi for most of my editing these days, and I have grown very fond of the movement keys (hjlk). Using these means I don’t have to move my hands from the keyboard to move the cursor, which reduces strain and increases speed. Of course I miss many other features, such as on-the-fly spell-checking (although Firefox 2+ has this), word completion, auto-save, auto-format, syntax highlighting, and bracket matching also.

I initially assumed that WordPress is too poorly thought-out to have something clever like an XML-RPC interface. Well, thankfully, I was wrong. WordPress in fact supports the Blogger, Moveable Type and a bunch of other XML-RPC interfaces. This meant I wasn’t going to have to screen-scrape and hack together some horrible and brittle HTTP client.

I do a fair bit of work in Python for my job, and I’ve come to rather like the language for whipping together smallish programs. Especially tasks involving processing files or URLS, Python particularly excels for these. While it would not be a big deal to use the Python XML-RPC library directly myself (its in the Python standard library since version 2.2), I figured I would use someone else’s code if at all possible. After a bit of searching, I found wordpresslib. This little .py file is available under the LGPL license and makes it trivial to perform a variety of Word Press operations from Python.

My rather trivial program simply allows you to submit a post in draft or published form from the shell. It accepts input either on STDIN or a specified file. You can also set a post title. For the hell of it, I’ve put it under a BSD license and packaged it up in a little tarball with wordress.py available for download here. Before using it, you need to set three variables in main.py - the URL of your Word Press install, your user and your password. Sample run:

$ echo this is a post | ./main.py -t 'a test post'

If you are seeing this post, that is proof that it works. There are many more features which could be implemented of course. Maybe I’ll bother maybe I won’t. I’ll see if I get the itch! Feel free to send me patches of course

For a long time I’ve considered Word Press to be an awful pile of poo. Simply consider that its written in PHP, a bad, buggy language, and is futher full of security holes of its own making. I was thinking about writing my own implementation in Python (maybe TurboGears or Django or my own thing) or C. If I was doing it in C, I could use the BSD-licensed undeadly.org code and plug in Clearsilver templates or somesuch. It would be incredibly fast and perhaps a bit of fun. The problem is not so much the software side of it - its trivial to throw a simple schema together to represent posts in SQLObject - but more the display and interface. Writing good forms and UI and so forth is highly time-consuming and I just couldn’t be bothered. HTML and CSS and JavaScript are hard. Word Press also has a ton of extra features which I probably wouldn’t have time to implement, yet could potentially be nice for my site. I decided I would leverage the scalable nature of the Internet Word Press community, all those hundreds of people working on little plugins and tweaks etc. Even if the average quality is pretty low, in the end it should work “well enough”. Much as I like polish and perfection, sometimes its nice to be lazy and accept the warm sludge vomited up by the PHP-abusing masses. So yeah, I went with Word Press.

Well, at least its running in a chroot()-jailed Apache, with some pretty strict PF rules blocking outbound connections by the Apache user in case yet-another-PHP hole comes out before we can patch our server.

I also decided to try out the reCAPTCHA WordPress plugin to protect my posts from spam-comments. reCAPTCHA is a very clever idea - take a huge amount of old books, OCR them, and then use the words that the OCR software failed to recognise as a starting point. Next make captcha-solving help digitize these old books. I like that it encourages a kind of “arms race” between those cracking captchas (effectively the same thing as OCR) and those making better captchas which in the end, no matter what, will benefit the effort of creating digital copies of old books.