Used AI to assist Bill from Soldersmoke

I'm a big fan of Bill Meara's Soldersmoke podcast and blog.  It's always a pleasure to hear from Bill and last week he contacted me asking for help with the website where he keeps an archive of the past editions of the podcast. 

Bill has been editing the HTML of the page manually using a web editor that had hit a page size limit of 1MB and would no longer let him add new podcasts so it had fallen behind. Here's how the page looked:

At first glance I thought this was an easy little job, get in via FTP and paste in the new episodes. When I looked more closely at the HTML, and ran a validator on it, my heart sank. It's a credit to the robustness of web browsers that they managed to render this as well as they did. The podcast has been around for 20 years and while some of the HTML was obsolete, worse was that Bill's copy pasting from his blog had all sorts of errors in it. Mostly table tags that weren't balanced etc.

The HTML had ballooned out and had parts that looked like they'd been generated in Microsoft Word's export to Web - huge runs of fonts and styles repeated over and over.

I thought this would take many days of manual clean up and editing so I turned to Claude code for the project. I downloaded the various existing pages to a local folder, explained what we wanted and mentioned the blog as a place to get missing information.

Claude installed BeautifulSoup and proceeded to write python scripts to extract just the needed information from the existing broken HTML. It also looked at the blog to get missing info, looked at YouTube and even got the modification dates of the mp3 files to figure out publish dates. The result is at soldersmoke.com and now looks like this:

The HTML now validates without any complaint and it's much more efficient and compact than before.

AI will put web developers out of work to some extent. On the other hand it's fantastic at doing what would have been an incredibly time consuming menial task. This kind of job might not have been done at all except for this marvelous new technology. It's not perfect but it's amazing what can be achieved.

Thanks to Bill for all his great content, it's great to see the podcast archive remains available and up to date.