waczdoc

A few days ago I wrote about a new tool called wacz-pdf, which will read a WACZ file and write out each web page as a PDF file. I got some quick feedback from Henry Wilkinson who suggested that it could be useful to allow a user to extract a Markdown file instead of a PDF. He also pointed out that the defuddle JavaScript library could be useful for doing it. defuddle is open source, and is part of the browser extension that Obsidian uses to save things from the web. Claude chewed on this idea for a bit. I happened to notice at some point that it decided to extract the HTML from the WARC and feed that to defuddle. That worked fine, but it wouldn’t work for all the web pages that are dynamically assembled with JavaScript. Since I didn’t have any WACZs nearby that had dynamic pages in them Claude decided there really weren’t many WACZs that contained pages like this. Even when I challenged that there were, Claude still opted to use replay as a fallback when absolutely necessary. Oh…