The hardest part for me when building a HN reader was getting each article metadata to render the meta image / descriptions. I know the main site doesn’t do this, but my favorite iOS hn reader app (Octal) does this so I’m used to it and quite like it.
I wanted to do it all locally / serverless but I struggled to find a good client friendly library for parsing meta tags that was 0 deps and typed so I wound up making a cloudflare worker to bypass cors and manually streamed the html to extract and parse the head element with a bit of regex and elbow grease.
It worked out, but I wish the HN api would cache and provide that stuff once to save the authors from the extra load of all the clients like mine that need to make extra requests to each articles home page.
