Parsing complex HTML is definitely challenging, even with PowerShell. I had hoped to use the free tier of ChatGPT and their latest 4o model to help build a PowerShell function for HTML parsing, but I kept running into system limits and the AI often misunderstood what I was asking for.
I finally gave in and purchased the $20 subscription so that I could expand on my 2017 blog post about automating vSphere Global Permissions and add support for listing global permissions, which came at a request from a recent user.
It turns out calling the private vSphere Global Permissions API via the vSphere MOB to show all current vSphere Global Permissions is extremely difficult due to the complex HTML that is rendered by the vSphere MOB. In fact, it took 25 iterations before I finally arrived at the solution using ChatGPT's 4o model. In several of the iterations, it ended up going backwards in progress, so that was pretty annoying!
Not sure if this is the new fancy "vibe coding" trend that I had experienced ... 😅