Media Catalog
A catalogue of our books, music and films, with OCR that reads the text off a photo of the cover.
We, like many folks I'm sure, have a large number of videos in our collection. They are mostly old VHS, DVD's and Blu-Ray and yes, we have accumulated a lot of them over the years. I never worried about them much but often thought it might be nice to have a list of them.
I came across an article online where a person described how they have found an open source package that helped them accomplish what I wanted to do. I looked at the software they used and some others and decided I wanted to do something different. Instead of setting up the software then sitting down and entering every item and all the details I wanted to take a photo of the cover art, upload it to the software and have the software look it up and download all the details of the video.
I told Claude what it was I wanted and it was done in minutes. This one was done a while back using Sonnet but it still worked well. It produced the site, I installed it on my home server, added an entry in a Caddy file and it was live online. When I say installed I mean I just copied the archive I downloaded from Claude, copied it to my server, unarchived it, ran npm install and restarted caddy.
The prompt was pretty simple. In other cases I have developed full requirements specifications I could give to a development team and given those to Claude but in this case I just gave a sort description and it worked. The site did as I asked but the OCR used to parse the photos didn't work well but Claude had compensated for that and when that failed the site went to a search screen where I could enter the name of the video and it would complete the search. After using it I realized the image upload was fun but didn't work well, the manual search, however, worked great. I asked Claude to expand the library to include TV shows and books as well as a simple login screen. I had to sign up to Google Books, OMDB and TMDB to get API keys but they are free and seem pretty complete. I gave Claude the keys in a project file and pointed it at the sites where the API was and it did all the rest. I have always found what I was looking for so far but I have a lot of movies and books to enter yet.