I use omnivore for longer articles and highlighting parts of the text. It also have a plugin to sync with obsidian. It’s really good, but I imagine self-hosting it can be tricky.
For a link-dump, I use Shiori. Could be anything vaguely interesting but I want to take a look later - works wonders for that.
And I have been a former pocket user, wallabag… But I stick with omnivore and Shiori.
Have you tried a low-code database, such as nocodb? I tried satisfy my need for structures data that way and it works for simple usecases. There are formulas too, but they (like spreadsheets) require some knowledge to set up.