Fixed extension

[varunneal]

See here: https://github.com/do-me/SemanticFinder/pull/34#issuecomment-1694345123

Additionally, a progress bar has been added for the summary feature on the website. Demo that here: https://varunnsrivastava.github.io/SemanticFinder/

[do-me]

I'm just reviewing the extension and found it to be great already - thanks for your efforts. It works as expected and can be used for different tabs. I noticed that it even manages to scroll to the right section within the iframe which is quite advanced!

Some smaller bugs I found:

• I already fixed the popup width (content was larger than the actual popup box leading to some cutoff)
• When working with large docs it's problematic to log all the content (if there are e.g. 100 pages or so), so I commented these lines out
• the debounce function is not working as expected. Instead of only firing function calls every e.g. 100ms with only the latest call it simply delays all calls, i.e. with every key stroke a call is fired but simply delayed by 100ms. It only becomes noticeable when working with larger documents however
• I don't remember if you said it's supposed to work for pdf's or not - if so, it throws an error atm. Tested with this pdf

If it's not supported, let's maybe add a friendly notice to the users that pdfs can't be read yet and instead the web version of SemanticFinder should be used.

I'd say the latter two are minor issues so I'll go ahead, update the logo on the web version and prepare some posts.

[varunneal]

Thanks for reviewing! I'll review the latter couple changes.

• Unfortunately, given how "hacky" the current pdf solution is, it only works if the url of the page ends with ".pdf". For example, this pdf would work to some extent (however scrolling is still broken).
• While I don't have an idea for a great fix, I'll try to add more friendly warnings and notices instead of the current error screen.
• Feel free to use your new logo/logos wherever you see fit.
• I'll add an extensions setting page soon where at the very least the # chars setting should be changeable

[do-me]

Awesome!

• I wasn't aware that some pdf's were already working too - that's neat. When posting on HN again, I'll add a few open points where we need help like accessing content or revamping the UI.
• I have an idea for a smaller logo that fits in the browser plugin bar too: just the magnifying glass with the circle. Will add it shortly.
• When adding the settings, it would be awesome if you could allow for switching models too, so that we don't need to build two versions anymore, one for EN and multilingual usage.