X-ray for teams and characters (single word or phrase highlights)

[michael-kucek]

Hello!

While the x-ray prompts are great for comprehensive analysis, I would love to have a way to run a prompt on a single word or phrase while providing the context of the book up to the current point.

Right now, I'm reading a book that has a city named Silverlight. When I use the dictionary or wikipedia prompts, it returns information about the old Microsoft Silverlight software instead of information based on the book.

Readwise Reader has a really great implementation of this (though they don't provide the prompt): https://docs.readwise.io/reader/guides/ghostreader/default-prompts#internal-x-ray-character-place-term

I just discovered this plugin today, and I'm loving it so far. I'm happy to help out where I can (I briefly used Lua at an old job), but I might need to be pointed in the right direction to get started.

[omer-faruq]

Using the content of the book every time you need to look up a word in a dictionary may use a lot of tokens since on each event you need to send the content again, and it may also be time-consuming. But you can enlarge the text that is used in the dictionary (before and after the word you highlighted) via modifying this part of the code ( giving a larger portion of the text would help)

Maybe we can have a button to have the ability to ask custom questions based on the text that's read. For now, you can use the recap and ask additional questions button on it.

Thanks for the link; we can use the ideas on it to improve this extension.

You are more than welcome to contribute to the project. I personally don't know much about the Lua language, but I am using the Windsurf (AI) to do modifications. Providing the whole KOReader folder as a workspace helps a lot to use the functionalities already existing in the KOReader for our extension. Of course, I'm reviewing the code and making recommendations to AI to fix or fixing some myself. That's my approach to this particular project.

[michael-kucek]

Thanks! I'll take a look at the code and see if I can come up with a more elegant solution than just throwing all the text at the LLM.

[omer-faruq]

You can use this feature to search for book-specific terms and characters: #138