Feature Idea: Resolve annotations without CFIs by searching for the text and generating a CFI on first click

[kit-lawless]

I often import highlights from Kindle clippings and creating a simple script to convert them into Foliate-compatible JSON is easy enough but generating the CFIs is harder. Annotations with no CFI show up in the side bar but nothing happens when you click on them. However, it seems to me that Foliate's search functionality could solve this problem.

Proposed Behaviour

When an annotation has no CFI, Foliate could:

1. Use the annotation’s text as a search query in the current book.
2. If there is exactly one match:
   • jump to that location
   • compute a CFI for the matched range
   • save the CFI back into the annotation (so subsequent clicks behave normally)
3. If there are multiple matches, optionally:
   • open the search UI pre-filled with the text, or
   • do nothing
4. If there are no matches:
   • do nothing

This would leverage existing functionality to make Foliate much friendlier to imported annotations and could be built upon in future for greater import/export functionality for annotations.

Questions

• Am I misunderstanding anything that would prevent this from working?
• Has anyone experimented with something similar before?
• Is this is line with the tool's design goals/philosophy?

Thanks for any feedback, just thought I would ask before I try prototyping this myself.

[johnfactotum]

I do think something like this would be useful. The main issue is in the implementation. Like how does it match for the text? Foliate's search module might work but it's not really designed for this. Does it search the whole book or only in some specific chapter? And it would probably be better if you can do it for all the annotations at once.

One could also argue that generating CFIs should be for the external tool that converts the annotations, and doesn't have to be implemented in Foliate. But it's still useful, for example, when the book is edited. CFI itself has text assertions (which are not supported by Foliate), so one can say that matching by text is something that is quite appropriate or even necessary for any CFI implementation.

[kit-lawless]

Thanks for the reply. I'll build an external tool first that parses Kindle clippings, searches the EPUB for the text, and generates CFIs to produce Foliate-compatible JSON. This way I can work through the edge cases (search scope, fuzzy matching, batch processing) without needing to modify Foliate itself. If it works well, I'll think about if/how this functionality would be brought into Foliate itself.