Generate epubcfi from unique text #1430
Comments
|
I should also say that this was my solution. Anything which makes more sense, makes more sense. |
|
Yes, it does make sense to use the text to correct the CFI. (There's actually something similar to this in the CFI spec. It's called a text location assertion, though such assertions contain not the highlighted text itself, but rather the text that precedes or follows it.) Anyway, to implement this, the main issue is not really related to CFI, which can be straightforwardly generated once you have the correct nodes and offsets. The problem mainly depends on how you would like to match the text, and that depends to some extent on how the text is generated from the book in the first place. foliate-js provides a |
|
@johnfactotum I have completed my project but the library that it depends on is not fully capable of generating cfi's. It puts Also your library is hard to understand for me. I don't know js. I use chatgpt and common sense. thank you. |
|
Ah, I see. Well, you're using epub-cfi-generator, which according to the readme, is simply using readium-cfi-js. So the bug your seeing is readium/readium-cfi-js#23, which has been fixed upstream. So you can either ask them to update their copy of readium-cfi-js, or just use readium-cfi-js yourself. If you want to use foliate-js, it can be indeed a bit more involved. But here's an example: import { makeBook } from './foliate-js/view.js'
import * as CFI from './foliate-js/epubcfi.js'
const file = /* a Blob or File object, or a string (a URL to the file) */;
const book = await makeBook(file)
const index = /* the index of the spine item, e.g. `0` for the first section */;
const range = /* a DOM Range object of the highlighted text */;
const baseCfi = book.sections[index].cfi ?? CFI.fake.fromIndex(index)
const cfi = CFI.joinIndir(baseCFI, CFI.fromRange(range))Alternatively, you can simply use the import './foliate-js/view.js'
const document = /* a DOM Document, which if you're in Node.js, you need to create with e.g. JSDOM */;
const view = document.createElement('foliate-view')
const file = /* a Blob or File object, or a string (a URL to the file) */;
await view.open(file)
// get CFI from index and range (same as the previous example)
view.getCFI(index, range)
// or search in the book
const generator = view.search({
query: 'your text here',
matchCase: true,
matchDiacritics: true,
matchWholeWords: false,
})
for await (const result of generator) {
console.log(result)
} |
My attempt: https://github.com/orhnk/Annot2CFI
There are some problems with miscalculated cfi's like:
The library I use cannot find that last node's possition correctly.
I'm going to use this program to fetch annotations from my kobo device (and hopefully a lot of people are going to use it :D)
What should I do?
Any integration ideas for your epubcfi.js? @johnfactotum
Or maybe foliate can have a feature like fixing up some epubcfi's by itself if it's easy to detect like the above unique text example?
A lot of the text people highlight from the book is unique. It's like... Why do you think they highlight it then?
Yours truly
The text was updated successfully, but these errors were encountered: