b00kwyrmy/docannotationstonote

This project is a plugin for the Supernote, a digital notebook device, that pulls the highlights, underlines, circles, and handwritten marks you make on a PDF and turns them into a clean, organized note. The idea is that when you read and mark up a document on the device, you should not have to manually retype or copy every important passage afterward. Instead, this plugin scans the marked-up document and rebuilds those marks into a note, keeping everything in the same order it appears in the original document. Different kinds of marks are handled differently. A highlighted passage becomes an image snippet in its original highlight color. Underlined text and bracketed passages are lifted out as their own entries, still in reading order. Circled or lassoed areas become either the enclosed text or an image crop if the area contains a diagram rather than text. Freehand handwriting and doodles are pulled out as ink on a clean background, keeping their original color. The plugin also removes duplicate marks and stray accidental dots, and it can detect headings within the extracted content. To use it, a person marks up a PDF as usual on their Supernote, then opens the plugin and chooses how much to extract, such as just the current page, specific pages, or the whole document. The plugin pulls out the marked content, and then the same plugin is run again inside the note where the person wants that content pasted in. Support for EPUB ebooks is still experimental. Unlike PDFs, EPUBs can reflow their text into different layouts, so the plugin has to work from a rendered image of each page and automatically correct for that shifting layout, recovering marks as image crops rather than editable text. This works reasonably well for books that are mostly plain text, but books with lots of images or unusual page layouts may not line up as reliably. The project is written mostly in Kotlin with a JavaScript and TypeScript frontend, and it is released under the MIT license.