Embed as many external resources such as images and CSS as possible into the HTML output files, rather than linking to external resources.
The use case for this option is to produce a single HTML file for an entire build that is fully self contained, and can therefore be given to consumers and viewed offline, much like a PDF.
Examples of embeddings done:
Examples of embedding that could be implemented in the future:
  • images are downloaded if needed and embedded as data: URLs.
    Doing this however has a downside: it would slow the page loading down. The root problem is that HTML was not designed to contain assets, and notably it doesn't have byte position indices that can tell it to skip blobs while parsing, and how to refer to them later on when they show up on the screen. This is kind of why EPUB exists: github.com/ourbigbook/ourbigbook/issues/158
    Images that are managed by the project itself and already locally present, such as those inside the project itself or due to media-providers usually don't require download.
    For images linked directly from the web, we maintain a local download cache, and skip downloads if the image is already in the cache.
    To re-download due to image updates, use either:
    • --asset-cache-update: download all images such that the local disk timestamp is older than the HTTP modification date with If-Modified-Since
    • --asset-cache-update-force: forcefully redownload all assets
Keep in mind that certain things can never be embedded, e.g.:
  • YouTube videos, since YouTube does not offer any download API