2020-11-27 Jean Louis wrote:
Now we have `evince' PDF viewer that can open PDF I think by page
number and by query but it cannot do the equivalent
`evince-store-link' so user has to think about the file name and page
number and so on.
However xpdf (evince predecessor in respect to PDF engine) allows to
define custom bindings in config file, so store link could be
implemented. It seems, the feature was lost on the way to user-friendly
interface. Unsure but maybe store link could be configured for browser
built-in PDF viewers.
We have browsers that each of them think for themselves. Each can
store bookmarks but hardly provide such to external programs.
I agree that bookmarks as they implemented in browser is something poor.
Forget external programs in the context of modern browsers (either you
like it or not). De facto, extensions should communicate with HTTP
servers, to protect users, access to filesystem is not allowed any more.
However some API to manipulate bookmarks exists, do not know if it is
really useful:
https://developer.mozilla.org/en-US/docs/Mozilla/Add-ons/WebExtensions/API/bookmarks
Almost none browser can store specific paragraph based bookmark. They
rely mostly on anchors as finely specified greed. Bookmarks could be
by HTML page number or query or paragraph.
In principle, any paragraph could be addressed using XPath
https://developer.mozilla.org/en-US/docs/Web/XPath/Introduction_to_using_XPath_in_JavaScript
but it is extremely fragile since link will be likely invalid after web
site redesign or modification of the text. There is no interface for end
users but web developers use it for UI tests if there is no more
reliable option.
There was text search addressing initiative "Scroll to text"
https://github.com/WICG/scroll-to-text-fragment/
The problem that URLs have no room for such extension points and some
web sites abuse #anchors to determine what content should be shown, so
addressing of particular element is not possible at all.
