https://bugs.documentfoundation.org/show_bug.cgi?id=35107
--- Comment #20 from [email protected] --- This one is pretty interesting. As Jacobo found back in 2014, we miss the actual object associated with the link, which should be a child of the paragraph. Yet this somewhat works with Orca already, because it's pretty forgiving and will announce links by their mere existence *given there is no object replacement characters (U+FFFC) in the range*. However if you have any U+FFFC in the result, it won't, and will try and figure out what to replace it with using the existing hyperlinks. And this situation exists: just add a comment to a line containing a link and it will not be presented as such anymore. So to properly fix this, we should: * Add children to the paragraph with ROLE_LINK and implementing Atspi.Hyperlink for each link * Add the appropriate name to the Hypertext entry * Report the new children in the paragraph Hyperlink's get_object() -- e.g. Atspi.Hypertext.get_link(paragraph, 0).get_object(0) * Place replacement characters in the text in place of the links (U+FFFC). Maybe this could be optional, but as Comments do this I'm afraid it's gonna be needed here too (and it's not a bad idea, other things like images or frames will eventually have to use this). Additionally, having the link reported on its own would allow to report e.g; state VISITED for visited links and such (if we have the info). For the expected results, if the APIs aren't clear enough, I guess looking at how browsers do this is a good inspiration (at least Firefox, most probably Chromium as well). As an aside as it's mentioned in the OP, Accerciser currently does not handle what LO provides very well so its output is incomplete and confusing. I have suggested a fix in https://gitlab.gnome.org/GNOME/accerciser/-/merge_requests/102 -- You are receiving this mail because: You are the assignee for the bug.
