bug#68254: EWW ‘readable’ by default

[Jim Porter <jporterbugs@gmail.com>]

On 3/17/2024 10:17 PM, Navajeeth via Bug reports for GNU Emacs, the 
Swiss army knife of text editors wrote:
> Thank you so much for the patch, @Jim! I dunno how to apply patches, but 
> I’ll learn and try yours out as soon as I can.

Even though I contribute to Emacs, I tend to use the latest proper 
release as my daily editor (so I'm using 29.2 now).[1] If I want to use 
a patch I wrote on an older release, I take the new version of all the 
relevant functions and then override the old ones using 'advice-add':

   (defun updated-eww-readable (&optional arg)
    ;; new implementation here
    )
   (advice-add 'eww-readable :override 'updated-eww-readable)

If the patch has merged to the master branch, I usually wrap that with 
'(when (< emacs-major-version 30) ...)' so that it doesn't do anything 
on the master builds, and also so I know to remove it when 30.1 comes 
out and I prune my init.el.

> I think a better way to go would be to have a /readability-off/ list for 
> the readability-minor-mode. In my experience, with the kind of sites I 
> open with EWW (textual sites without a lot of graphics or JavaScript), 
> the list of ones where ‘eww-readable’ doesn’t work is a lot smaller than 
> the ones where it does.

I was thinking about doing something like this. The list of regexps 
could include a way to say both "if this regexp matches, use readable 
mode" and "if this regexp matches, *don't* use readable mode". Then you 
could make the list look something like this:

  '(("^https://example.com/" . not-readable)
    ".*")

That would make every page except those from https://example.com use 
readable mode. I think that would be the most flexible for complex 
cases, while still being simple for the common case (a list of "plain" 
regexps for readable-mode pages). It would also make it easy to have 
most of a site (except for one section) use readable-mode.

[1] Mainly I just want to avoid having to worry about updating Emacs 
master and then ending up with a broken Emacs. Murphy's Law dictates 
that that will always occur at the worst possible time.