How to make Emacs recognize sentences after a set of parentheses/brackets?

Question

I am trying to use Emacs's sentence-end functionality to navigate around a file with citations (and double-spaces between sentences), and I'm having some trouble trying to get Emacs to recognize sentences with brackets/parentheses after a period. For example:

Some text.  "Some quote."  (Something in parentheses.)  Something with a citation.[cite]  Some more text.

Using Emacs's built-in forward-sentence and backward-sentence works fine with the first, second, and third double-space sentence breaks in the example above, but the fourth sentence break is not recognized.

Interestingly, enforcing (setq sentence-end-without-period t) seems to also ignore double spaces when a set of brackets are in place. In the example below, Emacs can delineate between sentences one, two, and three—but it will combine sentences four and five:

Sentence one  Sentence two.  Sentence three  Sentence four.[cite]  Sentence five.

Is there a way to make Emacs demarcate sentences by ALL instances of double-spacing, at least after parentheses/brackets? Thanks!

Answer 1

The docstring of a user option sentence-end-base says, that it is a variable set to a

Regexp matching the basic end of a sentence, not including following space.

The default regexp is [.?!…‽][]\"'”’)}»›]* currently. It translates into matching one of .?!…‽, including followed by any number of ]\"')}.

Hence, it should identify the sentences in the test snippets you have provided with a condition there's a period right before a closing paren/bracket.

[I agree we may expect sentences without a period to be identified having sentence-end-without-period disabled; this requires more look into how exactly the function sentence-end utilizes these two variables. Looks like it works only for the cases without paren/bracket involved. This might be a bug in Emacs behavior or --more probably-- documentation not explaining it well enough, so M-x report-emacs-bug perhaps.]

If the default behavior doesn't suit your needs, adjusting the value of sentence-end-base could be a solution. For example, consider any closing bracket (before spaces) to be a mark of an ending sentence with

(setq sentence-end-base "[.?!…‽][]\"'”’)}»›]*\\|]")

[Illustration below uses a bit different regexp syntax. All the illustrations match the behavior I have in Emacs.]