How to break CJK only after punctuation, not between characters?

Question

$ emacs file.html

We have one long line, with no ASCII whitespace to break on.
Before:

weeklyOSM 的想法是蒐集來自世界各地，與 OSM 相關的新聞，處理這些新聞並將其翻譯成盡可能多種語言。這應該有助於讓社群知道所有消息並克服語言障礙。

Do M-q (fill-paragraph), getting:
After:

weeklyOSM 的想法是蒐集來自世界各地，與 OSM 相關的新聞，處理這些新聞並
將其翻譯成盡可能多種語言。這應該有助於讓社群知道所有消息並克服語言障礙。

But I want it to break only after punctuation (，, 。) or whitespace

So what's the big deal about a newline injected between your Chinese characters?

The newlines become unsightly spaces when the HTML is rendered.

Well then just post-process it to get the newlines back out.

What a hassle.

So how would you like it broken?

weeklyOSM 的想法是蒐集來自世界各地，
與 OSM 相關的新聞，
處理這些新聞並將其翻譯成盡可能多種語言。
這應該有助於讓社群知道所有消息並克服語言障礙。

Well then just substitute newlines after each punctuation, s/，|。/$&\n/g in Perl.

Yes, but M-q will still break my longer stretches.

Well then don't use M-q.

But can't I just set some variable that would allow me to live a normal life, like my ASCII using friends do?

I mean great, some people think it is cool to add newlines (which I say are just as bad as spaces) between Chinese characters. Well how do I opt-out?

Answer 1

Don’t use fill-paragraph at all. Just use visual-line-mode instead. This visually wraps the long lines without inserting newline characters in the file. You can customize word-wrap-whitespace-characters to wrap at punctuation as well as whitespace if you need to.

Answer 2

Rather than trying to change and/or customize emacs, just go with the flow and indeed run your file through

perl -pwle 's/[；。，】]/$& /g' #etc. wide punctuation characters

and emacs will hopefully break at the spaces you have inserted.

As for long stretches, perhaps find a spot where a wide comma plus space can be inserted.

( https://www.unicode.org/reports/tr14/ but note when breaking in HTML source, the breaks become unsightly spaces when rendered.)