Bidi: inserting Japanese paragraphs in Arabic/Farsi document

Philippe Verdy verdy_p at
Mon Nov 21 12:27:10 CST 2016

2016-11-21 4:39 GMT+01:00 Eli Zaretskii <eliz at>:

> > From: Philippe Verdy <verdy_p at>
> > Date: Sun, 20 Nov 2016 21:19:40 +0100
> > Cc: Simon Cozens <simon at>,
> >       unicode Unicode Discussion <unicode at>
> >
> > Note that if you get :
> >
> > OWT-CIBARA "Japanese2 【Japanese1】" ENO-CIBARA
> >
> > this means that the first quotation mark is "transparent" and preserves
> the RTL direction.
> Yes.  It takes the direction of the paragraph, which is RTL.
> > And I don't see then how you can pair the final quotation mark, unless
> you consider it as "leading" the
> > ARABIC-TWO part (meaning that you don't pair these quotation marks at
> all: only brackets are paired and the
> > fragment【Japanese1】 is correct (you are using the new Bidi algorithm).
> The quotes don't need to pair, they just need both to have the
> paragraph direction.  And that's what happens, because text on each
> side of each quote has different directionality.  The UBA mandates
> that the quote (which is ON) takes the embedding direction in that
> case, and the embedding direction here is the base paragraph
> direction.

This is a reasonnable rule for most frequent cases, but I'm not sure this
works in the case of multiple levels of inclusions (with different
directions), where the paragraph direction is not relevant for quotation
marks in the inner levels.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Unicode mailing list