https://bugs.documentfoundation.org/show_bug.cgi?id=163878

--- Comment #3 from Hossein <[email protected]> ---
Created attachment 197585
  --> https://bugs.documentfoundation.org/attachment.cgi?id=197585&action=edit
Output of rendering Arabic and Persian yah from pango-view utility

> The text can still be read, but in my opinion this bug makes LibreOffice 
> unusable for professional Arabic texts.

The problem here is with two things:

1. Incorrect text language: the language set for the text is incorrectly set to
"Hindi". You can change that by going to "Character" and then setting "Arabic
(...)" in "Complex" section. You need to activate "Complex text layout" and set
it to "Arabic (...)" in "Tools > Options > Languages and Locales > General".

2. Fallback font: You have not set an appropriate Arabic font for the text. You
can use "Amiri" for example.

Other than that, I think even with a fallback font, shaping algorithm should
work correctly.

The fallback font here is "FreeSerif", and the digits font is "Lohit
Devanagari", because of the incorrect language, Hindi set for the text. You can
see the incorrect type of the digits for Arabic text.

The problem seems to be from the font itself. You can see how the underlying
HarfBuzz library renders the text:

$ pango-view --font="FreeSerif 100" --text 'في'

On the other hand, with Persian ی it is rendered correctly:

$ pango-view --font="FreeSerif 100" --text 'فی'

I think this is a font problem. FreeSerif font is provided by "GNU FreeFont",
and you can see the related bug here:

bug #65979: Arabic Alif Maqsura, Kaf and Lam Alif problems
https://savannah.gnu.org/bugs/?65979

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to