[HarfBuzz] hb_shape API is not working properly for some Bengali vowels

[Jonathan Kew]

On 03/06/2019 14:46, Vijendra Singh wrote:
> Hi All,
> 
> I am using HarfBuzz 2.10 for Indic languages in my application but 
> failed to get correct substituted glyph from hb_shape API for Bengali 
> text having vowels like e (U+09c7), ai (U+09c8), o (U+09cb), au (U+09cc) 
> in between the words in any Bangali font contains Bengali glyphs like 
> -Nirmala UI Regular font.
> 
> These glyphs are substituted fine if these are at the start of word and 
> if these glyphs are in between word then it’s Shirorekha glyph should be 
> return by the shaping API.
> 
> Actual result-
> 
> Expected result-
> 
> I have also tested same with latest (2.5.0) HarfBuzz library but no luck 
> result.
> 
> Kindly suggest whether the issue is in hb_shape API or in the fonts or I 
> have to do something at my end to get desire result.

It's helpful if you specify the exact character sequence involved when 
reporting a problem. From your actual/expected images, I'm assuming that 
word is <U+0986, U+09AC, U+09C7, U+09A6, U+09A8>.

If that sequence is right, then AFAICS it works as expected with the 
Nirmala font: the result I get looks like your "expected" image.

So I guess the problem is somewhere in your code. HarfBuzz shapes this 
text and font correctly.

JK