icu4c--> ushape.c missing character in shaping?

in our langauge we use arabic characters in writing with some differences, icu's ushape.c ( arabic shaper) only works with main arabic characters and dosn't shape my language specific characters ( i.e 0x6D5 etc) i'v changed ushape.c to work with my language and it worked well except for on character, that is 0x649, in arabic they have only 2 shapes, in my langauge we have 4 shapes for it.

i'v changed line 183

1                + 256 * 0x7F,/*0x0649*/

to

1+2+8             + 256 * 0x98 /*0x649*/

and changed line 121

static const UChar yehHamzaToYeh[] =
{
/* isolated*/ 0xFEEF,
/* final   */ 0xFEF0
};

to

static const UChar yehHamzaToYeh[] =
    {
        /* isolated */0xFEEF, 
                       0xFBE8, // my language specific
                      0xFBE9,// my language specific
        /* final */   0xFEF0 
   };

from ushape.c

no开发者_如何学JAVAw it can produce 3 shapes with no problem ( the beginning,isolated and final), but middle shape is displayed as a square ( missing character ) .

i tried replacing "* 0x98" with other numbers, but this best i can get.

what should i do ?

Uighur? I discussed with a couple of people about Uighur rendering, not this particular issue but in general.

When you said you get a square, what Unicode character do you get?

What you really should do is to file a bug with ICU and discuss it there. This is a feature request, not a usage question.

My rusty recollection is that for Uighur it makes different use of shaping, and you will want to basically have a different mode on the shaper.

ICU indeed seems to have problems for shaping with some languages, e.g. Urdu.

Your specific character 649 however is probably not the characters that you are looking for.

U+649 is alef maksura which looks identical to Farsi Yeh U+6cc which is shaped properly by ICU.

They do have different presentation forms: Alef maksura only has isolated and final form: U+feef U+fef0 Farsi yeh has all four forms: U+fbfc U+fbfd U+fbfe U+fbff