[HarfBuzz] hangul shaper patches

Jonathan Kew jfkthame at googlemail.com
Wed Jan 22 23:46:43 PST 2014


On 22/1/14 12:53, Behdad Esfahbod wrote:
> Thanks Jonathan.  I've merged these.
>
> A few points:
>
> /* Same order as the feature array below */
> enum {
>    NONE,
>
>    LJMO,
>    VJMO,
>    TJMO,
>
>    FIRST_HANGUL_FEATURE = LJMO,
>    HANGUL_FEATURE_COUNT = TJMO + 1
> };
>
> Do you really need the NONE?  I don't see where / how that's used.

The NONE isn't explicitly used, but is there as a placeholder for the 
zero value, so that we can tell whether info.hangul_shaping_feature() 
has actually been set on a glyph at all.

>
> I just want to note that by applying the jamo features only to one character
> at a time, we disallow contextual rules,

Is this true? I thought we ignored the mask when matching context, and 
only used it for the actual input string.

> but I guess that's what Uniscribe
> does also?

malgun.ttf has lookups that include backtrack and/or lookahead in some 
of its *jmo features. I haven't yet explicitly traced to see them to see 
what matches where, but the fact that we match uniscribe on the entire 
wikipedia wordlist suggests we're probably OK.

JK



More information about the HarfBuzz mailing list