Re: Automatic Nikud insertion

On 10/17/06, Jonathan Ben Avraham <[email protected]> wrote:
> Hi Dov,
> This is an excellent start. Of course, it's still a long way from the
> quality that we need for OpenOffice.
>
> 1. The HEBREW POINT DAGESH in the HEBREW LETTER SHIN of =D7=94=D7=A9=D7=
=9E=D7=99=D7=9D is
> misplaced.

This placement exists in the original FrankRuehl CLM font in the form
of a Shin/Dagesh ligature (U+FB49). My program broke apart this
ligature and expressed it as anchors instead. But the point here I
believe is that there needs to be human interaction.

> 2. The HEBREW POINT HOLAM after the HEBREW LETTER HET in =D7=97=D7=A9=D7=
=9A elides
> with the HEBREW POINT SHIN DOT above the HEBREW LETTER SHIN.

Yes. This has to be solved with a GPOS chained context rule. I still
haven't figured out how to define these. I would be very happy if
someone can show me.

> 3. The HEBREW POINT PATAH under the HEBREW LETTER AYIN in =D7=A2=D7=9C =
elides with
> the HEBREW LETTER AYIN itself.

This should be solved by doing a right align box, instead of center.
Of course this is font dependant. This might also be better fixed
manually.

> 4. The HEBREW PUNCTUATION MAQAF (if that's what it is and not a LATIN
> HYPHEN-MINUS in =D7=A2=D7=9C-=D7=A4=D7=A0=D7=99 is too far from the HEB=
REW LETTER LAMED.

It appears that machon mamre uses HYPHEN-MINUS instead of MAQAF. I'll
check it out.

> In addition  to the above placement problems, the HEBREW POINT HATAF SE=
GOL
> is slightly malformed - the horizontal dots of the segol portion of the
> HEBREW POINT HATAF SEGOL need to be separated a bit.

Glyph definition problem...

In summary, I believe that it is necessary to mix between automatic
heuristics and manual work. Once you get to chained sub rules,
fontforge does not provide any graphics tools for defining them. Also,
if you look at the font SBL Hebrew, it has hundreds if not thousands
of rules. I am convinced that most of these may be added
automatically. The idea is to minimize the need for human interaction
as much as possible.

By programming it is possible to express changes that are a lot of
dirty work to program by hand. E.g. in DejaVu the anchors of the Nikud
are in my opinion to close of to the base character. Of course you can
manually adjust them, but doing it by scripting is much faster. It's
the expressive vs visual paradigm all over again. I guess that's why I
like MetaFont/MetaPost/Asymptote. :-)

Regards,
Dov

----
Ivrix-discuss list. See http://ivrix.org.il.
To unsubscribe, please send mail to [email protected] wi=
th
only the following line in the message body (NOT SUBJECT!): unsubscribe