RE: ISFOC, ISCII, Devanagari and Unicode.

From: Marco.Cimarosti@icl.com
Date: Thu Apr 20 2000 - 12:42:12 EDT


Antoine Leca wrote:
> Jim Hargrave wrote:
> ...
> > If not, is it possible to convert these ISFOC "font codes"
> directly to/from Unicode?
>
> Yes, but that is horribly complex, because the conversion need a lot
> of context. For example, in Devanagari, a r that comes at the
> beginning
> of the conjunct is "typed" by the rendering engine very late,
> after all
> the "pieces" for the cluster consonnants have been drawn; furthermore,
> this piece (called the reph) is combined with the bindu if the later
> appears in the syllable. In the conversion, you are required to parse
> the all the glyphs that forms the syllable to find eg. the reph, then
> re-order the whole in Unicode order and output the correct codepoints.

It is indeed complicated, but not more than an Indic rendering engine. It
is, in fact, the reverse process of an Unicode Indic rendering engine:
converting from glyphs (and visual order) to logical characters (and
phonetic order).

However complicate, I am afraid that this is necessary: if the existing
glyph-encoded text cannot be converted to Unicode (or ISCII), you cannot ask
authors to abandon old methods.

Antoine Leca wrote privately (but allowing me to publish):
> > 2) Are Indic keyboard character-oriented or glyph oriented?
> Both exist; I believe the former are called "phonetic".

Thus: also visual-order input methods require a similar "de-rendering"
algorithm.

E.g., if users prefer to have a "repha" key (to be typed *after* a cluster),
the software must be able to convert this "visual" information to the
corresponding logical information needed in Unicode (a "ra + virama"
sequence at the *beginning* of the cluster).

_ Marco



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:02 EDT