Re: [Freetype] layout and all

[Antoine Leca]

Hi Somesh,

Somesh Agarwal wrote:
> 
>   I wan't to use freetype for writing strings in non-roman script
> language(devnagri).

Well, while this is certainly possible, this is *NOT* an easy task.
I am personnaly involved in that task for three years now, and while
I certainly lack time to achieve it, I can confirm this is very far
from being obvious.

Furthermore, there exists at least four different "kinds" of Devanagari-
enabled fonts: CDAC ISFOC, Verdhuis/ITRANS-based scheme, Apple GX/ILK,
and last but not least, Microsoft's Indic OpenType. This is, of course,
in addition to all these "hacked" Devanagari fonts that try to show
Devanagari based on the ASCII layout, for example ka is at character k,
kha is at K, the half form of k is something else, etc.

The hacked fonts need always a specific driver, because there is no
standard for them. In fact, ITRANS is sort of a library that allows
to use any of these, needing "only" to write a small description of
the encoding used for the font. ITRANS is not intended to be used with
Freetype, but it may be done, I think.

Verdhuis devnag is fully embeeded with TeX, I do not believe this is
what you are searching for.

Next, we have CDAC-ISFOC, which is really a sort of "hacked fonts", but
the encoding of the glyphs into the font is standardised (although this
is not very well published). It is certainly possible to write some
preprocessor that may take on input some Devanagari characters, and
on output gives the string of glyphs adequate for an ISFOC font, but
I am not aware of a OpenSource version of such a preprocessor.

The last two solutions, Apple's and Microsoft's, have published
specifications, but the fonts are presently very very few in number (but
this is growing, particularly for OpenType). They both rely on some
technologies (GX and OpenType, respectively), that sits "on top" of
normal font processing; these technologies are not implemented in
Freetype 2, while a preliminary support for OpenType exists in beta form
in Freetype 1.
The main difference is that the Apple "solution" is self-sufficient,
as long as you have a full-featured GX decoder (only MacOs does really,
while Java JDK and ICU is going to achieve it; Freetype does not have
it), while Microsoft's "solution" require an rather extensive
"preprocessing" that reorders the characters before the mapping from
characters to glyphs and the applications of the OpenType features.
The realisation of such a preprocessor is the task I am not able to
fullfill because of lacking time.


> For this I have to take care
> of how one character is placed on the screen after the first has being
> written .
> But I have not being able to go beyond being able to obtain bitmap and
> metrics of
> glyph from the font face file. Can someone tell how to place a string
> using freetype and
> what do we call technically this thing.

Hope the above (rather long) digression will answer your question.
If you still have queries (and perhaps a more precise problem), I shall
be pleased to further help you.


Antoine