[XeTeX] Velthuis to Roman translit

[Neal Delmonico]

Thanks for the suggestions.  See below.

On Thu, 03 Feb 2011 12:31:26 -0600, Dominik Wujastyk <wujastyk at gmail.com>  
wrote:

> Wait a minute.  Reading your note more carefully, I think the sed script  
> is
> giving you English text with unwanted "\ñ" strings here and there. Is  
> that
> right?  If so, you could just add a last line to the sed script to right
> that wrong ("s/\\ñ/ñ/g").

Yes.  This worked.  It turns out that there are only two occasions when  
the script overreaches.  "\~n" and "\.n"  Following your advice I added  
lines to the end of the script that turned the mistaken conversions back.   
So far I have not found any other problems.

>
> Yes, you are right about TEC/map files.  I think you should use the  
> existing
> velthuis map files and replace the Devanagari character codes for output
> with Latin, just as you say.  Once you get to understand the conjunct
> consonants coding, it can be stripped out.  Etc.  I don't know where the
> documentation for TEC files is, but probably on the SIL website.

I will look into this and see if I can make the appropriate changes.

One small question more.  Now that I have the files in Unicode encoding,  
how do I go from that to Devanagari?  Using the RomDev mapping just gives  
me more Romanized diacritics.   Here is what I am using:

\documentclass[10pt,titlepage]{book}
\usepackage{xltxtra}
\usepackage{polyglossia}
\setmainfont{Charis SIL}
\defaultfontfeatures{Mapping=tex-text}
\newfontfamily\sanskritfont[Script=Devanagari,Mapping=RomDev]{Sahadeva}
\newcommand{\skt}[1]{{\sanskritfont\textsanskrit{#1}}}

Thanks.

Neal

Using Opera's revolutionary email client: http://www.opera.com/mail/