2009/1/12 Sreenivasa Guttal <[email protected]>:
> Hi All,
>
> As I tried to look into code (devnag.c), I find that the in the table
> TabCom, index is assigned Type_Com value of 1. If I change this to 8,
> recompile and execute, my current code seems to work..
>
> Is there any reason, why index is not treated like other Tex commands like
> section etc?
>
> If not, should this be treated as bug with the above kind of fix?
>
It is not a problem of recognizing devanagari characters inside
\index{...}. There is much worse problem with sorting devanagari text.
I think that preprocessing the contents of \index{...} will make more
harm than profit. Take e.g. (assuming @modernhindi) the work kitaab.
If \index were changed to type 8, makeindex would see EktAb and sorti
it with respect to E which is wrong. Characters that should be treated
similarly when sorted may be converted to very different sequences.
For instance, kartaa will be output as ktA\0 while karanaa will be
krnA. Notice that repha in kartaa is converted by devnag to \0
appearing at the end of the word. The conjuncts are often converted to
cryptic sequences and some characters are formed of several pieces,
e.g. initial o is preprocessed to initial a followed by o matra. Thus
o.s.thya will be ao\309wW+.
Taking just a few words, the correct order in Hindi would be:
kam < karanaa < kartaa < karm < kaam < kitaab < kuttaa < kram < k.sa.n
After devnag preprocessing the list in the correct order would be
km < krnA < ktA\0 < km\0 < kAm < EktAb < \7{k}\381wA < \387wm < \322wZ
I feel this will be quite a difficult task for a sorting algorithm.
Deploying unmodified makeindex will not help either. For instance,
kaam would appear before kam which is wrong. Using kAm and instructing
makeindex to put uppercase letters after lowercase does not help
because kamii will then be put to a wrong position. Before changing
the type of \index in Tab_Comm it is necessary to solve the problem of
sorting. And remember that Sanskrit alphabetic order slightly differs
from Hindi. Implementation of the sort algorithm might be easier in
the IITK encoding used by CFILT, see
http://www.cfilt.iitb.ac.in/~hdict/webinterface_user/iitk_encoding.html
And the last note: quotation marks have special meaning for makeindex.
This is the reason why "s was changed to s in the ind file.
> Thanks,
> Sreenivasa
>
> On Mon, Jan 12, 2009 at 11:12 PM, Sreenivasa Guttal <[email protected]>
> wrote:
>>
>> Hi All,
>>
>> I tried using index feature with devnag and latex without success. It
>> seems to have issue in recognizing devnagari character within index.
>>
>> I have attached the relevant files with this mail.
>>
>> Appreciate any help.
>>
>> Thanks,
>> Sreenivasa
>>
>>
>
>
> _______________________________________________
> Devnag-general mailing list
> [email protected]
> http://lists.sarovar.org/cgi-bin/mailman/listinfo/devnag-general
>
--
Zdeněk Wagner
http://hroch486.icpf.cas.cz/wagner/
http://icebearsoft.euweb.cz
_______________________________________________
Devnag-general mailing list
[email protected]
http://lists.sarovar.org/cgi-bin/mailman/listinfo/devnag-general
