Re: intelligent lexically encoding

On Sep 08, 2005, at 12:39 , Jerzy Giergiel wrote:
> Neither of those fallbacks is OK, I want á converted to accent  
> stripped version of itself i.e. a. The second solution isn't very  
> helpful either, it's basically tr replacement table which is not  
> much fun to write when majority of upper 128 characters need to be  
> converted. There's gotta be a simpler and more elegant solution.   
> thanks anyway.

Well, it's not that hard to write a tr version if you let perl do the  
job.

#!/usr/bin/perl
use strict;
use charnames qw(:full);
my ($from, $to);
for my $ord (0x80..0xff){
     my $chr = chr $ord;
     my $name = charnames::viacode($ord);
     $name =~ /(SMALL|CAPITAL) LETTER ([A-Z]) WITH/i or next;
     my $az = $1 eq 'CAPITAL' ? uc($2) : $2;
     $from .= $chr;
     $to   .= $az;
}
binmode STDOUT => ":utf8";
print qq(tr[$from]\n  [$to];), "\n";
__END__

And here is the output.

tr[ÀÁÂÃÄÅÇÈÉÊËÌÍÎÏÑÒÓÔÕÖØÙÚÛÜÝàáâãäåçèéêëìíîïñòóôõöøùúûüýÿ]
   [AAAAAACEEEEIIIINOOOOOOUUUUYaaaaaaceeeeiiiinoooooouuuuyy];

In this kind of case, however, a simple tr/// won't cut it, however.   
Consider Schrödinger.  Usually you spell that 'Schroedinger", not  
"Shrodinger".  So you have to resort to s///g for most cases.

Dàñ thè Ëñçôdé Máìñtâíñêr