#!/usr/bin/perl -w 
#
# A script to convert tamil character clusters to
# their presentation format and assign them internal Yudit ids.
#
# Tamil clusters are formed by
# vowels = 12
# consonants = 22 (including grantha) 
# modifiers = pulli (VIRAMA), i, I, u, U, e, E, ai, o, O and au 
# and the aytham symbol
#
# In general, we need to output new glyphs for clusters
# formed by modifiers (i I u U and pulli). We assign unique ids 
# internal to Yudit for the output glyphs, following the Pango convention
# outlined in http://www.wholehog.fsnet.co.uk/robert/indic/fonts.htm
# 
# Thuraippah Vaseeharan <t_vasee@yahoo.com>
# Thu Dec 27 22:49:33 CST 2001
#
# CHANGES:
# merged Tamil.kmap.pl
# Mon Dec 31 19:18:06 CST 2001

$KMAP_FILE="../../mytool/kmap/Tamil.kmap" ;

open (KMAP_FILE, ">$KMAP_FILE") || die "unable to open kmap file for writing" ;
print KMAP_FILE "//autogenerated by Tamil.cluster.pl\n" ;

@consonant_keys = qw(k ng c j nj d N t q n p m y r R l L z v sh s h) ; 
%modifier_keys = ( '0bbe' => 'A', 
		   '0bcd' => 'f',
		   '0bbf' => 'i',
		   '0bc0' => 'I',
		   '0bc1' => 'u',
		   '0bc2' => 'U',
		   '0bc6' => 'e',
		   '0bc7' => 'E',
		   '0bc8' => 'G',
		   '0bca' => 'o',
		   '0bcb' => 'O',
		   '0bcc' => 'H'
		  ) ;

# The  notation for vowels modifiers ai(G)/au(H) 
# is not very obvious :-(
# They have been 'renamed' as G, H to prevent Yudit 'waiting' 
# for additional input, when typing the "a" vowel
@vowel_keys = qw(a A i I u U e E G o O H) ; 

$aytham = "0b83" ;
$pulli = "0bcd" ;
$aravu = "0bbe" ;
@vowels = qw(0b85 0b86 0b87 0b88 0b89 0b8a 0b8e 0b8f 0b90 0b92 0b93 0b94) ;
@consonants = qw(0b95 0b99 0b9a 0b9c 0b9e 0b9f 0ba3 0ba4 
		0ba8 0ba9 0baa 0bae 0baf 0bb0 0bb1 0bb2 
		0bb3 0bb4 0bb5 0bb7 0bb8 0bb9) ;

@modifiers_0 = qw(0bbf 0bc0 0bc1 0bc2) ; # i I u U 
@modifiers_1 = qw(0bc6 0bc7 0bc8) ; # e E ai
@modifiers_2 = qw(0bca 0bcb 0bcc) ; # o O au
%first_part = ( '0bca' => '0bc6', 
		'0bcb' => '0bc7',
		'0bcc' => '0bc6' ) ;
%last_part = (  '0bca' => '0bbe' ,
		'0bcb' => '0bbe' ,
		'0bcc' => '0bd7' ) ; 

$independent = 0x80060000 ;

print "COMM=begin tamil clusters \n" ;
print "COMM=autogenerated by Tamil.cluster.pl\n" ;

# uyirs 
$i= 0;
foreach $vowel (@vowels) {
	printf(KMAP_FILE "\"%s=0x%s\",\n", $vowel_keys[$i++], $vowel) ;
	printf("%s -> 0000%s %x\n", $vowel, $vowel, $independent++) ; 
}

# aytham 
printf(KMAP_FILE "\"%s=0x%s\",\n", "Q", $aytham) ;
printf("%s -> 0000%s %x\n", $aytham, $aytham, $independent++) ; 

# akara meys (base consonant)
$i= 0;
foreach $cons (@consonants) {
	printf(KMAP_FILE "\"%s%s=0x%s\",\n", $consonant_keys[$i++], "a", $cons) ;
	printf("%s -> 0000%s %x\n", $cons, $cons, $independent++) ; 
}

# Akaaram meys (long form of base consonant) 
$i = 0 ;
foreach $cons (@consonants) {
	printf(KMAP_FILE "\"%s%s=0x%s 0x%s\",\n", $consonant_keys[$i++], $modifier_keys{$aravu}, $cons, $aravu) ;
	printf("%s %s -> 0000%s 0000%s %x\n", $cons, $aravu, $cons, $aravu, $independent++) ; 
}

# meys (base consonant + pulli) and other uyir-meys (including ligatures and re-ordered forms)
$uyirmeys = 0x8006d000 ;
$meys = 0x8006f000 ;
$reordered = 0x8006e000 ;

$i = 0 ;
foreach $cons (@consonants) {
	# Tamil Meys: ligatures with pulli (virama)
	printf(KMAP_FILE "\"%s%s=0x%s 0x%s\",\n", $consonant_keys[$i], $modifier_keys{$pulli}, $cons, $pulli) ;
	printf("%s %s -> 0000%s 0000%s %x\n", $cons, $pulli, $cons, $pulli, $meys++) ; 
	# Tamil Uyir-Meys:
	# ligatures with i, I, u, U
	foreach $mod (@modifiers_0) {
		printf(KMAP_FILE "\"%s%s=0x%s 0x%s\",\n", $consonant_keys[$i], $modifier_keys{$mod}, $cons, $mod) ;
		printf("%s %s -> 0000%s 0000%s %x\n", $cons, $mod, $cons, $mod, $uyirmeys++) ;
	}
	# re-ordered forms
	# for (e E ai), just flip
	foreach $mod (@modifiers_1) {
		printf(KMAP_FILE "\"%s%s=0x%s 0x%s\",\n", $consonant_keys[$i], $modifier_keys{$mod}, $cons, $mod) ;
		printf("%s %s -> 0000%s 0000%s %x\n", $cons, $mod, $mod, $cons, $reordered++) ; 
	}
	# for (o O au), the modifier "splits" into two parts,
	# that surround the base consonant
	foreach $mod (@modifiers_2) {
	printf(KMAP_FILE "\"%s%s=0x%s 0x%s\",\n", $consonant_keys[$i], $modifier_keys{$mod}, $cons, $mod) ;
	printf("%s %s -> 0000%s 0000%s 0000%s %x\n", $cons, $mod, $first_part{$mod}, $cons, $last_part{$mod}, $reordered++) ;
	}
	$i++ ;
}
print "COMM=end tamil\n" ;