[p5sagit/p5-mst-13.2.git] / ext / Unicode / Normalize / mkheader

#!perl
#
# This auxiliary script makes five header files
# used for building XSUB of Unicode::Normalize.
#
# Usage:
#    <do 'mkheader'> in perl, or <perl mkheader> in command line
#
# Input files:
#    unicore/CombiningClass.pl (or unicode/CombiningClass.pl)
#    unicore/Decomposition.pl (or unicode/Decomposition.pl)
#    unicore/CompositionExclusions.txt (or unicode/CompExcl.txt)
#
# Output files:
#    unfcan.h
#    unfcpt.h
#    unfcmb.h
#    unfcmp.h
#    unfexc.h
#
use 5.006;
use strict;
use warnings;
use Carp;
use File::Spec;

BEGIN {
    unless ("A" eq pack('U', 0x41)) {
	die "Unicode::Normalize cannot stringify a Unicode code point\n";
    }
}

our $PACKAGE = 'Unicode::Normalize, mkheader';

our $Combin = do "unicore/CombiningClass.pl"
    || do "unicode/CombiningClass.pl"
    || croak "$PACKAGE: CombiningClass.pl not found";

our $Decomp = do "unicore/Decomposition.pl"
    || do "unicode/Decomposition.pl"
    || croak "$PACKAGE: Decomposition.pl not found";

our %Combin;	# $codepoint => $number    : combination class
our %Canon;	# $codepoint => \@codepoints : canonical decomp.
our %Compat;	# $codepoint => \@codepoints : compat. decomp.
# after _U_stringify(), ($codepoint => $hexstring) for %Canon and %Compat
our %Exclus;	# $codepoint => 1          : composition exclusions
our %Single;	# $codepoint => 1          : singletons
our %NonStD;	# $codepoint => 1          : non-starter decompositions

our %Comp1st;	# $codepoint => $listname  : may be composed with a next char.
our %Comp2nd;	# $codepoint => 1          : may be composed with a prev char.
our %CompList;	# $listname,$2nd  => $codepoint : composite

our $prefix = "UNF_";
our $structname = "${prefix}complist";

########## definition of Hangul constants ##########
use constant SBase  => 0xAC00;
use constant SFinal => 0xD7A3; # SBase -1 + SCount
use constant SCount =>  11172; # LCount * NCount
use constant NCount =>    588; # VCount * TCount
use constant LBase  => 0x1100;
use constant LFinal => 0x1112;
use constant LCount =>     19;
use constant VBase  => 0x1161;
use constant VFinal => 0x1175;
use constant VCount =>     21;
use constant TBase  => 0x11A7;
use constant TFinal => 0x11C2;
use constant TCount =>     28;

sub decomposeHangul {
    my $SIndex = $_[0] - SBase;
    my $LIndex = int( $SIndex / NCount);
    my $VIndex = int(($SIndex % NCount) / TCount);
    my $TIndex =      $SIndex % TCount;
    my @ret = (
       LBase + $LIndex,
       VBase + $VIndex,
      $TIndex ? (TBase + $TIndex) : (),
    );
    wantarray ? @ret : pack('U*', @ret);
     # any element in @ret greater than 0xFF, so no need of u2n conversion.
}

########## getting full decomposion ##########
{
    my($f, $fh);
    foreach my $d (@INC) {
	$f = File::Spec->catfile($d, "unicore", "CompositionExclusions.txt");
	last if open($fh, $f);
	$f = File::Spec->catfile($d, "unicode", "CompExcl.txt");
	last if open($fh, $f);
	$f = undef;
    }
    croak "$PACKAGE: neither unicore/CompositionExclusions.txt "
	. "nor unicode/CompExcl.txt is found in @INC" unless defined $f;

    while (<$fh>) {
	next if /^#/ or /^$/;
	s/#.*//;
	$Exclus{ hex($1) } = 1 if /([0-9A-Fa-f]+)/;
    }
    close $fh;
}

##
## converts string "hhhh hhhh hhhh" to a numeric list
##
sub _getHexArray { map hex, $_[0] =~ /([0-9A-Fa-f]+)/g }

while ($Combin =~ /(.+)/g) {
    my @tab = split /\t/, $1;
    my $ini = hex $tab[0];
    if ($tab[1] eq '') {
	$Combin{ $ini } = $tab[2];
    } else {
	$Combin{ $_ } = $tab[2] foreach $ini .. hex($tab[1]);
    }
}

while ($Decomp =~ /(.+)/g) {
    my @tab = split /\t/, $1;
    my $compat = $tab[2] =~ s/<[^>]+>//;
    my $dec = [ _getHexArray($tab[2]) ]; # decomposition
    my $ini = hex($tab[0]); # initial decomposable character

    my $listname =
	@$dec == 2 ? sprintf("${structname}_%06x", $dec->[0]) : 'USELESS';
		# %04x is bad since it'd place _3046 after _1d157.

    if ($tab[1] eq '') {
	$Compat{ $ini } = $dec;

	if (! $compat) {
	    $Canon{ $ini } = $dec;

	    if (@$dec == 2) {
		if ($Combin{ $dec->[0] }) {
		    $NonStD{ $ini } = 1;
		} else {
		    $CompList{ $listname }{ $dec->[1] } = $ini;
		    $Comp1st{ $dec->[0] } = $listname;
		    $Comp2nd{ $dec->[1] } = 1 if ! $Exclus{$ini};
		}
	    } elsif (@$dec == 1) {
		$Single{ $ini } = 1;
	    } else {
		croak("Weird Canonical Decomposition of U+$tab[0]");
	    }
	}
    } else {
	foreach my $u ($ini .. hex($tab[1])) {
	    $Compat{ $u } = $dec;

	    if (! $compat) {
		$Canon{ $u } = $dec;

		if (@$dec == 2) {
		    if ($Combin{ $dec->[0] }) {
			$NonStD{ $u } = 1;
		    } else {
			$CompList{ $listname }{ $dec->[1] } = $u;
			$Comp1st{ $dec->[0] } = $listname;
			$Comp2nd{ $dec->[1] } = 1 if ! $Exclus{$u};
		    }
		} elsif (@$dec == 1) {
		    $Single{ $u } = 1;
		} else {
		    croak("Weird Canonical Decomposition of U+$tab[0]");
		}
	    }
	}
    }
}

# modern HANGUL JUNGSEONG and HANGUL JONGSEONG jamo
foreach my $j (0x1161..0x1175, 0x11A8..0x11C2) {
    $Comp2nd{$j} = 1;
}

sub getCanonList {
    my @src = @_;
    my @dec = map {
	(SBase <= $_ && $_ <= SFinal) ? decomposeHangul($_)
	    : $Canon{$_} ? @{ $Canon{$_} } : $_
		} @src;
    return join(" ",@src) eq join(" ",@dec) ? @dec : getCanonList(@dec);
    # condition @src == @dec is not ok.
}

sub getCompatList {
    my @src = @_;
    my @dec = map {
	(SBase <= $_ && $_ <= SFinal) ? decomposeHangul($_)
	    : $Compat{$_} ? @{ $Compat{$_} } : $_
		} @src;
    return join(" ",@src) eq join(" ",@dec) ? @dec : getCompatList(@dec);
    # condition @src == @dec is not ok.
}

# exhaustive decomposition
foreach my $key (keys %Canon) {
    $Canon{$key}  = [ getCanonList($key) ];
}

# exhaustive decomposition
foreach my $key (keys %Compat) {
    $Compat{$key} = [ getCompatList($key) ];
}

sub _pack_U {
    return pack('U*', @_);
}

sub _U_stringify {
    sprintf '"%s"', join '',
	map sprintf("\\x%02x", $_), unpack 'C*', _pack_U(@_);
}

foreach my $hash (\%Canon, \%Compat) {
    foreach my $key (keys %$hash) {
	$hash->{$key} = _U_stringify( @{ $hash->{$key} } );
    }
}

########## writing header files ##########

my @boolfunc = (
    {
	name => "Exclusion",
	type => "bool",
	hash => \%Exclus,
    },
    {
	name => "Singleton",
	type => "bool",
	hash => \%Single,
    },
    {
	name => "NonStDecomp",
	type => "bool",
	hash => \%NonStD,
    },
    {
	name => "Comp2nd",
	type => "bool",
	hash => \%Comp2nd,
    },
);

my $file = "unfexc.h";
open FH, ">$file" or croak "$PACKAGE: $file can't be made";
binmode FH; select FH;

    print << 'EOF';
/*
 * This file is auto-generated by mkheader.
 * Any changes here will be lost!
 */
EOF

foreach my $tbl (@boolfunc) {
    my @temp = sort {$a <=> $b} keys %{$tbl->{hash}};
    my $type = $tbl->{type};
    my $name = $tbl->{name};
    print "$type is$name (UV uv)\n{\nreturn\n\t";

    while (@temp) {
	my $cur = shift @temp;
	if (@temp && $cur + 1 == $temp[0]) {
	    print "($cur <= uv && uv <= ";
	    while (@temp && $cur + 1 == $temp[0]) {
		$cur = shift @temp;
	    }
	    print "$cur)";
	    print "\n\t|| " if @temp;
	} else {
	    print "uv == $cur";
	    print "\n\t|| " if @temp;
	}
    }
    print "\n\t? TRUE : FALSE;\n}\n\n";
}

close FH;

####################################

my $compinit =
    "typedef struct { UV nextchar; UV composite; } $structname;\n\n";

foreach my $i (sort keys %CompList) {
    $compinit .= "$structname $i [] = {\n";
    $compinit .= join ",\n",
	map sprintf("\t{ %d, %d }", $_, $CompList{$i}{$_}),
	    sort {$a <=> $b } keys %{ $CompList{$i} };
    $compinit .= ",\n{0,0}\n};\n\n"; # with sentinel
}

my @tripletable = (
    {
	file => "unfcmb",
	name => "combin",
	type => "STDCHAR",
	hash => \%Combin,
	null =>  0,
    },
    {
	file => "unfcan",
	name => "canon",
	type => "char*",
	hash => \%Canon,
	null => "NULL",
    },
    {
	file => "unfcpt",
	name => "compat",
	type => "char*",
	hash => \%Compat,
	null => "NULL",
    },
    {
	file => "unfcmp",
	name => "compos",
	type => "$structname *",
	hash => \%Comp1st,
	null => "NULL",
	init => $compinit,
    },
);

foreach my $tbl (@tripletable) {
    my $file = "$tbl->{file}.h";
    my $head = "${prefix}$tbl->{name}";
    my $type = $tbl->{type};
    my $hash = $tbl->{hash};
    my $null = $tbl->{null};
    my $init = $tbl->{init};

    open FH, ">$file" or croak "$PACKAGE: $file can't be made";
    binmode FH; select FH;
    my %val;

    print FH << 'EOF';
/*
 * This file is auto-generated by mkheader.
 * Any changes here will be lost!
 */
EOF

    print $init if defined $init;

    foreach my $uv (keys %$hash) {
	croak sprintf("a Unicode code point 0x%04X over 0x10FFFF.", $uv)
	    unless $uv <= 0x10FFFF;
	my @c = unpack 'CCCC', pack 'N', $uv;
	$val{ $c[1] }{ $c[2] }{ $c[3] } = $hash->{$uv};
    }

    foreach my $p (sort { $a <=> $b } keys %val) {
	next if ! $val{ $p };
	for (my $r = 0; $r < 256; $r++) {
	    next if ! $val{ $p }{ $r };
	    printf "static $type ${head}_%02x_%02x [256] = {\n", $p, $r;
	    for (my $c = 0; $c < 256; $c++) {
		print "\t", defined $val{$p}{$r}{$c}
		    ? "($type)".$val{$p}{$r}{$c}
		    : $null;
		print ','  if $c != 255;
		print "\n" if $c % 8 == 7;
	    }
	    print "};\n\n";
	}
    }
    foreach my $p (sort { $a <=> $b } keys %val) {
	next if ! $val{ $p };
	printf "static $type* ${head}_%02x [256] = {\n", $p;
	for (my $r = 0; $r < 256; $r++) {
	    print $val{ $p }{ $r }
		? sprintf("${head}_%02x_%02x", $p, $r)
		: "NULL";
	    print ','  if $r != 255;
	    print "\n" if $val{ $p }{ $r } || ($r+1) % 8 == 0;
	}
	print "};\n\n";
    }
    print "static $type** $head [] = {\n";
    for (my $p = 0; $p <= 0x10; $p++) {
	print $val{ $p } ? sprintf("${head}_%02x", $p) : "NULL";
	print ','  if $p != 0x10;
	print "\n";
    }
    print "};\n\n";
    close FH;
}

1;
__END__
Commit	Line	Data
ac5ea531	1	#!perl
ac5ea531	2	#
628bbff0	3	# This auxiliary script makes five header files
628bbff0	4	# used for building XSUB of Unicode::Normalize.
ac5ea531	5	#
628bbff0	6	# Usage:
	7	# <do 'mkheader'> in perl, or <perl mkheader> in command line
	8	#
	9	# Input files:
	10	# unicore/CombiningClass.pl (or unicode/CombiningClass.pl)
	11	# unicore/Decomposition.pl (or unicode/Decomposition.pl)
	12	# unicore/CompositionExclusions.txt (or unicode/CompExcl.txt)
	13	#
	14	# Output files:
	15	# unfcan.h
	16	# unfcpt.h
	17	# unfcmb.h
	18	# unfcmp.h
	19	# unfexc.h
f027f502	20	#
ac5ea531	21	use 5.006;
	22	use strict;
	23	use warnings;
	24	use Carp;
6c941e0c	25	use File::Spec;
6c941e0c	26
9f1f04a1	27	BEGIN {
1efaba7f	28	unless ("A" eq pack('U', 0x41)) {
9f1f04a1	29	die "Unicode::Normalize cannot stringify a Unicode code point\n";
	30	}
	31	}
ac5ea531	32
	33	our $PACKAGE = 'Unicode::Normalize, mkheader';
	34
	35	our $Combin = do "unicore/CombiningClass.pl"
8f118dcd	36	\|\| do "unicode/CombiningClass.pl"
8f118dcd	37	\|\| croak "$PACKAGE: CombiningClass.pl not found";
ac5ea531	38
ac5ea531	39	our $Decomp = do "unicore/Decomposition.pl"
8f118dcd	40	\|\| do "unicode/Decomposition.pl"
8f118dcd	41	\|\| croak "$PACKAGE: Decomposition.pl not found";
ac5ea531	42
48287974	43	our %Combin; # $codepoint => $number : combination class
6c941e0c	44	our %Canon; # $codepoint => \@codepoints : canonical decomp.
	45	our %Compat; # $codepoint => \@codepoints : compat. decomp.
	46	# after _U_stringify(), ($codepoint => $hexstring) for %Canon and %Compat
48287974	47	our %Exclus; # $codepoint => 1 : composition exclusions
	48	our %Single; # $codepoint => 1 : singletons
	49	our %NonStD; # $codepoint => 1 : non-starter decompositions
	50
	51	our %Comp1st; # $codepoint => $listname : may be composed with a next char.
	52	our %Comp2nd; # $codepoint => 1 : may be composed with a prev char.
	53	our %CompList; # $listname,$2nd => $codepoint : composite
	54
	55	our $prefix = "UNF_";
	56	our $structname = "${prefix}complist";
ac5ea531	57
6c941e0c	58	########## definition of Hangul constants ##########
	59	use constant SBase => 0xAC00;
	60	use constant SFinal => 0xD7A3; # SBase -1 + SCount
	61	use constant SCount => 11172; # LCount * NCount
	62	use constant NCount => 588; # VCount * TCount
	63	use constant LBase => 0x1100;
	64	use constant LFinal => 0x1112;
	65	use constant LCount => 19;
	66	use constant VBase => 0x1161;
	67	use constant VFinal => 0x1175;
	68	use constant VCount => 21;
	69	use constant TBase => 0x11A7;
	70	use constant TFinal => 0x11C2;
	71	use constant TCount => 28;
	72
	73	sub decomposeHangul {
	74	my $SIndex = $_[0] - SBase;
	75	my $LIndex = int( $SIndex / NCount);
	76	my $VIndex = int(($SIndex % NCount) / TCount);
	77	my $TIndex = $SIndex % TCount;
	78	my @ret = (
	79	LBase + $LIndex,
	80	VBase + $VIndex,
	81	$TIndex ? (TBase + $TIndex) : (),
	82	);
	83	wantarray ? @ret : pack('U*', @ret);
	84	# any element in @ret greater than 0xFF, so no need of u2n conversion.
	85	}
	86
	87	########## getting full decomposion ##########
ac5ea531	88	{
8f118dcd	89	my($f, $fh);
8f118dcd	90	foreach my $d (@INC) {
8f118dcd	91	$f = File::Spec->catfile($d, "unicore", "CompositionExclusions.txt");
	92	last if open($fh, $f);
	93	$f = File::Spec->catfile($d, "unicode", "CompExcl.txt");
	94	last if open($fh, $f);
	95	$f = undef;
	96	}
48287974	97	croak "$PACKAGE: neither unicore/CompositionExclusions.txt "
	98	. "nor unicode/CompExcl.txt is found in @INC" unless defined $f;
	99
	100	while (<$fh>) {
	101	next if /^#/ or /^$/;
	102	s/#.*//;
	103	$Exclus{ hex($1) } = 1 if /([0-9A-Fa-f]+)/;
	104	}
8f118dcd	105	close $fh;
ac5ea531	106	}
ac5ea531	107
48287974	108	##
	109	## converts string "hhhh hhhh hhhh" to a numeric list
	110	##
	111	sub _getHexArray { map hex, $_[0] =~ /([0-9A-Fa-f]+)/g }
	112
8f118dcd	113	while ($Combin =~ /(.+)/g) {
	114	my @tab = split /\t/, $1;
	115	my $ini = hex $tab[0];
	116	if ($tab[1] eq '') {
	117	$Combin{ $ini } = $tab[2];
	118	} else {
	119	$Combin{ $_ } = $tab[2] foreach $ini .. hex($tab[1]);
	120	}
ac5ea531	121	}
ac5ea531	122
8f118dcd	123	while ($Decomp =~ /(.+)/g) {
	124	my @tab = split /\t/, $1;
	125	my $compat = $tab[2] =~ s/<[^>]+>//;
	126	my $dec = [ _getHexArray($tab[2]) ]; # decomposition
8f118dcd	127	my $ini = hex($tab[0]); # initial decomposable character
48287974	128
	129	my $listname =
	130	@$dec == 2 ? sprintf("${structname}_%06x", $dec->[0]) : 'USELESS';
	131	# %04x is bad since it'd place _3046 after _1d157.
	132
8f118dcd	133	if ($tab[1] eq '') {
	134	$Compat{ $ini } = $dec;
	135
	136	if (! $compat) {
48287974	137	$Canon{ $ini } = $dec;
8f118dcd	138
f027f502	139	if (@$dec == 2) {
8f118dcd	140	if ($Combin{ $dec->[0] }) {
	141	$NonStD{ $ini } = 1;
	142	} else {
48287974	143	$CompList{ $listname }{ $dec->[1] } = $ini;
	144	$Comp1st{ $dec->[0] } = $listname;
	145	$Comp2nd{ $dec->[1] } = 1 if ! $Exclus{$ini};
8f118dcd	146	}
f027f502	147	} elsif (@$dec == 1) {
8f118dcd	148	$Single{ $ini } = 1;
f027f502	149	} else {
f027f502	150	croak("Weird Canonical Decomposition of U+$tab[0]");
8f118dcd	151	}
	152	}
	153	} else {
48287974	154	foreach my $u ($ini .. hex($tab[1])) {
8f118dcd	155	$Compat{ $u } = $dec;
48287974	156
8f118dcd	157	if (! $compat) {
48287974	158	$Canon{ $u } = $dec;
8f118dcd	159
f027f502	160	if (@$dec == 2) {
8f118dcd	161	if ($Combin{ $dec->[0] }) {
	162	$NonStD{ $u } = 1;
	163	} else {
48287974	164	$CompList{ $listname }{ $dec->[1] } = $u;
	165	$Comp1st{ $dec->[0] } = $listname;
	166	$Comp2nd{ $dec->[1] } = 1 if ! $Exclus{$u};
8f118dcd	167	}
f027f502	168	} elsif (@$dec == 1) {
8f118dcd	169	$Single{ $u } = 1;
f027f502	170	} else {
f027f502	171	croak("Weird Canonical Decomposition of U+$tab[0]");
8f118dcd	172	}
	173	}
	174	}
ac5ea531	175	}
ac5ea531	176	}
ac5ea531	177
48287974	178	# modern HANGUL JUNGSEONG and HANGUL JONGSEONG jamo
	179	foreach my $j (0x1161..0x1175, 0x11A8..0x11C2) {
	180	$Comp2nd{$j} = 1;
ac5ea531	181	}
	182
	183	sub getCanonList {
8f118dcd	184	my @src = @_;
6c941e0c	185	my @dec = map {
	186	(SBase <= $_ && $_ <= SFinal) ? decomposeHangul($_)
	187	: $Canon{$_} ? @{ $Canon{$_} } : $_
	188	} @src;
8f118dcd	189	return join(" ",@src) eq join(" ",@dec) ? @dec : getCanonList(@dec);
8f118dcd	190	# condition @src == @dec is not ok.
ac5ea531	191	}
	192
	193	sub getCompatList {
8f118dcd	194	my @src = @_;
6c941e0c	195	my @dec = map {
	196	(SBase <= $_ && $_ <= SFinal) ? decomposeHangul($_)
	197	: $Compat{$_} ? @{ $Compat{$_} } : $_
	198	} @src;
8f118dcd	199	return join(" ",@src) eq join(" ",@dec) ? @dec : getCompatList(@dec);
8f118dcd	200	# condition @src == @dec is not ok.
ac5ea531	201	}
ac5ea531	202
48287974	203	# exhaustive decomposition
	204	foreach my $key (keys %Canon) {
	205	$Canon{$key} = [ getCanonList($key) ];
	206	}
	207
	208	# exhaustive decomposition
628bbff0	209	foreach my $key (keys %Compat) {
48287974	210	$Compat{$key} = [ getCompatList($key) ];
48287974	211	}
ac5ea531	212
9f1f04a1	213	sub _pack_U {
1efaba7f	214	return pack('U*', @_);
9f1f04a1	215	}
9f1f04a1	216
ac5ea531	217	sub _U_stringify {
8f118dcd	218	sprintf '"%s"', join '',
9f1f04a1	219	map sprintf("\\x%02x", $_), unpack 'C*', _pack_U(@_);
ac5ea531	220	}
	221
	222	foreach my $hash (\%Canon, \%Compat) {
8f118dcd	223	foreach my $key (keys %$hash) {
	224	$hash->{$key} = _U_stringify( @{ $hash->{$key} } );
	225	}
ac5ea531	226	}
ac5ea531	227
6c941e0c	228	########## writing header files ##########
ac5ea531	229
8f118dcd	230	my @boolfunc = (
	231	{
	232	name => "Exclusion",
	233	type => "bool",
	234	hash => \%Exclus,
	235	},
	236	{
	237	name => "Singleton",
	238	type => "bool",
	239	hash => \%Single,
	240	},
	241	{
	242	name => "NonStDecomp",
	243	type => "bool",
	244	hash => \%NonStD,
	245	},
	246	{
	247	name => "Comp2nd",
	248	type => "bool",
	249	hash => \%Comp2nd,
	250	},
	251	);
ac5ea531	252
	253	my $file = "unfexc.h";
	254	open FH, ">$file" or croak "$PACKAGE: $file can't be made";
	255	binmode FH; select FH;
	256
8f118dcd	257	print << 'EOF';
	258	/*
	259	* This file is auto-generated by mkheader.
	260	* Any changes here will be lost!
	261	*/
	262	EOF
ac5ea531	263
8f118dcd	264	foreach my $tbl (@boolfunc) {
	265	my @temp = sort {$a <=> $b} keys %{$tbl->{hash}};
	266	my $type = $tbl->{type};
	267	my $name = $tbl->{name};
	268	print "$type is$name (UV uv)\n{\nreturn\n\t";
	269
	270	while (@temp) {
	271	my $cur = shift @temp;
	272	if (@temp && $cur + 1 == $temp[0]) {
	273	print "($cur <= uv && uv <= ";
	274	while (@temp && $cur + 1 == $temp[0]) {
	275	$cur = shift @temp;
	276	}
	277	print "$cur)";
	278	print "\n\t\|\| " if @temp;
	279	} else {
	280	print "uv == $cur";
	281	print "\n\t\|\| " if @temp;
	282	}
ac5ea531	283	}
8f118dcd	284	print "\n\t? TRUE : FALSE;\n}\n\n";
ac5ea531	285	}
ac5ea531	286
ac5ea531	287	close FH;
	288
	289	####################################
	290
48287974	291	my $compinit =
	292	"typedef struct { UV nextchar; UV composite; } $structname;\n\n";
	293
	294	foreach my $i (sort keys %CompList) {
	295	$compinit .= "$structname $i [] = {\n";
	296	$compinit .= join ",\n",
	297	map sprintf("\t{ %d, %d }", $_, $CompList{$i}{$_}),
	298	sort {$a <=> $b } keys %{ $CompList{$i} };
	299	$compinit .= ",\n{0,0}\n};\n\n"; # with sentinel
	300	}
	301
ac5ea531	302	my @tripletable = (
8f118dcd	303	{
	304	file => "unfcmb",
	305	name => "combin",
	306	type => "STDCHAR",
	307	hash => \%Combin,
	308	null => 0,
	309	},
	310	{
	311	file => "unfcan",
	312	name => "canon",
	313	type => "char*",
	314	hash => \%Canon,
	315	null => "NULL",
	316	},
	317	{
	318	file => "unfcpt",
	319	name => "compat",
	320	type => "char*",
	321	hash => \%Compat,
	322	null => "NULL",
	323	},
	324	{
	325	file => "unfcmp",
	326	name => "compos",
	327	type => "$structname *",
	328	hash => \%Comp1st,
	329	null => "NULL",
	330	init => $compinit,
	331	},
ac5ea531	332	);
	333
	334	foreach my $tbl (@tripletable) {
8f118dcd	335	my $file = "$tbl->{file}.h";
	336	my $head = "${prefix}$tbl->{name}";
	337	my $type = $tbl->{type};
	338	my $hash = $tbl->{hash};
	339	my $null = $tbl->{null};
	340	my $init = $tbl->{init};
	341
	342	open FH, ">$file" or croak "$PACKAGE: $file can't be made";
	343	binmode FH; select FH;
	344	my %val;
	345
	346	print FH << 'EOF';
ac5ea531	347	/*
	348	* This file is auto-generated by mkheader.
	349	* Any changes here will be lost!
	350	*/
	351	EOF
	352
8f118dcd	353	print $init if defined $init;
	354
	355	foreach my $uv (keys %$hash) {
f027f502	356	croak sprintf("a Unicode code point 0x%04X over 0x10FFFF.", $uv)
f027f502	357	unless $uv <= 0x10FFFF;
8f118dcd	358	my @c = unpack 'CCCC', pack 'N', $uv;
	359	$val{ $c[1] }{ $c[2] }{ $c[3] } = $hash->{$uv};
	360	}
	361
	362	foreach my $p (sort { $a <=> $b } keys %val) {
	363	next if ! $val{ $p };
	364	for (my $r = 0; $r < 256; $r++) {
	365	next if ! $val{ $p }{ $r };
fe067ad9	366	printf "static $type ${head}_%02x_%02x [256] = {\n", $p, $r;
8f118dcd	367	for (my $c = 0; $c < 256; $c++) {
	368	print "\t", defined $val{$p}{$r}{$c}
	369	? "($type)".$val{$p}{$r}{$c}
	370	: $null;
	371	print ',' if $c != 255;
	372	print "\n" if $c % 8 == 7;
	373	}
	374	print "};\n\n";
	375	}
	376	}
	377	foreach my $p (sort { $a <=> $b } keys %val) {
	378	next if ! $val{ $p };
fe067ad9	379	printf "static $type* ${head}_%02x [256] = {\n", $p;
8f118dcd	380	for (my $r = 0; $r < 256; $r++) {
	381	print $val{ $p }{ $r }
	382	? sprintf("${head}_%02x_%02x", $p, $r)
	383	: "NULL";
	384	print ',' if $r != 255;
	385	print "\n" if $val{ $p }{ $r } \|\| ($r+1) % 8 == 0;
	386	}
	387	print "};\n\n";
ac5ea531	388	}
fe067ad9	389	print "static $type** $head [] = {\n";
8f118dcd	390	for (my $p = 0; $p <= 0x10; $p++) {
	391	print $val{ $p } ? sprintf("${head}_%02x", $p) : "NULL";
	392	print ',' if $p != 0x10;
	393	print "\n";
ac5ea531	394	}
ac5ea531	395	print "};\n\n";
8f118dcd	396	close FH;
ac5ea531	397	}
ac5ea531	398
628bbff0	399	1;
ac5ea531	400	__END__