[p5sagit/p5-mst-13.2.git] / lib / utf8_heavy.pl

package utf8;

sub DEBUG () { 0 }

sub DESTROY {}

sub croak { require Carp; Carp::croak(@_) }

sub SWASHNEW {
    my ($class, $type, $list, $minbits, $none) = @_;
    local $^D = 0 if $^D;

    print STDERR "SWASHNEW @_\n" if DEBUG;

    if ($type and ref ${"${class}::{$type}"} eq $class) {
	warn qq/Found \${"${class}::{$type}"}\n/ if DEBUG;
	return ${"${class}::{$type}"};	# Already there...
    }

    if ($type) {
	$type =~ s/^\s+//;
	$type =~ s/\s+$//;

	print "type = $type\n" if DEBUG;

	my $file;

	unless (defined $file) {
	    defined %utf8::Is || do "unicore/Is.pl";
	    if ($type =~ /^(?:Is)?[- _]?([A-Z].*)$/i) {
		my $istype = $1;
		print "istype = $istype\n" if DEBUG;
		unless ($list = do "unicore/Is/$istype.pl") {
		    if (exists $utf8::Is{$istype}) {
			$file = "unicore/Is/$utf8::Is{$istype}";
		    } else {
			my $isprefix = substr(lc($istype), 0, 2);
			print "isprefix = $isprefix\n" if DEBUG;
			if (exists $utf8::IsPat{$isprefix}) {
			    my $Is = $istype;
			    print "isprefix = $isprefix, Is = $Is\n" if DEBUG;
			    for my $k (keys %{$utf8::IsPat{$isprefix}}) {
				print "isprefix = $isprefix, Is = $Is, k = $k\n" if DEBUG;
				if ($Is =~ /^$k$/i) {
				    $file = "unicore/Is/$utf8::IsPat{$isprefix}->{$k}";
				    print "isprefix = $isprefix, Is = $Is, k = $k, file = $file\n" if DEBUG;
				    last;
				}
			    }
			}
		    }
		}
	    }
			
	    unless (defined $file) {
		defined %utf8::In || do "unicore/In.pl";
		$type = 'Lampersand' if $type =~ /^(?:Is)?L&$/;
		if ($type =~ /^(?:In)?[- _]?(?!herited$)(.+)/i) {
		    my $intype = $1;
		    print "intype = $intype\n" if DEBUG;
		    if (exists $utf8::Is{$istype}) {
			$file = "unicore/In/$utf8::In{$intype}";
		    } else {
			my $inprefix = substr(lc($intype), 0, 2);
			print "inprefix = $inprefix\n" if DEBUG;
			if (exists $utf8::InPat{$inprefix}) {
			    my $In = $intype;
			    print "inprefix = $inprefix, In = $In\n" if DEBUG;
			    for my $k (keys %{$utf8::InPat{$inprefix}}) {
				print "inprefix = $inprefix, In = $In, k = $k\n" if DEBUG;
				if ($In =~ /^$k$/i) {
				    $file = "unicore/In/$utf8::InPat{$inprefix}->{$k}";
				    print "inprefix = $inprefix, In = $In, k = $k, file = $file\n" if DEBUG;
				    last;
				}
			    }
			}
		    }
		}
	    }

	    unless (defined $file) {
		if ($type =~ /^To([A-Z][A-Za-z]+)$/) {
		    $file = "unicore/To/$1";
		}
	    }
	}

	if (defined $file) {
	    $list = do "$file.pl";
	}

	croak("Can't find Unicode character property \"$type\"")
	    unless $list;
    }

    my $extras;
    my $bits;
 
    if ($list) {
	my @tmp = split(/^/m, $list);
	my %seen;
	no warnings;
	$extras = join '', grep /^[^0-9a-fA-F]/, @tmp;
	$list = join '',
	    sort { hex $a <=> hex $b }
	    grep {/^([0-9a-fA-F]+)/ and not $seen{$1}++} @tmp; # XXX doesn't do ranges right
    }

    if ($none) {
	my $hextra = sprintf "%04x", $none + 1;
	$list =~ s/\tXXXX$/\t$hextra/mg;
    }

    if ($minbits < 32) {
	my $top = 0;
	while ($list =~ /^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
	    my $min = hex $1;
	    my $max = hex(defined $2 ? $2 : $1);
	    my $val = hex(defined $3 ? $3 : "");
	    $val += $max - $min if defined $3;
	    $top = $val if $val > $top;
	}
	$bits =
	    $top > 0xffff ? 32 :
	    $top > 0xff ? 16 :
	    $top > 1 ? 8 : 1
    }
    $bits = $minbits if $bits < $minbits;

    my @extras;
    for my $x ($extras) {
	pos $x = 0;
	while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
	    my $char = $1;
	    my $name = $2;
	    print STDERR "$1 => $2\n" if DEBUG;
	    if ($char =~ /[-+!]/) {
		my ($c,$t) = split(/::/, $name, 2);	# bogus use of ::, really
		my $subobj = $c->SWASHNEW($t, "", 0, 0, 0);
		push @extras, $name => $subobj;
		$bits = $subobj->{BITS} if $bits < $subobj->{BITS};
	    }
	}
    }

    print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG;

    ${"${class}::{$type}"} = bless {
	TYPE => $type,
	BITS => $bits,
	EXTRAS => $extras,
	LIST => $list,
	NONE => $none,
	@extras,
    } => $class;
}

# NOTE: utf8.c:swash_init() assumes entries are never modified once generated.

sub SWASHGET {
    # See utf8.c:Perl_swash_fetch for problems with this interface.
    my ($self, $start, $len) = @_;
    local $^D = 0 if $^D;
    my $type = $self->{TYPE};
    my $bits = $self->{BITS};
    my $none = $self->{NONE};
    print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if DEBUG;
    my $end = $start + $len;
    my $swatch = "";
    my $key;
    vec($swatch, $len - 1, $bits) = 0;	# Extend to correct length.
    if ($none) {
	for $key (0 .. $len - 1) { vec($swatch, $key, $bits) = $none }
    }

    for ($self->{LIST}) {
	pos $_ = 0;
	if ($bits > 1) {
	  LINE:
	    while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
		my $min = hex $1;
		my $max = (defined $2 ? hex $2 : $min);
		my $val = hex $3;
		next if $max < $start;
		print "$min $max $val\n" if DEBUG;
		if ($none) {
		    if ($min < $start) {
			$val += $start - $min if $val < $none;
			$min = $start;
		    }
		    for ($key = $min; $key <= $max; $key++) {
			last LINE if $key >= $end;
			print STDERR "$key => $val\n" if DEBUG;
			vec($swatch, $key - $start, $bits) = $val;
			++$val if $val < $none;
		    }
		}
		else {
		    if ($min < $start) {
			$val += $start - $min;
			$min = $start;
		    }
		    for ($key = $min; $key <= $max; $key++, $val++) {
			last LINE if $key >= $end;
			print STDERR "$key => $val\n" if DEBUG;
			vec($swatch, $key - $start, $bits) = $val;
		    }
		}
	    }
	}
	else {
	  LINE:
	    while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+))?/mg) {
		my $min = hex $1;
		my $max = (defined $2 ? hex $2 : $min);
		next if $max < $start;
		if ($min < $start) {
		    $min = $start;
		}
		for ($key = $min; $key <= $max; $key++) {
		    last LINE if $key >= $end;
		    print STDERR "$key => 1\n" if DEBUG;
		    vec($swatch, $key - $start, 1) = 1;
		}
	    }
	}
    }
    for my $x ($self->{EXTRAS}) {
	pos $x = 0;
	while ($x =~ /^([-+!])(.*)/mg) {
	    my $char = $1;
	    my $name = $2;
	    print STDERR "INDIRECT $1 $2\n" if DEBUG;
	    my $otherbits = $self->{$name}->{BITS};
	    croak("SWASHGET size mismatch") if $bits < $otherbits;
	    my $other = $self->{$name}->SWASHGET($start, $len);
	    if ($char eq '+') {
		if ($bits == 1 and $otherbits == 1) {
		    $swatch |= $other;
		}
		else {
		    for ($key = 0; $key < $len; $key++) {
			vec($swatch, $key, $bits) = vec($other, $key, $otherbits);
		    }
		}
	    }
	    elsif ($char eq '!') {
		if ($bits == 1 and $otherbits == 1) {
		    $swatch |= ~$other;
		}
		else {
		    for ($key = 0; $key < $len; $key++) {
			if (!vec($other, $key, $otherbits)) {
			    vec($swatch, $key, $bits) = 1;
			}
		    }
		}
	    }
	    elsif ($char eq '-') {
		if ($bits == 1 and $otherbits == 1) {
		    $swatch &= ~$other;
		}
		else {
		    for ($key = 0; $key < $len; $key++) {
			if (vec($other, $key, $otherbits)) {
			    vec($swatch, $key, $bits) = 0;
			}
		    }
		}
	    }
	}
    }
    if (DEBUG) {
	print STDERR "CELLS ";
	for ($key = 0; $key < $len; $key++) {
	    print STDERR vec($swatch, $key, $bits), " ";
	}
	print STDERR "\n";
    }
    $swatch;
}

1;
Commit	Line	Data
a0ed51b3	1	package utf8;
a0ed51b3	2
15732964	3	sub DEBUG () { 0 }
a0ed51b3	4
	5	sub DESTROY {}
	6
	7	sub croak { require Carp; Carp::croak(@_) }
	8
	9	sub SWASHNEW {
	10	my ($class, $type, $list, $minbits, $none) = @_;
	11	local $^D = 0 if $^D;
15732964	12
	13	print STDERR "SWASHNEW @_\n" if DEBUG;
	14
a0ed51b3	15	if ($type and ref ${"${class}::{$type}"} eq $class) {
15732964	16	warn qq/Found \${"${class}::{$type}"}\n/ if DEBUG;
a0ed51b3	17	return ${"${class}::{$type}"}; # Already there...
	18	}
	19
15732964	20	if ($type) {
d73e5302	21	$type =~ s/^\s+//;
d73e5302	22	$type =~ s/\s+$//;
15732964	23
d73e5302	24	print "type = $type\n" if DEBUG;
15732964	25
d73e5302	26	my $file;
15732964	27
d73e5302	28	unless (defined $file) {
	29	defined %utf8::Is \|\| do "unicore/Is.pl";
	30	if ($type =~ /^(?:Is)?[- _]?([A-Z].*)$/i) {
	31	my $istype = $1;
	32	print "istype = $istype\n" if DEBUG;
	33	unless ($list = do "unicore/Is/$istype.pl") {
	34	if (exists $utf8::Is{$istype}) {
	35	$file = "unicore/Is/$utf8::Is{$istype}";
	36	} else {
	37	my $isprefix = substr(lc($istype), 0, 2);
	38	print "isprefix = $isprefix\n" if DEBUG;
	39	if (exists $utf8::IsPat{$isprefix}) {
	40	my $Is = $istype;
	41	print "isprefix = $isprefix, Is = $Is\n" if DEBUG;
	42	for my $k (keys %{$utf8::IsPat{$isprefix}}) {
	43	print "isprefix = $isprefix, Is = $Is, k = $k\n" if DEBUG;
	44	if ($Is =~ /^$k$/i) {
	45	$file = "unicore/Is/$utf8::IsPat{$isprefix}->{$k}";
	46	print "isprefix = $isprefix, Is = $Is, k = $k, file = $file\n" if DEBUG;
	47	last;
	48	}
	49	}
	50	}
	51	}
	52	}
	53	}
	54
	55	unless (defined $file) {
	56	defined %utf8::In \|\| do "unicore/In.pl";
	57	$type = 'Lampersand' if $type =~ /^(?:Is)?L&$/;
	58	if ($type =~ /^(?:In)?[- _]?(?!herited$)(.+)/i) {
	59	my $intype = $1;
	60	print "intype = $intype\n" if DEBUG;
	61	if (exists $utf8::Is{$istype}) {
	62	$file = "unicore/In/$utf8::In{$intype}";
	63	} else {
	64	my $inprefix = substr(lc($intype), 0, 2);
	65	print "inprefix = $inprefix\n" if DEBUG;
	66	if (exists $utf8::InPat{$inprefix}) {
	67	my $In = $intype;
	68	print "inprefix = $inprefix, In = $In\n" if DEBUG;
	69	for my $k (keys %{$utf8::InPat{$inprefix}}) {
	70	print "inprefix = $inprefix, In = $In, k = $k\n" if DEBUG;
	71	if ($In =~ /^$k$/i) {
	72	$file = "unicore/In/$utf8::InPat{$inprefix}->{$k}";
	73	print "inprefix = $inprefix, In = $In, k = $k, file = $file\n" if DEBUG;
	74	last;
	75	}
	76	}
	77	}
d9efae67	78	}
	79	}
	80	}
15732964	81
d73e5302	82	unless (defined $file) {
	83	if ($type =~ /^To([A-Z][A-Za-z]+)$/) {
	84	$file = "unicore/To/$1";
	85	}
dee4fd60	86	}
15732964	87	}
886ba366	88
d73e5302	89	if (defined $file) {
	90	$list = do "$file.pl";
	91	}
	92
	93	croak("Can't find Unicode character property \"$type\"")
	94	unless $list;
886ba366	95	}
a0ed51b3	96
15732964	97	my $extras;
	98	my $bits;
	99
a0ed51b3	100	if ($list) {
	101	my @tmp = split(/^/m, $list);
	102	my %seen;
db376a24	103	no warnings;
a0ed51b3	104	$extras = join '', grep /^[^0-9a-fA-F]/, @tmp;
	105	$list = join '',
	106	sort { hex $a <=> hex $b }
	107	grep {/^([0-9a-fA-F]+)/ and not $seen{$1}++} @tmp; # XXX doesn't do ranges right
	108	}
	109
	110	if ($none) {
	111	my $hextra = sprintf "%04x", $none + 1;
	112	$list =~ s/\tXXXX$/\t$hextra/mg;
	113	}
	114
	115	if ($minbits < 32) {
	116	my $top = 0;
	117	while ($list =~ /^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
	118	my $min = hex $1;
	119	my $max = hex(defined $2 ? $2 : $1);
	120	my $val = hex(defined $3 ? $3 : "");
	121	$val += $max - $min if defined $3;
	122	$top = $val if $val > $top;
	123	}
	124	$bits =
	125	$top > 0xffff ? 32 :
	126	$top > 0xff ? 16 :
	127	$top > 1 ? 8 : 1
	128	}
	129	$bits = $minbits if $bits < $minbits;
	130
	131	my @extras;
	132	for my $x ($extras) {
	133	pos $x = 0;
1fef36c7	134	while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
a0ed51b3	135	my $char = $1;
a0ed51b3	136	my $name = $2;
6d6ae962	137	print STDERR "$1 => $2\n" if DEBUG;
a0ed51b3	138	if ($char =~ /[-+!]/) {
	139	my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really
	140	my $subobj = $c->SWASHNEW($t, "", 0, 0, 0);
	141	push @extras, $name => $subobj;
	142	$bits = $subobj->{BITS} if $bits < $subobj->{BITS};
	143	}
	144	}
	145	}
	146
15732964	147	print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG;
a0ed51b3	148
	149	${"${class}::{$type}"} = bless {
	150	TYPE => $type,
	151	BITS => $bits,
	152	EXTRAS => $extras,
	153	LIST => $list,
	154	NONE => $none,
	155	@extras,
	156	} => $class;
	157	}
	158
	159	# NOTE: utf8.c:swash_init() assumes entries are never modified once generated.
	160
	161	sub SWASHGET {
035d37be	162	# See utf8.c:Perl_swash_fetch for problems with this interface.
a0ed51b3	163	my ($self, $start, $len) = @_;
a0ed51b3	164	local $^D = 0 if $^D;
a0ed51b3	165	my $type = $self->{TYPE};
	166	my $bits = $self->{BITS};
	167	my $none = $self->{NONE};
15732964	168	print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if DEBUG;
a0ed51b3	169	my $end = $start + $len;
	170	my $swatch = "";
	171	my $key;
	172	vec($swatch, $len - 1, $bits) = 0; # Extend to correct length.
	173	if ($none) {
	174	for $key (0 .. $len - 1) { vec($swatch, $key, $bits) = $none }
	175	}
	176
	177	for ($self->{LIST}) {
	178	pos $_ = 0;
	179	if ($bits > 1) {
	180	LINE:
	181	while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
	182	my $min = hex $1;
	183	my $max = (defined $2 ? hex $2 : $min);
	184	my $val = hex $3;
	185	next if $max < $start;
6d6ae962	186	print "$min $max $val\n" if DEBUG;
a0ed51b3	187	if ($none) {
a0ed51b3	188	if ($min < $start) {
710250cb	189	$val += $start - $min if $val < $none;
a0ed51b3	190	$min = $start;
	191	}
	192	for ($key = $min; $key <= $max; $key++) {
	193	last LINE if $key >= $end;
6d6ae962	194	print STDERR "$key => $val\n" if DEBUG;
a0ed51b3	195	vec($swatch, $key - $start, $bits) = $val;
	196	++$val if $val < $none;
	197	}
	198	}
	199	else {
	200	if ($min < $start) {
	201	$val += $start - $min;
	202	$min = $start;
	203	}
	204	for ($key = $min; $key <= $max; $key++, $val++) {
	205	last LINE if $key >= $end;
6d6ae962	206	print STDERR "$key => $val\n" if DEBUG;
a0ed51b3	207	vec($swatch, $key - $start, $bits) = $val;
	208	}
	209	}
	210	}
	211	}
	212	else {
	213	LINE:
	214	while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+))?/mg) {
	215	my $min = hex $1;
	216	my $max = (defined $2 ? hex $2 : $min);
	217	next if $max < $start;
	218	if ($min < $start) {
	219	$min = $start;
	220	}
	221	for ($key = $min; $key <= $max; $key++) {
	222	last LINE if $key >= $end;
6d6ae962	223	print STDERR "$key => 1\n" if DEBUG;
a0ed51b3	224	vec($swatch, $key - $start, 1) = 1;
	225	}
	226	}
	227	}
	228	}
	229	for my $x ($self->{EXTRAS}) {
	230	pos $x = 0;
1fef36c7	231	while ($x =~ /^([-+!])(.*)/mg) {
a0ed51b3	232	my $char = $1;
a0ed51b3	233	my $name = $2;
15732964	234	print STDERR "INDIRECT $1 $2\n" if DEBUG;
1fef36c7	235	my $otherbits = $self->{$name}->{BITS};
	236	croak("SWASHGET size mismatch") if $bits < $otherbits;
	237	my $other = $self->{$name}->SWASHGET($start, $len);
	238	if ($char eq '+') {
	239	if ($bits == 1 and $otherbits == 1) {
	240	$swatch \|= $other;
a0ed51b3	241	}
1fef36c7	242	else {
	243	for ($key = 0; $key < $len; $key++) {
	244	vec($swatch, $key, $bits) = vec($other, $key, $otherbits);
a0ed51b3	245	}
1fef36c7	246	}
	247	}
	248	elsif ($char eq '!') {
	249	if ($bits == 1 and $otherbits == 1) {
	250	$swatch \|= ~$other;
	251	}
	252	else {
	253	for ($key = 0; $key < $len; $key++) {
	254	if (!vec($other, $key, $otherbits)) {
	255	vec($swatch, $key, $bits) = 1;
a0ed51b3	256	}
	257	}
	258	}
1fef36c7	259	}
	260	elsif ($char eq '-') {
	261	if ($bits == 1 and $otherbits == 1) {
	262	$swatch &= ~$other;
	263	}
	264	else {
	265	for ($key = 0; $key < $len; $key++) {
	266	if (vec($other, $key, $otherbits)) {
	267	vec($swatch, $key, $bits) = 0;
a0ed51b3	268	}
	269	}
	270	}
	271	}
	272	}
	273	}
15732964	274	if (DEBUG) {
a0ed51b3	275	print STDERR "CELLS ";
	276	for ($key = 0; $key < $len; $key++) {
	277	print STDERR vec($swatch, $key, $bits), " ";
	278	}
	279	print STDERR "\n";
	280	}
	281	$swatch;
	282	}
	283
	284	1;