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

package utf8;
use strict;
use warnings;

sub DEBUG () { 0 }

sub DESTROY {}

my %Cache;

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

##
## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape
##

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

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

    ##
    ## Get the list of codepoints for the type.
    ## Called from utf8.c
    ##
    ## Given a $type, our goal is to fill $list with the set of codepoint
    ## ranges.
    ##
    ## To make the parsing of $type clear, this code takes the a rather
    ## unorthodox approach of last'ing out of the block once we have the
    ## info we need. Were this to be a subroutine, the 'last' would just
    ## be a 'return'.
    ##
    my $file; ## file to load data from, and also part of the %Cache key.
    my $ListSorted = 0;

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

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

      GETFILE:
        {
            ##
            ## 'Is' is always optional, so if it's there, remove it.
            ## Same with 'Category=' and 'Script='.
            ##
            ## 'Block=' is replaced by 'In'.
            ##
            my $wasIs;

            ($wasIs = $type =~ s/^Is(?:\s+|[-_])?//i)
              or
            $type =~ s/^Category\s*=\s*//i
              or
            $type =~ s/^Script\s*=\s*//i
              or
            $type =~ s/^Block\s*=\s*/In/i;

            ##
            ## See if it's in the direct mapping table.
            ##
            require "unicore/Exact.pl";
            if (my $base = $utf8::Exact{$type}) {
                $file = "unicore/lib/$base.pl";
                last GETFILE;
            }

            ##
            ## If not there exactly, try the canonical form. The canonical
            ## form is lowercased, with any separators (\s+|[-_]) removed.
            ##
            my $canonical = lc $type;
            $canonical =~ s/(?<=[a-z\d])(?:\s+|[-_])(?=[a-z\d])//g;
            print "canonical = $canonical\n" if DEBUG;

            require "unicore/Canonical.pl";
            if (my $base = $utf8::Canonical{$canonical}) {
                $file = "unicore/lib/$base.pl";
                last GETFILE;
            }

	    ##
	    ## It could be a user-defined property.
	    ##

	    my $caller = caller(1);

	    if (defined $caller && $type =~ /^(?:\w+)$/) {
		my $prop = $caller . "::" . ( $wasIs ? "Is" : "" ) . $type;
		if (exists &{$prop}) {
		    no strict 'refs';
		    
		    $list = &{$prop};
		    last GETFILE;
		}
	    }

            ##
            ## Last attempt -- see if it's a "To" name (e.g. "ToLower")
            ##
            if ($type =~ /^To([A-Z][A-Za-z]+)$/)
            {
                $file = "unicore/To/$1.pl";
                ## would like to test to see if $file actually exists....
                last GETFILE;
            }

            ##
            ## If we reach this line, it's because we couldn't figure
            ## out what to do with $type. Ouch.
            ##

            return $type;
        }

	if (defined $file) {
	    print "found it (file='$file')\n" if DEBUG;

	    ##
	    ## If we reach here, it was due to a 'last GETFILE' above
	    ## (exception: user-defined properties), so we
	    ## have a filename, so now we load it if we haven't already.
	    ## If we have, return the cached results. The cache key is the
	    ## file to load.
	    ##
	    if ($Cache{$file} and ref($Cache{$file}) eq $class)
	    {
		print "Returning cached '$file' for \\p{$type}\n" if DEBUG;
		return $Cache{$class, $file};
	    }

	    $list = do $file;
	}

        $ListSorted = 1; ## we know that these lists are sorted
    }

    my $extras;
    my $bits;

    my $ORIG = $list;
    if ($list) {
	my @tmp = split(/^/m, $list);
	my %seen;
	no warnings;
	$extras = join '', grep /^[^0-9a-fA-F]/, @tmp;
	$list = join '',
	    map  { $_->[1] }
	    sort { $a->[0] <=> $b->[0] }
	    map  { /^([0-9a-fA-F]+)/; [ hex($1), $_ ] }
	    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 $subobj;
		if ($c eq 'utf8') {
		    $subobj = $c->SWASHNEW($t, "", 0, 0, 0);
		}
		elsif ($c =~ /^([0-9a-fA-F]+)/) {
		    $subobj = utf8->SWASHNEW("", $c, 0, 0, 0);
		}
		return $subobj unless ref $subobj;
		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;

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

    if ($file) {
        $Cache{$class, $file} = $SWASH;
    }

    return $SWASH;
}

# 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;
cf25bb62	2	use strict;
cf25bb62	3	use warnings;
a0ed51b3	4
15732964	5	sub DEBUG () { 0 }
a0ed51b3	6
	7	sub DESTROY {}
	8
5beb625e	9	my %Cache;
5beb625e	10
86916d66	11	sub croak { require Carp; Carp::croak(@_) }
86916d66	12
5beb625e	13	##
	14	## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape
	15	##
	16
a0ed51b3	17	sub SWASHNEW {
	18	my ($class, $type, $list, $minbits, $none) = @_;
	19	local $^D = 0 if $^D;
15732964	20
	21	print STDERR "SWASHNEW @_\n" if DEBUG;
	22
cf25bb62	23	##
	24	## Get the list of codepoints for the type.
	25	## Called from utf8.c
	26	##
	27	## Given a $type, our goal is to fill $list with the set of codepoint
5beb625e	28	## ranges.
cf25bb62	29	##
cf25bb62	30	## To make the parsing of $type clear, this code takes the a rather
2d92f8a0	31	## unorthodox approach of last'ing out of the block once we have the
cf25bb62	32	## info we need. Were this to be a subroutine, the 'last' would just
	33	## be a 'return'.
	34	##
5beb625e	35	my $file; ## file to load data from, and also part of the %Cache key.
	36	my $ListSorted = 0;
	37
cf25bb62	38	if ($type)
	39	{
	40	$type =~ s/^\s+//;
	41	$type =~ s/\s+$//;
	42
	43	print "type = $type\n" if DEBUG;
	44
cf25bb62	45	GETFILE:
	46	{
	47	##
5beb625e	48	## 'Is' is always optional, so if it's there, remove it.
5beb625e	49	## Same with 'Category=' and 'Script='.
cf25bb62	50	##
5beb625e	51	## 'Block=' is replaced by 'In'.
cf25bb62	52	##
2d92f8a0	53	my $wasIs;
	54
	55	($wasIs = $type =~ s/^Is(?:\s+\|[-_])?//i)
5beb625e	56	or
	57	$type =~ s/^Category\s=\s//i
	58	or
	59	$type =~ s/^Script\s=\s//i
	60	or
	61	$type =~ s/^Block\s=\s/In/i;
cf25bb62	62
5beb625e	63	##
	64	## See if it's in the direct mapping table.
	65	##
	66	require "unicore/Exact.pl";
	67	if (my $base = $utf8::Exact{$type}) {
	68	$file = "unicore/lib/$base.pl";
	69	last GETFILE;
cf25bb62	70	}
	71
	72	##
5beb625e	73	## If not there exactly, try the canonical form. The canonical
5beb625e	74	## form is lowercased, with any separators (\s+\|[-_]) removed.
cf25bb62	75	##
5beb625e	76	my $canonical = lc $type;
	77	$canonical =~ s/(?<=[a-z\d])(?:\s+\|[-_])(?=[a-z\d])//g;
	78	print "canonical = $canonical\n" if DEBUG;
cf25bb62	79
5beb625e	80	require "unicore/Canonical.pl";
	81	if (my $base = $utf8::Canonical{$canonical}) {
	82	$file = "unicore/lib/$base.pl";
	83	last GETFILE;
cf25bb62	84	}
cf25bb62	85
491fd90a	86	##
	87	## It could be a user-defined property.
	88	##
	89
2d92f8a0	90	my $caller = caller(1);
	91
	92	if (defined $caller && $type =~ /^(?:\w+)$/) {
	93	my $prop = $caller . "::" . ( $wasIs ? "Is" : "" ) . $type;
	94	if (exists &{$prop}) {
	95	no strict 'refs';
	96
	97	$list = &{$prop};
	98	last GETFILE;
	99	}
	100	}
491fd90a	101
cf25bb62	102	##
	103	## Last attempt -- see if it's a "To" name (e.g. "ToLower")
	104	##
	105	if ($type =~ /^To([A-Z][A-Za-z]+)$/)
	106	{
	107	$file = "unicore/To/$1.pl";
	108	## would like to test to see if $file actually exists....
	109	last GETFILE;
	110	}
	111
	112	##
	113	## If we reach this line, it's because we couldn't figure
	114	## out what to do with $type. Ouch.
	115	##
bc45ce41	116
bc45ce41	117	return $type;
cf25bb62	118	}
cf25bb62	119
491fd90a	120	if (defined $file) {
	121	print "found it (file='$file')\n" if DEBUG;
	122
	123	##
	124	## If we reach here, it was due to a 'last GETFILE' above
	125	## (exception: user-defined properties), so we
	126	## have a filename, so now we load it if we haven't already.
	127	## If we have, return the cached results. The cache key is the
	128	## file to load.
	129	##
	130	if ($Cache{$file} and ref($Cache{$file}) eq $class)
	131	{
	132	print "Returning cached '$file' for \\p{$type}\n" if DEBUG;
	133	return $Cache{$class, $file};
	134	}
	135
	136	$list = do $file;
	137	}
5beb625e	138
5beb625e	139	$ListSorted = 1; ## we know that these lists are sorted
886ba366	140	}
a0ed51b3	141
15732964	142	my $extras;
15732964	143	my $bits;
cf25bb62	144
5beb625e	145	my $ORIG = $list;
a0ed51b3	146	if ($list) {
	147	my @tmp = split(/^/m, $list);
	148	my %seen;
db376a24	149	no warnings;
a0ed51b3	150	$extras = join '', grep /^[^0-9a-fA-F]/, @tmp;
a0ed51b3	151	$list = join '',
df68b396	152	map { $_->[1] }
	153	sort { $a->[0] <=> $b->[0] }
	154	map { /^([0-9a-fA-F]+)/; [ hex($1), $_ ] }
	155	grep { /^([0-9a-fA-F]+)/ and not $seen{$1}++ } @tmp; # XXX doesn't do ranges right
a0ed51b3	156	}
	157
	158	if ($none) {
	159	my $hextra = sprintf "%04x", $none + 1;
	160	$list =~ s/\tXXXX$/\t$hextra/mg;
	161	}
	162
	163	if ($minbits < 32) {
	164	my $top = 0;
	165	while ($list =~ /^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
	166	my $min = hex $1;
	167	my $max = hex(defined $2 ? $2 : $1);
	168	my $val = hex(defined $3 ? $3 : "");
	169	$val += $max - $min if defined $3;
	170	$top = $val if $val > $top;
	171	}
	172	$bits =
	173	$top > 0xffff ? 32 :
	174	$top > 0xff ? 16 :
	175	$top > 1 ? 8 : 1
	176	}
	177	$bits = $minbits if $bits < $minbits;
	178
	179	my @extras;
	180	for my $x ($extras) {
	181	pos $x = 0;
1fef36c7	182	while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
a0ed51b3	183	my $char = $1;
a0ed51b3	184	my $name = $2;
6d6ae962	185	print STDERR "$1 => $2\n" if DEBUG;
a0ed51b3	186	if ($char =~ /[-+!]/) {
a0ed51b3	187	my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really
11ef8fdd	188	my $subobj;
	189	if ($c eq 'utf8') {
	190	$subobj = $c->SWASHNEW($t, "", 0, 0, 0);
	191	}
	192	elsif ($c =~ /^([0-9a-fA-F]+)/) {
	193	$subobj = utf8->SWASHNEW("", $c, 0, 0, 0);
	194	}
bc45ce41	195	return $subobj unless ref $subobj;
a0ed51b3	196	push @extras, $name => $subobj;
	197	$bits = $subobj->{BITS} if $bits < $subobj->{BITS};
	198	}
	199	}
	200	}
	201
15732964	202	print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG;
a0ed51b3	203
5beb625e	204	my $SWASH = bless {
a0ed51b3	205	TYPE => $type,
	206	BITS => $bits,
	207	EXTRAS => $extras,
	208	LIST => $list,
	209	NONE => $none,
	210	@extras,
	211	} => $class;
5beb625e	212
	213	if ($file) {
	214	$Cache{$class, $file} = $SWASH;
	215	}
	216
	217	return $SWASH;
a0ed51b3	218	}
	219
	220	# NOTE: utf8.c:swash_init() assumes entries are never modified once generated.
	221
	222	sub SWASHGET {
035d37be	223	# See utf8.c:Perl_swash_fetch for problems with this interface.
a0ed51b3	224	my ($self, $start, $len) = @_;
a0ed51b3	225	local $^D = 0 if $^D;
a0ed51b3	226	my $type = $self->{TYPE};
	227	my $bits = $self->{BITS};
	228	my $none = $self->{NONE};
15732964	229	print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if DEBUG;
a0ed51b3	230	my $end = $start + $len;
	231	my $swatch = "";
	232	my $key;
	233	vec($swatch, $len - 1, $bits) = 0; # Extend to correct length.
	234	if ($none) {
	235	for $key (0 .. $len - 1) { vec($swatch, $key, $bits) = $none }
	236	}
	237
	238	for ($self->{LIST}) {
	239	pos $_ = 0;
	240	if ($bits > 1) {
	241	LINE:
	242	while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
	243	my $min = hex $1;
	244	my $max = (defined $2 ? hex $2 : $min);
	245	my $val = hex $3;
	246	next if $max < $start;
6d6ae962	247	print "$min $max $val\n" if DEBUG;
a0ed51b3	248	if ($none) {
a0ed51b3	249	if ($min < $start) {
710250cb	250	$val += $start - $min if $val < $none;
a0ed51b3	251	$min = $start;
	252	}
	253	for ($key = $min; $key <= $max; $key++) {
	254	last LINE if $key >= $end;
6d6ae962	255	print STDERR "$key => $val\n" if DEBUG;
a0ed51b3	256	vec($swatch, $key - $start, $bits) = $val;
	257	++$val if $val < $none;
	258	}
	259	}
	260	else {
	261	if ($min < $start) {
	262	$val += $start - $min;
	263	$min = $start;
	264	}
	265	for ($key = $min; $key <= $max; $key++, $val++) {
	266	last LINE if $key >= $end;
6d6ae962	267	print STDERR "$key => $val\n" if DEBUG;
a0ed51b3	268	vec($swatch, $key - $start, $bits) = $val;
	269	}
	270	}
	271	}
	272	}
	273	else {
	274	LINE:
99a6b1f0	275	while (/^([0-9a-fA-F]+)(?:[ \t]+([0-9a-fA-F]+))?/mg) {
a0ed51b3	276	my $min = hex $1;
	277	my $max = (defined $2 ? hex $2 : $min);
	278	next if $max < $start;
	279	if ($min < $start) {
	280	$min = $start;
	281	}
	282	for ($key = $min; $key <= $max; $key++) {
	283	last LINE if $key >= $end;
6d6ae962	284	print STDERR "$key => 1\n" if DEBUG;
a0ed51b3	285	vec($swatch, $key - $start, 1) = 1;
	286	}
	287	}
	288	}
	289	}
	290	for my $x ($self->{EXTRAS}) {
	291	pos $x = 0;
1fef36c7	292	while ($x =~ /^([-+!])(.*)/mg) {
a0ed51b3	293	my $char = $1;
a0ed51b3	294	my $name = $2;
15732964	295	print STDERR "INDIRECT $1 $2\n" if DEBUG;
1fef36c7	296	my $otherbits = $self->{$name}->{BITS};
	297	croak("SWASHGET size mismatch") if $bits < $otherbits;
	298	my $other = $self->{$name}->SWASHGET($start, $len);
	299	if ($char eq '+') {
	300	if ($bits == 1 and $otherbits == 1) {
	301	$swatch \|= $other;
a0ed51b3	302	}
1fef36c7	303	else {
	304	for ($key = 0; $key < $len; $key++) {
	305	vec($swatch, $key, $bits) = vec($other, $key, $otherbits);
a0ed51b3	306	}
1fef36c7	307	}
	308	}
	309	elsif ($char eq '!') {
	310	if ($bits == 1 and $otherbits == 1) {
	311	$swatch \|= ~$other;
	312	}
	313	else {
	314	for ($key = 0; $key < $len; $key++) {
	315	if (!vec($other, $key, $otherbits)) {
	316	vec($swatch, $key, $bits) = 1;
a0ed51b3	317	}
	318	}
	319	}
1fef36c7	320	}
	321	elsif ($char eq '-') {
	322	if ($bits == 1 and $otherbits == 1) {
	323	$swatch &= ~$other;
	324	}
	325	else {
	326	for ($key = 0; $key < $len; $key++) {
	327	if (vec($other, $key, $otherbits)) {
	328	vec($swatch, $key, $bits) = 0;
a0ed51b3	329	}
	330	}
	331	}
	332	}
	333	}
	334	}
15732964	335	if (DEBUG) {
a0ed51b3	336	print STDERR "CELLS ";
	337	for ($key = 0; $key < $len; $key++) {
	338	print STDERR vec($swatch, $key, $bits), " ";
	339	}
	340	print STDERR "\n";
	341	}
	342	$swatch;
	343	}
	344
	345	1;