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

package utf8;
use strict;
use warnings;

sub DEBUG () { 0 }

sub DESTROY {}

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

my %Cache;

##
## "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
    ## unorthadox 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'.
            ##
            $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;
            }

            ##
            ## 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 $type. Ouch.
            ##
            croak("Can't find Unicode character property \"$type\"");
        }

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

        ##
        ## If we reach here, it was due to a 'last GETFILE' above, 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 '',
	    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;

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