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

package utf8;
use strict;
use warnings;

sub DEBUG () { 0 }

sub DESTROY {}

my %Cache;

our (%PropertyAlias, %PA_reverse, %PropValueAlias, %PVA_reverse, %PVA_abbr_map);

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

##
## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape.
## It's a data structure that encodes a set of Unicode characters.
##

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

	    my $caller1 = $type =~ s/(.+)::// ? $1 : caller(1);

	    if (defined $caller1 && $type =~ /^(?:\w+)$/) {
		my $prop = "${caller1}::$type";
		if (exists &{$prop}) {
		    no strict 'refs';
		    
		    $list = &{$prop};
		    last GETFILE;
		}
	    }

            my $wasIs;

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


	    ##
	    ## See if it's in some enumeration.
	    ##
	    require "unicore/PVA.pl";
	    if ($type =~ /^([\w\s]+)[:=]\s*(.*)/) {
		my ($enum, $val) = (lc $1, lc $2);
		$enum =~ tr/ _-//d;
		$val =~ tr/ _-//d;

		my $pa = $PropertyAlias{$enum} ? $enum : $PA_reverse{$enum};
		my $f = $PropValueAlias{$pa}{$val} ? $val : $PVA_reverse{$pa}{lc $val};

		if ($pa and $f) {
		    $pa = "gc_sc" if $pa eq "gc" or $pa eq "sc";
		    $file = "unicore/lib/$pa/$PVA_abbr_map{$pa}{lc $f}.pl";
		    last GETFILE;
		}
	    }
	    else {
		my $t = lc $type;
		$t =~ tr/ _-//d;

		if ($PropValueAlias{gc}{$t} or $PropValueAlias{sc}{$t}) {
		    $file = "unicore/lib/gc_sc/$PVA_abbr_map{gc_sc}{$t}.pl";
		    last GETFILE;
		}
	    }

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

	    ##
	    ## See if it's a user-level "To".
	    ##

	    my $caller0 = caller(0);

	    if (defined $caller0 && $type =~ /^To(?:\w+)$/) {
		my $map = $caller0 . "::" . $type;

		if (exists &{$map}) {
		    no strict 'refs';
		    
		    $list = &{$map};
		    last GETFILE;
		}
	    }

            ##
            ## Last attempt -- see if it's a standard "To" name
	    ## (e.g. "ToLower")  ToTitle is used by ucfirst().
	    ## The user-level way to access ToDigit() and ToFold()
	    ## is to use Unicode::UCD.
            ##
            if ($type =~ /^To(Digit|Fold|Lower|Title|Upper)$/)
            {
                $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 and mappings), 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 = 0;

    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 = defined $2 ? hex $2 : $min;
	    my $val = defined $3 ? hex $3 : 0;
	    $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') {
		    $subobj = utf8->SWASHNEW($t, "", 0, 0, 0);
		}
		elsif (exists &$name) {
		    $subobj = utf8->SWASHNEW($name, "", 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;
}

# Now SWASHGET is recasted into a C function S_swash_get (see utf8.c).

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
12ac2576	11	our (%PropertyAlias, %PA_reverse, %PropValueAlias, %PVA_reverse, %PVA_abbr_map);
12ac2576	12
86916d66	13	sub croak { require Carp; Carp::croak(@_) }
86916d66	14
5beb625e	15	##
3a2263fe	16	## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape.
3a2263fe	17	## It's a data structure that encodes a set of Unicode characters.
5beb625e	18	##
5beb625e	19
a0ed51b3	20	sub SWASHNEW {
	21	my ($class, $type, $list, $minbits, $none) = @_;
	22	local $^D = 0 if $^D;
15732964	23
	24	print STDERR "SWASHNEW @_\n" if DEBUG;
	25
cf25bb62	26	##
	27	## Get the list of codepoints for the type.
	28	## Called from utf8.c
	29	##
	30	## Given a $type, our goal is to fill $list with the set of codepoint
5beb625e	31	## ranges.
cf25bb62	32	##
cf25bb62	33	## To make the parsing of $type clear, this code takes the a rather
2d92f8a0	34	## unorthodox approach of last'ing out of the block once we have the
cf25bb62	35	## info we need. Were this to be a subroutine, the 'last' would just
	36	## be a 'return'.
	37	##
5beb625e	38	my $file; ## file to load data from, and also part of the %Cache key.
	39	my $ListSorted = 0;
	40
cf25bb62	41	if ($type)
	42	{
	43	$type =~ s/^\s+//;
	44	$type =~ s/\s+$//;
	45
	46	print "type = $type\n" if DEBUG;
	47
cf25bb62	48	GETFILE:
cf25bb62	49	{
12ac2576	50	##
	51	## It could be a user-defined property.
	52	##
	53
	54	my $caller1 = $type =~ s/(.+)::// ? $1 : caller(1);
	55
	56	if (defined $caller1 && $type =~ /^(?:\w+)$/) {
	57	my $prop = "${caller1}::$type";
	58	if (exists &{$prop}) {
	59	no strict 'refs';
	60
	61	$list = &{$prop};
	62	last GETFILE;
	63	}
	64	}
	65
2d92f8a0	66	my $wasIs;
	67
	68	($wasIs = $type =~ s/^Is(?:\s+\|[-_])?//i)
5beb625e	69	or
12ac2576	70	$type =~ s/^(?:(?:General(?:\s+\|_)?)?Category\|gc)\s[:=]\s//i
5beb625e	71	or
12ac2576	72	$type =~ s/^(?:Script\|sc)\s[:=]\s//i
5beb625e	73	or
12ac2576	74	$type =~ s/^Block\s[:=]\s/In/i;
	75
	76
	77	##
	78	## See if it's in some enumeration.
	79	##
7ebf06b3	80	require "unicore/PVA.pl";
12ac2576	81	if ($type =~ /^([\w\s]+)[:=]\s(.)/) {
12ac2576	82	my ($enum, $val) = (lc $1, lc $2);
	83	$enum =~ tr/ _-//d;
	84	$val =~ tr/ _-//d;
	85
	86	my $pa = $PropertyAlias{$enum} ? $enum : $PA_reverse{$enum};
	87	my $f = $PropValueAlias{$pa}{$val} ? $val : $PVA_reverse{$pa}{lc $val};
	88
	89	if ($pa and $f) {
	90	$pa = "gc_sc" if $pa eq "gc" or $pa eq "sc";
	91	$file = "unicore/lib/$pa/$PVA_abbr_map{$pa}{lc $f}.pl";
	92	last GETFILE;
	93	}
	94	}
	95	else {
	96	my $t = lc $type;
	97	$t =~ tr/ _-//d;
	98
	99	if ($PropValueAlias{gc}{$t} or $PropValueAlias{sc}{$t}) {
	100	$file = "unicore/lib/gc_sc/$PVA_abbr_map{gc_sc}{$t}.pl";
	101	last GETFILE;
	102	}
	103	}
cf25bb62	104
5beb625e	105	##
	106	## See if it's in the direct mapping table.
	107	##
	108	require "unicore/Exact.pl";
	109	if (my $base = $utf8::Exact{$type}) {
12ac2576	110	$file = "unicore/lib/gc_sc/$base.pl";
5beb625e	111	last GETFILE;
cf25bb62	112	}
	113
	114	##
5beb625e	115	## If not there exactly, try the canonical form. The canonical
5beb625e	116	## form is lowercased, with any separators (\s+\|[-_]) removed.
cf25bb62	117	##
5beb625e	118	my $canonical = lc $type;
	119	$canonical =~ s/(?<=[a-z\d])(?:\s+\|[-_])(?=[a-z\d])//g;
	120	print "canonical = $canonical\n" if DEBUG;
cf25bb62	121
5beb625e	122	require "unicore/Canonical.pl";
12ac2576	123	if (my $base = ($utf8::Canonical{$canonical} \|\| $utf8::Canonical{ lc $utf8::PropertyAlias{$canonical} })) {
12ac2576	124	$file = "unicore/lib/gc_sc/$base.pl";
5beb625e	125	last GETFILE;
cf25bb62	126	}
cf25bb62	127
491fd90a	128	##
3a2263fe	129	## See if it's a user-level "To".
	130	##
	131
	132	my $caller0 = caller(0);
	133
	134	if (defined $caller0 && $type =~ /^To(?:\w+)$/) {
	135	my $map = $caller0 . "::" . $type;
09e0265a	136
3a2263fe	137	if (exists &{$map}) {
	138	no strict 'refs';
	139
	140	$list = &{$map};
	141	last GETFILE;
	142	}
	143	}
	144
cf25bb62	145	##
3a2263fe	146	## Last attempt -- see if it's a standard "To" name
	147	## (e.g. "ToLower") ToTitle is used by ucfirst().
	148	## The user-level way to access ToDigit() and ToFold()
	149	## is to use Unicode::UCD.
cf25bb62	150	##
3a2263fe	151	if ($type =~ /^To(Digit\|Fold\|Lower\|Title\|Upper)$/)
cf25bb62	152	{
	153	$file = "unicore/To/$1.pl";
	154	## would like to test to see if $file actually exists....
	155	last GETFILE;
	156	}
	157
	158	##
	159	## If we reach this line, it's because we couldn't figure
	160	## out what to do with $type. Ouch.
	161	##
bc45ce41	162
bc45ce41	163	return $type;
cf25bb62	164	}
cf25bb62	165
491fd90a	166	if (defined $file) {
	167	print "found it (file='$file')\n" if DEBUG;
	168
	169	##
	170	## If we reach here, it was due to a 'last GETFILE' above
3a2263fe	171	## (exception: user-defined properties and mappings), so we
491fd90a	172	## have a filename, so now we load it if we haven't already.
	173	## If we have, return the cached results. The cache key is the
	174	## file to load.
	175	##
	176	if ($Cache{$file} and ref($Cache{$file}) eq $class)
	177	{
	178	print "Returning cached '$file' for \\p{$type}\n" if DEBUG;
	179	return $Cache{$class, $file};
	180	}
	181
	182	$list = do $file;
	183	}
5beb625e	184
5beb625e	185	$ListSorted = 1; ## we know that these lists are sorted
886ba366	186	}
a0ed51b3	187
15732964	188	my $extras;
485aafd9	189	my $bits = 0;
cf25bb62	190
5beb625e	191	my $ORIG = $list;
a0ed51b3	192	if ($list) {
	193	my @tmp = split(/^/m, $list);
	194	my %seen;
db376a24	195	no warnings;
a0ed51b3	196	$extras = join '', grep /^[^0-9a-fA-F]/, @tmp;
a0ed51b3	197	$list = join '',
df68b396	198	map { $_->[1] }
	199	sort { $a->[0] <=> $b->[0] }
	200	map { /^([0-9a-fA-F]+)/; [ hex($1), $_ ] }
	201	grep { /^([0-9a-fA-F]+)/ and not $seen{$1}++ } @tmp; # XXX doesn't do ranges right
a0ed51b3	202	}
	203
	204	if ($none) {
	205	my $hextra = sprintf "%04x", $none + 1;
	206	$list =~ s/\tXXXX$/\t$hextra/mg;
	207	}
	208
	209	if ($minbits < 32) {
	210	my $top = 0;
3a2263fe	211	while ($list =~ /^([0-9a-fA-F]+)(?:[\t]([0-9a-fA-F]+)?)(?:[ \t]([0-9a-fA-F]+))?/mg) {
a0ed51b3	212	my $min = hex $1;
3a2263fe	213	my $max = defined $2 ? hex $2 : $min;
3a2263fe	214	my $val = defined $3 ? hex $3 : 0;
a0ed51b3	215	$val += $max - $min if defined $3;
	216	$top = $val if $val > $top;
	217	}
	218	$bits =
	219	$top > 0xffff ? 32 :
	220	$top > 0xff ? 16 :
	221	$top > 1 ? 8 : 1
	222	}
	223	$bits = $minbits if $bits < $minbits;
	224
	225	my @extras;
	226	for my $x ($extras) {
	227	pos $x = 0;
1fef36c7	228	while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
a0ed51b3	229	my $char = $1;
a0ed51b3	230	my $name = $2;
6d6ae962	231	print STDERR "$1 => $2\n" if DEBUG;
09e0265a	232	if ($char =~ /[-+!&]/) {
a0ed51b3	233	my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really
11ef8fdd	234	my $subobj;
11ef8fdd	235	if ($c eq 'utf8') {
09e0265a	236	$subobj = utf8->SWASHNEW($t, "", 0, 0, 0);
	237	}
	238	elsif (exists &$name) {
	239	$subobj = utf8->SWASHNEW($name, "", 0, 0, 0);
11ef8fdd	240	}
	241	elsif ($c =~ /^([0-9a-fA-F]+)/) {
	242	$subobj = utf8->SWASHNEW("", $c, 0, 0, 0);
	243	}
bc45ce41	244	return $subobj unless ref $subobj;
a0ed51b3	245	push @extras, $name => $subobj;
	246	$bits = $subobj->{BITS} if $bits < $subobj->{BITS};
	247	}
	248	}
	249	}
	250
15732964	251	print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG;
a0ed51b3	252
5beb625e	253	my $SWASH = bless {
a0ed51b3	254	TYPE => $type,
	255	BITS => $bits,
	256	EXTRAS => $extras,
	257	LIST => $list,
	258	NONE => $none,
	259	@extras,
	260	} => $class;
5beb625e	261
	262	if ($file) {
	263	$Cache{$class, $file} = $SWASH;
	264	}
	265
	266	return $SWASH;
a0ed51b3	267	}
a0ed51b3	268
979f2922	269	# Now SWASHGET is recasted into a C function S_swash_get (see utf8.c).
a0ed51b3	270
a0ed51b3	271	1;