9 sub croak { require Carp; Carp::croak(@_) }
14 ## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape
18 my ($class, $type, $list, $minbits, $none) = @_;
21 print STDERR "SWASHNEW @_\n" if DEBUG;
24 ## Get the list of codepoints for the type.
27 ## Given a $type, our goal is to fill $list with the set of codepoint
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
35 my $file; ## file to load data from, and also part of the %Cache key.
43 print "type = $type\n" if DEBUG;
48 ## 'Is' is always optional, so if it's there, remove it.
49 ## Same with 'Category=' and 'Script='.
51 ## 'Block=' is replaced by 'In'.
53 $type =~ s/^Is(?:\s+|[-_])?//i
55 $type =~ s/^Category\s*=\s*//i
57 $type =~ s/^Script\s*=\s*//i
59 $type =~ s/^Block\s*=\s*/In/i;
62 ## See if it's in the direct mapping table.
64 require "unicore/Exact.pl";
65 if (my $base = $utf8::Exact{$type}) {
66 $file = "unicore/lib/$base.pl";
71 ## If not there exactly, try the canonical form. The canonical
72 ## form is lowercased, with any separators (\s+|[-_]) removed.
74 my $canonical = lc $type;
75 $canonical =~ s/(?<=[a-z\d])(?:\s+|[-_])(?=[a-z\d])//g;
76 print "canonical = $canonical\n" if DEBUG;
78 require "unicore/Canonical.pl";
79 if (my $base = $utf8::Canonical{$canonical}) {
80 $file = "unicore/lib/$base.pl";
85 ## It could be a user-defined property.
88 if ($type =~ /^I[ns](\w+)$/) {
89 my @caller = caller(1);
91 if (defined $caller[0]) {
92 my $prop = $caller[0] . "::" . $type;
94 if (exists &{$prop}) {
104 ## Last attempt -- see if it's a "To" name (e.g. "ToLower")
106 if ($type =~ /^To([A-Z][A-Za-z]+)$/)
108 $file = "unicore/To/$1.pl";
109 ## would like to test to see if $file actually exists....
114 ## If we reach this line, it's because we couldn't figure
115 ## out what to do with $type. Ouch.
122 print "found it (file='$file')\n" if DEBUG;
125 ## If we reach here, it was due to a 'last GETFILE' above
126 ## (exception: user-defined properties), so we
127 ## have a filename, so now we load it if we haven't already.
128 ## If we have, return the cached results. The cache key is the
131 if ($Cache{$file} and ref($Cache{$file}) eq $class)
133 print "Returning cached '$file' for \\p{$type}\n" if DEBUG;
134 return $Cache{$class, $file};
140 $ListSorted = 1; ## we know that these lists are sorted
148 my @tmp = split(/^/m, $list);
151 $extras = join '', grep /^[^0-9a-fA-F]/, @tmp;
153 sort { hex $a <=> hex $b }
154 grep {/^([0-9a-fA-F]+)/ and not $seen{$1}++} @tmp; # XXX doesn't do ranges right
158 my $hextra = sprintf "%04x", $none + 1;
159 $list =~ s/\tXXXX$/\t$hextra/mg;
164 while ($list =~ /^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
166 my $max = hex(defined $2 ? $2 : $1);
167 my $val = hex(defined $3 ? $3 : "");
168 $val += $max - $min if defined $3;
169 $top = $val if $val > $top;
176 $bits = $minbits if $bits < $minbits;
179 for my $x ($extras) {
181 while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
184 print STDERR "$1 => $2\n" if DEBUG;
185 if ($char =~ /[-+!]/) {
186 my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really
189 $subobj = $c->SWASHNEW($t, "", 0, 0, 0);
191 elsif ($c =~ /^([0-9a-fA-F]+)/) {
192 $subobj = utf8->SWASHNEW("", $c, 0, 0, 0);
194 return $subobj unless ref $subobj;
195 push @extras, $name => $subobj;
196 $bits = $subobj->{BITS} if $bits < $subobj->{BITS};
201 print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG;
213 $Cache{$class, $file} = $SWASH;
219 # NOTE: utf8.c:swash_init() assumes entries are never modified once generated.
222 # See utf8.c:Perl_swash_fetch for problems with this interface.
223 my ($self, $start, $len) = @_;
224 local $^D = 0 if $^D;
225 my $type = $self->{TYPE};
226 my $bits = $self->{BITS};
227 my $none = $self->{NONE};
228 print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if DEBUG;
229 my $end = $start + $len;
232 vec($swatch, $len - 1, $bits) = 0; # Extend to correct length.
234 for $key (0 .. $len - 1) { vec($swatch, $key, $bits) = $none }
237 for ($self->{LIST}) {
241 while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
243 my $max = (defined $2 ? hex $2 : $min);
245 next if $max < $start;
246 print "$min $max $val\n" if DEBUG;
249 $val += $start - $min if $val < $none;
252 for ($key = $min; $key <= $max; $key++) {
253 last LINE if $key >= $end;
254 print STDERR "$key => $val\n" if DEBUG;
255 vec($swatch, $key - $start, $bits) = $val;
256 ++$val if $val < $none;
261 $val += $start - $min;
264 for ($key = $min; $key <= $max; $key++, $val++) {
265 last LINE if $key >= $end;
266 print STDERR "$key => $val\n" if DEBUG;
267 vec($swatch, $key - $start, $bits) = $val;
274 while (/^([0-9a-fA-F]+)(?:[ \t]+([0-9a-fA-F]+))?/mg) {
276 my $max = (defined $2 ? hex $2 : $min);
277 next if $max < $start;
281 for ($key = $min; $key <= $max; $key++) {
282 last LINE if $key >= $end;
283 print STDERR "$key => 1\n" if DEBUG;
284 vec($swatch, $key - $start, 1) = 1;
289 for my $x ($self->{EXTRAS}) {
291 while ($x =~ /^([-+!])(.*)/mg) {
294 print STDERR "INDIRECT $1 $2\n" if DEBUG;
295 my $otherbits = $self->{$name}->{BITS};
296 croak("SWASHGET size mismatch") if $bits < $otherbits;
297 my $other = $self->{$name}->SWASHGET($start, $len);
299 if ($bits == 1 and $otherbits == 1) {
303 for ($key = 0; $key < $len; $key++) {
304 vec($swatch, $key, $bits) = vec($other, $key, $otherbits);
308 elsif ($char eq '!') {
309 if ($bits == 1 and $otherbits == 1) {
313 for ($key = 0; $key < $len; $key++) {
314 if (!vec($other, $key, $otherbits)) {
315 vec($swatch, $key, $bits) = 1;
320 elsif ($char eq '-') {
321 if ($bits == 1 and $otherbits == 1) {
325 for ($key = 0; $key < $len; $key++) {
326 if (vec($other, $key, $otherbits)) {
327 vec($swatch, $key, $bits) = 0;
335 print STDERR "CELLS ";
336 for ($key = 0; $key < $len; $key++) {
337 print STDERR vec($swatch, $key, $bits), " ";