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