Unicode categories continue:
[p5sagit/p5-mst-13.2.git] / lib / utf8_heavy.pl
CommitLineData
a0ed51b3 1package utf8;
2
15732964 3sub DEBUG () { 0 }
a0ed51b3 4
5sub DESTROY {}
6
7sub croak { require Carp; Carp::croak(@_) }
8
9sub SWASHNEW {
10 my ($class, $type, $list, $minbits, $none) = @_;
11 local $^D = 0 if $^D;
15732964 12
13 print STDERR "SWASHNEW @_\n" if DEBUG;
14
a0ed51b3 15 if ($type and ref ${"${class}::{$type}"} eq $class) {
15732964 16 warn qq/Found \${"${class}::{$type}"}\n/ if DEBUG;
a0ed51b3 17 return ${"${class}::{$type}"}; # Already there...
18 }
19
15732964 20 if ($type) {
d73e5302 21 $type =~ s/^\s+//;
22 $type =~ s/\s+$//;
15732964 23
d73e5302 24 print "type = $type\n" if DEBUG;
15732964 25
d73e5302 26 my $file;
15732964 27
d73e5302 28 unless (defined $file) {
29 defined %utf8::Is || do "unicore/Is.pl";
71d929cb 30 if ($type =~ /^(?:Is|Category\s*=\s*)?[- _]?([A-Z].*)$/i) {
d73e5302 31 my $istype = $1;
32 print "istype = $istype\n" if DEBUG;
33 unless ($list = do "unicore/Is/$istype.pl") {
34 if (exists $utf8::Is{$istype}) {
35 $file = "unicore/Is/$utf8::Is{$istype}";
36 } else {
37 my $isprefix = substr(lc($istype), 0, 2);
38 print "isprefix = $isprefix\n" if DEBUG;
39 if (exists $utf8::IsPat{$isprefix}) {
40 my $Is = $istype;
41 print "isprefix = $isprefix, Is = $Is\n" if DEBUG;
42 for my $k (keys %{$utf8::IsPat{$isprefix}}) {
43 print "isprefix = $isprefix, Is = $Is, k = $k\n" if DEBUG;
44 if ($Is =~ /^$k$/i) {
45 $file = "unicore/Is/$utf8::IsPat{$isprefix}->{$k}";
46 print "isprefix = $isprefix, Is = $Is, k = $k, file = $file\n" if DEBUG;
47 last;
48 }
49 }
50 }
51 }
52 }
53 }
54
55 unless (defined $file) {
56 defined %utf8::In || do "unicore/In.pl";
57 $type = 'Lampersand' if $type =~ /^(?:Is)?L&$/;
71d929cb 58 if ($type =~ /^(In|(?:Script|Block)\s*=\s*)?[- _]?(?!herited$)(.+)/i) {
59 my $incat = $1;
60 my $intype = $2;
61 print "incat = $incat, intype = $intype\n" if DEBUG;
62 if (exists $utf8::In{$intype}) {
d73e5302 63 $file = "unicore/In/$utf8::In{$intype}";
64 } else {
65 my $inprefix = substr(lc($intype), 0, 2);
66 print "inprefix = $inprefix\n" if DEBUG;
67 if (exists $utf8::InPat{$inprefix}) {
68 my $In = $intype;
69 print "inprefix = $inprefix, In = $In\n" if DEBUG;
70 for my $k (keys %{$utf8::InPat{$inprefix}}) {
71 print "inprefix = $inprefix, In = $In, k = $k\n" if DEBUG;
72 if ($In =~ /^$k$/i) {
71d929cb 73 my $i = $utf8::InPat{$inprefix}->{$k};
74 print "inprefix = $inprefix, In = $In, k = $k, i = $i\n" if DEBUG;
75 next if $incat =~ /^S/ &&
76 !exists $utf8::InScript{$i};
77 next if $incat =~ /^B/ &&
78 !exists $utf8::InBlock{$i};
79 $file = "unicore/In/$i";
d73e5302 80 print "inprefix = $inprefix, In = $In, k = $k, file = $file\n" if DEBUG;
81 last;
82 }
83 }
84 }
d9efae67 85 }
86 }
87 }
15732964 88
d73e5302 89 unless (defined $file) {
90 if ($type =~ /^To([A-Z][A-Za-z]+)$/) {
91 $file = "unicore/To/$1";
92 }
dee4fd60 93 }
15732964 94 }
886ba366 95
d73e5302 96 if (defined $file) {
97 $list = do "$file.pl";
98 }
99
100 croak("Can't find Unicode character property \"$type\"")
101 unless $list;
886ba366 102 }
a0ed51b3 103
15732964 104 my $extras;
105 my $bits;
106
a0ed51b3 107 if ($list) {
108 my @tmp = split(/^/m, $list);
109 my %seen;
db376a24 110 no warnings;
a0ed51b3 111 $extras = join '', grep /^[^0-9a-fA-F]/, @tmp;
112 $list = join '',
113 sort { hex $a <=> hex $b }
114 grep {/^([0-9a-fA-F]+)/ and not $seen{$1}++} @tmp; # XXX doesn't do ranges right
115 }
116
117 if ($none) {
118 my $hextra = sprintf "%04x", $none + 1;
119 $list =~ s/\tXXXX$/\t$hextra/mg;
120 }
121
122 if ($minbits < 32) {
123 my $top = 0;
124 while ($list =~ /^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
125 my $min = hex $1;
126 my $max = hex(defined $2 ? $2 : $1);
127 my $val = hex(defined $3 ? $3 : "");
128 $val += $max - $min if defined $3;
129 $top = $val if $val > $top;
130 }
131 $bits =
132 $top > 0xffff ? 32 :
133 $top > 0xff ? 16 :
134 $top > 1 ? 8 : 1
135 }
136 $bits = $minbits if $bits < $minbits;
137
138 my @extras;
139 for my $x ($extras) {
140 pos $x = 0;
1fef36c7 141 while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
a0ed51b3 142 my $char = $1;
143 my $name = $2;
6d6ae962 144 print STDERR "$1 => $2\n" if DEBUG;
a0ed51b3 145 if ($char =~ /[-+!]/) {
146 my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really
147 my $subobj = $c->SWASHNEW($t, "", 0, 0, 0);
148 push @extras, $name => $subobj;
149 $bits = $subobj->{BITS} if $bits < $subobj->{BITS};
150 }
151 }
152 }
153
15732964 154 print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG;
a0ed51b3 155
156 ${"${class}::{$type}"} = bless {
157 TYPE => $type,
158 BITS => $bits,
159 EXTRAS => $extras,
160 LIST => $list,
161 NONE => $none,
162 @extras,
163 } => $class;
164}
165
166# NOTE: utf8.c:swash_init() assumes entries are never modified once generated.
167
168sub SWASHGET {
035d37be 169 # See utf8.c:Perl_swash_fetch for problems with this interface.
a0ed51b3 170 my ($self, $start, $len) = @_;
171 local $^D = 0 if $^D;
a0ed51b3 172 my $type = $self->{TYPE};
173 my $bits = $self->{BITS};
174 my $none = $self->{NONE};
15732964 175 print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if DEBUG;
a0ed51b3 176 my $end = $start + $len;
177 my $swatch = "";
178 my $key;
179 vec($swatch, $len - 1, $bits) = 0; # Extend to correct length.
180 if ($none) {
181 for $key (0 .. $len - 1) { vec($swatch, $key, $bits) = $none }
182 }
183
184 for ($self->{LIST}) {
185 pos $_ = 0;
186 if ($bits > 1) {
187 LINE:
188 while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
189 my $min = hex $1;
190 my $max = (defined $2 ? hex $2 : $min);
191 my $val = hex $3;
192 next if $max < $start;
6d6ae962 193 print "$min $max $val\n" if DEBUG;
a0ed51b3 194 if ($none) {
195 if ($min < $start) {
710250cb 196 $val += $start - $min if $val < $none;
a0ed51b3 197 $min = $start;
198 }
199 for ($key = $min; $key <= $max; $key++) {
200 last LINE if $key >= $end;
6d6ae962 201 print STDERR "$key => $val\n" if DEBUG;
a0ed51b3 202 vec($swatch, $key - $start, $bits) = $val;
203 ++$val if $val < $none;
204 }
205 }
206 else {
207 if ($min < $start) {
208 $val += $start - $min;
209 $min = $start;
210 }
211 for ($key = $min; $key <= $max; $key++, $val++) {
212 last LINE if $key >= $end;
6d6ae962 213 print STDERR "$key => $val\n" if DEBUG;
a0ed51b3 214 vec($swatch, $key - $start, $bits) = $val;
215 }
216 }
217 }
218 }
219 else {
220 LINE:
221 while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+))?/mg) {
222 my $min = hex $1;
223 my $max = (defined $2 ? hex $2 : $min);
224 next if $max < $start;
225 if ($min < $start) {
226 $min = $start;
227 }
228 for ($key = $min; $key <= $max; $key++) {
229 last LINE if $key >= $end;
6d6ae962 230 print STDERR "$key => 1\n" if DEBUG;
a0ed51b3 231 vec($swatch, $key - $start, 1) = 1;
232 }
233 }
234 }
235 }
236 for my $x ($self->{EXTRAS}) {
237 pos $x = 0;
1fef36c7 238 while ($x =~ /^([-+!])(.*)/mg) {
a0ed51b3 239 my $char = $1;
240 my $name = $2;
15732964 241 print STDERR "INDIRECT $1 $2\n" if DEBUG;
1fef36c7 242 my $otherbits = $self->{$name}->{BITS};
243 croak("SWASHGET size mismatch") if $bits < $otherbits;
244 my $other = $self->{$name}->SWASHGET($start, $len);
245 if ($char eq '+') {
246 if ($bits == 1 and $otherbits == 1) {
247 $swatch |= $other;
a0ed51b3 248 }
1fef36c7 249 else {
250 for ($key = 0; $key < $len; $key++) {
251 vec($swatch, $key, $bits) = vec($other, $key, $otherbits);
a0ed51b3 252 }
1fef36c7 253 }
254 }
255 elsif ($char eq '!') {
256 if ($bits == 1 and $otherbits == 1) {
257 $swatch |= ~$other;
258 }
259 else {
260 for ($key = 0; $key < $len; $key++) {
261 if (!vec($other, $key, $otherbits)) {
262 vec($swatch, $key, $bits) = 1;
a0ed51b3 263 }
264 }
265 }
1fef36c7 266 }
267 elsif ($char eq '-') {
268 if ($bits == 1 and $otherbits == 1) {
269 $swatch &= ~$other;
270 }
271 else {
272 for ($key = 0; $key < $len; $key++) {
273 if (vec($other, $key, $otherbits)) {
274 vec($swatch, $key, $bits) = 0;
a0ed51b3 275 }
276 }
277 }
278 }
279 }
280 }
15732964 281 if (DEBUG) {
a0ed51b3 282 print STDERR "CELLS ";
283 for ($key = 0; $key < $len; $key++) {
284 print STDERR vec($swatch, $key, $bits), " ";
285 }
286 print STDERR "\n";
287 }
288 $swatch;
289}
290
2911;