Rewrite mktables from scratch.
[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 $file) {
29 defined %utf8::Is || do "unicore/Is.pl";
30 if ($type =~ /^(?:Is)?[- _]?([A-Z].*)$/i) {
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 $type =~ /^(?:Is)?L&$/;
58 if ($type =~ /^(?:In)?[- _]?(?!herited$)(.+)/i) {
59 my $intype = $1;
60 print "intype = $intype\n" if DEBUG;
61 if (exists $utf8::Is{$istype}) {
62 $file = "unicore/In/$utf8::In{$intype}";
63 } else {
64 my $inprefix = substr(lc($intype), 0, 2);
65 print "inprefix = $inprefix\n" if DEBUG;
66 if (exists $utf8::InPat{$inprefix}) {
67 my $In = $intype;
68 print "inprefix = $inprefix, In = $In\n" if DEBUG;
69 for my $k (keys %{$utf8::InPat{$inprefix}}) {
70 print "inprefix = $inprefix, In = $In, k = $k\n" if DEBUG;
71 if ($In =~ /^$k$/i) {
72 $file = "unicore/In/$utf8::InPat{$inprefix}->{$k}";
73 print "inprefix = $inprefix, In = $In, k = $k, file = $file\n" if DEBUG;
74 last;
75 }
76 }
77 }
d9efae67 78 }
79 }
80 }
15732964 81
d73e5302 82 unless (defined $file) {
83 if ($type =~ /^To([A-Z][A-Za-z]+)$/) {
84 $file = "unicore/To/$1";
85 }
dee4fd60 86 }
15732964 87 }
886ba366 88
d73e5302 89 if (defined $file) {
90 $list = do "$file.pl";
91 }
92
93 croak("Can't find Unicode character property \"$type\"")
94 unless $list;
886ba366 95 }
a0ed51b3 96
15732964 97 my $extras;
98 my $bits;
99
a0ed51b3 100 if ($list) {
101 my @tmp = split(/^/m, $list);
102 my %seen;
db376a24 103 no warnings;
a0ed51b3 104 $extras = join '', grep /^[^0-9a-fA-F]/, @tmp;
105 $list = join '',
106 sort { hex $a <=> hex $b }
107 grep {/^([0-9a-fA-F]+)/ and not $seen{$1}++} @tmp; # XXX doesn't do ranges right
108 }
109
110 if ($none) {
111 my $hextra = sprintf "%04x", $none + 1;
112 $list =~ s/\tXXXX$/\t$hextra/mg;
113 }
114
115 if ($minbits < 32) {
116 my $top = 0;
117 while ($list =~ /^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
118 my $min = hex $1;
119 my $max = hex(defined $2 ? $2 : $1);
120 my $val = hex(defined $3 ? $3 : "");
121 $val += $max - $min if defined $3;
122 $top = $val if $val > $top;
123 }
124 $bits =
125 $top > 0xffff ? 32 :
126 $top > 0xff ? 16 :
127 $top > 1 ? 8 : 1
128 }
129 $bits = $minbits if $bits < $minbits;
130
131 my @extras;
132 for my $x ($extras) {
133 pos $x = 0;
1fef36c7 134 while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
a0ed51b3 135 my $char = $1;
136 my $name = $2;
6d6ae962 137 print STDERR "$1 => $2\n" if DEBUG;
a0ed51b3 138 if ($char =~ /[-+!]/) {
139 my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really
140 my $subobj = $c->SWASHNEW($t, "", 0, 0, 0);
141 push @extras, $name => $subobj;
142 $bits = $subobj->{BITS} if $bits < $subobj->{BITS};
143 }
144 }
145 }
146
15732964 147 print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG;
a0ed51b3 148
149 ${"${class}::{$type}"} = bless {
150 TYPE => $type,
151 BITS => $bits,
152 EXTRAS => $extras,
153 LIST => $list,
154 NONE => $none,
155 @extras,
156 } => $class;
157}
158
159# NOTE: utf8.c:swash_init() assumes entries are never modified once generated.
160
161sub SWASHGET {
162 my ($self, $start, $len) = @_;
163 local $^D = 0 if $^D;
a0ed51b3 164 my $type = $self->{TYPE};
165 my $bits = $self->{BITS};
166 my $none = $self->{NONE};
15732964 167 print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if DEBUG;
a0ed51b3 168 my $end = $start + $len;
169 my $swatch = "";
170 my $key;
171 vec($swatch, $len - 1, $bits) = 0; # Extend to correct length.
172 if ($none) {
173 for $key (0 .. $len - 1) { vec($swatch, $key, $bits) = $none }
174 }
175
176 for ($self->{LIST}) {
177 pos $_ = 0;
178 if ($bits > 1) {
179 LINE:
180 while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
181 my $min = hex $1;
182 my $max = (defined $2 ? hex $2 : $min);
183 my $val = hex $3;
184 next if $max < $start;
6d6ae962 185 print "$min $max $val\n" if DEBUG;
a0ed51b3 186 if ($none) {
187 if ($min < $start) {
710250cb 188 $val += $start - $min if $val < $none;
a0ed51b3 189 $min = $start;
190 }
191 for ($key = $min; $key <= $max; $key++) {
192 last LINE if $key >= $end;
6d6ae962 193 print STDERR "$key => $val\n" if DEBUG;
a0ed51b3 194 vec($swatch, $key - $start, $bits) = $val;
195 ++$val if $val < $none;
196 }
197 }
198 else {
199 if ($min < $start) {
200 $val += $start - $min;
201 $min = $start;
202 }
203 for ($key = $min; $key <= $max; $key++, $val++) {
204 last LINE if $key >= $end;
6d6ae962 205 print STDERR "$key => $val\n" if DEBUG;
a0ed51b3 206 vec($swatch, $key - $start, $bits) = $val;
207 }
208 }
209 }
210 }
211 else {
212 LINE:
213 while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+))?/mg) {
214 my $min = hex $1;
215 my $max = (defined $2 ? hex $2 : $min);
216 next if $max < $start;
217 if ($min < $start) {
218 $min = $start;
219 }
220 for ($key = $min; $key <= $max; $key++) {
221 last LINE if $key >= $end;
6d6ae962 222 print STDERR "$key => 1\n" if DEBUG;
a0ed51b3 223 vec($swatch, $key - $start, 1) = 1;
224 }
225 }
226 }
227 }
228 for my $x ($self->{EXTRAS}) {
229 pos $x = 0;
1fef36c7 230 while ($x =~ /^([-+!])(.*)/mg) {
a0ed51b3 231 my $char = $1;
232 my $name = $2;
15732964 233 print STDERR "INDIRECT $1 $2\n" if DEBUG;
1fef36c7 234 my $otherbits = $self->{$name}->{BITS};
235 croak("SWASHGET size mismatch") if $bits < $otherbits;
236 my $other = $self->{$name}->SWASHGET($start, $len);
237 if ($char eq '+') {
238 if ($bits == 1 and $otherbits == 1) {
239 $swatch |= $other;
a0ed51b3 240 }
1fef36c7 241 else {
242 for ($key = 0; $key < $len; $key++) {
243 vec($swatch, $key, $bits) = vec($other, $key, $otherbits);
a0ed51b3 244 }
1fef36c7 245 }
246 }
247 elsif ($char eq '!') {
248 if ($bits == 1 and $otherbits == 1) {
249 $swatch |= ~$other;
250 }
251 else {
252 for ($key = 0; $key < $len; $key++) {
253 if (!vec($other, $key, $otherbits)) {
254 vec($swatch, $key, $bits) = 1;
a0ed51b3 255 }
256 }
257 }
1fef36c7 258 }
259 elsif ($char eq '-') {
260 if ($bits == 1 and $otherbits == 1) {
261 $swatch &= ~$other;
262 }
263 else {
264 for ($key = 0; $key < $len; $key++) {
265 if (vec($other, $key, $otherbits)) {
266 vec($swatch, $key, $bits) = 0;
a0ed51b3 267 }
268 }
269 }
270 }
271 }
272 }
15732964 273 if (DEBUG) {
a0ed51b3 274 print STDERR "CELLS ";
275 for ($key = 0; $key < $len; $key++) {
276 print STDERR vec($swatch, $key, $bits), " ";
277 }
278 print STDERR "\n";
279 }
280 $swatch;
281}
282
2831;