Commit | Line | Data |
a0ed51b3 |
1 | package utf8; |
2 | |
15732964 |
3 | sub DEBUG () { 0 } |
a0ed51b3 |
4 | |
5 | sub DESTROY {} |
6 | |
7 | sub croak { require Carp; Carp::croak(@_) } |
8 | |
9 | sub 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 | } |
fa689c23 |
54 | |
d73e5302 |
55 | unless (defined $file) { |
56 | defined %utf8::In || do "unicore/In.pl"; |
57 | $type = 'Lampersand' if $type =~ /^(?:Is)?L&$/; |
76ccdbe2 |
58 | $type = 'Assigned' if $type =~ /^(?:Is)?Assigned$/i; |
59 | $type = 'Unassigned' if $type =~ /^(?:Is)?Unassigned$/i; |
71d929cb |
60 | if ($type =~ /^(In|(?:Script|Block)\s*=\s*)?[- _]?(?!herited$)(.+)/i) { |
fa689c23 |
61 | my $incat = $1 || ''; |
71d929cb |
62 | my $intype = $2; |
63 | print "incat = $incat, intype = $intype\n" if DEBUG; |
64 | if (exists $utf8::In{$intype}) { |
d73e5302 |
65 | $file = "unicore/In/$utf8::In{$intype}"; |
66 | } else { |
67 | my $inprefix = substr(lc($intype), 0, 2); |
68 | print "inprefix = $inprefix\n" if DEBUG; |
69 | if (exists $utf8::InPat{$inprefix}) { |
70 | my $In = $intype; |
71 | print "inprefix = $inprefix, In = $In\n" if DEBUG; |
72 | for my $k (keys %{$utf8::InPat{$inprefix}}) { |
73 | print "inprefix = $inprefix, In = $In, k = $k\n" if DEBUG; |
74 | if ($In =~ /^$k$/i) { |
71d929cb |
75 | my $i = $utf8::InPat{$inprefix}->{$k}; |
76 | print "inprefix = $inprefix, In = $In, k = $k, i = $i\n" if DEBUG; |
77 | next if $incat =~ /^S/ && |
78 | !exists $utf8::InScript{$i}; |
79 | next if $incat =~ /^B/ && |
80 | !exists $utf8::InBlock{$i}; |
81 | $file = "unicore/In/$i"; |
d73e5302 |
82 | print "inprefix = $inprefix, In = $In, k = $k, file = $file\n" if DEBUG; |
83 | last; |
84 | } |
85 | } |
86 | } |
d9efae67 |
87 | } |
88 | } |
89 | } |
15732964 |
90 | |
d73e5302 |
91 | unless (defined $file) { |
92 | if ($type =~ /^To([A-Z][A-Za-z]+)$/) { |
93 | $file = "unicore/To/$1"; |
94 | } |
dee4fd60 |
95 | } |
15732964 |
96 | } |
886ba366 |
97 | |
d73e5302 |
98 | if (defined $file) { |
99 | $list = do "$file.pl"; |
100 | } |
101 | |
102 | croak("Can't find Unicode character property \"$type\"") |
103 | unless $list; |
886ba366 |
104 | } |
a0ed51b3 |
105 | |
15732964 |
106 | my $extras; |
107 | my $bits; |
108 | |
a0ed51b3 |
109 | if ($list) { |
110 | my @tmp = split(/^/m, $list); |
111 | my %seen; |
db376a24 |
112 | no warnings; |
a0ed51b3 |
113 | $extras = join '', grep /^[^0-9a-fA-F]/, @tmp; |
114 | $list = join '', |
115 | sort { hex $a <=> hex $b } |
116 | grep {/^([0-9a-fA-F]+)/ and not $seen{$1}++} @tmp; # XXX doesn't do ranges right |
117 | } |
118 | |
119 | if ($none) { |
120 | my $hextra = sprintf "%04x", $none + 1; |
121 | $list =~ s/\tXXXX$/\t$hextra/mg; |
122 | } |
123 | |
124 | if ($minbits < 32) { |
125 | my $top = 0; |
126 | while ($list =~ /^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) { |
127 | my $min = hex $1; |
128 | my $max = hex(defined $2 ? $2 : $1); |
129 | my $val = hex(defined $3 ? $3 : ""); |
130 | $val += $max - $min if defined $3; |
131 | $top = $val if $val > $top; |
132 | } |
133 | $bits = |
134 | $top > 0xffff ? 32 : |
135 | $top > 0xff ? 16 : |
136 | $top > 1 ? 8 : 1 |
137 | } |
138 | $bits = $minbits if $bits < $minbits; |
139 | |
140 | my @extras; |
141 | for my $x ($extras) { |
142 | pos $x = 0; |
1fef36c7 |
143 | while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) { |
a0ed51b3 |
144 | my $char = $1; |
145 | my $name = $2; |
6d6ae962 |
146 | print STDERR "$1 => $2\n" if DEBUG; |
a0ed51b3 |
147 | if ($char =~ /[-+!]/) { |
148 | my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really |
149 | my $subobj = $c->SWASHNEW($t, "", 0, 0, 0); |
150 | push @extras, $name => $subobj; |
151 | $bits = $subobj->{BITS} if $bits < $subobj->{BITS}; |
152 | } |
153 | } |
154 | } |
155 | |
15732964 |
156 | print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG; |
a0ed51b3 |
157 | |
158 | ${"${class}::{$type}"} = bless { |
159 | TYPE => $type, |
160 | BITS => $bits, |
161 | EXTRAS => $extras, |
162 | LIST => $list, |
163 | NONE => $none, |
164 | @extras, |
165 | } => $class; |
166 | } |
167 | |
168 | # NOTE: utf8.c:swash_init() assumes entries are never modified once generated. |
169 | |
170 | sub SWASHGET { |
035d37be |
171 | # See utf8.c:Perl_swash_fetch for problems with this interface. |
a0ed51b3 |
172 | my ($self, $start, $len) = @_; |
173 | local $^D = 0 if $^D; |
a0ed51b3 |
174 | my $type = $self->{TYPE}; |
175 | my $bits = $self->{BITS}; |
176 | my $none = $self->{NONE}; |
15732964 |
177 | print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if DEBUG; |
a0ed51b3 |
178 | my $end = $start + $len; |
179 | my $swatch = ""; |
180 | my $key; |
181 | vec($swatch, $len - 1, $bits) = 0; # Extend to correct length. |
182 | if ($none) { |
183 | for $key (0 .. $len - 1) { vec($swatch, $key, $bits) = $none } |
184 | } |
185 | |
186 | for ($self->{LIST}) { |
187 | pos $_ = 0; |
188 | if ($bits > 1) { |
189 | LINE: |
190 | while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) { |
191 | my $min = hex $1; |
192 | my $max = (defined $2 ? hex $2 : $min); |
193 | my $val = hex $3; |
194 | next if $max < $start; |
6d6ae962 |
195 | print "$min $max $val\n" if DEBUG; |
a0ed51b3 |
196 | if ($none) { |
197 | if ($min < $start) { |
710250cb |
198 | $val += $start - $min if $val < $none; |
a0ed51b3 |
199 | $min = $start; |
200 | } |
201 | for ($key = $min; $key <= $max; $key++) { |
202 | last LINE if $key >= $end; |
6d6ae962 |
203 | print STDERR "$key => $val\n" if DEBUG; |
a0ed51b3 |
204 | vec($swatch, $key - $start, $bits) = $val; |
205 | ++$val if $val < $none; |
206 | } |
207 | } |
208 | else { |
209 | if ($min < $start) { |
210 | $val += $start - $min; |
211 | $min = $start; |
212 | } |
213 | for ($key = $min; $key <= $max; $key++, $val++) { |
214 | last LINE if $key >= $end; |
6d6ae962 |
215 | print STDERR "$key => $val\n" if DEBUG; |
a0ed51b3 |
216 | vec($swatch, $key - $start, $bits) = $val; |
217 | } |
218 | } |
219 | } |
220 | } |
221 | else { |
222 | LINE: |
223 | while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+))?/mg) { |
224 | my $min = hex $1; |
225 | my $max = (defined $2 ? hex $2 : $min); |
226 | next if $max < $start; |
227 | if ($min < $start) { |
228 | $min = $start; |
229 | } |
230 | for ($key = $min; $key <= $max; $key++) { |
231 | last LINE if $key >= $end; |
6d6ae962 |
232 | print STDERR "$key => 1\n" if DEBUG; |
a0ed51b3 |
233 | vec($swatch, $key - $start, 1) = 1; |
234 | } |
235 | } |
236 | } |
237 | } |
238 | for my $x ($self->{EXTRAS}) { |
239 | pos $x = 0; |
1fef36c7 |
240 | while ($x =~ /^([-+!])(.*)/mg) { |
a0ed51b3 |
241 | my $char = $1; |
242 | my $name = $2; |
15732964 |
243 | print STDERR "INDIRECT $1 $2\n" if DEBUG; |
1fef36c7 |
244 | my $otherbits = $self->{$name}->{BITS}; |
245 | croak("SWASHGET size mismatch") if $bits < $otherbits; |
246 | my $other = $self->{$name}->SWASHGET($start, $len); |
247 | if ($char eq '+') { |
248 | if ($bits == 1 and $otherbits == 1) { |
249 | $swatch |= $other; |
a0ed51b3 |
250 | } |
1fef36c7 |
251 | else { |
252 | for ($key = 0; $key < $len; $key++) { |
253 | vec($swatch, $key, $bits) = vec($other, $key, $otherbits); |
a0ed51b3 |
254 | } |
1fef36c7 |
255 | } |
256 | } |
257 | elsif ($char eq '!') { |
258 | if ($bits == 1 and $otherbits == 1) { |
259 | $swatch |= ~$other; |
260 | } |
261 | else { |
262 | for ($key = 0; $key < $len; $key++) { |
263 | if (!vec($other, $key, $otherbits)) { |
264 | vec($swatch, $key, $bits) = 1; |
a0ed51b3 |
265 | } |
266 | } |
267 | } |
1fef36c7 |
268 | } |
269 | elsif ($char eq '-') { |
270 | if ($bits == 1 and $otherbits == 1) { |
271 | $swatch &= ~$other; |
272 | } |
273 | else { |
274 | for ($key = 0; $key < $len; $key++) { |
275 | if (vec($other, $key, $otherbits)) { |
276 | vec($swatch, $key, $bits) = 0; |
a0ed51b3 |
277 | } |
278 | } |
279 | } |
280 | } |
281 | } |
282 | } |
15732964 |
283 | if (DEBUG) { |
a0ed51b3 |
284 | print STDERR "CELLS "; |
285 | for ($key = 0; $key < $len; $key++) { |
286 | print STDERR vec($swatch, $key, $bits), " "; |
287 | } |
288 | print STDERR "\n"; |
289 | } |
290 | $swatch; |
291 | } |
292 | |
293 | 1; |