This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
s/SvPV/SvPVbyte/g, as suggested by Gisle Aas.
[perl5.git] / lib / utf8_heavy.pl
CommitLineData
a0ed51b3 1package utf8;
cf25bb62
JH
2use strict;
3use warnings;
a0ed51b3 4
15732964 5sub DEBUG () { 0 }
a0ed51b3
LW
6
7sub DESTROY {}
8
9sub croak { require Carp; Carp::croak(@_) }
10
11sub SWASHNEW {
12 my ($class, $type, $list, $minbits, $none) = @_;
13 local $^D = 0 if $^D;
15732964
JH
14
15 print STDERR "SWASHNEW @_\n" if DEBUG;
16
cf25bb62
JH
17 ## check to see if we've already got it.
18 {
19 no strict 'refs';
20 if ($type and ref ${"${class}::{$type}"} eq $class) {
21 warn qq/Found \${"${class}::{$type}"}\n/ if DEBUG;
22 return ${"${class}::{$type}"};
23 }
a0ed51b3
LW
24 }
25
cf25bb62
JH
26 ##
27 ## Get the list of codepoints for the type.
28 ## Called from utf8.c
29 ##
30 ## Given a $type, our goal is to fill $list with the set of codepoint
31 ## ranges. As we try various interpretations of $type, sometimes we'll
32 ## end up with the $list directly, and sometimes we'll end up with a
33 ## $file name that holds the list data.
34 ##
35 ## To make the parsing of $type clear, this code takes the a rather
36 ## unorthadox approach of last'ing out of the block once we have the
37 ## info we need. Were this to be a subroutine, the 'last' would just
38 ## be a 'return'.
39 ##
40 if ($type)
41 {
42 $type =~ s/^\s+//;
43 $type =~ s/\s+$//;
44
45 print "type = $type\n" if DEBUG;
46
47 my $file;
48 ## Figure out what file to load to get the data....
49 GETFILE:
50 {
51 ##
52 ## First, see if it's an "Is" name (the 'Is' is optional)
53 ##
54 ## Because we check "Is" names first, they have precidence over
55 ## "In" names. For example, "Greek" is both a script and a
56 ## block. "IsGreek" always gets the script, while "InGreek"
57 ## always gets the block. "Greek" gets the script because we
58 ## check "Is" names first.
59 ##
60 if ($type =~ m{^
61 ## "Is" prefix, or "Script=" or "Category="
62 (?: Is [- _]? | (?:Script|Category)\s*=\s* )?
63 ## name to check in the "Is" symbol table.
64 ([A-Z].*)
65 $
66 }ix)
67 {
68 my $istype = $1;
69 ##
70 ## Input ($type) Name To Check ($istype)
71 ## ------------- -----------------------
72 ## IsLu Lu
73 ## Lu Lu
74 ## Category = Lu Lu
75 ## Foo Foo
76 ## Script = Greek Greek
77 ##
78
79 print "istype = $istype\n" if DEBUG;
80
81 ## Load "Is" mapping data, if not yet loaded.
82 do "unicore/Is.pl" if not defined %utf8::Is;
83
84 ##
85 ## If the "Is" mapping data has an exact match, it points
86 ## to the file we need.
87 ##
88 if (exists $utf8::Is{$istype})
89 {
90 $file = "unicore/Is/$utf8::Is{$istype}.pl";
91 last GETFILE;
92 }
93
94 ##
95 ## Need to look at %utf8::IsPat (loaded from "unicore/Is.pl")
96 ## to see if there's a regex that matches this $istype.
97 ## If so, the associated name is the file we need.
98 ##
99 my $prefix = substr(lc($istype), 0, 2);
03138300 100 if (my $hashref = $utf8::IsPat{$prefix})
cf25bb62 101 {
03138300 102 while (my ($pat, $name) = each %{$hashref})
cf25bb62
JH
103 {
104 print "isprefix = $prefix, Is = $istype, pat = $pat\n" if DEBUG;
105 ##
106 ## The following regex probably need not be cached,
107 ## since every time there's a match, the results of
108 ## the entire call to SWASHNEW() is cached, so there's
109 ## a very limited number of times any one $pat will
110 ## be evaluated as a regex, at least with "reasonable"
111 ## code that doesn't try a baziilion \p{Random} names.
112 ##
113 if ($istype =~ /^$pat$/i)
114 {
115 $file = "unicore/Is/$name.pl";
03138300 116 keys %{$hashref}; ## reset the 'each' above
cf25bb62
JH
117 last GETFILE;
118 }
119 }
120 }
121 }
122
123 ##
124 ## Couldn't find via "Is" -- let's try via "In".....
125 ##
126 if ($type =~ m{^
127 ( In(?!herited$)[- _]? | Block\s*=\s*)?
128 ([A-Z].*)
129 $
130 }xi)
131 {
132 my $intype = $2;
133 print "intype = $intype\n" if DEBUG;
134
135 ##
136 ## Input ($type) Name To Check ($intype)
137 ## ------------- -----------------------
138 ## Inherited Inherited
139 ## InGreek Greek
140 ## Block = Greek Greek
141 ##
142
143 ## Load "In" mapping data, if not yet loaded.
144 do "unicore/In.pl" if not defined %utf8::In;
145
146 ## If there's a direct match, it points to the file we need
147 if (exists $utf8::In{$intype}) {
148 $file = "unicore/In/$utf8::In{$intype}.pl";
149 last GETFILE;
150 }
151
03138300
JH
152 ##
153 ## Need to look at %utf8::InPat (loaded from "unicore/In.pl")
154 ## to see if there's a regex that matches this $intype.
155 ## If so, the associated name is the file we need.
156 ##
cf25bb62 157 my $prefix = substr(lc($intype), 0, 2);
03138300 158 if (my $hashref = $utf8::InPat{$prefix})
cf25bb62
JH
159 {
160 print "inprefix = $prefix, In = $intype\n" if DEBUG;
03138300 161 while (my ($pat, $name) = each %{$hashref})
cf25bb62
JH
162 {
163 print "inprefix = $prefix, In = $intype, k = $pat\n" if DEBUG;
164 if ($intype =~ /^$pat$/i) {
165 $file = "unicore/In/$name.pl";
166 print "inprefix = $prefix, In = $intype, k = $pat, file = $file\n" if DEBUG;
03138300 167 keys %{$hashref}; ## reset the 'each' above
cf25bb62
JH
168 last GETFILE;
169 }
170 }
171 }
172 }
173
174 ##
175 ## Last attempt -- see if it's a "To" name (e.g. "ToLower")
176 ##
177 if ($type =~ /^To([A-Z][A-Za-z]+)$/)
178 {
179 $file = "unicore/To/$1.pl";
180 ## would like to test to see if $file actually exists....
181 last GETFILE;
182 }
183
184 ##
185 ## If we reach this line, it's because we couldn't figure
186 ## out what to do with $type. Ouch.
187 ##
188 croak("Can't find Unicode character property \"$type\"");
189 }
190
191 ##
192 ## If we reach here, it was due to a 'last GETFILE' above, so we
193 ## have a filename, so now we load it.
194 ##
195 $list = do $file;
886ba366 196 }
a0ed51b3 197
15732964
JH
198 my $extras;
199 my $bits;
cf25bb62 200
a0ed51b3
LW
201 if ($list) {
202 my @tmp = split(/^/m, $list);
203 my %seen;
db376a24 204 no warnings;
a0ed51b3
LW
205 $extras = join '', grep /^[^0-9a-fA-F]/, @tmp;
206 $list = join '',
207 sort { hex $a <=> hex $b }
208 grep {/^([0-9a-fA-F]+)/ and not $seen{$1}++} @tmp; # XXX doesn't do ranges right
209 }
210
211 if ($none) {
212 my $hextra = sprintf "%04x", $none + 1;
213 $list =~ s/\tXXXX$/\t$hextra/mg;
214 }
215
216 if ($minbits < 32) {
217 my $top = 0;
218 while ($list =~ /^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
219 my $min = hex $1;
220 my $max = hex(defined $2 ? $2 : $1);
221 my $val = hex(defined $3 ? $3 : "");
222 $val += $max - $min if defined $3;
223 $top = $val if $val > $top;
224 }
225 $bits =
226 $top > 0xffff ? 32 :
227 $top > 0xff ? 16 :
228 $top > 1 ? 8 : 1
229 }
230 $bits = $minbits if $bits < $minbits;
231
232 my @extras;
233 for my $x ($extras) {
234 pos $x = 0;
1fef36c7 235 while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
a0ed51b3
LW
236 my $char = $1;
237 my $name = $2;
6d6ae962 238 print STDERR "$1 => $2\n" if DEBUG;
a0ed51b3
LW
239 if ($char =~ /[-+!]/) {
240 my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really
241 my $subobj = $c->SWASHNEW($t, "", 0, 0, 0);
242 push @extras, $name => $subobj;
243 $bits = $subobj->{BITS} if $bits < $subobj->{BITS};
244 }
245 }
246 }
247
15732964 248 print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG;
a0ed51b3 249
cf25bb62 250 no strict 'refs';
a0ed51b3
LW
251 ${"${class}::{$type}"} = bless {
252 TYPE => $type,
253 BITS => $bits,
254 EXTRAS => $extras,
255 LIST => $list,
256 NONE => $none,
257 @extras,
258 } => $class;
259}
260
261# NOTE: utf8.c:swash_init() assumes entries are never modified once generated.
262
263sub SWASHGET {
035d37be 264 # See utf8.c:Perl_swash_fetch for problems with this interface.
a0ed51b3
LW
265 my ($self, $start, $len) = @_;
266 local $^D = 0 if $^D;
a0ed51b3
LW
267 my $type = $self->{TYPE};
268 my $bits = $self->{BITS};
269 my $none = $self->{NONE};
15732964 270 print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if DEBUG;
a0ed51b3
LW
271 my $end = $start + $len;
272 my $swatch = "";
273 my $key;
274 vec($swatch, $len - 1, $bits) = 0; # Extend to correct length.
275 if ($none) {
276 for $key (0 .. $len - 1) { vec($swatch, $key, $bits) = $none }
277 }
278
279 for ($self->{LIST}) {
280 pos $_ = 0;
281 if ($bits > 1) {
282 LINE:
283 while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) {
284 my $min = hex $1;
285 my $max = (defined $2 ? hex $2 : $min);
286 my $val = hex $3;
287 next if $max < $start;
6d6ae962 288 print "$min $max $val\n" if DEBUG;
a0ed51b3
LW
289 if ($none) {
290 if ($min < $start) {
710250cb 291 $val += $start - $min if $val < $none;
a0ed51b3
LW
292 $min = $start;
293 }
294 for ($key = $min; $key <= $max; $key++) {
295 last LINE if $key >= $end;
6d6ae962 296 print STDERR "$key => $val\n" if DEBUG;
a0ed51b3
LW
297 vec($swatch, $key - $start, $bits) = $val;
298 ++$val if $val < $none;
299 }
300 }
301 else {
302 if ($min < $start) {
303 $val += $start - $min;
304 $min = $start;
305 }
306 for ($key = $min; $key <= $max; $key++, $val++) {
307 last LINE if $key >= $end;
6d6ae962 308 print STDERR "$key => $val\n" if DEBUG;
a0ed51b3
LW
309 vec($swatch, $key - $start, $bits) = $val;
310 }
311 }
312 }
313 }
314 else {
315 LINE:
316 while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+))?/mg) {
317 my $min = hex $1;
318 my $max = (defined $2 ? hex $2 : $min);
319 next if $max < $start;
320 if ($min < $start) {
321 $min = $start;
322 }
323 for ($key = $min; $key <= $max; $key++) {
324 last LINE if $key >= $end;
6d6ae962 325 print STDERR "$key => 1\n" if DEBUG;
a0ed51b3
LW
326 vec($swatch, $key - $start, 1) = 1;
327 }
328 }
329 }
330 }
331 for my $x ($self->{EXTRAS}) {
332 pos $x = 0;
1fef36c7 333 while ($x =~ /^([-+!])(.*)/mg) {
a0ed51b3
LW
334 my $char = $1;
335 my $name = $2;
15732964 336 print STDERR "INDIRECT $1 $2\n" if DEBUG;
1fef36c7
GA
337 my $otherbits = $self->{$name}->{BITS};
338 croak("SWASHGET size mismatch") if $bits < $otherbits;
339 my $other = $self->{$name}->SWASHGET($start, $len);
340 if ($char eq '+') {
341 if ($bits == 1 and $otherbits == 1) {
342 $swatch |= $other;
a0ed51b3 343 }
1fef36c7
GA
344 else {
345 for ($key = 0; $key < $len; $key++) {
346 vec($swatch, $key, $bits) = vec($other, $key, $otherbits);
a0ed51b3 347 }
1fef36c7
GA
348 }
349 }
350 elsif ($char eq '!') {
351 if ($bits == 1 and $otherbits == 1) {
352 $swatch |= ~$other;
353 }
354 else {
355 for ($key = 0; $key < $len; $key++) {
356 if (!vec($other, $key, $otherbits)) {
357 vec($swatch, $key, $bits) = 1;
a0ed51b3
LW
358 }
359 }
360 }
1fef36c7
GA
361 }
362 elsif ($char eq '-') {
363 if ($bits == 1 and $otherbits == 1) {
364 $swatch &= ~$other;
365 }
366 else {
367 for ($key = 0; $key < $len; $key++) {
368 if (vec($other, $key, $otherbits)) {
369 vec($swatch, $key, $bits) = 0;
a0ed51b3
LW
370 }
371 }
372 }
373 }
374 }
375 }
15732964 376 if (DEBUG) {
a0ed51b3
LW
377 print STDERR "CELLS ";
378 for ($key = 0; $key < $len; $key++) {
379 print STDERR vec($swatch, $key, $bits), " ";
380 }
381 print STDERR "\n";
382 }
383 $swatch;
384}
385
3861;