This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Remove relics of regex swash use
[perl5.git] / lib / utf8_heavy.pl
CommitLineData
a0ed51b3 1package utf8;
cf25bb62
JH
2use strict;
3use warnings;
2783468b
KW
4use re "/aa"; # So we won't even try to look at above Latin1, potentially
5 # resulting in a recursive call
a0ed51b3 6
15732964 7sub DEBUG () { 0 }
4b17cb47 8$|=1 if DEBUG;
a0ed51b3
LW
9
10sub DESTROY {}
11
5beb625e
JH
12my %Cache;
13
86916d66
JH
14sub croak { require Carp; Carp::croak(@_) }
15
0426f635
KW
16# Digits may be separated by a single underscore
17my $digits = qr/ ( [0-9] _? )+ (?!:_) /x;
18
19# A sign can be surrounded by white space
20my $sign = qr/ \s* [+-]? \s* /x;
21
22my $f_float = qr/ $sign $digits+ \. $digits* # e.g., 5.0, 5.
23 | $sign $digits* \. $digits+/x; # 0.7, .7
24
25# A number may be an integer, a rational, or a float with an optional exponent
26# We (shudder) accept a signed denominator
27my $number = qr{ ^ $sign $digits+ $
28 | ^ $sign $digits+ \/ $sign $digits+ $
29 | ^ $f_float (?: [Ee] [+-]? $digits )? $}x;
30
c4ab2516 31sub _loose_name ($) {
45376db6
KW
32 # Given a lowercase property or property-value name, return its
33 # standardized version that is expected for look-up in the 'loose' hashes
34 # in Heavy.pl (hence, this depends on what mktables does). This squeezes
35 # out blanks, underscores and dashes. The complication stems from the
36 # grandfathered-in 'L_', which retains a single trailing underscore.
37
0fd86aa7 38 (my $loose = $_[0]) =~ s/[-_ \t]//g;
45376db6 39
cf2cd801 40 return $loose if $loose !~ / ^ (?: is | to )? l $/x;
45376db6
KW
41 return 'l_' if $_[0] =~ / l .* _ /x; # If original had a trailing '_'
42 return $loose;
43}
44
5beb625e 45##
3a2263fe
RGS
46## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape.
47## It's a data structure that encodes a set of Unicode characters.
5beb625e
JH
48##
49
99870f4d
KW
50{
51 # If a floating point number is within this distance from the value of a
52 # fraction, it is considered to be that fraction, even if many more digits
53 # are specified that don't exactly match.
54 my $min_floating_slop;
55
3dd10fe8
KW
56 # To guard against this program calling something that in turn ends up
57 # calling this program with the same inputs, and hence infinitely
58 # recursing, we keep a stack of the properties that are currently in
59 # progress, pushed upon entry, popped upon return.
60 my @recursed;
61
99870f4d
KW
62 sub SWASHNEW {
63 my ($class, $type, $list, $minbits, $none) = @_;
0fada5e1 64 my $user_defined = 0;
99870f4d
KW
65 local $^D = 0 if $^D;
66
3dd10fe8 67 $class = "" unless defined $class;
5bcf08c4
KW
68 print STDERR __LINE__, ": class=$class, type=$type, list=",
69 (defined $list) ? $list : ':undef:',
70 ", minbits=$minbits, none=$none\n" if DEBUG;
99870f4d
KW
71
72 ##
73 ## Get the list of codepoints for the type.
74 ## Called from swash_init (see utf8.c) or SWASHNEW itself.
75 ##
76 ## Callers of swash_init:
77 ## op.c:pmtrans -- for tr/// and y///
681d705c 78 ## Unicode::UCD::prop_invlist
62b3b855 79 ## Unicode::UCD::prop_invmap
99870f4d
KW
80 ##
81 ## Given a $type, our goal is to fill $list with the set of codepoint
82 ## ranges. If $type is false, $list passed is used.
83 ##
84 ## $minbits:
85 ## For binary properties, $minbits must be 1.
86 ## For character mappings (case and transliteration), $minbits must
87 ## be a number except 1.
88 ##
89 ## $list (or that filled according to $type):
90 ## Refer to perlunicode.pod, "User-Defined Character Properties."
91 ##
92 ## For binary properties, only characters with the property value
93 ## of True should be listed. The 3rd column, if any, will be ignored
94 ##
95 ## $none is undocumented, so I'm (khw) trying to do some documentation
96 ## of it now. It appears to be if there is a mapping in an input file
97 ## that maps to 'XXXX', then that is replaced by $none+1, expressed in
83fe8199 98 ## hexadecimal. It is used somehow in tr///.
99870f4d
KW
99 ##
100 ## To make the parsing of $type clear, this code takes the a rather
101 ## unorthodox approach of last'ing out of the block once we have the
102 ## info we need. Were this to be a subroutine, the 'last' would just
103 ## be a 'return'.
104 ##
fbe6f3cf
KW
105 # If a problem is found $type is returned;
106 # Upon success, a new (or cached) blessed object is returned with
107 # keys TYPE, BITS, EXTRAS, LIST, and NONE with values having the
108 # same meanings as the input parameters.
a6517ebf 109 # SPECIALS contains a reference to any special-treatment hash in the
13eda826 110 # property.
a6517ebf 111 # INVERT_IT is non-zero if the result should be inverted before use
0fada5e1 112 # USER_DEFINED is non-zero if the result came from a user-defined
99870f4d 113 my $file; ## file to load data from, and also part of the %Cache key.
99870f4d
KW
114
115 # Change this to get a different set of Unicode tables
116 my $unicore_dir = 'unicore';
a6517ebf 117 my $invert_it = 0;
7b953548
KW
118 my $list_is_from_mktables = 0; # Is $list returned from a mktables
119 # generated file? If so, we know it's
120 # well behaved.
99870f4d
KW
121
122 if ($type)
cf25bb62 123 {
3dd10fe8 124 # Verify that this isn't a recursive call for this property.
dbe7a391 125 # Can't use croak, as it may try to recurse to here itself.
3dd10fe8
KW
126 my $class_type = $class . "::$type";
127 if (grep { $_ eq $class_type } @recursed) {
128 CORE::die "panic: Infinite recursion in SWASHNEW for '$type'\n";
129 }
130 push @recursed, $class_type;
131
99870f4d
KW
132 $type =~ s/^\s+//;
133 $type =~ s/\s+$//;
134
4ff124d1
KW
135 # regcomp.c surrounds the property name with '__" and '_i' if this
136 # is to be caseless matching.
20b1a891 137 my $caseless = $type =~ s/^(.*)__(.*)_i$/$1$2/;
4ff124d1
KW
138
139 print STDERR __LINE__, ": type=$type, caseless=$caseless\n" if DEBUG;
99870f4d
KW
140
141 GETFILE:
142 {
143 ##
a44f4b74
KW
144 ## It could be a user-defined property. Look in current
145 ## package if no package given
99870f4d
KW
146 ##
147
20b1a891
KW
148
149 my $caller0 = caller(0);
345c18b8
KW
150 my $caller1 = $type =~ s/(.+):://
151 ? $1
152 : $caller0 eq 'main'
153 ? 'main'
154 : caller(1);
99870f4d 155
d658a8a8 156 if (defined $caller1 && $type =~ /^I[ns]\w+$/) {
99870f4d
KW
157 my $prop = "${caller1}::$type";
158 if (exists &{$prop}) {
0e9be77f
DM
159 # stolen from Scalar::Util::PP::tainted()
160 my $tainted;
161 {
162 local($@, $SIG{__DIE__}, $SIG{__WARN__});
163 local $^W = 0;
164 no warnings;
165 eval { kill 0 * $prop };
166 $tainted = 1 if $@ =~ /^Insecure/;
167 }
168 die "Insecure user-defined property \\p{$prop}\n"
169 if $tainted;
99870f4d 170 no strict 'refs';
4ff124d1 171 $list = &{$prop}($caseless);
0fada5e1 172 $user_defined = 1;
99870f4d
KW
173 last GETFILE;
174 }
175 }
176
a44f4b74
KW
177 # During Perl's compilation, this routine may be called before
178 # the tables are constructed. If so, we have a chicken/egg
179 # problem. If we die, the tables never get constructed, so
180 # keep going, but return an empty table so only what the code
181 # has compiled in internally (currently ASCII/Latin1 range
182 # matching) will work.
1363cbd0
FC
183 BEGIN {
184 # Poor man's constant, to avoid a run-time check.
185 $utf8::{miniperl}
186 = \! defined &DynaLoader::boot_DynaLoader;
187 }
188 if (miniperl) {
a44f4b74 189 eval "require '$unicore_dir/Heavy.pl'";
2db3e091
KW
190 if ($@) {
191 print STDERR __LINE__, ": '$@'\n" if DEBUG;
192 pop @recursed if @recursed;
193 return $type;
194 }
a44f4b74
KW
195 }
196 else {
197 require "$unicore_dir/Heavy.pl";
198 }
1363cbd0 199 BEGIN { delete $utf8::{miniperl} }
99870f4d 200
4ff124d1 201 # All property names are matched caselessly
49801ec2 202 my $property_and_table = CORE::lc $type;
99870f4d
KW
203 print STDERR __LINE__, ": $property_and_table\n" if DEBUG;
204
205 # See if is of the compound form 'property=value', where the
206 # value indicates the table we should use.
207 my ($property, $table, @remainder) =
208 split /\s*[:=]\s*/, $property_and_table, -1;
3dd10fe8
KW
209 if (@remainder) {
210 pop @recursed if @recursed;
211 return $type;
212 }
99870f4d
KW
213
214 my $prefix;
215 if (! defined $table) {
216
217 # Here, is the single form. The property becomes empty, and
218 # the whole value is the table.
219 $table = $property;
220 $prefix = $property = "";
221 } else {
222 print STDERR __LINE__, ": $property\n" if DEBUG;
223
224 # Here it is the compound property=table form. The property
807807b7
KW
225 # name is always loosely matched, and always can have an
226 # optional 'is' prefix (which isn't true in the single
227 # form).
c4ab2516 228 $property = _loose_name($property) =~ s/^is//r;
99870f4d
KW
229
230 # And convert to canonical form. Quit if not valid.
231 $property = $utf8::loose_property_name_of{$property};
3dd10fe8
KW
232 if (! defined $property) {
233 pop @recursed if @recursed;
234 return $type;
235 }
99870f4d
KW
236
237 $prefix = "$property=";
238
239 # If the rhs looks like it is a number...
240 print STDERR __LINE__, ": table=$table\n" if DEBUG;
99870f4d 241
0426f635
KW
242 if ($table =~ $number) {
243 print STDERR __LINE__, ": table=$table\n" if DEBUG;
99870f4d
KW
244
245 # Split on slash, in case it is a rational, like \p{1/5}
aa8f6cef 246 my @parts = split m{ \s* / \s* }x, $table, -1;
99870f4d
KW
247 print __LINE__, ": $type\n" if @parts > 2 && DEBUG;
248
99870f4d
KW
249 foreach my $part (@parts) {
250 print __LINE__, ": part=$part\n" if DEBUG;
251
252 $part =~ s/^\+\s*//; # Remove leading plus
253 $part =~ s/^-\s*/-/; # Remove blanks after unary
254 # minus
255
256 # Remove underscores between digits.
eb568297 257 $part =~ s/(?<= [0-9] ) _ (?= [0-9] ) //xg;
99870f4d
KW
258
259 # No leading zeros (but don't make a single '0'
260 # into a null string)
261 $part =~ s/ ^ ( -? ) 0+ /$1/x;
262 $part .= '0' if $part eq '-' || $part eq "";
263
264 # No trailing zeros after a decimal point
0426f635 265 $part =~ s/ ( \. [0-9]*? ) 0+ $ /$1/x;
99870f4d
KW
266
267 # Begin with a 0 if a leading decimal point
268 $part =~ s/ ^ ( -? ) \. /${1}0./x;
269
270 # Ensure not a trailing decimal point: turn into an
271 # integer
272 $part =~ s/ \. $ //x;
273
274 print STDERR __LINE__, ": part=$part\n" if DEBUG;
275 #return $type if $part eq "";
99870f4d
KW
276 }
277
278 # If a rational...
279 if (@parts == 2) {
280
281 # If denominator is negative, get rid of it, and ...
282 if ($parts[1] =~ s/^-//) {
283
284 # If numerator is also negative, convert the
3143528c
KW
285 # whole thing to positive, else move the minus
286 # to the numerator
99870f4d
KW
287 if ($parts[0] !~ s/^-//) {
288 $parts[0] = '-' . $parts[0];
289 }
290 }
291 $table = join '/', @parts;
292 }
293 elsif ($property ne 'nv' || $parts[0] !~ /\./) {
294
295 # Here is not numeric value, or doesn't have a
296 # decimal point. No further manipulation is
297 # necessary. (Note the hard-coded property name.
298 # This could fail if other properties eventually
299 # had fractions as well; perhaps the cjk ones
300 # could evolve to do that. This hard-coding could
301 # be fixed by mktables generating a list of
302 # properties that could have fractions.)
303 $table = $parts[0];
304 } else {
305
0426f635
KW
306 # Here is a floating point numeric_value. Convert
307 # to rational. Get a normalized form, like
308 # 5.00E-01, and look that up in the hash
99870f4d 309
0426f635
KW
310 my $float = sprintf "%.*e",
311 $utf8::e_precision,
312 0 + $parts[0];
99870f4d 313
0426f635
KW
314 if (exists $utf8::nv_floating_to_rational{$float}) {
315 $table = $utf8::nv_floating_to_rational{$float};
316 } else {
317 pop @recursed if @recursed;
318 return $type;
99870f4d
KW
319 }
320 }
321 print STDERR __LINE__, ": $property=$table\n" if DEBUG;
322 }
323 }
324
325 # Combine lhs (if any) and rhs to get something that matches
326 # the syntax of the lookups.
327 $property_and_table = "$prefix$table";
328 print STDERR __LINE__, ": $property_and_table\n" if DEBUG;
329
330 # First try stricter matching.
331 $file = $utf8::stricter_to_file_of{$property_and_table};
332
333 # If didn't find it, try again with looser matching by editing
334 # out the applicable characters on the rhs and looking up
335 # again.
e1dc048d 336 my $strict_property_and_table;
99870f4d 337 if (! defined $file) {
e1dc048d
KW
338
339 # This isn't used unless the name begins with 'to'
340 $strict_property_and_table = $property_and_table =~ s/^to//r;
c4ab2516 341 $table = _loose_name($table);
99870f4d
KW
342 $property_and_table = "$prefix$table";
343 print STDERR __LINE__, ": $property_and_table\n" if DEBUG;
344 $file = $utf8::loose_to_file_of{$property_and_table};
345 }
346
347 # Add the constant and go fetch it in.
348 if (defined $file) {
a6517ebf 349
eb092534
KW
350 # If the file name contains a !, it means to invert. The
351 # 0+ makes sure result is numeric
352 $invert_it = 0 + $file =~ s/!//;
a6517ebf 353
99870f4d
KW
354 if ($utf8::why_deprecated{$file}) {
355 warnings::warnif('deprecated', "Use of '$type' in \\p{} or \\P{} is deprecated because: $utf8::why_deprecated{$file};");
356 }
4ff124d1
KW
357
358 if ($caseless
359 && exists $utf8::caseless_equivalent{$property_and_table})
360 {
361 $file = $utf8::caseless_equivalent{$property_and_table};
362 }
eb092534
KW
363
364 # The pseudo-directory '#' means that there really isn't a
365 # file to read, the data is in-line as part of the string;
366 # we extract it below.
367 $file = "$unicore_dir/lib/$file.pl" unless $file =~ m!^#/!;
99870f4d
KW
368 last GETFILE;
369 }
370 print STDERR __LINE__, ": didn't find $property_and_table\n" if DEBUG;
371
372 ##
99870f4d
KW
373 ## Last attempt -- see if it's a standard "To" name
374 ## (e.g. "ToLower") ToTitle is used by ucfirst().
375 ## The user-level way to access ToDigit() and ToFold()
376 ## is to use Unicode::UCD.
377 ##
cf2cd801 378 # Only check if caller wants non-binary
e1dc048d
KW
379 if ($minbits != 1) {
380 if ($property_and_table =~ s/^to//) {
cf2cd801 381 # Look input up in list of properties for which we have
e1dc048d 382 # mapping files. First do it with the strict approach
5a8dc5ca
KW
383 if (defined ($file = $utf8::strict_property_to_file_of{
384 $strict_property_and_table}))
e1dc048d
KW
385 {
386 $type = $utf8::file_to_swash_name{$file};
5a8dc5ca
KW
387 print STDERR __LINE__, ": type set to $type\n"
388 if DEBUG;
e1dc048d
KW
389 $file = "$unicore_dir/$file.pl";
390 last GETFILE;
391 }
392 elsif (defined ($file =
cf2cd801 393 $utf8::loose_property_to_file_of{$property_and_table}))
5a8dc5ca
KW
394 {
395 $type = $utf8::file_to_swash_name{$file};
396 print STDERR __LINE__, ": type set to $type\n"
397 if DEBUG;
398 $file = "$unicore_dir/$file.pl";
399 last GETFILE;
400 } # If that fails see if there is a corresponding binary
401 # property file
402 elsif (defined ($file =
403 $utf8::loose_to_file_of{$property_and_table}))
404 {
6aec79bd 405
5a8dc5ca
KW
406 # Here, there is no map file for the property we
407 # are trying to get the map of, but this is a
408 # binary property, and there is a file for it that
d963b40d
KW
409 # can easily be translated to a mapping, so use
410 # that, treating this as a binary property.
411 # Setting 'minbits' here causes it to be stored as
412 # such in the cache, so if someone comes along
413 # later looking for just a binary, they get it.
5a8dc5ca
KW
414 $minbits = 1;
415
416 # The 0+ makes sure is numeric
417 $invert_it = 0 + $file =~ s/!//;
418 $file = "$unicore_dir/lib/$file.pl"
419 unless $file =~ m!^#/!;
420 last GETFILE;
421 }
cf2cd801 422 }
e1dc048d 423 }
99870f4d
KW
424
425 ##
426 ## If we reach this line, it's because we couldn't figure
427 ## out what to do with $type. Ouch.
428 ##
429
3dd10fe8 430 pop @recursed if @recursed;
99870f4d 431 return $type;
c9db9632 432 } # end of GETFILE block
cf25bb62 433
99870f4d
KW
434 if (defined $file) {
435 print STDERR __LINE__, ": found it (file='$file')\n" if DEBUG;
436
437 ##
438 ## If we reach here, it was due to a 'last GETFILE' above
439 ## (exception: user-defined properties and mappings), so we
440 ## have a filename, so now we load it if we haven't already.
eb092534
KW
441
442 # The pseudo-directory '#' means the result isn't really a
443 # file, but is in-line, with semi-colons to be turned into
444 # new-lines. Since it is in-line there is no advantage to
445 # caching the result
446 if ($file =~ s!^#/!!) {
3854b4b8 447 $list = $utf8::inline_definitions[$file];
eb092534
KW
448 }
449 else {
450 # Here, we have an actual file to read in and load, but it
451 # may already have been read-in and cached. The cache key
452 # is the class and file to load, and whether the results
453 # need to be inverted.
1f9f7d4c
KW
454 my $found = $Cache{$class, $file, $invert_it};
455 if ($found and ref($found) eq $class) {
456 print STDERR __LINE__, ": Returning cached swash for '$class,$file,$invert_it' for \\p{$type}\n" if DEBUG;
457 pop @recursed if @recursed;
458 return $found;
459 }
99870f4d 460
1f9f7d4c
KW
461 local $@;
462 local $!;
463 $list = do $file; die $@ if $@;
eb092534
KW
464 }
465
7b953548 466 $list_is_from_mktables = 1;
0a0ffbce 467 }
c9db9632
KW
468 } # End of $type is non-null
469
470 # Here, either $type was null, or we found the requested property and
471 # read it into $list
cf25bb62 472
7b953548 473 my $extras = "";
c9db9632 474
99870f4d
KW
475 my $bits = $minbits;
476
7b953548
KW
477 # mktables lists don't have extras, like '&utf8::prop', so don't need
478 # to separate them; also lists are already sorted, so don't need to do
479 # that.
480 if ($list && ! $list_is_from_mktables) {
0e9be77f 481 my $taint = substr($list,0,0); # maintain taint
9b225781 482
cb6d3474
KW
483 # Separate the extras from the code point list, and make sure
484 # user-defined properties and tr/// are well-behaved for
485 # downstream code.
486 if ($user_defined || $none) {
9b225781
KW
487 my @tmp = split(/^/m, $list);
488 my %seen;
489 no warnings;
490
491 # The extras are anything that doesn't begin with a hex digit.
492 $extras = join '', $taint, grep /^[^0-9a-fA-F]/, @tmp;
493
494 # Remove the extras, and sort the remaining entries by the
495 # numeric value of their beginning hex digits, removing any
496 # duplicates.
497 $list = join '', $taint,
498 map { $_->[1] }
499 sort { $a->[0] <=> $b->[0] }
55e767ae
RGS
500 map { /^([0-9a-fA-F]+)/ && !$seen{$1}++ ? [ CORE::hex($1), $_ ] : () }
501 @tmp; # XXX doesn't do ranges right
4de6d205
KW
502 }
503 else {
504 # mktables has gone to some trouble to make non-user defined
505 # properties well-behaved, so we can skip the effort we do for
506 # user-defined ones. Any extras are at the very beginning of
507 # the string.
508
509 # This regex splits out the first lines of $list into $1 and
510 # strips them off from $list, until we get one that begins
511 # with a hex number, alone on the line, or followed by a tab.
512 # Either portion may be empty.
513 $list =~ s/ \A ( .*? )
514 (?: \z | (?= ^ [0-9a-fA-F]+ (?: \t | $) ) )
515 //msx;
516
517 $extras = "$taint$1";
518 }
99870f4d 519 }
a0ed51b3 520
99870f4d
KW
521 if ($none) {
522 my $hextra = sprintf "%04x", $none + 1;
523 $list =~ s/\tXXXX$/\t$hextra/mg;
524 }
a0ed51b3 525
99870f4d
KW
526 if ($minbits != 1 && $minbits < 32) { # not binary property
527 my $top = 0;
528 while ($list =~ /^([0-9a-fA-F]+)(?:[\t]([0-9a-fA-F]+)?)(?:[ \t]([0-9a-fA-F]+))?/mg) {
529 my $min = CORE::hex $1;
530 my $max = defined $2 ? CORE::hex $2 : $min;
531 my $val = defined $3 ? CORE::hex $3 : 0;
532 $val += $max - $min if defined $3;
533 $top = $val if $val > $top;
534 }
535 my $topbits =
536 $top > 0xffff ? 32 :
537 $top > 0xff ? 16 : 8;
538 $bits = $topbits if $bits < $topbits;
539 }
a0ed51b3 540
99870f4d
KW
541 my @extras;
542 if ($extras) {
543 for my $x ($extras) {
0e9be77f 544 my $taint = substr($x,0,0); # maintain taint
99870f4d
KW
545 pos $x = 0;
546 while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
0e9be77f
DM
547 my $char = "$1$taint";
548 my $name = "$2$taint";
549 print STDERR __LINE__, ": char [$char] => name [$name]\n"
550 if DEBUG;
99870f4d
KW
551 if ($char =~ /[-+!&]/) {
552 my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really
553 my $subobj;
554 if ($c eq 'utf8') {
555 $subobj = utf8->SWASHNEW($t, "", $minbits, 0);
556 }
557 elsif (exists &$name) {
558 $subobj = utf8->SWASHNEW($name, "", $minbits, 0);
559 }
560 elsif ($c =~ /^([0-9a-fA-F]+)/) {
561 $subobj = utf8->SWASHNEW("", $c, $minbits, 0);
562 }
25df5e40 563 print STDERR __LINE__, ": returned from getting sub object for $name\n" if DEBUG;
3dd10fe8
KW
564 if (! ref $subobj) {
565 pop @recursed if @recursed && $type;
566 return $subobj;
567 }
99870f4d
KW
568 push @extras, $name => $subobj;
569 $bits = $subobj->{BITS} if $bits < $subobj->{BITS};
04e9cbbb
KW
570 $user_defined = $subobj->{USER_DEFINED}
571 if $subobj->{USER_DEFINED};
99870f4d
KW
572 }
573 }
574 }
575 }
a0ed51b3 576
99870f4d 577 if (DEBUG) {
0fada5e1 578 print STDERR __LINE__, ": CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none, INVERT_IT => $invert_it, USER_DEFINED => $user_defined";
99870f4d
KW
579 print STDERR "\nLIST =>\n$list" if defined $list;
580 print STDERR "\nEXTRAS =>\n$extras" if defined $extras;
581 print STDERR "\n";
582 }
a0ed51b3 583
99870f4d
KW
584 my $SWASH = bless {
585 TYPE => $type,
586 BITS => $bits,
587 EXTRAS => $extras,
588 LIST => $list,
589 NONE => $none,
0fada5e1 590 USER_DEFINED => $user_defined,
99870f4d
KW
591 @extras,
592 } => $class;
593
594 if ($file) {
4a7e937e 595 $Cache{$class, $file, $invert_it} = $SWASH;
64870e05
KW
596 if ($type
597 && exists $utf8::SwashInfo{$type}
598 && exists $utf8::SwashInfo{$type}{'specials_name'})
599 {
600 my $specials_name = $utf8::SwashInfo{$type}{'specials_name'};
601 no strict "refs";
69b36ee0 602 print STDERR "\nspecials_name => $specials_name\n" if DEBUG;
64870e05
KW
603 $SWASH->{'SPECIALS'} = \%$specials_name;
604 }
a6517ebf 605 $SWASH->{'INVERT_IT'} = $invert_it;
99870f4d 606 }
5beb625e 607
3dd10fe8
KW
608 pop @recursed if @recursed && $type;
609
99870f4d 610 return $SWASH;
5beb625e 611 }
a0ed51b3
LW
612}
613
b0e3252e 614# Now SWASHGET is recasted into a C function S_swatch_get (see utf8.c).
a0ed51b3
LW
615
6161;