This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
uni/cache.t: Fix to handle regex compile time Uni props
[perl5.git] / lib / utf8_heavy.pl
CommitLineData
a0ed51b3 1package utf8;
cf25bb62
JH
2use strict;
3use warnings;
a0ed51b3 4
15732964 5sub DEBUG () { 0 }
4b17cb47 6$|=1 if DEBUG;
a0ed51b3
LW
7
8sub DESTROY {}
9
5beb625e
JH
10my %Cache;
11
86916d66
JH
12sub croak { require Carp; Carp::croak(@_) }
13
c4ab2516 14sub _loose_name ($) {
45376db6
KW
15 # Given a lowercase property or property-value name, return its
16 # standardized version that is expected for look-up in the 'loose' hashes
17 # in Heavy.pl (hence, this depends on what mktables does). This squeezes
18 # out blanks, underscores and dashes. The complication stems from the
19 # grandfathered-in 'L_', which retains a single trailing underscore.
20
21 my $loose = $_[0] =~ s/[-\s_]//rg;
22
cf2cd801 23 return $loose if $loose !~ / ^ (?: is | to )? l $/x;
45376db6
KW
24 return 'l_' if $_[0] =~ / l .* _ /x; # If original had a trailing '_'
25 return $loose;
26}
27
5beb625e 28##
3a2263fe
RGS
29## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape.
30## It's a data structure that encodes a set of Unicode characters.
5beb625e
JH
31##
32
99870f4d
KW
33{
34 # If a floating point number is within this distance from the value of a
35 # fraction, it is considered to be that fraction, even if many more digits
36 # are specified that don't exactly match.
37 my $min_floating_slop;
38
3dd10fe8
KW
39 # To guard against this program calling something that in turn ends up
40 # calling this program with the same inputs, and hence infinitely
41 # recursing, we keep a stack of the properties that are currently in
42 # progress, pushed upon entry, popped upon return.
43 my @recursed;
44
99870f4d
KW
45 sub SWASHNEW {
46 my ($class, $type, $list, $minbits, $none) = @_;
0fada5e1 47 my $user_defined = 0;
99870f4d
KW
48 local $^D = 0 if $^D;
49
3dd10fe8 50 $class = "" unless defined $class;
5bcf08c4
KW
51 print STDERR __LINE__, ": class=$class, type=$type, list=",
52 (defined $list) ? $list : ':undef:',
53 ", minbits=$minbits, none=$none\n" if DEBUG;
99870f4d
KW
54
55 ##
56 ## Get the list of codepoints for the type.
57 ## Called from swash_init (see utf8.c) or SWASHNEW itself.
58 ##
59 ## Callers of swash_init:
60 ## op.c:pmtrans -- for tr/// and y///
61 ## regexec.c:regclass_swash -- for /[]/, \p, and \P
62 ## utf8.c:is_utf8_common -- for common Unicode properties
63 ## utf8.c:to_utf8_case -- for lc, uc, ucfirst, etc. and //i
681d705c 64 ## Unicode::UCD::prop_invlist
62b3b855 65 ## Unicode::UCD::prop_invmap
99870f4d
KW
66 ##
67 ## Given a $type, our goal is to fill $list with the set of codepoint
68 ## ranges. If $type is false, $list passed is used.
69 ##
70 ## $minbits:
71 ## For binary properties, $minbits must be 1.
72 ## For character mappings (case and transliteration), $minbits must
73 ## be a number except 1.
74 ##
75 ## $list (or that filled according to $type):
76 ## Refer to perlunicode.pod, "User-Defined Character Properties."
77 ##
78 ## For binary properties, only characters with the property value
79 ## of True should be listed. The 3rd column, if any, will be ignored
80 ##
81 ## $none is undocumented, so I'm (khw) trying to do some documentation
82 ## of it now. It appears to be if there is a mapping in an input file
83 ## that maps to 'XXXX', then that is replaced by $none+1, expressed in
83fe8199 84 ## hexadecimal. It is used somehow in tr///.
99870f4d
KW
85 ##
86 ## To make the parsing of $type clear, this code takes the a rather
87 ## unorthodox approach of last'ing out of the block once we have the
88 ## info we need. Were this to be a subroutine, the 'last' would just
89 ## be a 'return'.
90 ##
fbe6f3cf
KW
91 # If a problem is found $type is returned;
92 # Upon success, a new (or cached) blessed object is returned with
93 # keys TYPE, BITS, EXTRAS, LIST, and NONE with values having the
94 # same meanings as the input parameters.
a6517ebf
KW
95 # SPECIALS contains a reference to any special-treatment hash in the
96 # INVERT_IT is non-zero if the result should be inverted before use
0fada5e1
KW
97 # USER_DEFINED is non-zero if the result came from a user-defined
98 # property.
99870f4d 99 my $file; ## file to load data from, and also part of the %Cache key.
99870f4d
KW
100
101 # Change this to get a different set of Unicode tables
102 my $unicore_dir = 'unicore';
a6517ebf 103 my $invert_it = 0;
7b953548
KW
104 my $list_is_from_mktables = 0; # Is $list returned from a mktables
105 # generated file? If so, we know it's
106 # well behaved.
99870f4d
KW
107
108 if ($type)
cf25bb62 109 {
3dd10fe8 110 # Verify that this isn't a recursive call for this property.
dbe7a391 111 # Can't use croak, as it may try to recurse to here itself.
3dd10fe8
KW
112 my $class_type = $class . "::$type";
113 if (grep { $_ eq $class_type } @recursed) {
114 CORE::die "panic: Infinite recursion in SWASHNEW for '$type'\n";
115 }
116 push @recursed, $class_type;
117
99870f4d
KW
118 $type =~ s/^\s+//;
119 $type =~ s/\s+$//;
120
4ff124d1
KW
121 # regcomp.c surrounds the property name with '__" and '_i' if this
122 # is to be caseless matching.
20b1a891 123 my $caseless = $type =~ s/^(.*)__(.*)_i$/$1$2/;
4ff124d1
KW
124
125 print STDERR __LINE__, ": type=$type, caseless=$caseless\n" if DEBUG;
99870f4d
KW
126
127 GETFILE:
128 {
129 ##
a44f4b74
KW
130 ## It could be a user-defined property. Look in current
131 ## package if no package given
99870f4d
KW
132 ##
133
20b1a891
KW
134
135 my $caller0 = caller(0);
136 my $caller1 = $type =~ s/(.+)::// ? $1 : $caller0 eq 'main' ?
137 'main' : caller(1);
99870f4d 138
d658a8a8 139 if (defined $caller1 && $type =~ /^I[ns]\w+$/) {
99870f4d
KW
140 my $prop = "${caller1}::$type";
141 if (exists &{$prop}) {
0e9be77f
DM
142 # stolen from Scalar::Util::PP::tainted()
143 my $tainted;
144 {
145 local($@, $SIG{__DIE__}, $SIG{__WARN__});
146 local $^W = 0;
147 no warnings;
148 eval { kill 0 * $prop };
149 $tainted = 1 if $@ =~ /^Insecure/;
150 }
151 die "Insecure user-defined property \\p{$prop}\n"
152 if $tainted;
99870f4d 153 no strict 'refs';
4ff124d1 154 $list = &{$prop}($caseless);
0fada5e1 155 $user_defined = 1;
99870f4d
KW
156 last GETFILE;
157 }
158 }
159
a44f4b74
KW
160 # During Perl's compilation, this routine may be called before
161 # the tables are constructed. If so, we have a chicken/egg
162 # problem. If we die, the tables never get constructed, so
163 # keep going, but return an empty table so only what the code
164 # has compiled in internally (currently ASCII/Latin1 range
165 # matching) will work.
1363cbd0
FC
166 BEGIN {
167 # Poor man's constant, to avoid a run-time check.
168 $utf8::{miniperl}
169 = \! defined &DynaLoader::boot_DynaLoader;
170 }
171 if (miniperl) {
a44f4b74
KW
172 eval "require '$unicore_dir/Heavy.pl'";
173 last GETFILE if $@;
174 }
175 else {
176 require "$unicore_dir/Heavy.pl";
177 }
1363cbd0 178 BEGIN { delete $utf8::{miniperl} }
99870f4d 179
4ff124d1 180 # All property names are matched caselessly
49801ec2 181 my $property_and_table = CORE::lc $type;
99870f4d
KW
182 print STDERR __LINE__, ": $property_and_table\n" if DEBUG;
183
184 # See if is of the compound form 'property=value', where the
185 # value indicates the table we should use.
186 my ($property, $table, @remainder) =
187 split /\s*[:=]\s*/, $property_and_table, -1;
3dd10fe8
KW
188 if (@remainder) {
189 pop @recursed if @recursed;
190 return $type;
191 }
99870f4d
KW
192
193 my $prefix;
194 if (! defined $table) {
195
196 # Here, is the single form. The property becomes empty, and
197 # the whole value is the table.
198 $table = $property;
199 $prefix = $property = "";
200 } else {
201 print STDERR __LINE__, ": $property\n" if DEBUG;
202
203 # Here it is the compound property=table form. The property
807807b7
KW
204 # name is always loosely matched, and always can have an
205 # optional 'is' prefix (which isn't true in the single
206 # form).
c4ab2516 207 $property = _loose_name($property) =~ s/^is//r;
99870f4d
KW
208
209 # And convert to canonical form. Quit if not valid.
210 $property = $utf8::loose_property_name_of{$property};
3dd10fe8
KW
211 if (! defined $property) {
212 pop @recursed if @recursed;
213 return $type;
214 }
99870f4d
KW
215
216 $prefix = "$property=";
217
218 # If the rhs looks like it is a number...
219 print STDERR __LINE__, ": table=$table\n" if DEBUG;
220 if ($table =~ qr{ ^ [ \s 0-9 _ + / . -]+ $ }x) {
221 print STDERR __LINE__, ": table=$table\n" if DEBUG;
222
223 # Don't allow leading nor trailing slashes
3dd10fe8
KW
224 if ($table =~ / ^ \/ | \/ $ /x) {
225 pop @recursed if @recursed;
226 return $type;
227 }
99870f4d
KW
228
229 # Split on slash, in case it is a rational, like \p{1/5}
230 my @parts = split qr{ \s* / \s* }x, $table, -1;
231 print __LINE__, ": $type\n" if @parts > 2 && DEBUG;
232
233 # Can have maximum of one slash
3dd10fe8
KW
234 if (@parts > 2) {
235 pop @recursed if @recursed;
236 return $type;
237 }
99870f4d
KW
238
239 foreach my $part (@parts) {
240 print __LINE__, ": part=$part\n" if DEBUG;
241
242 $part =~ s/^\+\s*//; # Remove leading plus
243 $part =~ s/^-\s*/-/; # Remove blanks after unary
244 # minus
245
246 # Remove underscores between digits.
247 $part =~ s/( ?<= [0-9] ) _ (?= [0-9] ) //xg;
248
249 # No leading zeros (but don't make a single '0'
250 # into a null string)
251 $part =~ s/ ^ ( -? ) 0+ /$1/x;
252 $part .= '0' if $part eq '-' || $part eq "";
253
254 # No trailing zeros after a decimal point
255 $part =~ s/ ( \. .*? ) 0+ $ /$1/x;
256
257 # Begin with a 0 if a leading decimal point
258 $part =~ s/ ^ ( -? ) \. /${1}0./x;
259
260 # Ensure not a trailing decimal point: turn into an
261 # integer
262 $part =~ s/ \. $ //x;
263
264 print STDERR __LINE__, ": part=$part\n" if DEBUG;
265 #return $type if $part eq "";
266
267 # Result better look like a number. (This test is
268 # needed because, for example could have a plus in
269 # the middle.)
3dd10fe8
KW
270 if ($part !~ / ^ -? [0-9]+ ( \. [0-9]+)? $ /x) {
271 pop @recursed if @recursed;
272 return $type;
273 }
99870f4d
KW
274 }
275
276 # If a rational...
277 if (@parts == 2) {
278
279 # If denominator is negative, get rid of it, and ...
280 if ($parts[1] =~ s/^-//) {
281
282 # If numerator is also negative, convert the
283 # whole thing to positive, or move the minus to
284 # the numerator
285 if ($parts[0] !~ s/^-//) {
286 $parts[0] = '-' . $parts[0];
287 }
288 }
289 $table = join '/', @parts;
290 }
291 elsif ($property ne 'nv' || $parts[0] !~ /\./) {
292
293 # Here is not numeric value, or doesn't have a
294 # decimal point. No further manipulation is
295 # necessary. (Note the hard-coded property name.
296 # This could fail if other properties eventually
297 # had fractions as well; perhaps the cjk ones
298 # could evolve to do that. This hard-coding could
299 # be fixed by mktables generating a list of
300 # properties that could have fractions.)
301 $table = $parts[0];
302 } else {
303
304 # Here is a floating point numeric_value. Try to
305 # convert to rational. First see if is in the list
306 # of known ones.
307 if (exists $utf8::nv_floating_to_rational{$parts[0]}) {
308 $table = $utf8::nv_floating_to_rational{$parts[0]};
309 } else {
310
311 # Here not in the list. See if is close
312 # enough to something in the list. First
313 # determine what 'close enough' means. It has
314 # to be as tight as what mktables says is the
315 # maximum slop, and as tight as how many
316 # digits we were passed. That is, if the user
317 # said .667, .6667, .66667, etc. we match as
318 # many digits as they passed until get to
319 # where it doesn't matter any more due to the
320 # machine's precision. If they said .6666668,
321 # we fail.
322 (my $fraction = $parts[0]) =~ s/^.*\.//;
323 my $epsilon = 10 ** - (length($fraction));
324 if ($epsilon > $utf8::max_floating_slop) {
325 $epsilon = $utf8::max_floating_slop;
326 }
327
328 # But it can't be tighter than the minimum
329 # precision for this machine. If haven't
330 # already calculated that minimum, do so now.
331 if (! defined $min_floating_slop) {
332
333 # Keep going down an order of magnitude
334 # until find that adding this quantity to
335 # 1 remains 1; but put an upper limit on
336 # this so in case this algorithm doesn't
337 # work properly on some platform, that we
338 # won't loop forever.
339 my $count = 0;
340 $min_floating_slop = 1;
341 while (1+ $min_floating_slop != 1
342 && $count++ < 50)
343 {
344 my $next = $min_floating_slop / 10;
345 last if $next == 0; # If underflows,
346 # use previous one
347 $min_floating_slop = $next;
348 print STDERR __LINE__, ": min_float_slop=$min_floating_slop\n" if DEBUG;
349 }
350
351 # Back off a couple orders of magnitude,
352 # just to be safe.
353 $min_floating_slop *= 100;
354 }
355
356 if ($epsilon < $min_floating_slop) {
357 $epsilon = $min_floating_slop;
358 }
359 print STDERR __LINE__, ": fraction=.$fraction; epsilon=$epsilon\n" if DEBUG;
360
361 undef $table;
362
363 # And for each possible rational in the table,
364 # see if it is within epsilon of the input.
365 foreach my $official
366 (keys %utf8::nv_floating_to_rational)
367 {
368 print STDERR __LINE__, ": epsilon=$epsilon, official=$official, diff=", abs($parts[0] - $official), "\n" if DEBUG;
369 if (abs($parts[0] - $official) < $epsilon) {
370 $table =
371 $utf8::nv_floating_to_rational{$official};
372 last;
373 }
374 }
375
376 # Quit if didn't find one.
3dd10fe8
KW
377 if (! defined $table) {
378 pop @recursed if @recursed;
379 return $type;
380 }
99870f4d
KW
381 }
382 }
383 print STDERR __LINE__, ": $property=$table\n" if DEBUG;
384 }
385 }
386
387 # Combine lhs (if any) and rhs to get something that matches
388 # the syntax of the lookups.
389 $property_and_table = "$prefix$table";
390 print STDERR __LINE__, ": $property_and_table\n" if DEBUG;
391
392 # First try stricter matching.
393 $file = $utf8::stricter_to_file_of{$property_and_table};
394
395 # If didn't find it, try again with looser matching by editing
396 # out the applicable characters on the rhs and looking up
397 # again.
398 if (! defined $file) {
c4ab2516 399 $table = _loose_name($table);
99870f4d
KW
400 $property_and_table = "$prefix$table";
401 print STDERR __LINE__, ": $property_and_table\n" if DEBUG;
402 $file = $utf8::loose_to_file_of{$property_and_table};
403 }
404
405 # Add the constant and go fetch it in.
406 if (defined $file) {
a6517ebf
KW
407
408 # A beginning ! means to invert
409 $invert_it = $file =~ s/^!//;
410
99870f4d
KW
411 if ($utf8::why_deprecated{$file}) {
412 warnings::warnif('deprecated', "Use of '$type' in \\p{} or \\P{} is deprecated because: $utf8::why_deprecated{$file};");
413 }
4ff124d1
KW
414
415 if ($caseless
416 && exists $utf8::caseless_equivalent{$property_and_table})
417 {
418 $file = $utf8::caseless_equivalent{$property_and_table};
419 }
99870f4d
KW
420 $file= "$unicore_dir/lib/$file.pl";
421 last GETFILE;
422 }
423 print STDERR __LINE__, ": didn't find $property_and_table\n" if DEBUG;
424
425 ##
99870f4d
KW
426 ## Last attempt -- see if it's a standard "To" name
427 ## (e.g. "ToLower") ToTitle is used by ucfirst().
428 ## The user-level way to access ToDigit() and ToFold()
429 ## is to use Unicode::UCD.
430 ##
cf2cd801 431 # Only check if caller wants non-binary
6aec79bd
KW
432 my $retried = 0;
433 if ($minbits != 1 && $property_and_table =~ s/^to//) {{
cf2cd801
KW
434 # Look input up in list of properties for which we have
435 # mapping files.
436 if (defined ($file =
437 $utf8::loose_property_to_file_of{$property_and_table}))
438 {
439 $type = $utf8::file_to_swash_name{$file};
440 print STDERR __LINE__, ": type set to $type\n" if DEBUG;
441 $file = "$unicore_dir/$file.pl";
442 last GETFILE;
6aec79bd
KW
443 } # If that fails see if there is a corresponding binary
444 # property file
445 elsif (defined ($file =
446 $utf8::loose_to_file_of{$property_and_table}))
447 {
448
449 # Here, there is no map file for the property we are
450 # trying to get the map of, but this is a binary
451 # property, and there is a file for it that can easily
452 # be translated to a mapping.
453
454 # In the case of properties that are forced to binary,
455 # they are a combination. We return the actual
456 # mapping instead of the binary. If the input is
457 # something like 'Tocjkkiicore', it will be found in
458 # %loose_property_to_file_of above as => 'To/kIICore'.
459 # But the form like ToIskiicore won't be. To fix
460 # this, it was easiest to do it here. These
461 # properties are the complements of the default
462 # property, so there is an entry in %loose_to_file_of
463 # that is 'iskiicore' => '!kIICore/N', If we find such
464 # an entry, strip off things and try again, which
465 # should find the entry in %loose_property_to_file_of.
466 # Actual binary properties that are of this form, such
467 # as this entry: 'ishrkt' => '!Perl/Any' will also be
468 # retried, but won't be in %loose_property_to_file_of,
469 # and instead the next time through, it will find
470 # 'hrkt' => '!Perl/Any' and proceed.
471 redo if ! $retried
472 && $file =~ /^!/
473 && $property_and_table =~ s/^is//;
474
475 # This is a binary property. Setting this here causes
476 # it to be stored as such in the cache, so if someone
477 # comes along later looking for just a binary, they
478 # get it.
479 $minbits = 1;
480
481 $invert_it = $file =~ s/^!//;
482 $file = "$unicore_dir/lib/$file.pl";
483 last GETFILE;
cf2cd801 484 }
6aec79bd 485 } }
99870f4d
KW
486
487 ##
488 ## If we reach this line, it's because we couldn't figure
489 ## out what to do with $type. Ouch.
490 ##
491
3dd10fe8 492 pop @recursed if @recursed;
99870f4d 493 return $type;
c9db9632 494 } # end of GETFILE block
cf25bb62 495
99870f4d
KW
496 if (defined $file) {
497 print STDERR __LINE__, ": found it (file='$file')\n" if DEBUG;
498
499 ##
500 ## If we reach here, it was due to a 'last GETFILE' above
501 ## (exception: user-defined properties and mappings), so we
502 ## have a filename, so now we load it if we haven't already.
503 ## If we have, return the cached results. The cache key is the
504 ## class and file to load.
505 ##
506 my $found = $Cache{$class, $file};
507 if ($found and ref($found) eq $class) {
a6517ebf 508 print STDERR __LINE__, ": Returning cached '$file' for \\p{$type}; invert_it=$invert_it\n" if DEBUG;
3dd10fe8 509 pop @recursed if @recursed;
a6517ebf 510 $found->{'INVERT_IT'} = $invert_it;
99870f4d
KW
511 return $found;
512 }
513
514 local $@;
515 local $!;
516 $list = do $file; die $@ if $@;
7b953548 517 $list_is_from_mktables = 1;
0a0ffbce 518 }
c9db9632
KW
519 } # End of $type is non-null
520
521 # Here, either $type was null, or we found the requested property and
522 # read it into $list
cf25bb62 523
7b953548 524 my $extras = "";
c9db9632 525
99870f4d
KW
526 my $bits = $minbits;
527
7b953548
KW
528 # mktables lists don't have extras, like '&utf8::prop', so don't need
529 # to separate them; also lists are already sorted, so don't need to do
530 # that.
531 if ($list && ! $list_is_from_mktables) {
0e9be77f 532 my $taint = substr($list,0,0); # maintain taint
9b225781 533
4de6d205
KW
534 # Separate the extras from the code point list, and for
535 # user-defined properties, make sure the latter are well-behaved
9b225781 536 # for downstream code.
4de6d205 537 if ($user_defined) {
9b225781
KW
538 my @tmp = split(/^/m, $list);
539 my %seen;
540 no warnings;
541
542 # The extras are anything that doesn't begin with a hex digit.
543 $extras = join '', $taint, grep /^[^0-9a-fA-F]/, @tmp;
544
545 # Remove the extras, and sort the remaining entries by the
546 # numeric value of their beginning hex digits, removing any
547 # duplicates.
548 $list = join '', $taint,
549 map { $_->[1] }
550 sort { $a->[0] <=> $b->[0] }
551 map { /^([0-9a-fA-F]+)/; [ CORE::hex($1), $_ ] }
552 grep { /^([0-9a-fA-F]+)/ and not $seen{$1}++ } @tmp; # XXX doesn't do ranges right
4de6d205
KW
553 }
554 else {
555 # mktables has gone to some trouble to make non-user defined
556 # properties well-behaved, so we can skip the effort we do for
557 # user-defined ones. Any extras are at the very beginning of
558 # the string.
559
560 # This regex splits out the first lines of $list into $1 and
561 # strips them off from $list, until we get one that begins
562 # with a hex number, alone on the line, or followed by a tab.
563 # Either portion may be empty.
564 $list =~ s/ \A ( .*? )
565 (?: \z | (?= ^ [0-9a-fA-F]+ (?: \t | $) ) )
566 //msx;
567
568 $extras = "$taint$1";
569 }
99870f4d 570 }
a0ed51b3 571
99870f4d
KW
572 if ($none) {
573 my $hextra = sprintf "%04x", $none + 1;
574 $list =~ s/\tXXXX$/\t$hextra/mg;
575 }
a0ed51b3 576
99870f4d
KW
577 if ($minbits != 1 && $minbits < 32) { # not binary property
578 my $top = 0;
579 while ($list =~ /^([0-9a-fA-F]+)(?:[\t]([0-9a-fA-F]+)?)(?:[ \t]([0-9a-fA-F]+))?/mg) {
580 my $min = CORE::hex $1;
581 my $max = defined $2 ? CORE::hex $2 : $min;
582 my $val = defined $3 ? CORE::hex $3 : 0;
583 $val += $max - $min if defined $3;
584 $top = $val if $val > $top;
585 }
586 my $topbits =
587 $top > 0xffff ? 32 :
588 $top > 0xff ? 16 : 8;
589 $bits = $topbits if $bits < $topbits;
590 }
a0ed51b3 591
99870f4d
KW
592 my @extras;
593 if ($extras) {
594 for my $x ($extras) {
0e9be77f 595 my $taint = substr($x,0,0); # maintain taint
99870f4d
KW
596 pos $x = 0;
597 while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
0e9be77f
DM
598 my $char = "$1$taint";
599 my $name = "$2$taint";
600 print STDERR __LINE__, ": char [$char] => name [$name]\n"
601 if DEBUG;
99870f4d
KW
602 if ($char =~ /[-+!&]/) {
603 my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really
604 my $subobj;
605 if ($c eq 'utf8') {
606 $subobj = utf8->SWASHNEW($t, "", $minbits, 0);
607 }
608 elsif (exists &$name) {
609 $subobj = utf8->SWASHNEW($name, "", $minbits, 0);
610 }
611 elsif ($c =~ /^([0-9a-fA-F]+)/) {
612 $subobj = utf8->SWASHNEW("", $c, $minbits, 0);
613 }
25df5e40 614 print STDERR __LINE__, ": returned from getting sub object for $name\n" if DEBUG;
3dd10fe8
KW
615 if (! ref $subobj) {
616 pop @recursed if @recursed && $type;
617 return $subobj;
618 }
99870f4d
KW
619 push @extras, $name => $subobj;
620 $bits = $subobj->{BITS} if $bits < $subobj->{BITS};
04e9cbbb
KW
621 $user_defined = $subobj->{USER_DEFINED}
622 if $subobj->{USER_DEFINED};
99870f4d
KW
623 }
624 }
625 }
626 }
a0ed51b3 627
99870f4d 628 if (DEBUG) {
0fada5e1 629 print STDERR __LINE__, ": CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none, INVERT_IT => $invert_it, USER_DEFINED => $user_defined";
99870f4d
KW
630 print STDERR "\nLIST =>\n$list" if defined $list;
631 print STDERR "\nEXTRAS =>\n$extras" if defined $extras;
632 print STDERR "\n";
633 }
a0ed51b3 634
99870f4d
KW
635 my $SWASH = bless {
636 TYPE => $type,
637 BITS => $bits,
638 EXTRAS => $extras,
639 LIST => $list,
640 NONE => $none,
0fada5e1 641 USER_DEFINED => $user_defined,
99870f4d
KW
642 @extras,
643 } => $class;
644
645 if ($file) {
646 $Cache{$class, $file} = $SWASH;
64870e05
KW
647 if ($type
648 && exists $utf8::SwashInfo{$type}
649 && exists $utf8::SwashInfo{$type}{'specials_name'})
650 {
651 my $specials_name = $utf8::SwashInfo{$type}{'specials_name'};
652 no strict "refs";
69b36ee0 653 print STDERR "\nspecials_name => $specials_name\n" if DEBUG;
64870e05
KW
654 $SWASH->{'SPECIALS'} = \%$specials_name;
655 }
a6517ebf 656 $SWASH->{'INVERT_IT'} = $invert_it;
99870f4d 657 }
5beb625e 658
3dd10fe8
KW
659 pop @recursed if @recursed && $type;
660
99870f4d 661 return $SWASH;
5beb625e 662 }
a0ed51b3
LW
663}
664
b0e3252e 665# Now SWASHGET is recasted into a C function S_swatch_get (see utf8.c).
a0ed51b3
LW
666
6671;