This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Bump Locale::Maketext's version
[perl5.git] / dist / Locale-Maketext / lib / Locale / Maketext.pm
CommitLineData
e9c9ffca 1
9378c581
JH
2package Locale::Maketext;
3use strict;
4use vars qw( @ISA $VERSION $MATCH_SUPERS $USING_LANGUAGE_TAGS
f666394a 5$USE_LITERALS $MATCH_SUPERS_TIGHTLY);
9378c581 6use Carp ();
ed897e29
TR
7use I18N::LangTags ();
8use I18N::LangTags::Detect ();
9378c581
JH
9
10#--------------------------------------------------------------------------
11
12BEGIN { unless(defined &DEBUG) { *DEBUG = sub () {0} } }
f666394a 13# define the constant 'DEBUG' at compile-time
9378c581 14
87d86da5
TR
15# turn on utf8 if we have it (this is what GutsLoader.pm used to do essentially )
16# use if (exists $INC{'utf8.pm'} || eval 'use utf8'), 'utf8';
17BEGIN {
18
19 # if we have it || we can load it
20 if ( exists $INC{'utf8.pm'} || eval { local $SIG{'__DIE__'};require utf8; } ) {
21 utf8->import();
22 DEBUG and warn " utf8 on for _compile()\n";
23 }
24 else {
25 DEBUG and warn " utf8 not available for _compile() ($INC{'utf8.pm'})\n$@\n";
26 }
27}
28
29
6274c42e 30$VERSION = '1.19';
9378c581
JH
31@ISA = ();
32
33$MATCH_SUPERS = 1;
14be35aa
AMS
34$MATCH_SUPERS_TIGHTLY = 1;
35$USING_LANGUAGE_TAGS = 1;
f666394a
RGS
36# Turning this off is somewhat of a security risk in that little or no
37# checking will be done on the legality of tokens passed to the
38# eval("use $module_name") in _try_use. If you turn this off, you have
39# to do your own taint checking.
9378c581
JH
40
41$USE_LITERALS = 1 unless defined $USE_LITERALS;
f666394a 42# a hint for compiling bracket-notation things.
9378c581
JH
43
44my %isa_scan = ();
45
46###########################################################################
47
48sub quant {
f666394a 49 my($handle, $num, @forms) = @_;
9378c581 50
f666394a
RGS
51 return $num if @forms == 0; # what should this mean?
52 return $forms[2] if @forms > 2 and $num == 0; # special zeroth case
9378c581 53
f666394a
RGS
54 # Normal case:
55 # Note that the formatting of $num is preserved.
56 return( $handle->numf($num) . ' ' . $handle->numerate($num, @forms) );
57 # Most human languages put the number phrase before the qualified phrase.
9378c581
JH
58}
59
60
61sub numerate {
f666394a
RGS
62 # return this lexical item in a form appropriate to this number
63 my($handle, $num, @forms) = @_;
64 my $s = ($num == 1);
65
66 return '' unless @forms;
67 if(@forms == 1) { # only the headword form specified
68 return $s ? $forms[0] : ($forms[0] . 's'); # very cheap hack.
69 }
70 else { # sing and plural were specified
71 return $s ? $forms[0] : $forms[1];
72 }
9378c581
JH
73}
74
75#--------------------------------------------------------------------------
76
77sub numf {
f666394a
RGS
78 my($handle, $num) = @_[0,1];
79 if($num < 10_000_000_000 and $num > -10_000_000_000 and $num == int($num)) {
80 $num += 0; # Just use normal integer stringification.
81 # Specifically, don't let %G turn ten million into 1E+007
82 }
83 else {
84 $num = CORE::sprintf('%G', $num);
85 # "CORE::" is there to avoid confusion with the above sub sprintf.
86 }
87 while( $num =~ s/^([-+]?\d+)(\d{3})/$1,$2/s ) {1} # right from perlfaq5
88 # The initial \d+ gobbles as many digits as it can, and then we
89 # backtrack so it un-eats the rightmost three, and then we
90 # insert the comma there.
91
92 $num =~ tr<.,><,.> if ref($handle) and $handle->{'numf_comma'};
93 # This is just a lame hack instead of using Number::Format
94 return $num;
9378c581
JH
95}
96
97sub sprintf {
f666394a
RGS
98 no integer;
99 my($handle, $format, @params) = @_;
100 return CORE::sprintf($format, @params);
9378c581
JH
101 # "CORE::" is there to avoid confusion with myself!
102}
103
104#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#
105
106use integer; # vroom vroom... applies to the whole rest of the module
107
108sub language_tag {
f666394a
RGS
109 my $it = ref($_[0]) || $_[0];
110 return undef unless $it =~ m/([^':]+)(?:::)?$/s;
111 $it = lc($1);
112 $it =~ tr<_><->;
113 return $it;
9378c581
JH
114}
115
116sub encoding {
f666394a
RGS
117 my $it = $_[0];
118 return(
119 (ref($it) && $it->{'encoding'})
120 || 'iso-8859-1' # Latin-1
121 );
122}
9378c581
JH
123
124#--------------------------------------------------------------------------
125
126sub fallback_languages { return('i-default', 'en', 'en-US') }
127
128sub fallback_language_classes { return () }
129
130#--------------------------------------------------------------------------
131
132sub fail_with { # an actual attribute method!
f666394a
RGS
133 my($handle, @params) = @_;
134 return unless ref($handle);
135 $handle->{'fail'} = $params[0] if @params;
136 return $handle->{'fail'};
9378c581
JH
137}
138
139#--------------------------------------------------------------------------
140
141sub failure_handler_auto {
f666394a
RGS
142 # Meant to be used like:
143 # $handle->fail_with('failure_handler_auto')
144
145 my $handle = shift;
146 my $phrase = shift;
147
148 $handle->{'failure_lex'} ||= {};
149 my $lex = $handle->{'failure_lex'};
150
f7d5028f 151 my $value ||= ($lex->{$phrase} ||= $handle->_compile($phrase));
f666394a
RGS
152
153 # Dumbly copied from sub maketext:
154 return ${$value} if ref($value) eq 'SCALAR';
155 return $value if ref($value) ne 'CODE';
156 {
157 local $SIG{'__DIE__'};
158 eval { $value = &$value($handle, @_) };
159 }
160 # If we make it here, there was an exception thrown in the
161 # call to $value, and so scream:
162 if($@) {
f666394a 163 # pretty up the error message
9961f4dd 164 $@ =~ s{\s+at\s+\(eval\s+\d+\)\s+line\s+(\d+)\.?\n?}
f666394a
RGS
165 {\n in bracket code [compiled line $1],}s;
166 #$err =~ s/\n?$/\n/s;
9961f4dd 167 Carp::croak "Error in maketexting \"$phrase\":\n$@ as used";
f666394a
RGS
168 # Rather unexpected, but suppose that the sub tried calling
169 # a method that didn't exist.
170 }
171 else {
172 return $value;
173 }
9378c581
JH
174}
175
176#==========================================================================
177
178sub new {
f666394a
RGS
179 # Nothing fancy!
180 my $class = ref($_[0]) || $_[0];
181 my $handle = bless {}, $class;
182 $handle->init;
183 return $handle;
9378c581
JH
184}
185
186sub init { return } # no-op
187
188###########################################################################
189
190sub maketext {
f666394a
RGS
191 # Remember, this can fail. Failure is controllable many ways.
192 Carp::croak 'maketext requires at least one parameter' unless @_ > 1;
193
194 my($handle, $phrase) = splice(@_,0,2);
195 Carp::confess('No handle/phrase') unless (defined($handle) && defined($phrase));
196
9961f4dd
TR
197 # backup $@ in case it it's still being used in the calling code.
198 # If no failures, we'll re-set it back to what it was later.
199 my $at = $@;
f666394a 200
bac7bf84
FC
201 # Copy @_ case one of its elements is $@.
202 @_ = @_;
203
f666394a
RGS
204 # Look up the value:
205
206 my $value;
ace47d68
TR
207 if (exists $handle->{'_external_lex_cache'}{$phrase}) {
208 DEBUG and warn "* Using external lex cache version of \"$phrase\"\n";
209 $value = $handle->{'_external_lex_cache'}{$phrase};
210 }
211 else {
212 foreach my $h_r (
213 @{ $isa_scan{ref($handle) || $handle} || $handle->_lex_refs }
214 ) {
215 DEBUG and warn "* Looking up \"$phrase\" in $h_r\n";
216 if(exists $h_r->{$phrase}) {
217 DEBUG and warn " Found \"$phrase\" in $h_r\n";
218 unless(ref($value = $h_r->{$phrase})) {
219 # Nonref means it's not yet compiled. Compile and replace.
220 if ($handle->{'use_external_lex_cache'}) {
221 $value = $handle->{'_external_lex_cache'}{$phrase} = $handle->_compile($value);
222 }
223 else {
224 $value = $h_r->{$phrase} = $handle->_compile($value);
225 }
226 }
227 last;
f666394a 228 }
ace47d68
TR
229 # extending packages need to be able to localize _AUTO and if readonly can't "local $h_r->{'_AUTO'} = 1;"
230 # but they can "local $handle->{'_external_lex_cache'}{'_AUTO'} = 1;"
231 elsif($phrase !~ m/^_/s and ($handle->{'use_external_lex_cache'} ? ( exists $handle->{'_external_lex_cache'}{'_AUTO'} ? $handle->{'_external_lex_cache'}{'_AUTO'} : $h_r->{'_AUTO'} ) : $h_r->{'_AUTO'})) {
232 # it's an auto lex, and this is an autoable key!
233 DEBUG and warn " Automaking \"$phrase\" into $h_r\n";
234 if ($handle->{'use_external_lex_cache'}) {
235 $value = $handle->{'_external_lex_cache'}{$phrase} = $handle->_compile($phrase);
236 }
237 else {
238 $value = $h_r->{$phrase} = $handle->_compile($phrase);
239 }
240 last;
241 }
242 DEBUG>1 and print " Not found in $h_r, nor automakable\n";
243 # else keep looking
f666394a 244 }
9378c581 245 }
f666394a
RGS
246
247 unless(defined($value)) {
ad56049d 248 DEBUG and warn "! Lookup of \"$phrase\" in/under ", ref($handle) || $handle, " fails.\n";
f666394a 249 if(ref($handle) and $handle->{'fail'}) {
ad56049d 250 DEBUG and warn "WARNING0: maketext fails looking for <$phrase>\n";
f666394a
RGS
251 my $fail;
252 if(ref($fail = $handle->{'fail'}) eq 'CODE') { # it's a sub reference
9961f4dd 253 $@ = $at; # Put $@ back in case we altered it along the way.
f666394a
RGS
254 return &{$fail}($handle, $phrase, @_);
255 # If it ever returns, it should return a good value.
256 }
257 else { # It's a method name
9961f4dd 258 $@ = $at; # Put $@ back in case we altered it along the way.
f666394a
RGS
259 return $handle->$fail($phrase, @_);
260 # If it ever returns, it should return a good value.
261 }
262 }
263 else {
264 # All we know how to do is this;
265 Carp::croak("maketext doesn't know how to say:\n$phrase\nas needed");
266 }
267 }
268
9961f4dd
TR
269 if(ref($value) eq 'SCALAR'){
270 $@ = $at; # Put $@ back in case we altered it along the way.
271 return $$value ;
272 }
273 if(ref($value) ne 'CODE'){
274 $@ = $at; # Put $@ back in case we altered it along the way.
275 return $value ;
276 }
f666394a
RGS
277
278 {
279 local $SIG{'__DIE__'};
280 eval { $value = &$value($handle, @_) };
281 }
282 # If we make it here, there was an exception thrown in the
283 # call to $value, and so scream:
284 if ($@) {
f666394a 285 # pretty up the error message
9961f4dd 286 $@ =~ s{\s+at\s+\(eval\s+\d+\)\s+line\s+(\d+)\.?\n?}
f666394a
RGS
287 {\n in bracket code [compiled line $1],}s;
288 #$err =~ s/\n?$/\n/s;
9961f4dd 289 Carp::croak "Error in maketexting \"$phrase\":\n$@ as used";
f666394a
RGS
290 # Rather unexpected, but suppose that the sub tried calling
291 # a method that didn't exist.
292 }
293 else {
9961f4dd 294 $@ = $at; # Put $@ back in case we altered it along the way.
f666394a 295 return $value;
9378c581 296 }
9961f4dd 297 $@ = $at; # Put $@ back in case we altered it along the way.
9378c581
JH
298}
299
300###########################################################################
301
302sub get_handle { # This is a constructor and, yes, it CAN FAIL.
f666394a
RGS
303 # Its class argument has to be the base class for the current
304 # application's l10n files.
305
306 my($base_class, @languages) = @_;
307 $base_class = ref($base_class) || $base_class;
308 # Complain if they use __PACKAGE__ as a project base class?
309
310 if( @languages ) {
ad56049d 311 DEBUG and warn 'Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
f666394a
RGS
312 if($USING_LANGUAGE_TAGS) { # An explicit language-list was given!
313 @languages =
314 map {; $_, I18N::LangTags::alternate_language_tags($_) }
315 # Catch alternation
316 map I18N::LangTags::locale2language_tag($_),
317 # If it's a lg tag, fine, pass thru (untainted)
318 # If it's a locale ID, try converting to a lg tag (untainted),
319 # otherwise nix it.
320 @languages;
ad56049d 321 DEBUG and warn 'Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
f666394a
RGS
322 }
323 }
324 else {
325 @languages = $base_class->_ambient_langprefs;
1189c245 326 }
f666394a
RGS
327
328 @languages = $base_class->_langtag_munging(@languages);
329
330 my %seen;
331 foreach my $module_name ( map { $base_class . '::' . $_ } @languages ) {
332 next unless length $module_name; # sanity
333 next if $seen{$module_name}++ # Already been here, and it was no-go
334 || !&_try_use($module_name); # Try to use() it, but can't it.
335 return($module_name->new); # Make it!
336 }
337
338 return undef; # Fail!
14be35aa
AMS
339}
340
341###########################################################################
342
343sub _langtag_munging {
f666394a
RGS
344 my($base_class, @languages) = @_;
345
346 # We have all these DEBUG statements because otherwise it's hard as hell
347 # to diagnose ifwhen something goes wrong.
348
ad56049d 349 DEBUG and warn 'Lgs1: ', map("<$_>", @languages), "\n";
f666394a
RGS
350
351 if($USING_LANGUAGE_TAGS) {
ad56049d 352 DEBUG and warn 'Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
f666394a
RGS
353 @languages = $base_class->_add_supers( @languages );
354
355 push @languages, I18N::LangTags::panic_languages(@languages);
ad56049d 356 DEBUG and warn "After adding panic languages:\n",
f666394a
RGS
357 ' Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
358
359 push @languages, $base_class->fallback_languages;
360 # You are free to override fallback_languages to return empty-list!
ad56049d 361 DEBUG and warn 'Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
f666394a
RGS
362
363 @languages = # final bit of processing to turn them into classname things
364 map {
365 my $it = $_; # copy
366 $it =~ tr<-A-Z><_a-z>; # lc, and turn - to _
367 $it =~ tr<_a-z0-9><>cd; # remove all but a-z0-9_
368 $it;
369 } @languages
370 ;
ad56049d 371 DEBUG and warn "Nearing end of munging:\n",
f666394a
RGS
372 ' Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
373 }
374 else {
ad56049d 375 DEBUG and warn "Bypassing language-tags.\n",
f666394a
RGS
376 ' Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
377 }
378
ad56049d 379 DEBUG and warn "Before adding fallback classes:\n",
f666394a
RGS
380 ' Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
381
382 push @languages, $base_class->fallback_language_classes;
383 # You are free to override that to return whatever.
384
ad56049d 385 DEBUG and warn "Finally:\n",
f666394a
RGS
386 ' Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
387
388 return @languages;
14be35aa
AMS
389}
390
391###########################################################################
392
393sub _ambient_langprefs {
f666394a 394 return I18N::LangTags::Detect::detect();
14be35aa
AMS
395}
396
397###########################################################################
398
399sub _add_supers {
f666394a
RGS
400 my($base_class, @languages) = @_;
401
402 if (!$MATCH_SUPERS) {
403 # Nothing
ad56049d 404 DEBUG and warn "Bypassing any super-matching.\n",
f666394a
RGS
405 ' Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
406
407 }
408 elsif( $MATCH_SUPERS_TIGHTLY ) {
ad56049d 409 DEBUG and warn "Before adding new supers tightly:\n",
f666394a
RGS
410 ' Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
411 @languages = I18N::LangTags::implicate_supers( @languages );
ad56049d 412 DEBUG and warn "After adding new supers tightly:\n",
f666394a
RGS
413 ' Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
414
415 }
416 else {
ad56049d 417 DEBUG and warn "Before adding supers to end:\n",
f666394a
RGS
418 ' Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
419 @languages = I18N::LangTags::implicate_supers_strictly( @languages );
ad56049d 420 DEBUG and warn "After adding supers to end:\n",
f666394a
RGS
421 ' Lgs@', __LINE__, ': ', map("<$_>", @languages), "\n";
422 }
423
424 return @languages;
9378c581
JH
425}
426
427###########################################################################
428#
429# This is where most people should stop reading.
430#
431###########################################################################
432
9378c581 433my %tried = ();
f666394a 434# memoization of whether we've used this module, or found it unusable.
9378c581
JH
435
436sub _try_use { # Basically a wrapper around "require Modulename"
f666394a
RGS
437 # "Many men have tried..." "They tried and failed?" "They tried and died."
438 return $tried{$_[0]} if exists $tried{$_[0]}; # memoization
439
440 my $module = $_[0]; # ASSUME sane module name!
441 { no strict 'refs';
364c63cf 442 no warnings 'once';
f666394a 443 return($tried{$module} = 1)
9e20cf84 444 if %{$module . '::Lexicon'} or @{$module . '::ISA'};
f666394a
RGS
445 # weird case: we never use'd it, but there it is!
446 }
447
ad56049d 448 DEBUG and warn " About to use $module ...\n";
9961f4dd
TR
449
450 local $SIG{'__DIE__'};
451 local $@;
452 eval "require $module"; # used to be "use $module", but no point in that.
453
f666394a 454 if($@) {
ad56049d 455 DEBUG and warn "Error using $module \: $@\n";
f666394a
RGS
456 return $tried{$module} = 0;
457 }
458 else {
ad56049d 459 DEBUG and warn " OK, $module is used\n";
f666394a
RGS
460 return $tried{$module} = 1;
461 }
9378c581
JH
462}
463
464#--------------------------------------------------------------------------
465
466sub _lex_refs { # report the lexicon references for this handle's class
f666394a
RGS
467 # returns an arrayREF!
468 no strict 'refs';
469 no warnings 'once';
470 my $class = ref($_[0]) || $_[0];
ad56049d 471 DEBUG and warn "Lex refs lookup on $class\n";
f666394a
RGS
472 return $isa_scan{$class} if exists $isa_scan{$class}; # memoization!
473
474 my @lex_refs;
475 my $seen_r = ref($_[1]) ? $_[1] : {};
476
477 if( defined( *{$class . '::Lexicon'}{'HASH'} )) {
478 push @lex_refs, *{$class . '::Lexicon'}{'HASH'};
ad56049d 479 DEBUG and warn '%' . $class . '::Lexicon contains ',
f666394a
RGS
480 scalar(keys %{$class . '::Lexicon'}), " entries\n";
481 }
482
483 # Implements depth(height?)-first recursive searching of superclasses.
484 # In hindsight, I suppose I could have just used Class::ISA!
485 foreach my $superclass (@{$class . '::ISA'}) {
ad56049d 486 DEBUG and warn " Super-class search into $superclass\n";
f666394a
RGS
487 next if $seen_r->{$superclass}++;
488 push @lex_refs, @{&_lex_refs($superclass, $seen_r)}; # call myself
489 }
490
491 $isa_scan{$class} = \@lex_refs; # save for next time
492 return \@lex_refs;
9378c581
JH
493}
494
495sub clear_isa_scan { %isa_scan = (); return; } # end on a note of simplicity!
496
87d86da5
TR
497#--------------------------------------------------------------------------
498
499sub _compile {
500 # This big scary routine compiles an entry.
501 # It returns either a coderef if there's brackety bits in this, or
502 # otherwise a ref to a scalar.
503
152eae84
TR
504 my $string_to_compile = $_[1]; # There are taint issues using regex on @_ - perlbug 60378,27344
505
506 # The while() regex is more expensive than this check on strings that don't need a compile.
507 # this op causes a ~2% speed hit for strings that need compile and a 250% speed improvement
508 # on strings that don't need compiling.
509 return \"$string_to_compile" if($string_to_compile !~ m/[\[~\]]/ms); # return a string ref if chars [~] are not in the string
510
87d86da5
TR
511 my $target = ref($_[0]) || $_[0];
512
513 my(@code);
514 my(@c) = (''); # "chunks" -- scratch.
515 my $call_count = 0;
516 my $big_pile = '';
517 {
518 my $in_group = 0; # start out outside a group
519 my($m, @params); # scratch
520
e9c9ffca 521 while($string_to_compile =~ # Iterate over chunks.
152eae84
TR
522 m/(
523 [^\~\[\]]+ # non-~[] stuff (Capture everything else here)
87d86da5
TR
524 |
525 ~. # ~[, ~], ~~, ~other
526 |
527 \[ # [ presumably opening a group
528 |
529 \] # ] presumably closing a group
530 |
531 ~ # terminal ~ ?
532 |
533 $
534 )/xgs
535 ) {
536 DEBUG>2 and warn qq{ "$1"\n};
537
538 if($1 eq '[' or $1 eq '') { # "[" or end
539 # Whether this is "[" or end, force processing of any
540 # preceding literal.
541 if($in_group) {
542 if($1 eq '') {
e9c9ffca 543 $target->_die_pointing($string_to_compile, 'Unterminated bracket group');
87d86da5
TR
544 }
545 else {
e9c9ffca 546 $target->_die_pointing($string_to_compile, 'You can\'t nest bracket groups');
87d86da5
TR
547 }
548 }
549 else {
550 if ($1 eq '') {
551 DEBUG>2 and warn " [end-string]\n";
552 }
553 else {
554 $in_group = 1;
555 }
e9c9ffca 556 die "How come \@c is empty?? in <$string_to_compile>" unless @c; # sanity
87d86da5
TR
557 if(length $c[-1]) {
558 # Now actually processing the preceding literal
559 $big_pile .= $c[-1];
560 if($USE_LITERALS and (
561 (ord('A') == 65)
562 ? $c[-1] !~ m/[^\x20-\x7E]/s
563 # ASCII very safe chars
564 : $c[-1] !~ m/[^ !"\#\$%&'()*+,\-.\/0-9:;<=>?\@A-Z[\\\]^_`a-z{|}~\x07]/s
565 # EBCDIC very safe chars
566 )) {
567 # normal case -- all very safe chars
568 $c[-1] =~ s/'/\\'/g;
569 push @code, q{ '} . $c[-1] . "',\n";
570 $c[-1] = ''; # reuse this slot
571 }
572 else {
573 push @code, ' $c[' . $#c . "],\n";
574 push @c, ''; # new chunk
575 }
576 }
577 # else just ignore the empty string.
578 }
579
580 }
581 elsif($1 eq ']') { # "]"
582 # close group -- go back in-band
583 if($in_group) {
584 $in_group = 0;
585
586 DEBUG>2 and warn " --Closing group [$c[-1]]\n";
587
588 # And now process the group...
589
590 if(!length($c[-1]) or $c[-1] =~ m/^\s+$/s) {
591 DEBUG>2 and warn " -- (Ignoring)\n";
592 $c[-1] = ''; # reset out chink
593 next;
594 }
595
596 #$c[-1] =~ s/^\s+//s;
597 #$c[-1] =~ s/\s+$//s;
598 ($m,@params) = split(/,/, $c[-1], -1); # was /\s*,\s*/
599
600 # A bit of a hack -- we've turned "~,"'s into DELs, so turn
601 # 'em into real commas here.
602 if (ord('A') == 65) { # ASCII, etc
603 foreach($m, @params) { tr/\x7F/,/ }
604 }
605 else { # EBCDIC (1047, 0037, POSIX-BC)
606 # Thanks to Peter Prymmer for the EBCDIC handling
607 foreach($m, @params) { tr/\x07/,/ }
608 }
609
610 # Special-case handling of some method names:
611 if($m eq '_*' or $m =~ m/^_(-?\d+)$/s) {
612 # Treat [_1,...] as [,_1,...], etc.
613 unshift @params, $m;
614 $m = '';
615 }
616 elsif($m eq '*') {
617 $m = 'quant'; # "*" for "times": "4 cars" is 4 times "cars"
618 }
619 elsif($m eq '#') {
620 $m = 'numf'; # "#" for "number": [#,_1] for "the number _1"
621 }
622
623 # Most common case: a simple, legal-looking method name
624 if($m eq '') {
625 # 0-length method name means to just interpolate:
626 push @code, ' (';
627 }
628 elsif($m =~ /^\w+(?:\:\:\w+)*$/s
629 and $m !~ m/(?:^|\:)\d/s
630 # exclude starting a (sub)package or symbol with a digit
631 ) {
632 # Yes, it even supports the demented (and undocumented?)
633 # $obj->Foo::bar(...) syntax.
634 $target->_die_pointing(
e9c9ffca 635 $string_to_compile, q{Can't use "SUPER::" in a bracket-group method},
87d86da5
TR
636 2 + length($c[-1])
637 )
638 if $m =~ m/^SUPER::/s;
639 # Because for SUPER:: to work, we'd have to compile this into
640 # the right package, and that seems just not worth the bother,
641 # unless someone convinces me otherwise.
642
643 push @code, ' $_[0]->' . $m . '(';
644 }
645 else {
646 # TODO: implement something? or just too icky to consider?
647 $target->_die_pointing(
e9c9ffca 648 $string_to_compile,
87d86da5
TR
649 "Can't use \"$m\" as a method name in bracket group",
650 2 + length($c[-1])
651 );
652 }
653
654 pop @c; # we don't need that chunk anymore
655 ++$call_count;
656
657 foreach my $p (@params) {
658 if($p eq '_*') {
659 # Meaning: all parameters except $_[0]
660 $code[-1] .= ' @_[1 .. $#_], ';
661 # and yes, that does the right thing for all @_ < 3
662 }
663 elsif($p =~ m/^_(-?\d+)$/s) {
664 # _3 meaning $_[3]
665 $code[-1] .= '$_[' . (0 + $1) . '], ';
666 }
667 elsif($USE_LITERALS and (
668 (ord('A') == 65)
669 ? $p !~ m/[^\x20-\x7E]/s
670 # ASCII very safe chars
671 : $p !~ m/[^ !"\#\$%&'()*+,\-.\/0-9:;<=>?\@A-Z[\\\]^_`a-z{|}~\x07]/s
672 # EBCDIC very safe chars
673 )) {
674 # Normal case: a literal containing only safe characters
675 $p =~ s/'/\\'/g;
676 $code[-1] .= q{'} . $p . q{', };
677 }
678 else {
679 # Stow it on the chunk-stack, and just refer to that.
680 push @c, $p;
681 push @code, ' $c[' . $#c . '], ';
682 }
683 }
684 $code[-1] .= "),\n";
685
686 push @c, '';
687 }
688 else {
e9c9ffca 689 $target->_die_pointing($string_to_compile, q{Unbalanced ']'});
87d86da5
TR
690 }
691
692 }
693 elsif(substr($1,0,1) ne '~') {
694 # it's stuff not containing "~" or "[" or "]"
695 # i.e., a literal blob
696 $c[-1] .= $1;
697
698 }
699 elsif($1 eq '~~') { # "~~"
700 $c[-1] .= '~';
701
702 }
703 elsif($1 eq '~[') { # "~["
704 $c[-1] .= '[';
705
706 }
707 elsif($1 eq '~]') { # "~]"
708 $c[-1] .= ']';
709
710 }
711 elsif($1 eq '~,') { # "~,"
712 if($in_group) {
713 # This is a hack, based on the assumption that no-one will actually
714 # want a DEL inside a bracket group. Let's hope that's it's true.
715 if (ord('A') == 65) { # ASCII etc
716 $c[-1] .= "\x7F";
717 }
718 else { # EBCDIC (cp 1047, 0037, POSIX-BC)
719 $c[-1] .= "\x07";
720 }
721 }
722 else {
723 $c[-1] .= '~,';
724 }
725
726 }
727 elsif($1 eq '~') { # possible only at string-end, it seems.
728 $c[-1] .= '~';
729
730 }
731 else {
732 # It's a "~X" where X is not a special character.
733 # Consider it a literal ~ and X.
734 $c[-1] .= $1;
735 }
736 }
737 }
738
739 if($call_count) {
740 undef $big_pile; # Well, nevermind that.
741 }
742 else {
743 # It's all literals! Ahwell, that can happen.
744 # So don't bother with the eval. Return a SCALAR reference.
745 return \$big_pile;
746 }
747
748 die q{Last chunk isn't null??} if @c and length $c[-1]; # sanity
749 DEBUG and warn scalar(@c), " chunks under closure\n";
750 if(@code == 0) { # not possible?
751 DEBUG and warn "Empty code\n";
752 return \'';
753 }
754 elsif(@code > 1) { # most cases, presumably!
755 unshift @code, "join '',\n";
756 }
757 unshift @code, "use strict; sub {\n";
758 push @code, "}\n";
759
760 DEBUG and warn @code;
761 my $sub = eval(join '', @code);
762 die "$@ while evalling" . join('', @code) if $@; # Should be impossible.
763 return $sub;
764}
765
766#--------------------------------------------------------------------------
767
768sub _die_pointing {
769 # This is used by _compile to throw a fatal error
770 my $target = shift; # class name
771 # ...leaving $_[0] the error-causing text, and $_[1] the error message
772
773 my $i = index($_[0], "\n");
774
775 my $pointy;
776 my $pos = pos($_[0]) - (defined($_[2]) ? $_[2] : 0) - 1;
777 if($pos < 1) {
778 $pointy = "^=== near there\n";
779 }
780 else { # we need to space over
781 my $first_tab = index($_[0], "\t");
782 if($pos > 2 and ( -1 == $first_tab or $first_tab > pos($_[0]))) {
783 # No tabs, or the first tab is harmlessly after where we will point to,
784 # AND we're far enough from the margin that we can draw a proper arrow.
785 $pointy = ('=' x $pos) . "^ near there\n";
786 }
787 else {
788 # tabs screw everything up!
789 $pointy = substr($_[0],0,$pos);
790 $pointy =~ tr/\t //cd;
c4a6f826 791 # make everything into whitespace, but preserving tabs
87d86da5
TR
792 $pointy .= "^=== near there\n";
793 }
794 }
795
796 my $errmsg = "$_[1], in\:\n$_[0]";
797
798 if($i == -1) {
799 # No newline.
800 $errmsg .= "\n" . $pointy;
801 }
802 elsif($i == (length($_[0]) - 1) ) {
803 # Already has a newline at end.
804 $errmsg .= $pointy;
805 }
806 else {
807 # don't bother with the pointy bit, I guess.
808 }
809 Carp::croak( "$errmsg via $target, as used" );
810}
811
9378c581 8121;