2 # Time-stamp: "2000-11-14 22:27:26 MST"
5 package Locale::Maketext;
7 use vars qw( @ISA $VERSION $MATCH_SUPERS $USING_LANGUAGE_TAGS
10 use I18N::LangTags 0.21 ();
12 #--------------------------------------------------------------------------
14 BEGIN { unless(defined &DEBUG) { *DEBUG = sub () {0} } }
15 # define the constant 'DEBUG' at compile-time
21 $USING_LANGUAGE_TAGS = 1;
22 # Turning this off is somewhat of a security risk in that little or no
23 # checking will be done on the legality of tokens passed to the
24 # eval("use $module_name") in _try_use. If you turn this off, you have
25 # to do your own taint checking.
27 $USE_LITERALS = 1 unless defined $USE_LITERALS;
28 # a hint for compiling bracket-notation things.
32 ###########################################################################
35 my($handle, $num, @forms) = @_;
37 return $num if @forms == 0; # what should this mean?
38 return $forms[2] if @forms > 2 and $num == 0; # special zeroth case
41 # Note that the formatting of $num is preserved.
42 return( $handle->numf($num) . ' ' . $handle->numerate($num, @forms) );
43 # Most human languages put the number phrase before the qualified phrase.
48 # return this lexical item in a form appropriate to this number
49 my($handle, $num, @forms) = @_;
52 return '' unless @forms;
53 if(@forms == 1) { # only the headword form specified
54 return $s ? $forms[0] : ($forms[0] . 's'); # very cheap hack.
55 } else { # sing and plural were specified
56 return $s ? $forms[0] : $forms[1];
60 #--------------------------------------------------------------------------
63 my($handle, $num) = @_[0,1];
64 if($num < 10_000_000_000 and $num > -10_000_000_000 and $num == int($num)) {
65 $num += 0; # Just use normal integer stringification.
66 # Specifically, don't let %G turn ten million into 1E+007
68 $num = CORE::sprintf("%G", $num);
69 # "CORE::" is there to avoid confusion with the above sub sprintf.
71 while( $num =~ s/^([-+]?\d+)(\d{3})/$1,$2/s ) {1} # right from perlfaq5
72 # The initial \d+ gobbles as many digits as it can, and then we
73 # backtrack so it un-eats the rightmost three, and then we
74 # insert the comma there.
76 $num =~ tr<.,><,.> if ref($handle) and $handle->{'numf_comma'};
77 # This is just a lame hack instead of using Number::Format
83 my($handle, $format, @params) = @_;
84 return CORE::sprintf($format, @params);
85 # "CORE::" is there to avoid confusion with myself!
88 #=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#
90 use integer; # vroom vroom... applies to the whole rest of the module
93 my $it = ref($_[0]) || $_[0];
94 return undef unless $it =~ m/([^':]+)(?:::)?$/s;
103 (ref($it) && $it->{'encoding'})
104 || "iso-8859-1" # Latin-1
108 #--------------------------------------------------------------------------
110 sub fallback_languages { return('i-default', 'en', 'en-US') }
112 sub fallback_language_classes { return () }
114 #--------------------------------------------------------------------------
116 sub fail_with { # an actual attribute method!
117 my($handle, @params) = @_;
118 return unless ref($handle);
119 $handle->{'fail'} = $params[0] if @params;
120 return $handle->{'fail'};
123 #--------------------------------------------------------------------------
125 sub failure_handler_auto {
126 # Meant to be used like:
127 # $handle->fail_with('failure_handler_auto')
129 my($handle, $phrase, @params) = @_;
130 $handle->{'failure_lex'} ||= {};
131 my $lex = $handle->{'failure_lex'};
134 $lex->{$phrase} ||= ($value = $handle->_compile($phrase));
136 # Dumbly copied from sub maketext:
138 local $SIG{'__DIE__'};
139 eval { $value = &$value($handle, @_) };
141 # If we make it here, there was an exception thrown in the
142 # call to $value, and so scream:
145 # pretty up the error message
146 $err =~ s<\s+at\s+\(eval\s+\d+\)\s+line\s+(\d+)\.?\n?>
147 <\n in bracket code [compiled line $1],>s;
148 #$err =~ s/\n?$/\n/s;
149 Carp::croak "Error in maketexting \"$phrase\":\n$err as used";
150 # Rather unexpected, but suppose that the sub tried calling
151 # a method that didn't exist.
157 #==========================================================================
161 my $class = ref($_[0]) || $_[0];
162 my $handle = bless {}, $class;
167 sub init { return } # no-op
169 ###########################################################################
172 # Remember, this can fail. Failure is controllable many ways.
173 Carp::croak "maketext requires at least one parameter" unless @_ > 1;
175 my($handle, $phrase) = splice(@_,0,2);
181 @{ $isa_scan{ref($handle) || $handle} || $handle->_lex_refs }
183 print "* Looking up \"$phrase\" in $h_r\n" if DEBUG;
184 if(exists $h_r->{$phrase}) {
185 print " Found \"$phrase\" in $h_r\n" if DEBUG;
186 unless(ref($value = $h_r->{$phrase})) {
187 # Nonref means it's not yet compiled. Compile and replace.
188 $value = $h_r->{$phrase} = $handle->_compile($value);
191 } elsif($phrase !~ m/^_/s and $h_r->{'_AUTO'}) {
192 # it's an auto lex, and this is an autoable key!
193 print " Automaking \"$phrase\" into $h_r\n" if DEBUG;
195 $value = $h_r->{$phrase} = $handle->_compile($phrase);
198 print " Not found in $h_r, nor automakable\n" if DEBUG > 1;
202 unless(defined($value)) {
203 print "! Lookup of \"$phrase\" in/under ", ref($handle) || $handle,
204 " fails.\n" if DEBUG;
205 if(ref($handle) and $handle->{'fail'}) {
206 print "WARNING0: maketext fails looking for <$phrase>\n" if DEBUG;
208 if(ref($fail = $handle->{'fail'}) eq 'CODE') { # it's a sub reference
209 return &{$fail}($handle, $phrase, @_);
210 # If it ever returns, it should return a good value.
211 } else { # It's a method name
212 return $handle->$fail($phrase, @_);
213 # If it ever returns, it should return a good value.
216 # All we know how to do is this;
217 Carp::croak("maketext doesn't know how to say:\n$phrase\nas needed");
221 return $$value if ref($value) eq 'SCALAR';
222 return $value unless ref($value) eq 'CODE';
225 local $SIG{'__DIE__'};
226 eval { $value = &$value($handle, @_) };
228 # If we make it here, there was an exception thrown in the
229 # call to $value, and so scream:
232 # pretty up the error message
233 $err =~ s<\s+at\s+\(eval\s+\d+\)\s+line\s+(\d+)\.?\n?>
234 <\n in bracket code [compiled line $1],>s;
235 #$err =~ s/\n?$/\n/s;
236 Carp::croak "Error in maketexting \"$phrase\":\n$err as used";
237 # Rather unexpected, but suppose that the sub tried calling
238 # a method that didn't exist.
244 ###########################################################################
246 sub get_handle { # This is a constructor and, yes, it CAN FAIL.
247 # Its class argument has to be the base class for the current
248 # application's l10n files.
249 my($base_class, @languages) = @_;
250 $base_class = ref($base_class) || $base_class;
251 # Complain if they use __PACKAGE__ as a project base class?
253 unless(@languages) { # Calling with no args is magical! wooo, magic!
254 if(length( $ENV{'REQUEST_METHOD'} || '' )) { # I'm a CGI
255 my $in = $ENV{'HTTP_ACCEPT_LANGUAGE'} || '';
256 # supposedly that works under mod_perl, too.
257 $in =~ s<\([\)]*\)><>g; # Kill parens'd things -- just a hack.
258 @languages = &I18N::LangTags::extract_language_tags($in) if length $in;
259 # ...which untaints, incidentally.
261 } else { # Not running as a CGI: try to puzzle out from the environment
262 if(length( $ENV{'LANG'} || '' )) {
263 push @languages, split m/[,:]/, $ENV{'LANG'};
264 # LANG can be only /one/ locale as far as I know, but what the hey.
266 if(length( $ENV{'LANGUAGE'} || '' )) {
267 push @languages, split m/[,:]/, $ENV{'LANGUAGE'};
269 print "Noting ENV LANG ", join(',', @languages),"\n" if DEBUG;
270 # Those are really locale IDs, but they get xlated a few lines down.
272 if(&_try_use('Win32::Locale')) {
273 # If we have that module installed...
274 push @languages, Win32::Locale::get_language()
275 if defined &Win32::Locale::get_language;
280 #------------------------------------------------------------------------
281 print "Lgs1: ", map("<$_>", @languages), "\n" if DEBUG;
283 if($USING_LANGUAGE_TAGS) {
284 @languages = map &I18N::LangTags::locale2language_tag($_), @languages;
285 # if it's a lg tag, fine, pass thru (untainted)
286 # if it's a locale ID, try converting to a lg tag (untainted),
289 push @languages, map I18N::LangTags::super_languages($_), @languages
292 @languages = map { $_, I18N::LangTags::alternate_language_tags($_) }
293 @languages; # catch alternation
295 push @languages, I18N::LangTags::panic_languages(@languages)
296 if defined &I18N::LangTags::panic_languages;
298 push @languages, $base_class->fallback_languages;
299 # You are free to override fallback_languages to return empty-list!
301 @languages = # final bit of processing:
304 $it =~ tr<-A-Z><_a-z>; # lc, and turn - to _
305 $it =~ tr<_a-z0-9><>cd; # remove all but a-z0-9_
310 print "Lgs2: ", map("<$_>", @languages), "\n" if DEBUG > 1;
312 push @languages, $base_class->fallback_language_classes;
313 # You are free to override that to return whatever.
317 foreach my $module_name ( map { $base_class . "::" . $_ } @languages )
319 next unless length $module_name; # sanity
320 next if $seen{$module_name}++ # Already been here, and it was no-go
321 || !&_try_use($module_name); # Try to use() it, but can't it.
322 return($module_name->new); # Make it!
325 return undef; # Fail!
328 ###########################################################################
330 # This is where most people should stop reading.
332 ###########################################################################
335 # This big scarp routine compiles an entry.
336 # It returns either a coderef if there's brackety bits in this, or
337 # otherwise a ref to a scalar.
339 my $target = ref($_[0]) || $_[0];
342 my(@c) = (''); # "chunks" -- scratch.
346 my $in_group = 0; # start out outside a group
347 my($m, @params); # scratch
349 while($_[1] =~ # Iterate over chunks.
351 [^\~\[\]]+ # non-~[] stuff
353 ~. # ~[, ~], ~~, ~other
355 \[ # [ presumably opening a group
357 \] # ] presumably closing a group
364 print " \"$1\"\n" if DEBUG > 2;
366 if($1 eq '[' or $1 eq '') { # "[" or end
367 # Whether this is "[" or end, force processing of any
371 $target->_die_pointing($_[1], "Unterminated bracket group");
373 $target->_die_pointing($_[1], "You can't nest bracket groups");
377 print " [end-string]\n" if DEBUG > 2;
381 die "How come \@c is empty?? in <$_[1]>" unless @c; # sanity
383 # Now actually processing the preceding literal
385 if($USE_LITERALS and (
387 ? $c[-1] !~ m<[^\x20-\x7E]>s
388 # ASCII very safe chars
389 : $c[-1] !~ m/[^ !"\#\$%&'()*+,\-.\/0-9:;<=>?\@A-Z[\\\]^_`a-z{|}~\x07]/s
390 # EBCDIC very safe chars
392 # normal case -- all very safe chars
394 push @code, q{ '} . $c[-1] . "',\n";
395 $c[-1] = ''; # reuse this slot
397 push @code, ' $c[' . $#c . "],\n";
398 push @c, ''; # new chunk
401 # else just ignore the empty string.
404 } elsif($1 eq ']') { # "]"
405 # close group -- go back in-band
409 print " --Closing group [$c[-1]]\n" if DEBUG > 2;
411 # And now process the group...
413 if(!length($c[-1]) or $c[-1] =~ m/^\s+$/s) {
414 DEBUG > 2 and print " -- (Ignoring)\n";
415 $c[-1] = ''; # reset out chink
419 #$c[-1] =~ s/^\s+//s;
420 #$c[-1] =~ s/\s+$//s;
421 ($m,@params) = split(",", $c[-1], -1); # was /\s*,\s*/
423 # A bit of a hack -- we've turned "~,"'s into DELs, so turn
424 # 'em into real commas here.
425 if (ord('A') == 65) { # ASCII, etc
426 foreach($m, @params) { tr/\x7F/,/ }
427 } else { # EBCDIC (1047, 0037, POSIX-BC)
428 # Thanks to Peter Prymmer for the EBCDIC handling
429 foreach($m, @params) { tr/\x07/,/ }
432 # Special-case handling of some method names:
433 if($m eq '_*' or $m =~ m<^_(-?\d+)$>s) {
434 # Treat [_1,...] as [,_1,...], etc.
438 $m = 'quant'; # "*" for "times": "4 cars" is 4 times "cars"
440 $m = 'numf'; # "#" for "number": [#,_1] for "the number _1"
443 # Most common case: a simple, legal-looking method name
445 # 0-length method name means to just interpolate:
447 } elsif($m =~ m<^\w+(?:\:\:\w+)*$>s
448 and $m !~ m<(?:^|\:)\d>s
449 # exclude starting a (sub)package or symbol with a digit
451 # Yes, it even supports the demented (and undocumented?)
452 # $obj->Foo::bar(...) syntax.
453 $target->_die_pointing(
454 $_[1], "Can't (yet?) use \"SUPER::\" in a bracket-group method",
457 if $m =~ m/^SUPER::/s;
458 # Because for SUPER:: to work, we'd have to compile this into
459 # the right package, and that seems just not worth the bother,
460 # unless someone convinces me otherwise.
462 push @code, ' $_[0]->' . $m . '(';
464 # TODO: implement something? or just too icky to consider?
465 $target->_die_pointing(
467 "Can't use \"$m\" as a method name in bracket group",
472 pop @c; # we don't need that chunk anymore
475 foreach my $p (@params) {
477 # Meaning: all parameters except $_[0]
478 $code[-1] .= ' @_[1 .. $#_], ';
479 # and yes, that does the right thing for all @_ < 3
480 } elsif($p =~ m<^_(-?\d+)$>s) {
482 $code[-1] .= '$_[' . (0 + $1) . '], ';
483 } elsif($USE_LITERALS and (
485 ? $p !~ m<[^\x20-\x7E]>s
486 # ASCII very safe chars
487 : $p !~ m/[^ !"\#\$%&'()*+,\-.\/0-9:;<=>?\@A-Z[\\\]^_`a-z{|}~\x07]/s
488 # EBCDIC very safe chars
490 # Normal case: a literal containing only safe characters
492 $code[-1] .= q{'} . $p . q{', };
494 # Stow it on the chunk-stack, and just refer to that.
496 push @code, ' $c[' . $#c . "], ";
503 $target->_die_pointing($_[1], "Unbalanced ']'");
506 } elsif(substr($1,0,1) ne '~') {
507 # it's stuff not containing "~" or "[" or "]"
508 # i.e., a literal blob
511 } elsif($1 eq '~~') { # "~~"
514 } elsif($1 eq '~[') { # "~["
517 } elsif($1 eq '~]') { # "~]"
520 } elsif($1 eq '~,') { # "~,"
522 # This is a hack, based on the assumption that no-one will actually
523 # want a DEL inside a bracket group. Let's hope that's it's true.
524 if (ord('A') == 65) { # ASCII etc
526 } else { # EBCDIC (cp 1047, 0037, POSIX-BC)
533 } elsif($1 eq '~') { # possible only at string-end, it seems.
537 # It's a "~X" where X is not a special character.
538 # Consider it a literal ~ and X.
545 undef $big_pile; # Well, nevermind that.
547 # It's all literals! Ahwell, that can happen.
548 # So don't bother with the eval. Return a SCALAR reference.
552 die "Last chunk isn't null??" if @c and length $c[-1]; # sanity
553 print scalar(@c), " chunks under closure\n" if DEBUG;
554 if(@code == 0) { # not possible?
555 print "Empty code\n" if DEBUG;
557 } elsif(@code > 1) { # most cases, presumably!
558 unshift @code, "join '',\n";
560 unshift @code, "use strict; sub {\n";
563 print @code if DEBUG;
564 my $sub = eval(join '', @code);
565 die "$@ while evalling" . join('', @code) if $@; # Should be impossible.
569 # - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
572 # This is used by _compile to throw a fatal error
573 my $target = shift; # class name
574 # ...leaving $_[0] the error-causing text, and $_[1] the error message
576 my $i = index($_[0], "\n");
579 my $pos = pos($_[0]) - (defined($_[2]) ? $_[2] : 0) - 1;
581 $pointy = "^=== near there\n";
582 } else { # we need to space over
583 my $first_tab = index($_[0], "\t");
584 if($pos > 2 and ( -1 == $first_tab or $first_tab > pos($_[0]))) {
585 # No tabs, or the first tab is harmlessly after where we will point to,
586 # AND we're far enough from the margin that we can draw a proper arrow.
587 $pointy = ('=' x $pos) . "^ near there\n";
589 # tabs screw everything up!
590 $pointy = substr($_[0],0,$pos);
591 $pointy =~ tr/\t //cd;
592 # make everything into whitespace, but preseving tabs
593 $pointy .= "^=== near there\n";
597 my $errmsg = "$_[1], in\:\n$_[0]";
601 $errmsg .= "\n" . $pointy;
602 } elsif($i == (length($_[0]) - 1) ) {
603 # Already has a newline at end.
606 # don't bother with the pointy bit, I guess.
608 Carp::croak( "$errmsg via $target, as used" );
611 ###########################################################################
614 # memoization of whether we've used this module, or found it unusable.
616 sub _try_use { # Basically a wrapper around "require Modulename"
617 # "Many men have tried..." "They tried and failed?" "They tried and died."
618 return $tried{$_[0]} if exists $tried{$_[0]}; # memoization
620 my $module = $_[0]; # ASSUME sane module name!
622 return($tried{$module} = 1)
623 if defined(%{$module . "::Lexicon"}) or defined(@{$module . "::ISA"});
624 # weird case: we never use'd it, but there it is!
627 print " About to use $module ...\n" if DEBUG;
629 local $SIG{'__DIE__'};
630 eval "require $module"; # used to be "use $module", but no point in that.
633 print "Error using $module \: $@\n" if DEBUG > 1;
634 return $tried{$module} = 0;
636 print " OK, $module is used\n" if DEBUG;
637 return $tried{$module} = 1;
641 #--------------------------------------------------------------------------
643 sub _lex_refs { # report the lexicon references for this handle's class
644 # returns an arrayREF!
646 my $class = ref($_[0]) || $_[0];
647 print "Lex refs lookup on $class\n" if DEBUG > 1;
648 return $isa_scan{$class} if exists $isa_scan{$class}; # memoization!
651 my $seen_r = ref($_[1]) ? $_[1] : {};
653 if( defined( *{$class . '::Lexicon'}{'HASH'} )) {
654 push @lex_refs, *{$class . '::Lexicon'}{'HASH'};
655 print "%" . $class . "::Lexicon contains ",
656 scalar(keys %{$class . '::Lexicon'}), " entries\n" if DEBUG;
659 # Implements depth(height?)-first recursive searching of superclasses.
660 # In hindsight, I suppose I could have just used Class::ISA!
661 foreach my $superclass (@{$class . "::ISA"}) {
662 print " Super-class search into $superclass\n" if DEBUG;
663 next if $seen_r->{$superclass}++;
664 push @lex_refs, @{&_lex_refs($superclass, $seen_r)}; # call myself
667 $isa_scan{$class} = \@lex_refs; # save for next time
671 sub clear_isa_scan { %isa_scan = (); return; } # end on a note of simplicity!
673 ###########################################################################