use warnings;
sub DEBUG () { 0 }
+$|=1 if DEBUG;
sub DESTROY {}
my $loose = $_[0] =~ s/[-\s_]//rg;
- return $loose if $loose !~ / ^ (?: is )? l $/x;
+ return $loose if $loose !~ / ^ (?: is | to )? l $/x;
return 'l_' if $_[0] =~ / l .* _ /x; # If original had a trailing '_'
return $loose;
}
## regexec.c:regclass_swash -- for /[]/, \p, and \P
## utf8.c:is_utf8_common -- for common Unicode properties
## utf8.c:to_utf8_case -- for lc, uc, ucfirst, etc. and //i
+ ## Unicode::UCD::prop_invlist
+ ## Unicode::UCD::prop_invmap
##
## Given a $type, our goal is to fill $list with the set of codepoint
## ranges. If $type is false, $list passed is used.
## The user-level way to access ToDigit() and ToFold()
## is to use Unicode::UCD.
##
- if ($type =~ /^To(Digit|Fold|Lower|Title|Upper)$/) {
+ # Only check if caller wants non-binary
+ my $retried = 0;
+ if ($minbits != 1 && $property_and_table =~ s/^to//) {{
+ # Look input up in list of properties for which we have
+ # mapping files.
+ if (defined ($file =
+ $utf8::loose_property_to_file_of{$property_and_table}))
+ {
+ $type = $utf8::file_to_swash_name{$file};
+ print STDERR __LINE__, ": type set to $type\n" if DEBUG;
+ $file = "$unicore_dir/$file.pl";
+ last GETFILE;
+ } # If that fails see if there is a corresponding binary
+ # property file
+ elsif (defined ($file =
+ $utf8::loose_to_file_of{$property_and_table}))
+ {
- # Fail if wanting a binary property, as these aren't.
- if ($minbits == 1) {
- pop @recursed if @recursed;
- return $type;
+ # Here, there is no map file for the property we are
+ # trying to get the map of, but this is a binary
+ # property, and there is a file for it that can easily
+ # be translated to a mapping.
+
+ # In the case of properties that are forced to binary,
+ # they are a combination. We return the actual
+ # mapping instead of the binary. If the input is
+ # something like 'Tocjkkiicore', it will be found in
+ # %loose_property_to_file_of above as => 'To/kIICore'.
+ # But the form like ToIskiicore won't be. To fix
+ # this, it was easiest to do it here. These
+ # properties are the complements of the default
+ # property, so there is an entry in %loose_to_file_of
+ # that is 'iskiicore' => '!kIICore/N', If we find such
+ # an entry, strip off things and try again, which
+ # should find the entry in %loose_property_to_file_of.
+ # Actual binary properties that are of this form, such
+ # as this entry: 'ishrkt' => '!Perl/Any' will also be
+ # retried, but won't be in %loose_property_to_file_of,
+ # and instead the next time through, it will find
+ # 'hrkt' => '!Perl/Any' and proceed.
+ redo if ! $retried
+ && $file =~ /^!/
+ && $property_and_table =~ s/^is//;
+
+ # This is a binary property. Setting this here causes
+ # it to be stored as such in the cache, so if someone
+ # comes along later looking for just a binary, they
+ # get it.
+ $minbits = 1;
+
+ $invert_it = $file =~ s/^!//;
+ $file = "$unicore_dir/lib/$file.pl";
+ last GETFILE;
}
- $file = "$unicore_dir/To/$1.pl";
- ## would like to test to see if $file actually exists....
- last GETFILE;
- }
+ } }
##
## If we reach this line, it's because we couldn't figure
pop @recursed if @recursed;
return $type;
- }
+ } # end of GETFILE block
if (defined $file) {
print STDERR __LINE__, ": found it (file='$file')\n" if DEBUG;
}
$ListSorted = 1; ## we know that these lists are sorted
- }
+ } # End of $type is non-null
+
+ # Here, either $type was null, or we found the requested property and
+ # read it into $list
my $extras;
+
my $bits = $minbits;
if ($list) {
}
push @extras, $name => $subobj;
$bits = $subobj->{BITS} if $bits < $subobj->{BITS};
+ $user_defined = $subobj->{USER_DEFINED}
+ if $subobj->{USER_DEFINED};
}
}
}