[perl5.git] / lib / utf8_heavy.pl

package utf8;
use strict;
use warnings;

sub DEBUG () { 0 }

sub DESTROY {}

my %Cache;

sub croak { require Carp; Carp::croak(@_) }

##
## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape.
## It's a data structure that encodes a set of Unicode characters.
##

{
    # If a floating point number is within this distance from the value of a
    # fraction, it is considered to be that fraction, even if many more digits
    # are specified that don't exactly match.
    my $min_floating_slop;

    # To guard against this program calling something that in turn ends up
    # calling this program with the same inputs, and hence infinitely
    # recursing, we keep a stack of the properties that are currently in
    # progress, pushed upon entry, popped upon return.
    my @recursed;

    sub SWASHNEW {
        my ($class, $type, $list, $minbits, $none) = @_;
        local $^D = 0 if $^D;

        $class = "" unless defined $class;
        print STDERR __LINE__, ": class=$class, type=$type, list=",
                                (defined $list) ? $list : ':undef:',
                                ", minbits=$minbits, none=$none\n" if DEBUG;

        ##
        ## Get the list of codepoints for the type.
        ## Called from swash_init (see utf8.c) or SWASHNEW itself.
        ##
        ## Callers of swash_init:
        ##     op.c:pmtrans             -- for tr/// and y///
        ##     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
        ##
        ## Given a $type, our goal is to fill $list with the set of codepoint
        ## ranges. If $type is false, $list passed is used.
        ##
        ## $minbits:
        ##     For binary properties, $minbits must be 1.
        ##     For character mappings (case and transliteration), $minbits must
        ##     be a number except 1.
        ##
        ## $list (or that filled according to $type):
        ##     Refer to perlunicode.pod, "User-Defined Character Properties."
        ##     
        ##     For binary properties, only characters with the property value
        ##     of True should be listed. The 3rd column, if any, will be ignored
        ##
        ## $none is undocumented, so I'm (khw) trying to do some documentation
        ## of it now.  It appears to be if there is a mapping in an input file
        ## that maps to 'XXXX', then that is replaced by $none+1, expressed in
        ## hexadecimal.  The only place I found it possibly used was in
        ## S_pmtrans in op.c.
        ##
        ## To make the parsing of $type clear, this code takes the a rather
        ## unorthodox approach of last'ing out of the block once we have the
        ## info we need. Were this to be a subroutine, the 'last' would just
        ## be a 'return'.
        ##
        my $file; ## file to load data from, and also part of the %Cache key.
        my $ListSorted = 0;

        # Change this to get a different set of Unicode tables
        my $unicore_dir = 'unicore';

        if ($type)
        {

            # Verify that this isn't a recursive call for this property.
            # Can't use croak, as it may try to recurse here itself.
            my $class_type = $class . "::$type";
            if (grep { $_ eq $class_type } @recursed) {
                CORE::die "panic: Infinite recursion in SWASHNEW for '$type'\n";
            }
            push @recursed, $class_type;

            $type =~ s/^\s+//;
            $type =~ s/\s+$//;

            # regcomp.c surrounds the property name with '__" and '_i' if this
            # is to be caseless matching.
            my $caseless = $type =~ s/^__(.*)_i$/$1/;

            print STDERR __LINE__, ": type=$type, caseless=$caseless\n" if DEBUG;

        GETFILE:
            {
                ##
                ## It could be a user-defined property.  Look in current
                ## package if no package given
                ##

                my $caller1 = $type =~ s/(.+)::// ? $1 : caller(1);

                if (defined $caller1 && $type =~ /^I[ns]\w+$/) {
                    my $prop = "${caller1}::$type";
                    if (exists &{$prop}) {
                        # stolen from Scalar::Util::PP::tainted()
                        my $tainted;
                        {
                            local($@, $SIG{__DIE__}, $SIG{__WARN__});
                            local $^W = 0;
                            no warnings;
                            eval { kill 0 * $prop };
                            $tainted = 1 if $@ =~ /^Insecure/;
                        }
                        die "Insecure user-defined property \\p{$prop}\n"
                            if $tainted;
                        no strict 'refs';
                        $list = &{$prop}($caseless);
                        last GETFILE;
                    }
                }

                # During Perl's compilation, this routine may be called before
                # the tables are constructed.  If so, we have a chicken/egg
                # problem.  If we die, the tables never get constructed, so
                # keep going, but return an empty table so only what the code
                # has compiled in internally (currently ASCII/Latin1 range
                # matching) will work.
                BEGIN {
                    # Poor man's constant, to avoid a run-time check.
                    $utf8::{miniperl}
                        = \! defined &DynaLoader::boot_DynaLoader;
                }
                if (miniperl) {
                    eval "require '$unicore_dir/Heavy.pl'";
                    last GETFILE if $@;
                }
                else {
                    require "$unicore_dir/Heavy.pl";
                }
                BEGIN { delete $utf8::{miniperl} }

                # All property names are matched caselessly
                my $property_and_table = lc $type;
                print STDERR __LINE__, ": $property_and_table\n" if DEBUG;

                # See if is of the compound form 'property=value', where the
                # value indicates the table we should use.
                my ($property, $table, @remainder) =
                                    split /\s*[:=]\s*/, $property_and_table, -1;
                if (@remainder) {
                    pop @recursed if @recursed;
                    return $type;
                }

                my $prefix;
                if (! defined $table) {
                        
                    # Here, is the single form.  The property becomes empty, and
                    # the whole value is the table.
                    $table = $property;
                    $prefix = $property = "";
                } else {
                    print STDERR __LINE__, ": $property\n" if DEBUG;

                    # Here it is the compound property=table form.  The property
                    # name is always loosely matched, which means remove any of
                    # these:
                    $property =~ s/[_\s-]//g;

                    # And convert to canonical form.  Quit if not valid.
                    $property = $utf8::loose_property_name_of{$property};
                    if (! defined $property) {
                        pop @recursed if @recursed;
                        return $type;
                    }

                    $prefix = "$property=";

                    # If the rhs looks like it is a number...
                    print STDERR __LINE__, ": table=$table\n" if DEBUG;
                    if ($table =~ qr{ ^ [ \s 0-9 _  + / . -]+ $ }x) {
                        print STDERR __LINE__, ": table=$table\n" if DEBUG;

                        # Don't allow leading nor trailing slashes 
                        if ($table =~ / ^ \/ | \/ $ /x) {
                            pop @recursed if @recursed;
                            return $type;
                        }

                        # Split on slash, in case it is a rational, like \p{1/5}
                        my @parts = split qr{ \s* / \s* }x, $table, -1;
                        print __LINE__, ": $type\n" if @parts > 2 && DEBUG;

                        # Can have maximum of one slash
                        if (@parts > 2) {
                            pop @recursed if @recursed;
                            return $type;
                        }

                        foreach my $part (@parts) {
                            print __LINE__, ": part=$part\n" if DEBUG;

                            $part =~ s/^\+\s*//;    # Remove leading plus
                            $part =~ s/^-\s*/-/;    # Remove blanks after unary
                                                    # minus

                            # Remove underscores between digits.
                            $part =~ s/( ?<= [0-9] ) _ (?= [0-9] ) //xg;

                            # No leading zeros (but don't make a single '0'
                            # into a null string)
                            $part =~ s/ ^ ( -? ) 0+ /$1/x;
                            $part .= '0' if $part eq '-' || $part eq "";

                            # No trailing zeros after a decimal point
                            $part =~ s/ ( \. .*? ) 0+ $ /$1/x;

                            # Begin with a 0 if a leading decimal point
                            $part =~ s/ ^ ( -? ) \. /${1}0./x;

                            # Ensure not a trailing decimal point: turn into an
                            # integer
                            $part =~ s/ \. $ //x;

                            print STDERR __LINE__, ": part=$part\n" if DEBUG;
                            #return $type if $part eq "";
                            
                            # Result better look like a number.  (This test is
                            # needed because, for example could have a plus in
                            # the middle.)
                            if ($part !~ / ^ -? [0-9]+ ( \. [0-9]+)? $ /x) {
                                pop @recursed if @recursed;
                                return $type;
                            }
                        }

                        #  If a rational...
                        if (@parts == 2) {

                            # If denominator is negative, get rid of it, and ...
                            if ($parts[1] =~ s/^-//) {

                                # If numerator is also negative, convert the
                                # whole thing to positive, or move the minus to
                                # the numerator
                                if ($parts[0] !~ s/^-//) {
                                    $parts[0] = '-' . $parts[0];
                                }
                            }
                            $table = join '/', @parts;
                        }
                        elsif ($property ne 'nv' || $parts[0] !~ /\./) {

                            # Here is not numeric value, or doesn't have a
                            # decimal point.  No further manipulation is
                            # necessary.  (Note the hard-coded property name.
                            # This could fail if other properties eventually
                            # had fractions as well; perhaps the cjk ones
                            # could evolve to do that.  This hard-coding could
                            # be fixed by mktables generating a list of
                            # properties that could have fractions.)
                            $table = $parts[0];
                        } else {

                            # Here is a floating point numeric_value.  Try to
                            # convert to rational.  First see if is in the list
                            # of known ones.
                            if (exists $utf8::nv_floating_to_rational{$parts[0]}) {
                                $table = $utf8::nv_floating_to_rational{$parts[0]};
                            } else {

                                # Here not in the list.  See if is close
                                # enough to something in the list.  First
                                # determine what 'close enough' means.  It has
                                # to be as tight as what mktables says is the
                                # maximum slop, and as tight as how many
                                # digits we were passed.  That is, if the user
                                # said .667, .6667, .66667, etc.  we match as
                                # many digits as they passed until get to
                                # where it doesn't matter any more due to the
                                # machine's precision.  If they said .6666668,
                                # we fail.
                                (my $fraction = $parts[0]) =~ s/^.*\.//;
                                my $epsilon = 10 ** - (length($fraction));
                                if ($epsilon > $utf8::max_floating_slop) {
                                    $epsilon = $utf8::max_floating_slop;
                                }

                                # But it can't be tighter than the minimum
                                # precision for this machine.  If haven't
                                # already calculated that minimum, do so now.
                                if (! defined $min_floating_slop) {

                                    # Keep going down an order of magnitude
                                    # until find that adding this quantity to
                                    # 1 remains 1; but put an upper limit on
                                    # this so in case this algorithm doesn't
                                    # work properly on some platform, that we
                                    # won't loop forever.
                                    my $count = 0;
                                    $min_floating_slop = 1;
                                    while (1+ $min_floating_slop != 1
                                           && $count++ < 50)
                                    {
                                        my $next = $min_floating_slop / 10;
                                        last if $next == 0; # If underflows,
                                                            # use previous one
                                        $min_floating_slop = $next;
                                        print STDERR __LINE__, ": min_float_slop=$min_floating_slop\n" if DEBUG;
                                    }

                                    # Back off a couple orders of magnitude,
                                    # just to be safe.
                                    $min_floating_slop *= 100;
                                }
                                    
                                if ($epsilon < $min_floating_slop) {
                                    $epsilon = $min_floating_slop;
                                }
                                print STDERR __LINE__, ": fraction=.$fraction; epsilon=$epsilon\n" if DEBUG;

                                undef $table;

                                # And for each possible rational in the table,
                                # see if it is within epsilon of the input.
                                foreach my $official
                                        (keys %utf8::nv_floating_to_rational)
                                {
                                    print STDERR __LINE__, ": epsilon=$epsilon, official=$official, diff=", abs($parts[0] - $official), "\n" if DEBUG;
                                    if (abs($parts[0] - $official) < $epsilon) {
                                      $table =
                                      $utf8::nv_floating_to_rational{$official};
                                        last;
                                    }
                                }

                                # Quit if didn't find one.
                                if (! defined $table) {
                                    pop @recursed if @recursed;
                                    return $type;
                                }
                            }
                        }
                        print STDERR __LINE__, ": $property=$table\n" if DEBUG;
                    }
                }

                # Combine lhs (if any) and rhs to get something that matches
                # the syntax of the lookups.
                $property_and_table = "$prefix$table";
                print STDERR __LINE__, ": $property_and_table\n" if DEBUG;

                # First try stricter matching.
                $file = $utf8::stricter_to_file_of{$property_and_table};

                # If didn't find it, try again with looser matching by editing
                # out the applicable characters on the rhs and looking up
                # again.
                if (! defined $file) {
                    $table =~ s/ [_\s-] //xg;
                    $property_and_table = "$prefix$table";
                    print STDERR __LINE__, ": $property_and_table\n" if DEBUG;
                    $file = $utf8::loose_to_file_of{$property_and_table};
                }

                # Add the constant and go fetch it in.
                if (defined $file) {
                    if ($utf8::why_deprecated{$file}) {
                        warnings::warnif('deprecated', "Use of '$type' in \\p{} or \\P{} is deprecated because: $utf8::why_deprecated{$file};");
                    }

                    if ($caseless
                        && exists $utf8::caseless_equivalent{$property_and_table})
                    {
                        $file = $utf8::caseless_equivalent{$property_and_table};
                    }
                    $file= "$unicore_dir/lib/$file.pl";
                    last GETFILE;
                }
                print STDERR __LINE__, ": didn't find $property_and_table\n" if DEBUG;

                ##
                ## See if it's a user-level "To".
                ##

                my $caller0 = caller(0);

                if (defined $caller0 && $type =~ /^To(?:\w+)$/) {
                    my $map = $caller0 . "::" . $type;

                    if (exists &{$map}) {
                        no strict 'refs';
                        
                        $list = &{$map};
                        warnings::warnif('deprecated', "User-defined case-mapping '$type' is deprecated");
                        last GETFILE;
                    }
                }

                ##
                ## Last attempt -- see if it's a standard "To" name
                ## (e.g. "ToLower")  ToTitle is used by ucfirst().
                ## The user-level way to access ToDigit() and ToFold()
                ## is to use Unicode::UCD.
                ##
                if ($type =~ /^To(Digit|Fold|Lower|Title|Upper)$/) {
                    $file = "$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
                ## out what to do with $type. Ouch.
                ##

                pop @recursed if @recursed;
                return $type;
            }

            if (defined $file) {
                print STDERR __LINE__, ": found it (file='$file')\n" if DEBUG;

                ##
                ## If we reach here, it was due to a 'last GETFILE' above
                ## (exception: user-defined properties and mappings), so we
                ## have a filename, so now we load it if we haven't already.
                ## If we have, return the cached results. The cache key is the
                ## class and file to load.
                ##
                my $found = $Cache{$class, $file};
                if ($found and ref($found) eq $class) {
                    print STDERR __LINE__, ": Returning cached '$file' for \\p{$type}\n" if DEBUG;
                    pop @recursed if @recursed;
                    return $found;
                }

                local $@;
                local $!;
                $list = do $file; die $@ if $@;
            }

            $ListSorted = 1; ## we know that these lists are sorted
        }

        my $extras;
        my $bits = $minbits;

        if ($list) {
            my $taint = substr($list,0,0); # maintain taint
            my @tmp = split(/^/m, $list);
            my %seen;
            no warnings;
            $extras = join '', $taint, grep /^[^0-9a-fA-F]/, @tmp;
            $list = join '', $taint,
                map  { $_->[1] }
                sort { $a->[0] <=> $b->[0] }
                map  { /^([0-9a-fA-F]+)/; [ CORE::hex($1), $_ ] }
                grep { /^([0-9a-fA-F]+)/ and not $seen{$1}++ } @tmp; # XXX doesn't do ranges right
        }

        if ($none) {
            my $hextra = sprintf "%04x", $none + 1;
            $list =~ s/\tXXXX$/\t$hextra/mg;
        }

        if ($minbits != 1 && $minbits < 32) { # not binary property
            my $top = 0;
            while ($list =~ /^([0-9a-fA-F]+)(?:[\t]([0-9a-fA-F]+)?)(?:[ \t]([0-9a-fA-F]+))?/mg) {
                my $min = CORE::hex $1;
                my $max = defined $2 ? CORE::hex $2 : $min;
                my $val = defined $3 ? CORE::hex $3 : 0;
                $val += $max - $min if defined $3;
                $top = $val if $val > $top;
            }
            my $topbits =
                $top > 0xffff ? 32 :
                $top > 0xff ? 16 : 8;
            $bits = $topbits if $bits < $topbits;
        }

        my @extras;
        if ($extras) {
            for my $x ($extras) {
                my $taint = substr($x,0,0); # maintain taint
                pos $x = 0;
                while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
                    my $char = "$1$taint";
                    my $name = "$2$taint";
                    print STDERR __LINE__, ": char [$char] => name [$name]\n"
                        if DEBUG;
                    if ($char =~ /[-+!&]/) {
                        my ($c,$t) = split(/::/, $name, 2);	# bogus use of ::, really
                        my $subobj;
                        if ($c eq 'utf8') {
                            $subobj = utf8->SWASHNEW($t, "", $minbits, 0);
                        }
                        elsif (exists &$name) {
                            $subobj = utf8->SWASHNEW($name, "", $minbits, 0);
                        }
                        elsif ($c =~ /^([0-9a-fA-F]+)/) {
                            $subobj = utf8->SWASHNEW("", $c, $minbits, 0);
                        }
                        if (! ref $subobj) {
                            pop @recursed if @recursed && $type;
                            return $subobj;
                        }
                        push @extras, $name => $subobj;
                        $bits = $subobj->{BITS} if $bits < $subobj->{BITS};
                    }
                }
            }
        }

        if (DEBUG) {
            print STDERR __LINE__, ": CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none";
            print STDERR "\nLIST =>\n$list" if defined $list;
            print STDERR "\nEXTRAS =>\n$extras" if defined $extras;
            print STDERR "\n";
        }

        my $SWASH = bless {
            TYPE => $type,
            BITS => $bits,
            EXTRAS => $extras,
            LIST => $list,
            NONE => $none,
            @extras,
        } => $class;

        if ($file) {
            $Cache{$class, $file} = $SWASH;
        }

        pop @recursed if @recursed && $type;

        return $SWASH;
    }
}

# Now SWASHGET is recasted into a C function S_swash_get (see utf8.c).

1;
Commit	Line	Data
a0ed51b3	1	package utf8;
cf25bb62 JH	2	use strict;
cf25bb62 JH	3	use warnings;
a0ed51b3	4
15732964	5	sub DEBUG () { 0 }
a0ed51b3 LW	6
	7	sub DESTROY {}
	8
5beb625e JH	9	my %Cache;
5beb625e JH	10
86916d66 JH	11	sub croak { require Carp; Carp::croak(@_) }
86916d66 JH	12
5beb625e	13	##
3a2263fe RGS	14	## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape.
3a2263fe RGS	15	## It's a data structure that encodes a set of Unicode characters.
5beb625e JH	16	##
5beb625e JH	17
99870f4d KW	18	{
	19	# If a floating point number is within this distance from the value of a
	20	# fraction, it is considered to be that fraction, even if many more digits
	21	# are specified that don't exactly match.
	22	my $min_floating_slop;
	23
3dd10fe8 KW	24	# To guard against this program calling something that in turn ends up
	25	# calling this program with the same inputs, and hence infinitely
	26	# recursing, we keep a stack of the properties that are currently in
	27	# progress, pushed upon entry, popped upon return.
	28	my @recursed;
	29
99870f4d KW	30	sub SWASHNEW {
	31	my ($class, $type, $list, $minbits, $none) = @_;
	32	local $^D = 0 if $^D;
	33
3dd10fe8	34	$class = "" unless defined $class;
5bcf08c4 KW	35	print STDERR __LINE__, ": class=$class, type=$type, list=",
	36	(defined $list) ? $list : ':undef:',
	37	", minbits=$minbits, none=$none\n" if DEBUG;
99870f4d KW	38
	39	##
	40	## Get the list of codepoints for the type.
	41	## Called from swash_init (see utf8.c) or SWASHNEW itself.
	42	##
	43	## Callers of swash_init:
	44	## op.c:pmtrans -- for tr/// and y///
	45	## regexec.c:regclass_swash -- for /[]/, \p, and \P
	46	## utf8.c:is_utf8_common -- for common Unicode properties
	47	## utf8.c:to_utf8_case -- for lc, uc, ucfirst, etc. and //i
	48	##
	49	## Given a $type, our goal is to fill $list with the set of codepoint
	50	## ranges. If $type is false, $list passed is used.
	51	##
	52	## $minbits:
	53	## For binary properties, $minbits must be 1.
	54	## For character mappings (case and transliteration), $minbits must
	55	## be a number except 1.
	56	##
	57	## $list (or that filled according to $type):
	58	## Refer to perlunicode.pod, "User-Defined Character Properties."
	59	##
	60	## For binary properties, only characters with the property value
	61	## of True should be listed. The 3rd column, if any, will be ignored
	62	##
	63	## $none is undocumented, so I'm (khw) trying to do some documentation
	64	## of it now. It appears to be if there is a mapping in an input file
	65	## that maps to 'XXXX', then that is replaced by $none+1, expressed in
	66	## hexadecimal. The only place I found it possibly used was in
	67	## S_pmtrans in op.c.
	68	##
	69	## To make the parsing of $type clear, this code takes the a rather
	70	## unorthodox approach of last'ing out of the block once we have the
	71	## info we need. Were this to be a subroutine, the 'last' would just
	72	## be a 'return'.
	73	##
	74	my $file; ## file to load data from, and also part of the %Cache key.
	75	my $ListSorted = 0;
	76
	77	# Change this to get a different set of Unicode tables
	78	my $unicore_dir = 'unicore';
	79
	80	if ($type)
cf25bb62	81	{
3dd10fe8 KW	82
	83	# Verify that this isn't a recursive call for this property.
	84	# Can't use croak, as it may try to recurse here itself.
	85	my $class_type = $class . "::$type";
	86	if (grep { $_ eq $class_type } @recursed) {
	87	CORE::die "panic: Infinite recursion in SWASHNEW for '$type'\n";
	88	}
	89	push @recursed, $class_type;
	90
99870f4d KW	91	$type =~ s/^\s+//;
	92	$type =~ s/\s+$//;
	93
4ff124d1 KW	94	# regcomp.c surrounds the property name with '__" and '_i' if this
	95	# is to be caseless matching.
	96	my $caseless = $type =~ s/^__(.*)_i$/$1/;
	97
	98	print STDERR __LINE__, ": type=$type, caseless=$caseless\n" if DEBUG;
99870f4d KW	99
	100	GETFILE:
	101	{
	102	##
a44f4b74 KW	103	## It could be a user-defined property. Look in current
a44f4b74 KW	104	## package if no package given
99870f4d KW	105	##
	106
	107	my $caller1 = $type =~ s/(.+)::// ? $1 : caller(1);
	108
d658a8a8	109	if (defined $caller1 && $type =~ /^I[ns]\w+$/) {
99870f4d KW	110	my $prop = "${caller1}::$type";
99870f4d KW	111	if (exists &{$prop}) {
0e9be77f DM	112	# stolen from Scalar::Util::PP::tainted()
	113	my $tainted;
	114	{
	115	local($@, $SIG{__DIE__}, $SIG{__WARN__});
	116	local $^W = 0;
	117	no warnings;
	118	eval { kill 0 * $prop };
	119	$tainted = 1 if $@ =~ /^Insecure/;
	120	}
	121	die "Insecure user-defined property \\p{$prop}\n"
	122	if $tainted;
99870f4d	123	no strict 'refs';
4ff124d1	124	$list = &{$prop}($caseless);
99870f4d KW	125	last GETFILE;
	126	}
	127	}
	128
a44f4b74 KW	129	# During Perl's compilation, this routine may be called before
	130	# the tables are constructed. If so, we have a chicken/egg
	131	# problem. If we die, the tables never get constructed, so
	132	# keep going, but return an empty table so only what the code
	133	# has compiled in internally (currently ASCII/Latin1 range
	134	# matching) will work.
1363cbd0 FC	135	BEGIN {
	136	# Poor man's constant, to avoid a run-time check.
	137	$utf8::{miniperl}
	138	= \! defined &DynaLoader::boot_DynaLoader;
	139	}
	140	if (miniperl) {
a44f4b74 KW	141	eval "require '$unicore_dir/Heavy.pl'";
	142	last GETFILE if $@;
	143	}
	144	else {
	145	require "$unicore_dir/Heavy.pl";
	146	}
1363cbd0	147	BEGIN { delete $utf8::{miniperl} }
99870f4d	148
4ff124d1	149	# All property names are matched caselessly
99870f4d KW	150	my $property_and_table = lc $type;
	151	print STDERR __LINE__, ": $property_and_table\n" if DEBUG;
	152
	153	# See if is of the compound form 'property=value', where the
	154	# value indicates the table we should use.
	155	my ($property, $table, @remainder) =
	156	split /\s[:=]\s/, $property_and_table, -1;
3dd10fe8 KW	157	if (@remainder) {
	158	pop @recursed if @recursed;
	159	return $type;
	160	}
99870f4d KW	161
	162	my $prefix;
	163	if (! defined $table) {
	164
	165	# Here, is the single form. The property becomes empty, and
	166	# the whole value is the table.
	167	$table = $property;
	168	$prefix = $property = "";
	169	} else {
	170	print STDERR __LINE__, ": $property\n" if DEBUG;
	171
	172	# Here it is the compound property=table form. The property
	173	# name is always loosely matched, which means remove any of
	174	# these:
	175	$property =~ s/[_\s-]//g;
	176
	177	# And convert to canonical form. Quit if not valid.
	178	$property = $utf8::loose_property_name_of{$property};
3dd10fe8 KW	179	if (! defined $property) {
	180	pop @recursed if @recursed;
	181	return $type;
	182	}
99870f4d KW	183
	184	$prefix = "$property=";
	185
	186	# If the rhs looks like it is a number...
	187	print STDERR __LINE__, ": table=$table\n" if DEBUG;
	188	if ($table =~ qr{ ^ [ \s 0-9 _ + / . -]+ $ }x) {
	189	print STDERR __LINE__, ": table=$table\n" if DEBUG;
	190
	191	# Don't allow leading nor trailing slashes
3dd10fe8 KW	192	if ($table =~ / ^ \/ \| \/ $ /x) {
	193	pop @recursed if @recursed;
	194	return $type;
	195	}
99870f4d KW	196
	197	# Split on slash, in case it is a rational, like \p{1/5}
	198	my @parts = split qr{ \s* / \s* }x, $table, -1;
	199	print __LINE__, ": $type\n" if @parts > 2 && DEBUG;
	200
	201	# Can have maximum of one slash
3dd10fe8 KW	202	if (@parts > 2) {
	203	pop @recursed if @recursed;
	204	return $type;
	205	}
99870f4d KW	206
	207	foreach my $part (@parts) {
	208	print __LINE__, ": part=$part\n" if DEBUG;
	209
	210	$part =~ s/^\+\s*//; # Remove leading plus
	211	$part =~ s/^-\s*/-/; # Remove blanks after unary
	212	# minus
	213
	214	# Remove underscores between digits.
	215	$part =~ s/( ?<= [0-9] ) _ (?= [0-9] ) //xg;
	216
	217	# No leading zeros (but don't make a single '0'
	218	# into a null string)
	219	$part =~ s/ ^ ( -? ) 0+ /$1/x;
	220	$part .= '0' if $part eq '-' \|\| $part eq "";
	221
	222	# No trailing zeros after a decimal point
	223	$part =~ s/ ( \. .*? ) 0+ $ /$1/x;
	224
	225	# Begin with a 0 if a leading decimal point
	226	$part =~ s/ ^ ( -? ) \. /${1}0./x;
	227
	228	# Ensure not a trailing decimal point: turn into an
	229	# integer
	230	$part =~ s/ \. $ //x;
	231
	232	print STDERR __LINE__, ": part=$part\n" if DEBUG;
	233	#return $type if $part eq "";
	234
	235	# Result better look like a number. (This test is
	236	# needed because, for example could have a plus in
	237	# the middle.)
3dd10fe8 KW	238	if ($part !~ / ^ -? [0-9]+ ( \. [0-9]+)? $ /x) {
	239	pop @recursed if @recursed;
	240	return $type;
	241	}
99870f4d KW	242	}
	243
	244	# If a rational...
	245	if (@parts == 2) {
	246
	247	# If denominator is negative, get rid of it, and ...
	248	if ($parts[1] =~ s/^-//) {
	249
	250	# If numerator is also negative, convert the
	251	# whole thing to positive, or move the minus to
	252	# the numerator
	253	if ($parts[0] !~ s/^-//) {
	254	$parts[0] = '-' . $parts[0];
	255	}
	256	}
	257	$table = join '/', @parts;
	258	}
	259	elsif ($property ne 'nv' \|\| $parts[0] !~ /\./) {
	260
	261	# Here is not numeric value, or doesn't have a
	262	# decimal point. No further manipulation is
	263	# necessary. (Note the hard-coded property name.
	264	# This could fail if other properties eventually
	265	# had fractions as well; perhaps the cjk ones
	266	# could evolve to do that. This hard-coding could
	267	# be fixed by mktables generating a list of
	268	# properties that could have fractions.)
	269	$table = $parts[0];
	270	} else {
	271
	272	# Here is a floating point numeric_value. Try to
	273	# convert to rational. First see if is in the list
	274	# of known ones.
	275	if (exists $utf8::nv_floating_to_rational{$parts[0]}) {
	276	$table = $utf8::nv_floating_to_rational{$parts[0]};
	277	} else {
	278
	279	# Here not in the list. See if is close
	280	# enough to something in the list. First
	281	# determine what 'close enough' means. It has
	282	# to be as tight as what mktables says is the
	283	# maximum slop, and as tight as how many
	284	# digits we were passed. That is, if the user
	285	# said .667, .6667, .66667, etc. we match as
	286	# many digits as they passed until get to
	287	# where it doesn't matter any more due to the
	288	# machine's precision. If they said .6666668,
	289	# we fail.
	290	(my $fraction = $parts[0]) =~ s/^.*\.//;
	291	my $epsilon = 10 ** - (length($fraction));
	292	if ($epsilon > $utf8::max_floating_slop) {
	293	$epsilon = $utf8::max_floating_slop;
	294	}
	295
	296	# But it can't be tighter than the minimum
	297	# precision for this machine. If haven't
	298	# already calculated that minimum, do so now.
	299	if (! defined $min_floating_slop) {
	300
	301	# Keep going down an order of magnitude
	302	# until find that adding this quantity to
	303	# 1 remains 1; but put an upper limit on
	304	# this so in case this algorithm doesn't
	305	# work properly on some platform, that we
306	# won't loop forever.
307	my $count = 0;
308	$min_floating_slop = 1;
309	while (1+ $min_floating_slop != 1
310	&& $count++ < 50)
311	{
312	my $next = $min_floating_slop / 10;
313	last if $next == 0; # If underflows,
314	# use previous one
315	$min_floating_slop = $next;
316	print STDERR __LINE__, ": min_float_slop=$min_floating_slop\n" if DEBUG;
317	}
318
319	# Back off a couple orders of magnitude,
320	# just to be safe.
321	$min_floating_slop *= 100;
322	}
323
324	if ($epsilon < $min_floating_slop) {
325	$epsilon = $min_floating_slop;
326	}
327	print STDERR __LINE__, ": fraction=.$fraction; epsilon=$epsilon\n" if DEBUG;
328
329	undef $table;
330
331	# And for each possible rational in the table,
332	# see if it is within epsilon of the input.
333	foreach my $official
334	(keys %utf8::nv_floating_to_rational)
335	{
336	print STDERR __LINE__, ": epsilon=$epsilon, official=$official, diff=", abs($parts[0] - $official), "\n" if DEBUG;
337	if (abs($parts[0] - $official) < $epsilon) {
338	$table =
339	$utf8::nv_floating_to_rational{$official};
340	last;
341	}
342	}
343
344	# Quit if didn't find one.
3dd10fe8 KW	345	if (! defined $table) {
	346	pop @recursed if @recursed;
	347	return $type;
	348	}
99870f4d KW	349	}
	350	}
	351	print STDERR __LINE__, ": $property=$table\n" if DEBUG;
	352	}
	353	}
	354
	355	# Combine lhs (if any) and rhs to get something that matches
	356	# the syntax of the lookups.
	357	$property_and_table = "$prefix$table";
	358	print STDERR __LINE__, ": $property_and_table\n" if DEBUG;
	359
	360	# First try stricter matching.
	361	$file = $utf8::stricter_to_file_of{$property_and_table};
	362
	363	# If didn't find it, try again with looser matching by editing
	364	# out the applicable characters on the rhs and looking up
	365	# again.
	366	if (! defined $file) {
	367	$table =~ s/ [_\s-] //xg;
	368	$property_and_table = "$prefix$table";
	369	print STDERR __LINE__, ": $property_and_table\n" if DEBUG;
	370	$file = $utf8::loose_to_file_of{$property_and_table};
	371	}
	372
	373	# Add the constant and go fetch it in.
	374	if (defined $file) {
	375	if ($utf8::why_deprecated{$file}) {
	376	warnings::warnif('deprecated', "Use of '$type' in \\p{} or \\P{} is deprecated because: $utf8::why_deprecated{$file};");
	377	}
4ff124d1 KW	378
	379	if ($caseless
	380	&& exists $utf8::caseless_equivalent{$property_and_table})
	381	{
	382	$file = $utf8::caseless_equivalent{$property_and_table};
	383	}
99870f4d KW	384	$file= "$unicore_dir/lib/$file.pl";
	385	last GETFILE;
	386	}
	387	print STDERR __LINE__, ": didn't find $property_and_table\n" if DEBUG;
	388
	389	##
	390	## See if it's a user-level "To".
	391	##
	392
	393	my $caller0 = caller(0);
	394
	395	if (defined $caller0 && $type =~ /^To(?:\w+)$/) {
	396	my $map = $caller0 . "::" . $type;
	397
	398	if (exists &{$map}) {
	399	no strict 'refs';
	400
	401	$list = &{$map};
3a3263a0	402	warnings::warnif('deprecated', "User-defined case-mapping '$type' is deprecated");
99870f4d KW	403	last GETFILE;
	404	}
	405	}
	406
	407	##
	408	## Last attempt -- see if it's a standard "To" name
	409	## (e.g. "ToLower") ToTitle is used by ucfirst().
	410	## The user-level way to access ToDigit() and ToFold()
	411	## is to use Unicode::UCD.
	412	##
	413	if ($type =~ /^To(Digit\|Fold\|Lower\|Title\|Upper)$/) {
	414	$file = "$unicore_dir/To/$1.pl";
	415	## would like to test to see if $file actually exists....
	416	last GETFILE;
	417	}
	418
	419	##
	420	## If we reach this line, it's because we couldn't figure
	421	## out what to do with $type. Ouch.
	422	##
	423
3dd10fe8	424	pop @recursed if @recursed;
99870f4d	425	return $type;
cf25bb62 JH	426	}
cf25bb62 JH	427
99870f4d KW	428	if (defined $file) {
	429	print STDERR __LINE__, ": found it (file='$file')\n" if DEBUG;
	430
	431	##
	432	## If we reach here, it was due to a 'last GETFILE' above
	433	## (exception: user-defined properties and mappings), so we
	434	## have a filename, so now we load it if we haven't already.
	435	## If we have, return the cached results. The cache key is the
	436	## class and file to load.
	437	##
	438	my $found = $Cache{$class, $file};
	439	if ($found and ref($found) eq $class) {
	440	print STDERR __LINE__, ": Returning cached '$file' for \\p{$type}\n" if DEBUG;
3dd10fe8	441	pop @recursed if @recursed;
99870f4d KW	442	return $found;
	443	}
	444
	445	local $@;
	446	local $!;
	447	$list = do $file; die $@ if $@;
0a0ffbce	448	}
cf25bb62	449
99870f4d	450	$ListSorted = 1; ## we know that these lists are sorted
cf25bb62 JH	451	}
cf25bb62 JH	452
99870f4d KW	453	my $extras;
	454	my $bits = $minbits;
	455
99870f4d	456	if ($list) {
0e9be77f	457	my $taint = substr($list,0,0); # maintain taint
99870f4d KW	458	my @tmp = split(/^/m, $list);
	459	my %seen;
	460	no warnings;
0e9be77f DM	461	$extras = join '', $taint, grep /^[^0-9a-fA-F]/, @tmp;
0e9be77f DM	462	$list = join '', $taint,
99870f4d KW	463	map { $_->[1] }
	464	sort { $a->[0] <=> $b->[0] }
	465	map { /^([0-9a-fA-F]+)/; [ CORE::hex($1), $_ ] }
	466	grep { /^([0-9a-fA-F]+)/ and not $seen{$1}++ } @tmp; # XXX doesn't do ranges right
	467	}
a0ed51b3	468
99870f4d KW	469	if ($none) {
	470	my $hextra = sprintf "%04x", $none + 1;
	471	$list =~ s/\tXXXX$/\t$hextra/mg;
	472	}
a0ed51b3	473
99870f4d KW	474	if ($minbits != 1 && $minbits < 32) { # not binary property
	475	my $top = 0;
	476	while ($list =~ /^([0-9a-fA-F]+)(?:[\t]([0-9a-fA-F]+)?)(?:[ \t]([0-9a-fA-F]+))?/mg) {
	477	my $min = CORE::hex $1;
	478	my $max = defined $2 ? CORE::hex $2 : $min;
	479	my $val = defined $3 ? CORE::hex $3 : 0;
	480	$val += $max - $min if defined $3;
	481	$top = $val if $val > $top;
	482	}
	483	my $topbits =
	484	$top > 0xffff ? 32 :
	485	$top > 0xff ? 16 : 8;
	486	$bits = $topbits if $bits < $topbits;
	487	}
a0ed51b3	488
99870f4d KW	489	my @extras;
	490	if ($extras) {
	491	for my $x ($extras) {
0e9be77f	492	my $taint = substr($x,0,0); # maintain taint
99870f4d KW	493	pos $x = 0;
99870f4d KW	494	while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
0e9be77f DM	495	my $char = "$1$taint";
	496	my $name = "$2$taint";
	497	print STDERR __LINE__, ": char [$char] => name [$name]\n"
	498	if DEBUG;
99870f4d KW	499	if ($char =~ /[-+!&]/) {
	500	my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really
	501	my $subobj;
	502	if ($c eq 'utf8') {
	503	$subobj = utf8->SWASHNEW($t, "", $minbits, 0);
	504	}
	505	elsif (exists &$name) {
	506	$subobj = utf8->SWASHNEW($name, "", $minbits, 0);
	507	}
	508	elsif ($c =~ /^([0-9a-fA-F]+)/) {
	509	$subobj = utf8->SWASHNEW("", $c, $minbits, 0);
	510	}
3dd10fe8 KW	511	if (! ref $subobj) {
	512	pop @recursed if @recursed && $type;
	513	return $subobj;
	514	}
99870f4d KW	515	push @extras, $name => $subobj;
	516	$bits = $subobj->{BITS} if $bits < $subobj->{BITS};
	517	}
	518	}
	519	}
	520	}
a0ed51b3	521
99870f4d KW	522	if (DEBUG) {
	523	print STDERR __LINE__, ": CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none";
	524	print STDERR "\nLIST =>\n$list" if defined $list;
	525	print STDERR "\nEXTRAS =>\n$extras" if defined $extras;
	526	print STDERR "\n";
	527	}
a0ed51b3	528
99870f4d KW	529	my $SWASH = bless {
	530	TYPE => $type,
	531	BITS => $bits,
	532	EXTRAS => $extras,
	533	LIST => $list,
	534	NONE => $none,
	535	@extras,
	536	} => $class;
	537
	538	if ($file) {
	539	$Cache{$class, $file} = $SWASH;
	540	}
5beb625e	541
3dd10fe8 KW	542	pop @recursed if @recursed && $type;
3dd10fe8 KW	543
99870f4d	544	return $SWASH;
5beb625e	545	}
a0ed51b3 LW	546	}
a0ed51b3 LW	547
979f2922	548	# Now SWASHGET is recasted into a C function S_swash_get (see utf8.c).
a0ed51b3 LW	549
a0ed51b3 LW	550	1;