X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/3dd10fe8450c405da35678d844f7a7fbb1dafc9f..50de2b3a27dc3b906062d35ae0f9733aa106baf3:/lib/utf8_heavy.pl diff --git a/lib/utf8_heavy.pl b/lib/utf8_heavy.pl index ab31922..08d901f 100644 --- a/lib/utf8_heavy.pl +++ b/lib/utf8_heavy.pl @@ -3,6 +3,7 @@ use strict; use warnings; sub DEBUG () { 0 } +$|=1 if DEBUG; sub DESTROY {} @@ -10,6 +11,20 @@ my %Cache; sub croak { require Carp; Carp::croak(@_) } +sub _loose_name ($) { + # Given a lowercase property or property-value name, return its + # standardized version that is expected for look-up in the 'loose' hashes + # in Heavy.pl (hence, this depends on what mktables does). This squeezes + # out blanks, underscores and dashes. The complication stems from the + # grandfathered-in 'L_', which retains a single trailing underscore. + + my $loose = $_[0] =~ s/[-\s_]//rg; + + return $loose if $loose !~ / ^ (?: is | to )? l $/x; + return 'l_' if $_[0] =~ / l .* _ /x; # If original had a trailing '_' + return $loose; +} + ## ## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape. ## It's a data structure that encodes a set of Unicode characters. @@ -29,10 +44,13 @@ sub croak { require Carp; Carp::croak(@_) } sub SWASHNEW { my ($class, $type, $list, $minbits, $none) = @_; + my $user_defined = 0; local $^D = 0 if $^D; $class = "" unless defined $class; - print STDERR __LINE__, ": class=$class, type=$type, list=$list, minbits=$minbits, none=$none\n" if DEBUG; + 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. @@ -43,6 +61,8 @@ sub croak { require Carp; Carp::croak(@_) } ## 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. @@ -61,19 +81,27 @@ sub croak { require Carp; Carp::croak(@_) } ## $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. + ## hexadecimal. It is used somehow in tr///. ## ## 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'. ## + # If a problem is found $type is returned; + # Upon success, a new (or cached) blessed object is returned with + # keys TYPE, BITS, EXTRAS, LIST, and NONE with values having the + # same meanings as the input parameters. + # SPECIALS contains a reference to any special-treatment hash in the + # INVERT_IT is non-zero if the result should be inverted before use + # USER_DEFINED is non-zero if the result came from a user-defined + # property. 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'; + my $invert_it = 0; if ($type) { @@ -89,7 +117,11 @@ sub croak { require Carp; Carp::croak(@_) } $type =~ s/^\s+//; $type =~ s/\s+$//; - print STDERR __LINE__, ": type = $type\n" if DEBUG; + # 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: { @@ -100,12 +132,23 @@ sub croak { require Carp; Carp::croak(@_) } my $caller1 = $type =~ s/(.+)::// ? $1 : caller(1); - if (defined $caller1 && $type =~ /^(?:\w+)$/) { + 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}; + $list = &{$prop}($caseless); + $user_defined = 1; last GETFILE; } } @@ -130,8 +173,8 @@ sub croak { require Carp; Carp::croak(@_) } } BEGIN { delete $utf8::{miniperl} } - # Everything is caseless matching - my $property_and_table = lc $type; + # All property names are matched caselessly + my $property_and_table = CORE::lc $type; print STDERR __LINE__, ": $property_and_table\n" if DEBUG; # See if is of the compound form 'property=value', where the @@ -154,9 +197,10 @@ sub croak { require Carp; Carp::croak(@_) } 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; + # name is always loosely matched, and always can have an + # optional 'is' prefix (which isn't true in the single + # form). + $property = _loose_name($property) =~ s/^is//r; # And convert to canonical form. Quit if not valid. $property = $utf8::loose_property_name_of{$property}; @@ -348,7 +392,7 @@ sub croak { require Carp; Carp::croak(@_) } # out the applicable characters on the rhs and looking up # again. if (! defined $file) { - $table =~ s/ [_\s-] //xg; + $table = _loose_name($table); $property_and_table = "$prefix$table"; print STDERR __LINE__, ": $property_and_table\n" if DEBUG; $file = $utf8::loose_to_file_of{$property_and_table}; @@ -356,42 +400,85 @@ sub croak { require Carp; Carp::croak(@_) } # Add the constant and go fetch it in. if (defined $file) { + + # A beginning ! means to invert + $invert_it = $file =~ s/^!//; + if ($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}; - 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; - } + # 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})) + { + + # 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; + } + } } ## ## If we reach this line, it's because we couldn't figure @@ -400,7 +487,7 @@ sub croak { require Carp; Carp::croak(@_) } pop @recursed if @recursed; return $type; - } + } # end of GETFILE block if (defined $file) { print STDERR __LINE__, ": found it (file='$file')\n" if DEBUG; @@ -414,8 +501,9 @@ sub croak { require Carp; Carp::croak(@_) } ## my $found = $Cache{$class, $file}; if ($found and ref($found) eq $class) { - print STDERR __LINE__, ": Returning cached '$file' for \\p{$type}\n" if DEBUG; + print STDERR __LINE__, ": Returning cached '$file' for \\p{$type}; invert_it=$invert_it\n" if DEBUG; pop @recursed if @recursed; + $found->{'INVERT_IT'} = $invert_it; return $found; } @@ -425,18 +513,22 @@ sub croak { require Carp; Carp::croak(@_) } } $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; - my $ORIG = $list; if ($list) { + my $taint = substr($list,0,0); # maintain taint my @tmp = split(/^/m, $list); my %seen; no warnings; - $extras = join '', grep /^[^0-9a-fA-F]/, @tmp; - $list = join '', + $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), $_ ] } @@ -466,11 +558,13 @@ sub croak { require Carp; Carp::croak(@_) } my @extras; if ($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; - my $name = $2; - print STDERR __LINE__, ": $1 => $2\n" if DEBUG; + 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; @@ -489,13 +583,15 @@ sub croak { require Carp; Carp::croak(@_) } } push @extras, $name => $subobj; $bits = $subobj->{BITS} if $bits < $subobj->{BITS}; + $user_defined = $subobj->{USER_DEFINED} + if $subobj->{USER_DEFINED}; } } } } if (DEBUG) { - print STDERR __LINE__, ": CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none"; + print STDERR __LINE__, ": CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none, INVERT_IT => $invert_it, USER_DEFINED => $user_defined"; print STDERR "\nLIST =>\n$list" if defined $list; print STDERR "\nEXTRAS =>\n$extras" if defined $extras; print STDERR "\n"; @@ -507,11 +603,22 @@ sub croak { require Carp; Carp::croak(@_) } EXTRAS => $extras, LIST => $list, NONE => $none, + USER_DEFINED => $user_defined, @extras, } => $class; if ($file) { $Cache{$class, $file} = $SWASH; + if ($type + && exists $utf8::SwashInfo{$type} + && exists $utf8::SwashInfo{$type}{'specials_name'}) + { + my $specials_name = $utf8::SwashInfo{$type}{'specials_name'}; + no strict "refs"; + print STDERR "\nspecials_name => $specials_name\n" if DEBUG; + $SWASH->{'SPECIALS'} = \%$specials_name; + } + $SWASH->{'INVERT_IT'} = $invert_it; } pop @recursed if @recursed && $type;