+#!perl
package CharClass::Matcher;
use strict;
use 5.008;
use warnings;
use warnings FATAL => 'all';
use Text::Wrap qw(wrap);
-use Encode;
use Data::Dumper;
$Data::Dumper::Useqq= 1;
our $hex_fmt= "0x%02X";
+sub DEBUG () { 0 }
+$|=1 if DEBUG;
+
+sub ASCII_PLATFORM { (ord('A') == 65) }
+
require 'regen/regen_lib.pl';
=head1 NAME
Dynamically generates macros for detecting special charclasses
in latin-1, utf8, and codepoint forms. Macros can be set to return
-the length (in bytes) of the matched codepoint, or the codepoint itself.
+the length (in bytes) of the matched codepoint, and/or the codepoint itself.
-To regenerate regcharclass.h, run this script from perl-root. No arguments
+To regenerate F<regcharclass.h>, run this script from perl-root. No arguments
are necessary.
-Using WHATEVER as an example the following macros will be produced:
+Using WHATEVER as an example the following macros can be produced, depending
+on the input parameters (how to get each is described by internal comments at
+the C<__DATA__> line):
=over 4
-=item is_WHATEVER(s,is_utf8)
+=item C<is_WHATEVER(s,is_utf8)>
-=item is_WHATEVER_safe(s,e,is_utf8)
+=item C<is_WHATEVER_safe(s,e,is_utf8)>
-Do a lookup as appropriate based on the is_utf8 flag. When possible
-comparisons involving octect<128 are done before checking the is_utf8
+Do a lookup as appropriate based on the C<is_utf8> flag. When possible
+comparisons involving octect<128 are done before checking the C<is_utf8>
flag, hopefully saving time.
-=item is_WHATEVER_utf8(s)
+The version without the C<_safe> suffix should be used only when the input is
+known to be well-formed.
+
+=item C<is_WHATEVER_utf8(s)>
-=item is_WHATEVER_utf8_safe(s,e)
+=item C<is_WHATEVER_utf8_safe(s,e)>
Do a lookup assuming the string is encoded in (normalized) UTF8.
-=item is_WHATEVER_latin1(s)
+The version without the C<_safe> suffix should be used only when the input is
+known to be well-formed.
-=item is_WHATEVER_latin1_safe(s,e)
+=item C<is_WHATEVER_latin1(s)>
+
+=item C<is_WHATEVER_latin1_safe(s,e)>
Do a lookup assuming the string is encoded in latin-1 (aka plan octets).
-=item is_WHATEVER_cp(cp)
+The version without the C<_safe> suffix should be used only when it is known
+that C<s> contains at least one character.
+
+=item C<is_WHATEVER_cp(cp)>
Check to see if the string matches a given codepoint (hypothetically a
U32). The condition is constructed as as to "break out" as early as
Thus if the character is X+1 only two comparisons will be done. Making
matching lookups slower, but non-matching faster.
-=back
+=item C<what_len_WHATEVER_FOO(arg1, ..., len)>
+
+A variant form of each of the macro types described above can be generated, in
+which the code point is returned by the macro, and an extra parameter (in the
+final position) is added, which is a pointer for the macro to set the byte
+length of the returned code point.
+
+These forms all have a C<what_len> prefix instead of the C<is_>, for example
+C<what_len_WHATEVER_safe(s,e,is_utf8,len)> and
+C<what_len_WHATEVER_utf8(s,len)>.
+
+These forms should not be used I<except> on small sets of mostly widely
+separated code points; otherwise the code generated is inefficient. For these
+cases, it is best to use the C<is_> forms, and then find the code point with
+C<utf8_to_uvchr_buf>(). This program can fail with a "deep recursion"
+message on the worst of the inappropriate sets. Examine the generated macro
+to see if it is acceptable.
+
+=item C<what_WHATEVER_FOO(arg1, ...)>
-Additionally it is possible to generate C<what_> variants that return
-the codepoint read instead of the number of octets read, this can be
-done by suffixing '-cp' to the type description.
+A variant form of each of the C<is_> macro types described above can be generated, in
+which the code point and not the length is returned by the macro. These have
+the same caveat as L</what_len_WHATEVER_FOO(arg1, ..., len)>, plus they should
+not be used where the set contains a NULL, as 0 is returned for two different
+cases: a) the set doesn't include the input code point; b) the set does
+include it, and it is a NULL.
+
+=back
=head2 CODE FORMAT
=head1 AUTHOR
-Author: Yves Orton (demerphq) 2007
+Author: Yves Orton (demerphq) 2007. Maintained by Perl5 Porters.
=head1 BUGS
# represent the string in some given encoding with specific conditions.
#
# $cp - list of codepoints that make up the string.
-# $n - list of octets that make up the string if all codepoints < 128
+# $n - list of octets that make up the string if all codepoints are invariant
+# regardless of if the string is in UTF-8 or not.
# $l - list of octets that make up the string in latin1 encoding if all
-# codepoints < 256, and at least one codepoint is >127.
-# $u - list of octets that make up the string in utf8 if any codepoint >127
+# codepoints < 256, and at least one codepoint is UTF-8 variant.
+# $u - list of octets that make up the string in utf8 if any codepoint is
+# UTF-8 variant
#
# High CP | Defined
#-----------+----------
-# 0 - 127 : $n
+# 0 - 127 : $n (127/128 are the values for ASCII platforms)
# 128 - 255 : $l, $u
# 256 - ... : $u
#
my $str= shift;
my $max= 0;
my @cp;
+ my @cp_high;
+ my $only_has_invariants = 1;
for my $ch ( split //, $str ) {
my $cp= ord $ch;
push @cp, $cp;
+ push @cp_high, $cp if $cp > 255;
$max= $cp if $max < $cp;
+ if (! ASCII_PLATFORM && $only_has_invariants) {
+ if ($cp > 255) {
+ $only_has_invariants = 0;
+ }
+ else {
+ my $temp = chr($cp);
+ utf8::upgrade($temp);
+ my @utf8 = unpack "U0C*", $temp;
+ $only_has_invariants = (@utf8 == 1 && $utf8[0] == $cp);
+ }
+ }
}
my ( $n, $l, $u );
- if ( $max < 128 ) {
+ $only_has_invariants = $max < 128 if ASCII_PLATFORM;
+ if ($only_has_invariants) {
$n= [@cp];
} else {
$l= [@cp] if $max && $max < 256;
- my $copy= $str; # must copy string, FB_CROAK makes encode destructive
- $u= eval { Encode::encode( "utf8", $copy, Encode::FB_CROAK ) };
- # $u is utf8 but with the utf8 flag OFF
- # therefore "C*" gets us the values of the bytes involved.
- $u= [ unpack "C*", $u ] if defined $u;
+ $u= $str;
+ utf8::upgrade($u);
+ $u= [ unpack "U0C*", $u ] if defined $u;
}
- return ( \@cp, $n, $l, $u );
+ return ( \@cp, \@cp_high, $n, $l, $u );
}
#
sub __clean {
my ( $expr )= @_;
+
+ #return $expr;
+
our $parens;
$parens= qr/ (?> \( (?> (?: (?> [^()]+ ) | (??{ $parens }) )* ) \) ) /x;
- #print "$parens\n$expr\n";
+ ## remove redundant parens
1 while $expr =~ s/ \( \s* ( $parens ) \s* \) /$1/gx;
- 1 while $expr =~ s/ \( \s* ($parens) \s* \? \s*
- \( \s* ($parens) \s* \? \s* ($parens|[^:]+?) \s* : \s* ($parens|[^)]+?) \s* \)
- \s* : \s* \4 \s* \)/( ( $1 && $2 ) ? $3 : 0 )/gx;
+
+
+ # repeatedly simplify conditions like
+ # ( (cond1) ? ( (cond2) ? X : Y ) : Y )
+ # into
+ # ( ( (cond1) && (cond2) ) ? X : Y )
+ # Also similarly handles expressions like:
+ # : (cond1) ? ( (cond2) ? X : Y ) : Y )
+ # Note the inclusion of the close paren in ([:()]) and the open paren in ([()]) is
+ # purely to ensure we have a balanced set of parens in the expression which makes
+ # it easier to understand the pattern in an editor that understands paren's, we do
+ # not expect either of these cases to actually fire. - Yves
+ 1 while $expr =~ s/
+ ([:()]) \s*
+ ($parens) \s*
+ \? \s*
+ \( \s* ($parens) \s*
+ \? \s* ($parens|[^()?:\s]+?) \s*
+ : \s* ($parens|[^()?:\s]+?) \s*
+ \) \s*
+ : \s* \5 \s*
+ ([()])
+ /$1 ( $2 && $3 ) ? $4 : $5 $6/gx;
+ #$expr=~s/\(\(U8\*\)s\)\[(\d+)\]/S$1/g if length $expr > 8000;
+ #$expr=~s/\s+//g if length $expr > 8000;
+
+ die "Expression too long" if length $expr > 8000;
+
return $expr;
}
#
# Each string is then stored in the 'strs' subhash as a hash record
# made up of the results of __uni_latin1, using the keynames
-# 'low','latin1','utf8', as well as the synthesized 'LATIN1' and
+# 'low','latin1','utf8', as well as the synthesized 'LATIN1', 'high', and
# 'UTF8' which hold a merge of 'low' and their lowercase equivelents.
#
# Size data is tracked per type in the 'size' subhash.
push @{$opt{txt}}, sprintf "0x%X", $cp;
}
next;
- } elsif ( $str =~ /^0x/ ) {
+ } elsif ($str =~ s/ ^ N (?= 0x ) //x ) {
+ # Otherwise undocumented, a leading N means is already in the
+ # native character set; don't convert.
$str= chr eval $str;
+ } elsif ( $str =~ /^0x/ ) {
+ $str= eval $str;
+
+ # Convert from Unicode/ASCII to native, if necessary
+ $str = utf8::unicode_to_native($str) if ! ASCII_PLATFORM
+ && $str <= 0xFF;
+ $str = chr $str;
} elsif ( $str =~ / \s* \\p \{ ( .*? ) \} /x) {
my $property = $1;
use Unicode::UCD qw(prop_invlist);
# Replace this element on the list with the property's expansion
for (my $i = 0; $i < @invlist; $i += 2) {
foreach my $cp ($invlist[$i] .. $invlist[$i+1] - 1) {
- push @{$opt{txt}}, sprintf "0x%X", $cp;
+
+ # prop_invlist() returns native values; add leading 'N'
+ # to indicate that.
+ push @{$opt{txt}}, sprintf "N0x%X", $cp;
}
}
next;
- } elsif ( /\S/ ) {
- die "Unparsable line: $txt\n";
- } else {
+ } elsif ($str =~ / ^ do \s+ ( .* ) /x) {
+ die "do '$1' failed: $!$@" if ! do $1 or $@;
+ next;
+ } elsif ($str =~ / ^ & \s* ( .* ) /x) { # user-furnished sub() call
+ my @results = eval "$1";
+ die "eval '$1' failed: $@" if $@;
+ push @{$opt{txt}}, @results;
next;
+ } else {
+ die "Unparsable line: $txt\n";
}
- my ( $cp, $low, $latin1, $utf8 )= __uni_latin1( $str );
+ my ( $cp, $cp_high, $low, $latin1, $utf8 )= __uni_latin1( $str );
my $UTF8= $low || $utf8;
my $LATIN1= $low || $latin1;
+ my $high = (scalar grep { $_ < 256 } @$cp) ? 0 : $utf8;
#die Dumper($txt,$cp,$low,$latin1,$utf8)
# if $txt=~/NEL/ or $utf8 and @$utf8>3;
- @{ $self->{strs}{$str} }{qw( str txt low utf8 latin1 cp UTF8 LATIN1 )}=
- ( $str, $txt, $low, $utf8, $latin1, $cp, $UTF8, $LATIN1 );
+ @{ $self->{strs}{$str} }{qw( str txt low utf8 latin1 high cp cp_high UTF8 LATIN1 )}=
+ ( $str, $txt, $low, $utf8, $latin1, $high, $cp, $cp_high, $UTF8, $LATIN1 );
my $rec= $self->{strs}{$str};
- foreach my $key ( qw(low utf8 latin1 cp UTF8 LATIN1) ) {
+ foreach my $key ( qw(low utf8 latin1 high cp cp_high UTF8 LATIN1) ) {
$self->{size}{$key}{ 0 + @{ $self->{strs}{$str}{$key} } }++
if $self->{strs}{$str}{$key};
}
return 0 + keys( %trie ) ? \%trie : undef;
}
+sub pop_count ($) {
+ my $word = shift;
+
+ # This returns a list of the positions of the bits in the input word that
+ # are 1.
+
+ my @positions;
+ my $position = 0;
+ while ($word) {
+ push @positions, $position if $word & 1;
+ $position++;
+ $word >>= 1;
+ }
+ return @positions;
+}
+
# my $optree= _optree()
#
# recursively convert a trie to an optree where every node represents
$else= 0 unless defined $else;
$depth= 0 unless defined $depth;
- my @conds= sort { $a <=> $b } grep { length $_ } keys %$trie;
- if ( $trie->{''} ) {
+ # if we have an emptry string as a key it means we are in an
+ # accepting state and unless we can match further on should
+ # return the value of the '' key.
+ if (exists $trie->{''} ) {
+ # we can now update the "else" value, anything failing to match
+ # after this point should return the value from this.
if ( $ret_type eq 'cp' ) {
$else= $self->{strs}{ $trie->{''} }{cp}[0];
$else= sprintf "$self->{val_fmt}", $else if $else > 9;
$else= "len=$depth, $else";
}
}
+ # extract the meaningful keys from the trie, filter out '' as
+ # it means we are an accepting state (end of sequence).
+ my @conds= sort { $a <=> $b } grep { length $_ } keys %$trie;
+
+ # if we havent any keys there is no further we can match and we
+ # can return the "else" value.
return $else if !@conds;
- my $node= {};
- my $root= $node;
- my ( $yes_res, $as_code, @cond );
- my $test= $test_type eq 'cp' ? "cp" : "((U8*)s)[$depth]";
- my $Update= sub {
- $node->{vals}= [@cond];
+
+
+ my $test= $test_type =~ /^cp/ ? "cp" : "((U8*)s)[$depth]";
+ # first we loop over the possible keys/conditions and find out what they look like
+ # we group conditions with the same optree together.
+ my %dmp_res;
+ my @res_order;
+ local $Data::Dumper::Sortkeys=1;
+ foreach my $cond ( @conds ) {
+
+ # get the optree for this child/condition
+ my $res= $self->_optree( $trie->{$cond}, $test_type, $ret_type, $else, $depth + 1 );
+ # convert it to a string with Dumper
+ my $res_code= Dumper( $res );
+
+ push @{$dmp_res{$res_code}{vals}}, $cond;
+ if (!$dmp_res{$res_code}{optree}) {
+ $dmp_res{$res_code}{optree}= $res;
+ push @res_order, $res_code;
+ }
+ }
+
+ # now that we have deduped the optrees we construct a new optree containing the merged
+ # results.
+ my %root;
+ my $node= \%root;
+ foreach my $res_code_idx (0 .. $#res_order) {
+ my $res_code= $res_order[$res_code_idx];
+ $node->{vals}= $dmp_res{$res_code}{vals};
$node->{test}= $test;
- $node->{yes}= $yes_res;
+ $node->{yes}= $dmp_res{$res_code}{optree};
$node->{depth}= $depth;
- $node->{no}= shift;
- };
- while ( @conds ) {
- my $cond= shift @conds;
- my $res=
- $self->_optree( $trie->{$cond}, $test_type, $ret_type, $else,
- $depth + 1 );
- my $res_code= Dumper( $res );
- if ( !$yes_res || $res_code ne $as_code ) {
- if ( $yes_res ) {
- $Update->( {} );
- $node= $node->{no};
- }
- ( $yes_res, $as_code )= ( $res, $res_code );
- @cond= ( $cond );
+ if ($res_code_idx < $#res_order) {
+ $node= $node->{no}= {};
} else {
- push @cond, $cond;
+ $node->{no}= $else;
}
}
- $Update->( $else );
- return $root;
+
+ # return the optree.
+ return \%root;
}
# my $optree= optree(%opts);
my %opt= @_;
my $trie= $self->make_trie( $opt{type}, $opt{max_depth} );
$opt{ret_type} ||= 'len';
- my $test_type= $opt{type} eq 'cp' ? 'cp' : 'depth';
+ my $test_type= $opt{type} =~ /^cp/ ? 'cp' : 'depth';
return $self->_optree( $trie, $test_type, $opt{ret_type}, $opt{else}, 0 );
}
my $type= $opt{type};
die "Can't do a length_optree on type 'cp', makes no sense."
- if $type eq 'cp';
+ if $type =~ /^cp/;
my ( @size, $method );
return $else;
}
+sub calculate_mask(@) {
+ # Look at the input list of byte values. This routine returns an array of
+ # mask/base pairs to generate that list.
+
+ my @list = @_;
+ my $list_count = @list;
+
+ # Consider a set of byte values, A, B, C .... If we want to determine if
+ # <c> is one of them, we can write c==A || c==B || c==C .... If the
+ # values are consecutive, we can shorten that to A<=c && c<=Z, which uses
+ # far fewer branches. If only some of them are consecutive we can still
+ # save some branches by creating range tests for just those that are
+ # consecutive. _cond_as_str() does this work for looking for ranges.
+ #
+ # Another approach is to look at the bit patterns for A, B, C .... and see
+ # if they have some commonalities. That's what this function does. For
+ # example, consider a set consisting of the bytes
+ # 0xF0, 0xF1, 0xF2, and 0xF3. We could write:
+ # 0xF0 <= c && c <= 0xF4
+ # But the following mask/compare also works, and has just one test:
+ # (c & 0xFC) == 0xF0
+ # The reason it works is that the set consists of exactly those bytes
+ # whose first 4 bits are 1, and the next two are 0. (The value of the
+ # other 2 bits is immaterial in determining if a byte is in the set or
+ # not.) The mask masks out those 2 irrelevant bits, and the comparison
+ # makes sure that the result matches all bytes which match those 6
+ # material bits exactly. In other words, the set of bytes contains
+ # exactly those whose bottom two bit positions are either 0 or 1. The
+ # same principle applies to bit positions that are not necessarily
+ # adjacent. And it can be applied to bytes that differ in 1 through all 8
+ # bit positions. In order to be a candidate for this optimization, the
+ # number of bytes in the set must be a power of 2.
+ #
+ # Consider a different example, the set 0x53, 0x54, 0x73, and 0x74. That
+ # requires 4 tests using either ranges or individual values, and even
+ # though the number in the set is a power of 2, it doesn't qualify for the
+ # mask optimization described above because the number of bits that are
+ # different is too large for that. However, the set can be expressed as
+ # two branches with masks thusly:
+ # (c & 0xDF) == 0x53 || (c & 0xDF) == 0x54
+ # a branch savings of 50%. This is done by splitting the set into two
+ # subsets each of which has 2 elements, and within each set the values
+ # differ by 1 byte.
+ #
+ # This function attempts to find some way to save some branches using the
+ # mask technique. If not, it returns an empty list; if so, it
+ # returns a list consisting of
+ # [ [compare1, mask1], [compare2, mask2], ...
+ # [compare_n, undef], [compare_m, undef], ...
+ # ]
+ # The <mask> is undef in the above for those bytes that must be tested
+ # for individually.
+ #
+ # This function does not attempt to find the optimal set. To do so would
+ # probably require testing all possible combinations, and keeping track of
+ # the current best one.
+ #
+ # There are probably much better algorithms, but this is the one I (khw)
+ # came up with. We start with doing a bit-wise compare of every byte in
+ # the set with every other byte. The results are sorted into arrays of
+ # all those that differ by the same bit positions. These are stored in a
+ # hash with the each key being the bits they differ in. Here is the hash
+ # for the 0x53, 0x54, 0x73, 0x74 set:
+ # {
+ # 4 => {
+ # "0,1,2,5" => [
+ # 83,
+ # 116,
+ # 84,
+ # 115
+ # ]
+ # },
+ # 3 => {
+ # "0,1,2" => [
+ # 83,
+ # 84,
+ # 115,
+ # 116
+ # ]
+ # }
+ # 1 => {
+ # 5 => [
+ # 83,
+ # 115,
+ # 84,
+ # 116
+ # ]
+ # },
+ # }
+ #
+ # The set consisting of values which differ in the 4 bit positions 0, 1,
+ # 2, and 5 from some other value in the set consists of all 4 values.
+ # Likewise all 4 values differ from some other value in the 3 bit
+ # positions 0, 1, and 2; and all 4 values differ from some other value in
+ # the single bit position 5. The keys at the uppermost level in the above
+ # hash, 1, 3, and 4, give the number of bit positions that each sub-key
+ # below it has. For example, the 4 key could have as its value an array
+ # consisting of "0,1,2,5", "0,1,2,6", and "3,4,6,7", if the inputs were
+ # such. The best optimization will group the most values into a single
+ # mask. The most values will be the ones that differ in the most
+ # positions, the ones with the largest value for the topmost key. These
+ # keys, are thus just for convenience of sorting by that number, and do
+ # not have any bearing on the core of the algorithm.
+ #
+ # We start with an element from largest number of differing bits. The
+ # largest in this case is 4 bits, and there is only one situation in this
+ # set which has 4 differing bits, "0,1,2,5". We look for any subset of
+ # this set which has 16 values that differ in these 4 bits. There aren't
+ # any, because there are only 4 values in the entire set. We then look at
+ # the next possible thing, which is 3 bits differing in positions "0,1,2".
+ # We look for a subset that has 8 values that differ in these 3 bits.
+ # Again there are none. So we go to look for the next possible thing,
+ # which is a subset of 2**1 values that differ only in bit position 5. 83
+ # and 115 do, so we calculate a mask and base for those and remove them
+ # from every set. Since there is only the one set remaining, we remove
+ # them from just this one. We then look to see if there is another set of
+ # 2 values that differ in bit position 5. 84 and 116 do, so we calculate
+ # a mask and base for those and remove them from every set (again only
+ # this set remains in this example). The set is now empty, and there are
+ # no more sets to look at, so we are done.
+
+ if ($list_count == 256) { # All 256 is trivially masked
+ return (0, 0);
+ }
+
+ my %hash;
+
+ # Generate bits-differing lists for each element compared against each
+ # other element
+ for my $i (0 .. $list_count - 2) {
+ for my $j ($i + 1 .. $list_count - 1) {
+ my @bits_that_differ = pop_count($list[$i] ^ $list[$j]);
+ my $differ_count = @bits_that_differ;
+ my $key = join ",", @bits_that_differ;
+ push @{$hash{$differ_count}{$key}}, $list[$i] unless grep { $_ == $list[$i] } @{$hash{$differ_count}{$key}};
+ push @{$hash{$differ_count}{$key}}, $list[$j];
+ }
+ }
+
+ print STDERR __LINE__, ": calculate_mask() called: List of values grouped by differing bits: ", Dumper \%hash if DEBUG;
+
+ my @final_results;
+ foreach my $count (reverse sort { $a <=> $b } keys %hash) {
+ my $need = 2 ** $count; # Need 8 values for 3 differing bits, etc
+ foreach my $bits (sort keys $hash{$count}) {
+
+ print STDERR __LINE__, ": For $count bit(s) difference ($bits), need $need; have ", scalar @{$hash{$count}{$bits}}, "\n" if DEBUG;
+
+ # Look only as long as there are at least as many elements in the
+ # subset as are needed
+ while ((my $cur_count = @{$hash{$count}{$bits}}) >= $need) {
+
+ print STDERR __LINE__, ": Looking at bit positions ($bits): ", Dumper $hash{$count}{$bits} if DEBUG;
+
+ # Start with the first element in it
+ my $try_base = $hash{$count}{$bits}[0];
+ my @subset = $try_base;
+
+ # If it succeeds, we return a mask and a base to compare
+ # against the masked value. That base will be the AND of
+ # every element in the subset. Initialize to the one element
+ # we have so far.
+ my $compare = $try_base;
+
+ # We are trying to find a subset of this that has <need>
+ # elements that differ in the bit positions given by the
+ # string $bits, which is comma separated.
+ my @bits = split ",", $bits;
+
+ TRY: # Look through the remainder of the list for other
+ # elements that differ only by these bit positions.
+
+ for (my $i = 1; $i < $cur_count; $i++) {
+ my $try_this = $hash{$count}{$bits}[$i];
+ my @positions = pop_count($try_base ^ $try_this);
+
+ print STDERR __LINE__, ": $try_base vs $try_this: is (", join(',', @positions), ") a subset of ($bits)?" if DEBUG;;
+
+ foreach my $pos (@positions) {
+ unless (grep { $pos == $_ } @bits) {
+ print STDERR " No\n" if DEBUG;
+ my $remaining = $cur_count - $i - 1;
+ if ($remaining && @subset + $remaining < $need) {
+ print STDERR __LINE__, ": Can stop trying $try_base, because even if all the remaining $remaining values work, they wouldn't add up to the needed $need when combined with the existing ", scalar @subset, " ones\n" if DEBUG;
+ last TRY;
+ }
+ next TRY;
+ }
+ }
+
+ print STDERR " Yes\n" if DEBUG;
+ push @subset, $try_this;
+
+ # Add this to the mask base, in case it ultimately
+ # succeeds,
+ $compare &= $try_this;
+ }
+
+ print STDERR __LINE__, ": subset (", join(", ", @subset), ") has ", scalar @subset, " elements; needs $need\n" if DEBUG;
+
+ if (@subset < $need) {
+ shift @{$hash{$count}{$bits}};
+ next; # Try with next value
+ }
+
+ # Create the mask
+ my $mask = 0;
+ foreach my $position (@bits) {
+ $mask |= 1 << $position;
+ }
+ $mask = ~$mask & 0xFF;
+ push @final_results, [$compare, $mask];
+
+ printf STDERR "%d: Got it: compare=%d=0x%X; mask=%X\n", __LINE__, $compare, $compare, $mask if DEBUG;
+
+ # These values are now spoken for. Remove them from future
+ # consideration
+ foreach my $remove_count (sort keys %hash) {
+ foreach my $bits (sort keys %{$hash{$remove_count}}) {
+ foreach my $to_remove (@subset) {
+ @{$hash{$remove_count}{$bits}} = grep { $_ != $to_remove } @{$hash{$remove_count}{$bits}};
+ }
+ }
+ }
+ }
+ }
+ }
+
+ # Any values that remain in the list are ones that have to be tested for
+ # individually.
+ my @individuals;
+ foreach my $count (reverse sort { $a <=> $b } keys %hash) {
+ foreach my $bits (sort keys $hash{$count}) {
+ foreach my $remaining (@{$hash{$count}{$bits}}) {
+
+ # If we already know about this value, just ignore it.
+ next if grep { $remaining == $_ } @individuals;
+
+ # Otherwise it needs to be returned as something to match
+ # individually
+ push @final_results, [$remaining, undef];
+ push @individuals, $remaining;
+ }
+ }
+ }
+
+ # Sort by increasing numeric value
+ @final_results = sort { $a->[0] <=> $b->[0] } @final_results;
+
+ print STDERR __LINE__, ": Final return: ", Dumper \@final_results if DEBUG;
+
+ return @final_results;
+}
+
# _cond_as_str
# turn a list of conditions into a text expression
# - merges ranges of conditions, and joins the result with ||
sub _cond_as_str {
- my ( $self, $op, $combine )= @_;
+ my ( $self, $op, $combine, $opts_ref )= @_;
my $cond= $op->{vals};
my $test= $op->{test};
+ my $is_cp_ret = $opts_ref->{ret_type} eq "cp";
return "( $test )" if !defined $cond;
- # rangify the list
+ # rangify the list.
my @ranges;
my $Update= sub {
- if ( @ranges ) {
+ # We skip this if there are optimizations that
+ # we can apply (below) to the individual ranges
+ if ( ($is_cp_ret || $combine) && @ranges && ref $ranges[-1]) {
if ( $ranges[-1][0] == $ranges[-1][1] ) {
$ranges[-1]= $ranges[-1][0];
} elsif ( $ranges[-1][0] + 1 == $ranges[-1][1] ) {
}
}
};
- for my $cond ( @$cond ) {
- if ( !@ranges || $cond != $ranges[-1][1] + 1 ) {
+ for my $condition ( @$cond ) {
+ if ( !@ranges || $condition != $ranges[-1][1] + 1 ) {
$Update->();
- push @ranges, [ $cond, $cond ];
+ push @ranges, [ $condition, $condition ];
} else {
$ranges[-1][1]++;
}
}
$Update->();
+
return $self->_combine( $test, @ranges )
if $combine;
- @ranges= map {
- ref $_
- ? sprintf(
- "( $self->{val_fmt} <= $test && $test <= $self->{val_fmt} )",
- @$_ )
- : sprintf( "$self->{val_fmt} == $test", $_ );
- } @ranges;
- return "( " . join( " || ", @ranges ) . " )";
+
+ if ($is_cp_ret) {
+ @ranges= map {
+ ref $_
+ ? sprintf(
+ "( $self->{val_fmt} <= $test && $test <= $self->{val_fmt} )",
+ @$_ )
+ : sprintf( "$self->{val_fmt} == $test", $_ );
+ } @ranges;
+
+ return "( " . join( " || ", @ranges ) . " )";
+ }
+
+ # If the input set has certain characteristics, we can optimize tests
+ # for it. This doesn't apply if returning the code point, as we want
+ # each element of the set individually. The code above is for this
+ # simpler case.
+
+ return 1 if @$cond == 256; # If all bytes match, is trivially true
+
+ my @masks;
+ if (@ranges > 1) {
+
+ # See if the entire set shares optimizable characterstics, and if so,
+ # return the optimization. We delay checking for this on sets with
+ # just a single range, as there may be better optimizations available
+ # in that case.
+ @masks = calculate_mask(@$cond);
+
+ # Stringify the output of calculate_mask()
+ if (@masks) {
+ my @return;
+ foreach my $mask_ref (@masks) {
+ if (defined $mask_ref->[1]) {
+ push @return, sprintf "( ( $test & $self->{val_fmt} ) == $self->{val_fmt} )", $mask_ref->[1], $mask_ref->[0];
+ }
+ else { # An undefined mask means to use the value as-is
+ push @return, sprintf "$test == $self->{val_fmt}", $mask_ref->[0];
+ }
+ }
+
+ # The best possible case below for specifying this set of values via
+ # ranges is 1 branch per range. If our mask method yielded better
+ # results, there is no sense trying something that is bound to be
+ # worse.
+ if (@return < @ranges) {
+ return "( " . join( " || ", @return ) . " )";
+ }
+
+ @masks = @return;
+ }
+ }
+
+ # Here, there was no entire-class optimization that was clearly better
+ # than doing things by ranges. Look at each range.
+ my $range_count_extra = 0;
+ for (my $i = 0; $i < @ranges; $i++) {
+ if (! ref $ranges[$i]) { # Trivial case: no range
+ $ranges[$i] = sprintf "$self->{val_fmt} == $test", $ranges[$i];
+ }
+ elsif ($ranges[$i]->[0] == $ranges[$i]->[1]) {
+ $ranges[$i] = # Trivial case: single element range
+ sprintf "$self->{val_fmt} == $test", $ranges[$i]->[0];
+ }
+ else {
+ my $output = "";
+
+ # Well-formed UTF-8 continuation bytes on ascii platforms must be
+ # in the range 0x80 .. 0xBF. If we know that the input is
+ # well-formed (indicated by not trying to be 'safe'), we can omit
+ # tests that verify that the input is within either of these
+ # bounds. (No legal UTF-8 character can begin with anything in
+ # this range, so we don't have to worry about this being a
+ # continuation byte or not.)
+ if (ASCII_PLATFORM
+ && ! $opts_ref->{safe}
+ && $opts_ref->{type} =~ / ^ (?: utf8 | high ) $ /xi)
+ {
+ my $lower_limit_is_80 = ($ranges[$i]->[0] == 0x80);
+ my $upper_limit_is_BF = ($ranges[$i]->[1] == 0xBF);
+
+ # If the range is the entire legal range, it matches any legal
+ # byte, so we can omit both tests. (This should happen only
+ # if the number of ranges is 1.)
+ if ($lower_limit_is_80 && $upper_limit_is_BF) {
+ return 1;
+ }
+ elsif ($lower_limit_is_80) { # Just use the upper limit test
+ $output = sprintf("( $test <= $self->{val_fmt} )",
+ $ranges[$i]->[1]);
+ }
+ elsif ($upper_limit_is_BF) { # Just use the lower limit test
+ $output = sprintf("( $test >= $self->{val_fmt} )",
+ $ranges[$i]->[0]);
+ }
+ }
+
+ # If we didn't change to omit a test above, see if the number of
+ # elements is a power of 2 (only a single bit in the
+ # representation of its count will be set) and if so, it may be
+ # that a mask/compare optimization is possible.
+ if ($output eq ""
+ && pop_count($ranges[$i]->[1] - $ranges[$i]->[0] + 1) == 1)
+ {
+ my @list;
+ push @list, $_ for ($ranges[$i]->[0] .. $ranges[$i]->[1]);
+ my @this_masks = calculate_mask(@list);
+
+ # Use the mask if there is just one for the whole range.
+ # Otherwise there is no savings over the two branches that can
+ # define the range.
+ if (@this_masks == 1 && defined $this_masks[0][1]) {
+ $output = sprintf "( $test & $self->{val_fmt} ) == $self->{val_fmt}", $this_masks[0][1], $this_masks[0][0];
+ }
+ }
+
+ if ($output ne "") { # Prefer any optimization
+ $ranges[$i] = $output;
+ }
+ else {
+ # No optimization happened. We need a test that the code
+ # point is within both bounds. But, if the bounds are
+ # adjacent code points, it is cleaner to say
+ # 'first == test || second == test'
+ # than it is to say
+ # 'first <= test && test <= second'
+
+ $range_count_extra++; # This range requires 2 branches to
+ # represent
+ if ($ranges[$i]->[0] + 1 == $ranges[$i]->[1]) {
+ $ranges[$i] = "( "
+ . join( " || ", ( map
+ { sprintf "$self->{val_fmt} == $test", $_ }
+ @{$ranges[$i]} ) )
+ . " )";
+ }
+ else { # Full bounds checking
+ $ranges[$i] = sprintf("( $self->{val_fmt} <= $test && $test <= $self->{val_fmt} )", $ranges[$i]->[0], $ranges[$i]->[1]);
+ }
+ }
+ }
+ }
+
+ # We have generated the list of bytes in two ways; one trying to use masks
+ # to cut the number of branches down, and the other to look at individual
+ # ranges (some of which could be cut down by using a mask for just it).
+ # We return whichever method uses the fewest branches.
+ return "( "
+ . join( " || ", (@masks && @masks < @ranges + $range_count_extra)
+ ? @masks
+ : @ranges)
+ . " )";
}
# _combine
$gtv= sprintf "$self->{val_fmt}", $item;
}
if ( @cond ) {
- return "( $cstr || ( $gtv < $test &&\n"
- . $self->_combine( $test, @cond ) . " ) )";
+ my $combine= $self->_combine( $test, @cond );
+ if (@cond >1) {
+ return "( $cstr || ( $gtv < $test &&\n"
+ . $combine . " ) )";
+ } else {
+ return "( $cstr || $combine )";
+ }
} else {
return $cstr;
}
# _render()
# recursively convert an optree to text with reasonably neat formatting
sub _render {
- my ( $self, $op, $combine, $brace )= @_;
+ my ( $self, $op, $combine, $brace, $opts_ref, $def, $submacros )= @_;
+ return 0 if ! defined $op; # The set is empty
if ( !ref $op ) {
return $op;
}
- my $cond= $self->_cond_as_str( $op, $combine );
- my $yes= $self->_render( $op->{yes}, $combine, 1 );
- my $no= $self->_render( $op->{no}, $combine, 0 );
+ my $cond= $self->_cond_as_str( $op, $combine, $opts_ref );
+ #no warnings 'recursion'; # This would allow really really inefficient
+ # code to be generated. See pod
+ my $yes= $self->_render( $op->{yes}, $combine, 1, $opts_ref, $def, $submacros );
+ return $yes if $cond eq '1';
+
+ my $no= $self->_render( $op->{no}, $combine, 0, $opts_ref, $def, $submacros );
return "( $cond )" if $yes eq '1' and $no eq '0';
my ( $lb, $rb )= $brace ? ( "( ", " )" ) : ( "", "" );
return "$lb$cond ? $yes : $no$rb"
$yes= " " . $yes;
}
- return "$lb$cond ?$yes$ind: $no$rb";
+ my $str= "$lb$cond ?$yes$ind: $no$rb";
+ if (length $str > 6000) {
+ push @$submacros, sprintf "#define $def\n( %s )", "_part" . (my $yes_idx= 0+@$submacros), $yes;
+ push @$submacros, sprintf "#define $def\n( %s )", "_part" . (my $no_idx= 0+@$submacros), $no;
+ return sprintf "%s%s ? $def : $def%s", $lb, $cond, "_part$yes_idx", "_part$no_idx", $rb;
+ }
+ return $str;
}
# $expr=render($op,$combine)
# longer lists such as that resulting from type 'cp' output.
# Currently only used for type 'cp' macros.
sub render {
- my ( $self, $op, $combine )= @_;
- my $str= "( " . $self->_render( $op, $combine ) . " )";
- return __clean( $str );
+ my ( $self, $op, $combine, $opts_ref, $def_fmt )= @_;
+
+ my @submacros;
+ my $macro= sprintf "#define $def_fmt\n( %s )", "", $self->_render( $op, $combine, 0, $opts_ref, $def_fmt, \@submacros );
+
+ return join "\n\n", map { "/*** GENERATED CODE ***/\n" . __macro( __clean( $_ ) ) } @submacros, $macro;
}
# make_macro
# make a macro of a given type.
# calls into make_trie and (generic_|length_)optree as needed
# Opts are:
-# type : 'cp','generic','low','latin1','utf8','LATIN1','UTF8'
+# type : 'cp','cp_high', 'generic','high','low','latin1','utf8','LATIN1','UTF8'
# ret_type : 'cp' or 'len'
# safe : add length guards to macro
#
my %opts= @_;
my $type= $opts{type} || 'generic';
die "Can't do a 'cp' on multi-codepoint character class '$self->{op}'"
- if $type eq 'cp'
+ if $type =~ /^cp/
and $self->{has_multi};
- my $ret_type= $opts{ret_type} || ( $opts{type} eq 'cp' ? 'cp' : 'len' );
+ my $ret_type= $opts{ret_type} || ( $opts{type} =~ /^cp/ ? 'cp' : 'len' );
my $method;
if ( $opts{safe} ) {
$method= 'length_optree';
} else {
$method= 'optree';
}
- my $optree= $self->$method( %opts, type => $type, ret_type => $ret_type );
- my $text= $self->render( $optree, $type eq 'cp' );
- my @args= $type eq 'cp' ? 'cp' : 's';
+ my @args= $type =~ /^cp/ ? 'cp' : 's';
push @args, "e" if $opts{safe};
push @args, "is_utf8" if $type eq 'generic';
push @args, "len" if $ret_type eq 'both';
my $ext= $type eq 'generic' ? '' : '_' . lc( $type );
$ext .= "_safe" if $opts{safe};
my $argstr= join ",", @args;
- return "/*** GENERATED CODE ***/\n"
- . __macro( "#define $pfx$self->{op}$ext($argstr)\n$text" );
+ my $def_fmt="$pfx$self->{op}$ext%s($argstr)";
+ my $optree= $self->$method( %opts, type => $type, ret_type => $ret_type );
+ return $self->render( $optree, ($type =~ /^cp/) ? 1 : 0, \%opts, $def_fmt );
}
# if we arent being used as a module (highly likely) then process
copyright => [2007, 2011] );
print $out_fh "\n#ifndef H_REGCHARCLASS /* Guard against nested #includes */\n#define H_REGCHARCLASS 1\n\n";
- my ( $op, $title, @txt, @types, @mods );
+ my ( $op, $title, @txt, @types, %mods );
my $doit= sub {
return unless $op;
+
+ # Skip if to compile on a different platform.
+ return if delete $mods{only_ascii_platform} && ! ASCII_PLATFORM;
+ return if delete $mods{only_ebcdic_platform} && ord 'A' != 193;
+
print $out_fh "/*\n\t$op: $title\n\n";
print $out_fh join "\n", ( map { "\t$_" } @txt ), "*/", "";
my $obj= __PACKAGE__->new( op => $op, title => $title, txt => \@txt );
- #die Dumper(\@types,\@mods);
+ #die Dumper(\@types,\%mods);
+
+ my @mods;
+ push @mods, 'safe' if delete $mods{safe};
+ unshift @mods, 'fast' if delete $mods{fast} || ! @mods; # Default to 'fast'
+ # do this one
+ # first, as
+ # traditional
+ if (%mods) {
+ die "Unknown modifiers: ", join ", ", map { "'$_'" } sort keys %mods;
+ }
foreach my $type_spec ( @types ) {
my ( $type, $ret )= split /-/, $type_spec;
$ret ||= 'len';
foreach my $mod ( @mods ) {
- next if $mod eq 'safe' and $type eq 'cp';
+ next if $mod eq 'safe' and $type =~ /^cp/;
+ delete $mods{$mod};
my $macro= $obj->make_macro(
type => $type,
ret_type => $ret,
};
while ( <DATA> ) {
- s/^\s*#//;
+ s/^ \s* (?: \# .* ) ? $ //x; # squeeze out comment and blanks
next unless /\S/;
chomp;
- if ( /^([A-Z]+)/ ) {
- $doit->();
+ if ( /^[A-Z]/ ) {
+ $doit->(); # This starts a new definition; do the previous one
( $op, $title )= split /\s*:\s*/, $_, 2;
@txt= ();
} elsif ( s/^=>// ) {
my ( $type, $modifier )= split /:/, $_;
@types= split ' ', $type;
- @mods= split ' ', $modifier;
+ undef %mods;
+ map { $mods{$_} = 1 } split ' ', $modifier;
} else {
push @txt, "$_";
}
}
}
+# The form of the input is a series of definitions to make macros for.
+# The first line gives the base name of the macro, followed by a colon, and
+# then text to be used in comments associated with the macro that are its
+# title or description. In all cases the first (perhaps only) parameter to
+# the macro is a pointer to the first byte of the code point it is to test to
+# see if it is in the class determined by the macro. In the case of non-UTF8,
+# the code point consists only of a single byte.
+#
+# The second line must begin with a '=>' and be followed by the types of
+# macro(s) to be generated; these are specified below. A colon follows the
+# types, followed by the modifiers, also specified below. At least one
+# modifier is required.
+#
+# The subsequent lines give what code points go into the class defined by the
+# macro. Multiple characters may be specified via a string like "\x0D\x0A",
+# enclosed in quotes. Otherwise the lines consist of one of:
+# 1) a single Unicode code point, prefaced by 0x
+# 2) a single range of Unicode code points separated by a minus (and
+# optional space)
+# 3) a single Unicode property specified in the standard Perl form
+# "\p{...}"
+# 4) a line like 'do path'. This will do a 'do' on the file given by
+# 'path'. It is assumed that this does nothing but load subroutines
+# (See item 5 below). The reason 'require path' is not used instead is
+# because 'do' doesn't assume that path is in @INC.
+# 5) a subroutine call
+# &pkg::foo(arg1, ...)
+# where pkg::foo was loaded by a 'do' line (item 4). The subroutine
+# returns an array of entries of forms like items 1-3 above. This
+# allows more complex inputs than achievable from the other input types.
#
-# Valid types: generic, LATIN1, UTF8, low, latin1, utf8
-# default return value is octects read.
-# append -cp to make it codepoint matched.
-# modifiers come after the colon, valid possibilities
-# being 'fast' and 'safe'.
+# A blank line or one whose first non-blank character is '#' is a comment.
+# The definition of the macro is terminated by a line unlike those described.
#
-# Accepts a single code point per line, prefaced by '0x'
-# or a range of two code points separated by a minus (and optional space)
-# or a single \p{} per line.
+# Valid types:
+# low generate a macro whose name is 'is_BASE_low' and defines a
+# class that includes only ASCII-range chars. (BASE is the
+# input macro base name.)
+# latin1 generate a macro whose name is 'is_BASE_latin1' and defines a
+# class that includes only upper-Latin1-range chars. It is not
+# designed to take a UTF-8 input parameter.
+# high generate a macro whose name is 'is_BASE_high' and defines a
+# class that includes all relevant code points that are above
+# the Latin1 range. This is for very specialized uses only.
+# It is designed to take only an input UTF-8 parameter.
+# utf8 generate a macro whose name is 'is_BASE_utf8' and defines a
+# class that includes all relevant characters that aren't ASCII.
+# It is designed to take only an input UTF-8 parameter.
+# LATIN1 generate a macro whose name is 'is_BASE_latin1' and defines a
+# class that includes both ASCII and upper-Latin1-range chars.
+# It is not designed to take a UTF-8 input parameter.
+# UTF8 generate a macro whose name is 'is_BASE_utf8' and defines a
+# class that can include any code point, adding the 'low' ones
+# to what 'utf8' works on. It is designed to take only an input
+# UTF-8 parameter.
+# generic generate a macro whose name is 'is_BASE". It has a 2nd,
+# boolean, parameter which indicates if the first one points to
+# a UTF-8 string or not. Thus it works in all circumstances.
+# cp generate a macro whose name is 'is_BASE_cp' and defines a
+# class that returns true if the UV parameter is a member of the
+# class; false if not.
+# cp_high like cp, but it is assumed that it is known that the UV
+# parameter is above Latin1. The name of the generated macro is
+# 'is_BASE_cp_high'. This is different from high-cp, derived
+# below.
+# A macro of the given type is generated for each type listed in the input.
+# The default return value is the number of octets read to generate the match.
+# Append "-cp" to the type to have it instead return the matched codepoint.
+# The macro name is changed to 'what_BASE...'. See pod for
+# caveats
+# Appending '-both" instead adds an extra parameter to the end of the argument
+# list, which is a pointer as to where to store the number of
+# bytes matched, while also returning the code point. The macro
+# name is changed to 'what_len_BASE...'. See pod for caveats
#
-# This is no longer used, but retained in case it is needed some day. Put the
-# lines below under __DATA__
+# Valid modifiers:
+# safe The input string is not necessarily valid UTF-8. In
+# particular an extra parameter (always the 2nd) to the macro is
+# required, which points to one beyond the end of the string.
+# The macro will make sure not to read off the end of the
+# string. In the case of non-UTF8, it makes sure that the
+# string has at least one byte in it. The macro name has
+# '_safe' appended to it.
+# fast The input string is valid UTF-8. No bounds checking is done,
+# and the macro can make assumptions that lead to faster
+# execution.
+# only_ascii_platform Skip this definition if this program is being run on
+# a non-ASCII platform.
+# only_ebcdic_platform Skip this definition if this program is being run on
+# a non-EBCDIC platform.
+# No modifier need be specified; fast is assumed for this case. If both
+# 'fast', and 'safe' are specified, two macros will be created for each
+# 'type'.
+#
+# If run on a non-ASCII platform will automatically convert the Unicode input
+# to native. The documentation above is slightly wrong in this case. 'low'
+# actually refers to code points whose UTF-8 representation is the same as the
+# non-UTF-8 version (invariants); and 'latin1' refers to all the rest of the
+# code points less than 256.
+
+1; # in the unlikely case we are being used as a module
+
+__DATA__
+# This is no longer used, but retained in case it is needed some day.
# TRICKYFOLD: Problematic fold case letters. When adding to this list, also should add them to regcomp.c and fold_grind.t
# => generic cp generic-cp generic-both :fast safe
# 0x00DF # LATIN SMALL LETTER SHARP S
# 0x1FD3 # GREEK SMALL LETTER IOTA WITH DIALYTIKA AND OXIA; maps same as 0390
# 0x1FE3 # GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND OXIA; maps same as 03B0
-1; # in the unlikely case we are being used as a module
-
-__DATA__
LNBREAK: Line Break: \R
=> generic UTF8 LATIN1 :fast safe
"\x0D\x0A" # CRLF - Network (Windows) line ending
\p{VertSpace}
HORIZWS: Horizontal Whitespace: \h \H
-=> generic UTF8 LATIN1 cp :fast safe
+=> generic UTF8 LATIN1 high cp cp_high :fast safe
\p{HorizSpace}
VERTWS: Vertical Whitespace: \v \V
-=> generic UTF8 LATIN1 cp :fast safe
+=> generic UTF8 high LATIN1 cp cp_high :fast safe
\p{VertSpace}
+
+XDIGIT: Hexadecimal digits
+=> UTF8 high cp_high :fast
+\p{XDigit}
+
+XPERLSPACE: \p{XPerlSpace}
+=> generic UTF8 high cp_high :fast
+\p{XPerlSpace}
+
+REPLACEMENT: Unicode REPLACEMENT CHARACTER
+=> UTF8 :safe
+0xFFFD
+
+NONCHAR: Non character code points
+=> UTF8 :fast
+\p{Nchar}
+
+SURROGATE: Surrogate characters
+=> UTF8 :fast
+\p{Gc=Cs}
+
+GCB_L: Grapheme_Cluster_Break=L
+=> UTF8 :fast
+\p{_X_GCB_L}
+
+GCB_LV_LVT_V: Grapheme_Cluster_Break=(LV or LVT or V)
+=> UTF8 :fast
+\p{_X_LV_LVT_V}
+
+GCB_Prepend: Grapheme_Cluster_Break=Prepend
+=> UTF8 :fast
+\p{_X_GCB_Prepend}
+
+GCB_RI: Grapheme_Cluster_Break=RI
+=> UTF8 :fast
+\p{_X_RI}
+
+GCB_SPECIAL_BEGIN_START: Grapheme_Cluster_Break=special_begin_starts
+=> UTF8 :fast
+\p{_X_Special_Begin_Start}
+
+GCB_T: Grapheme_Cluster_Break=T
+=> UTF8 :fast
+\p{_X_GCB_T}
+
+GCB_V: Grapheme_Cluster_Break=V
+=> UTF8 :fast
+\p{_X_GCB_V}
+
+# This program was run with this enabled, and the results copied to utf8.h;
+# then this was commented out because it takes so long to figure out these 2
+# million code points. The results would not change unless utf8.h decides it
+# wants a maximum other than 4 bytes, or this program creates better
+# optimizations
+#UTF8_CHAR: Matches utf8 from 1 to 4 bytes
+#=> UTF8 :safe only_ascii_platform
+#0x0 - 0x1FFFFF
+
+# This hasn't been commented out, because we haven't an EBCDIC platform to run
+# it on, and the 3 types of EBCDIC allegedly supported by Perl would have
+# different results
+UTF8_CHAR: Matches utf8 from 1 to 5 bytes
+=> UTF8 :safe only_ebcdic_platform
+0x0 - 0x3FFFFF:
+
+QUOTEMETA: Meta-characters that \Q should quote
+=> high :fast
+\p{_Perl_Quotemeta}
+
+MULTI_CHAR_FOLD: multi-char strings that are folded to by a single character
+=> UTF8 :safe
+do regen/regcharclass_multi_char_folds.pl
+
+# 1 => All folds
+®charclass_multi_char_folds::multi_char_folds(1)
+
+MULTI_CHAR_FOLD: multi-char strings that are folded to by a single character
+=> LATIN1 :safe
+
+®charclass_multi_char_folds::multi_char_folds(0)
+# 0 => Latin1-only
+
+PATWS: pattern white space
+=> generic cp : fast safe
+\p{PatWS}
https://perl5.git.perl.org / perl5.git / blobdiff