X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2Futf8_heavy.pl;h=de4d01d95c6887de6379e433c847842a47eee7c9;hb=4bd7b839c96352aa075f714f4c1aaa963c29018e;hp=9e18772b615662c97d81bdc9af4e199e4636c91a;hpb=76ccdbe266e63a2a3ac21a782e44a6b13093ac7f;p=p5sagit%2Fp5-mst-13.2.git diff --git a/lib/utf8_heavy.pl b/lib/utf8_heavy.pl index 9e18772..de4d01d 100644 --- a/lib/utf8_heavy.pl +++ b/lib/utf8_heavy.pl @@ -1,119 +1,219 @@ package utf8; +use strict; +use warnings; sub DEBUG () { 0 } sub DESTROY {} +my %Cache; + +our (%PropertyAlias, %PA_reverse, %PropValueAlias, %PVA_reverse, %PVA_abbr_map); + 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. +## + sub SWASHNEW { my ($class, $type, $list, $minbits, $none) = @_; local $^D = 0 if $^D; print STDERR "SWASHNEW @_\n" if DEBUG; - if ($type and ref ${"${class}::{$type}"} eq $class) { - warn qq/Found \${"${class}::{$type}"}\n/ if DEBUG; - return ${"${class}::{$type}"}; # Already there... - } + ## + ## 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. + ## + ## 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 $ListSorted = 0; + + if ($type) + { + $type =~ s/^\s+//; + $type =~ s/\s+$//; + + print STDERR "type = $type\n" if DEBUG; + + GETFILE: + { + ## + ## It could be a user-defined property. + ## + + my $caller1 = $type =~ s/(.+)::// ? $1 : caller(1); - if ($type) { - $type =~ s/^\s+//; - $type =~ s/\s+$//; - - print "type = $type\n" if DEBUG; - - my $file; - - unless (defined $file) { - defined %utf8::Is || do "unicore/Is.pl"; - if ($type =~ /^(?:Is|Category\s*=\s*)?[- _]?([A-Z].*)$/i) { - my $istype = $1; - print "istype = $istype\n" if DEBUG; - unless ($list = do "unicore/Is/$istype.pl") { - if (exists $utf8::Is{$istype}) { - $file = "unicore/Is/$utf8::Is{$istype}"; - } else { - my $isprefix = substr(lc($istype), 0, 2); - print "isprefix = $isprefix\n" if DEBUG; - if (exists $utf8::IsPat{$isprefix}) { - my $Is = $istype; - print "isprefix = $isprefix, Is = $Is\n" if DEBUG; - for my $k (keys %{$utf8::IsPat{$isprefix}}) { - print "isprefix = $isprefix, Is = $Is, k = $k\n" if DEBUG; - if ($Is =~ /^$k$/i) { - $file = "unicore/Is/$utf8::IsPat{$isprefix}->{$k}"; - print "isprefix = $isprefix, Is = $Is, k = $k, file = $file\n" if DEBUG; - last; - } - } - } - } + if (defined $caller1 && $type =~ /^(?:\w+)$/) { + my $prop = "${caller1}::$type"; + if (exists &{$prop}) { + no strict 'refs'; + + $list = &{$prop}; + last GETFILE; } } - - unless (defined $file) { - defined %utf8::In || do "unicore/In.pl"; - $type = 'Lampersand' if $type =~ /^(?:Is)?L&$/; - $type = 'Assigned' if $type =~ /^(?:Is)?Assigned$/i; - $type = 'Unassigned' if $type =~ /^(?:Is)?Unassigned$/i; - if ($type =~ /^(In|(?:Script|Block)\s*=\s*)?[- _]?(?!herited$)(.+)/i) { - my $incat = $1; - my $intype = $2; - print "incat = $incat, intype = $intype\n" if DEBUG; - if (exists $utf8::In{$intype}) { - $file = "unicore/In/$utf8::In{$intype}"; - } else { - my $inprefix = substr(lc($intype), 0, 2); - print "inprefix = $inprefix\n" if DEBUG; - if (exists $utf8::InPat{$inprefix}) { - my $In = $intype; - print "inprefix = $inprefix, In = $In\n" if DEBUG; - for my $k (keys %{$utf8::InPat{$inprefix}}) { - print "inprefix = $inprefix, In = $In, k = $k\n" if DEBUG; - if ($In =~ /^$k$/i) { - my $i = $utf8::InPat{$inprefix}->{$k}; - print "inprefix = $inprefix, In = $In, k = $k, i = $i\n" if DEBUG; - next if $incat =~ /^S/ && - !exists $utf8::InScript{$i}; - next if $incat =~ /^B/ && - !exists $utf8::InBlock{$i}; - $file = "unicore/In/$i"; - print "inprefix = $inprefix, In = $In, k = $k, file = $file\n" if DEBUG; - last; - } - } - } - } + + my $wasIs; + + ($wasIs = $type =~ s/^Is(?:\s+|[-_])?//i) + or + $type =~ s/^(?:(?:General(?:\s+|_)?)?Category|gc)\s*[:=]\s*//i + or + $type =~ s/^(?:Script|sc)\s*[:=]\s*//i + or + $type =~ s/^Block\s*[:=]\s*/In/i; + + + ## + ## See if it's in some enumeration. + ## + require "unicore/PVA.pl"; + if ($type =~ /^([\w\s]+)[:=]\s*(.*)/) { + my ($enum, $val) = (lc $1, lc $2); + $enum =~ tr/ _-//d; + $val =~ tr/ _-//d; + + my $pa = $PropertyAlias{$enum} ? $enum : $PA_reverse{$enum}; + my $f = $PropValueAlias{$pa}{$val} ? $val : $PVA_reverse{$pa}{lc $val}; + + if ($pa and $f) { + $pa = "gc_sc" if $pa eq "gc" or $pa eq "sc"; + $file = "unicore/lib/$pa/$PVA_abbr_map{$pa}{lc $f}.pl"; + last GETFILE; } } + else { + my $t = lc $type; + $t =~ tr/ _-//d; - unless (defined $file) { - if ($type =~ /^To([A-Z][A-Za-z]+)$/) { - $file = "unicore/To/$1"; + if ($PropValueAlias{gc}{$t} or $PropValueAlias{sc}{$t}) { + $file = "unicore/lib/gc_sc/$PVA_abbr_map{gc_sc}{$t}.pl"; + last GETFILE; } } - } + + ## + ## See if it's in the direct mapping table. + ## + require "unicore/Exact.pl"; + if (my $base = $utf8::Exact{$type}) { + $file = "unicore/lib/gc_sc/$base.pl"; + last GETFILE; + } + + ## + ## If not there exactly, try the canonical form. The canonical + ## form is lowercased, with any separators (\s+|[-_]) removed. + ## + my $canonical = lc $type; + $canonical =~ s/(?<=[a-z\d])(?:\s+|[-_])(?=[a-z\d])//g; + print STDERR "canonical = $canonical\n" if DEBUG; + + require "unicore/Canonical.pl"; + if (my $base = ($utf8::Canonical{$canonical} || $utf8::Canonical{ lc $utf8::PropertyAlias{$canonical} })) { + $file = "unicore/lib/gc_sc/$base.pl"; + last GETFILE; + } + + ## + ## 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/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. + ## + + return $type; + } if (defined $file) { - $list = do "$file.pl"; + print STDERR "found it (file='$file')\n" if DEBUG; + + ## + ## If we reach here, it was due to a 'last GETFILE' above + ## (exception: user-defined properties and mappings), so we + ## have a filename, so now we load it if we haven't already. + ## If we have, return the cached results. The cache key is the + ## file to load. + ## + if ($Cache{$file} and ref($Cache{$file}) eq $class) { + print STDERR "Returning cached '$file' for \\p{$type}\n" if DEBUG; + return $Cache{$class, $file}; + } + + $list = do $file; } - croak("Can't find Unicode character property \"$type\"") - unless $list; + $ListSorted = 1; ## we know that these lists are sorted } my $extras; - my $bits; - + my $bits = $minbits; + + my $ORIG = $list; if ($list) { my @tmp = split(/^/m, $list); my %seen; no warnings; $extras = join '', grep /^[^0-9a-fA-F]/, @tmp; $list = join '', - sort { hex $a <=> hex $b } - grep {/^([0-9a-fA-F]+)/ and not $seen{$1}++} @tmp; # XXX doesn't do ranges right + map { $_->[1] } + sort { $a->[0] <=> $b->[0] } + map { /^([0-9a-fA-F]+)/; [ hex($1), $_ ] } + grep { /^([0-9a-fA-F]+)/ and not $seen{$1}++ } @tmp; # XXX doesn't do ranges right } if ($none) { @@ -121,21 +221,20 @@ sub SWASHNEW { $list =~ s/\tXXXX$/\t$hextra/mg; } - if ($minbits < 32) { + 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) { + while ($list =~ /^([0-9a-fA-F]+)(?:[\t]([0-9a-fA-F]+)?)(?:[ \t]([0-9a-fA-F]+))?/mg) { my $min = hex $1; - my $max = hex(defined $2 ? $2 : $1); - my $val = hex(defined $3 ? $3 : ""); + my $max = defined $2 ? hex $2 : $min; + my $val = defined $3 ? hex $3 : 0; $val += $max - $min if defined $3; $top = $val if $val > $top; } - $bits = + my $topbits = $top > 0xffff ? 32 : - $top > 0xff ? 16 : - $top > 1 ? 8 : 1 + $top > 0xff ? 16 : 8; + $bits = $topbits if $bits < $topbits; } - $bits = $minbits if $bits < $minbits; my @extras; for my $x ($extras) { @@ -144,9 +243,19 @@ sub SWASHNEW { my $char = $1; my $name = $2; print STDERR "$1 => $2\n" if DEBUG; - if ($char =~ /[-+!]/) { + if ($char =~ /[-+!&]/) { my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really - my $subobj = $c->SWASHNEW($t, "", 0, 0, 0); + 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); + } + return $subobj unless ref $subobj; push @extras, $name => $subobj; $bits = $subobj->{BITS} if $bits < $subobj->{BITS}; } @@ -155,7 +264,7 @@ sub SWASHNEW { print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG; - ${"${class}::{$type}"} = bless { + my $SWASH = bless { TYPE => $type, BITS => $bits, EXTRAS => $extras, @@ -163,131 +272,14 @@ sub SWASHNEW { NONE => $none, @extras, } => $class; -} - -# NOTE: utf8.c:swash_init() assumes entries are never modified once generated. -sub SWASHGET { - # See utf8.c:Perl_swash_fetch for problems with this interface. - my ($self, $start, $len) = @_; - local $^D = 0 if $^D; - my $type = $self->{TYPE}; - my $bits = $self->{BITS}; - my $none = $self->{NONE}; - print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if DEBUG; - my $end = $start + $len; - my $swatch = ""; - my $key; - vec($swatch, $len - 1, $bits) = 0; # Extend to correct length. - if ($none) { - for $key (0 .. $len - 1) { vec($swatch, $key, $bits) = $none } + if ($file) { + $Cache{$class, $file} = $SWASH; } - for ($self->{LIST}) { - pos $_ = 0; - if ($bits > 1) { - LINE: - while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) { - my $min = hex $1; - my $max = (defined $2 ? hex $2 : $min); - my $val = hex $3; - next if $max < $start; - print "$min $max $val\n" if DEBUG; - if ($none) { - if ($min < $start) { - $val += $start - $min if $val < $none; - $min = $start; - } - for ($key = $min; $key <= $max; $key++) { - last LINE if $key >= $end; - print STDERR "$key => $val\n" if DEBUG; - vec($swatch, $key - $start, $bits) = $val; - ++$val if $val < $none; - } - } - else { - if ($min < $start) { - $val += $start - $min; - $min = $start; - } - for ($key = $min; $key <= $max; $key++, $val++) { - last LINE if $key >= $end; - print STDERR "$key => $val\n" if DEBUG; - vec($swatch, $key - $start, $bits) = $val; - } - } - } - } - else { - LINE: - while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+))?/mg) { - my $min = hex $1; - my $max = (defined $2 ? hex $2 : $min); - next if $max < $start; - if ($min < $start) { - $min = $start; - } - for ($key = $min; $key <= $max; $key++) { - last LINE if $key >= $end; - print STDERR "$key => 1\n" if DEBUG; - vec($swatch, $key - $start, 1) = 1; - } - } - } - } - for my $x ($self->{EXTRAS}) { - pos $x = 0; - while ($x =~ /^([-+!])(.*)/mg) { - my $char = $1; - my $name = $2; - print STDERR "INDIRECT $1 $2\n" if DEBUG; - my $otherbits = $self->{$name}->{BITS}; - croak("SWASHGET size mismatch") if $bits < $otherbits; - my $other = $self->{$name}->SWASHGET($start, $len); - if ($char eq '+') { - if ($bits == 1 and $otherbits == 1) { - $swatch |= $other; - } - else { - for ($key = 0; $key < $len; $key++) { - vec($swatch, $key, $bits) = vec($other, $key, $otherbits); - } - } - } - elsif ($char eq '!') { - if ($bits == 1 and $otherbits == 1) { - $swatch |= ~$other; - } - else { - for ($key = 0; $key < $len; $key++) { - if (!vec($other, $key, $otherbits)) { - vec($swatch, $key, $bits) = 1; - } - } - } - } - elsif ($char eq '-') { - if ($bits == 1 and $otherbits == 1) { - $swatch &= ~$other; - } - else { - for ($key = 0; $key < $len; $key++) { - if (vec($other, $key, $otherbits)) { - vec($swatch, $key, $bits) = 0; - } - } - } - } - } - } - if (DEBUG) { - print STDERR "CELLS "; - for ($key = 0; $key < $len; $key++) { - print STDERR vec($swatch, $key, $bits), " "; - } - print STDERR "\n"; - } - $swatch; + return $SWASH; } +# Now SWASHGET is recasted into a C function S_swash_get (see utf8.c). + 1;