X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2Futf8_heavy.pl;h=b6fdeb997b3407bb003b5d3d884cc86c4ed47270;hb=8af3c3489f0c9ca8d37ebae27991ac4f34dfdb2f;hp=3f14afece222415fac44222bed1c0313b7451a9e;hpb=03138300d2ba00b7ff75af983974ea0770374802;p=p5sagit%2Fp5-mst-13.2.git diff --git a/lib/utf8_heavy.pl b/lib/utf8_heavy.pl index 3f14afe..b6fdeb9 100644 --- a/lib/utf8_heavy.pl +++ b/lib/utf8_heavy.pl @@ -6,37 +6,38 @@ 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 $^D; print STDERR "SWASHNEW @_\n" if DEBUG; - ## check to see if we've already got it. - { - no strict 'refs'; - if ($type and ref ${"${class}::{$type}"} eq $class) { - warn qq/Found \${"${class}::{$type}"}\n/ if DEBUG; - return ${"${class}::{$type}"}; - } - } - ## ## Get the list of codepoints for the type. ## Called from utf8.c ## ## Given a $type, our goal is to fill $list with the set of codepoint - ## ranges. As we try various interpretations of $type, sometimes we'll - ## end up with the $list directly, and sometimes we'll end up with a - ## $file name that holds the list data. + ## ranges. ## ## To make the parsing of $type clear, this code takes the a rather - ## unorthadox approach of last'ing out of the block once we have the + ## 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+//; @@ -44,137 +45,110 @@ sub SWASHNEW { print "type = $type\n" if DEBUG; - my $file; - ## Figure out what file to load to get the data.... GETFILE: { + ## + ## It could be a user-defined property. + ## + + my $caller1 = $type =~ s/(.+)::// ? $1 : caller(1); + + if (defined $caller1 && $type =~ /^(?:\w+)$/) { + my $prop = "${caller1}::$type"; + if (exists &{$prop}) { + no strict 'refs'; + + $list = &{$prop}; + last GETFILE; + } + } + + 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; + $t =~ tr/ _-//d; + + if ($PropValueAlias{gc}{$t} or $PropValueAlias{sc}{$t}) { + $file = "unicore/lib/gc_sc/$PVA_abbr_map{gc_sc}{$t}.pl"; + last GETFILE; + } + } + ## - ## First, see if it's an "Is" name (the 'Is' is optional) - ## - ## Because we check "Is" names first, they have precidence over - ## "In" names. For example, "Greek" is both a script and a - ## block. "IsGreek" always gets the script, while "InGreek" - ## always gets the block. "Greek" gets the script because we - ## check "Is" names first. + ## See if it's in the direct mapping table. ## - if ($type =~ m{^ - ## "Is" prefix, or "Script=" or "Category=" - (?: Is [- _]? | (?:Script|Category)\s*=\s* )? - ## name to check in the "Is" symbol table. - ([A-Z].*) - $ - }ix) - { - my $istype = $1; - ## - ## Input ($type) Name To Check ($istype) - ## ------------- ----------------------- - ## IsLu Lu - ## Lu Lu - ## Category = Lu Lu - ## Foo Foo - ## Script = Greek Greek - ## - - print "istype = $istype\n" if DEBUG; - - ## Load "Is" mapping data, if not yet loaded. - do "unicore/Is.pl" if not defined %utf8::Is; - - ## - ## If the "Is" mapping data has an exact match, it points - ## to the file we need. - ## - if (exists $utf8::Is{$istype}) - { - $file = "unicore/Is/$utf8::Is{$istype}.pl"; - last GETFILE; - } - - ## - ## Need to look at %utf8::IsPat (loaded from "unicore/Is.pl") - ## to see if there's a regex that matches this $istype. - ## If so, the associated name is the file we need. - ## - my $prefix = substr(lc($istype), 0, 2); - if (my $hashref = $utf8::IsPat{$prefix}) - { - while (my ($pat, $name) = each %{$hashref}) - { - print "isprefix = $prefix, Is = $istype, pat = $pat\n" if DEBUG; - ## - ## The following regex probably need not be cached, - ## since every time there's a match, the results of - ## the entire call to SWASHNEW() is cached, so there's - ## a very limited number of times any one $pat will - ## be evaluated as a regex, at least with "reasonable" - ## code that doesn't try a baziilion \p{Random} names. - ## - if ($istype =~ /^$pat$/i) - { - $file = "unicore/Is/$name.pl"; - keys %{$hashref}; ## reset the 'each' above - last GETFILE; - } - } - } + require "unicore/Exact.pl"; + if (my $base = $utf8::Exact{$type}) { + $file = "unicore/lib/gc_sc/$base.pl"; + last GETFILE; } ## - ## Couldn't find via "Is" -- let's try via "In"..... + ## If not there exactly, try the canonical form. The canonical + ## form is lowercased, with any separators (\s+|[-_]) removed. ## - if ($type =~ m{^ - ( In(?!herited$)[- _]? | Block\s*=\s*)? - ([A-Z].*) - $ - }xi) - { - my $intype = $2; - print "intype = $intype\n" if DEBUG; - - ## - ## Input ($type) Name To Check ($intype) - ## ------------- ----------------------- - ## Inherited Inherited - ## InGreek Greek - ## Block = Greek Greek - ## - - ## Load "In" mapping data, if not yet loaded. - do "unicore/In.pl" if not defined %utf8::In; - - ## If there's a direct match, it points to the file we need - if (exists $utf8::In{$intype}) { - $file = "unicore/In/$utf8::In{$intype}.pl"; - last GETFILE; - } - - ## - ## Need to look at %utf8::InPat (loaded from "unicore/In.pl") - ## to see if there's a regex that matches this $intype. - ## If so, the associated name is the file we need. - ## - my $prefix = substr(lc($intype), 0, 2); - if (my $hashref = $utf8::InPat{$prefix}) - { - print "inprefix = $prefix, In = $intype\n" if DEBUG; - while (my ($pat, $name) = each %{$hashref}) - { - print "inprefix = $prefix, In = $intype, k = $pat\n" if DEBUG; - if ($intype =~ /^$pat$/i) { - $file = "unicore/In/$name.pl"; - print "inprefix = $prefix, In = $intype, k = $pat, file = $file\n" if DEBUG; - keys %{$hashref}; ## reset the 'each' above - last GETFILE; - } - } - } + my $canonical = lc $type; + $canonical =~ s/(?<=[a-z\d])(?:\s+|[-_])(?=[a-z\d])//g; + print "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 "To" name (e.g. "ToLower") + ## 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 ($type =~ /^To([A-Z][A-Za-z]+)$/) + if ($type =~ /^To(Digit|Fold|Lower|Title|Upper)$/) { $file = "unicore/To/$1.pl"; ## would like to test to see if $file actually exists.... @@ -185,27 +159,46 @@ sub SWASHNEW { ## If we reach this line, it's because we couldn't figure ## out what to do with $type. Ouch. ## - croak("Can't find Unicode character property \"$type\""); + + return $type; } - ## - ## If we reach here, it was due to a 'last GETFILE' above, so we - ## have a filename, so now we load it. - ## - $list = do $file; + if (defined $file) { + print "found it (file='$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 $class) + { + print "Returning cached '$file' for \\p{$type}\n" if DEBUG; + return $Cache{$class, $file}; + } + + $list = do $file; + } + + $ListSorted = 1; ## we know that these lists are sorted } my $extras; - my $bits; + my $bits = 0; + 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) { @@ -215,10 +208,10 @@ sub SWASHNEW { if ($minbits < 32) { 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; } @@ -236,9 +229,19 @@ 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, "", 0, 0, 0); + } + elsif (exists &$name) { + $subobj = utf8->SWASHNEW($name, "", 0, 0, 0); + } + elsif ($c =~ /^([0-9a-fA-F]+)/) { + $subobj = utf8->SWASHNEW("", $c, 0, 0, 0); + } + return $subobj unless ref $subobj; push @extras, $name => $subobj; $bits = $subobj->{BITS} if $bits < $subobj->{BITS}; } @@ -247,8 +250,7 @@ sub SWASHNEW { print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG; - no strict 'refs'; - ${"${class}::{$type}"} = bless { + my $SWASH = bless { TYPE => $type, BITS => $bits, EXTRAS => $extras, @@ -256,6 +258,12 @@ sub SWASHNEW { NONE => $none, @extras, } => $class; + + if ($file) { + $Cache{$class, $file} = $SWASH; + } + + return $SWASH; } # NOTE: utf8.c:swash_init() assumes entries are never modified once generated. @@ -280,10 +288,15 @@ sub SWASHGET { 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 $3; + while (/^([0-9a-fA-F]+)(?:[ \t]([0-9a-fA-F]+)?)?(?:[ \t]([0-9a-fA-F]+))?/mg) { + chomp; + my ($a, $b, $c) = ($1, $2, $3); + croak "$type: illegal mapping '$_'" + if $type =~ /^To/ && + !(defined $a && defined $c); + my $min = hex $a; + my $max = defined $b ? hex $b : $min; + my $val = defined $c ? hex $c : 0; next if $max < $start; print "$min $max $val\n" if DEBUG; if ($none) { @@ -313,9 +326,10 @@ sub SWASHGET { } else { LINE: - while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+))?/mg) { + while (/^([0-9a-fA-F]+)(?:[ \t]+([0-9a-fA-F]+))?/mg) { + chomp; my $min = hex $1; - my $max = (defined $2 ? hex $2 : $min); + my $max = defined $2 ? hex $2 : $min; next if $max < $start; if ($min < $start) { $min = $start; @@ -330,7 +344,7 @@ sub SWASHGET { } for my $x ($self->{EXTRAS}) { pos $x = 0; - while ($x =~ /^([-+!])(.*)/mg) { + while ($x =~ /^([-+!&])(.*)/mg) { my $char = $1; my $name = $2; print STDERR "INDIRECT $1 $2\n" if DEBUG; @@ -371,6 +385,18 @@ sub SWASHGET { } } } + elsif ($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) {