X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2Futf8_heavy.pl;h=1839d2144f125697a486c0f01951a379475e8dff;hb=83272a45226e83bd136d713158e9b44ace2dbc8d;hp=e82c06b52056c270753ad6d4d02ca0228848eb8a;hpb=cf25bb620ab43f67a0edb45a418782cd53e76918;p=p5sagit%2Fp5-mst-13.2.git diff --git a/lib/utf8_heavy.pl b/lib/utf8_heavy.pl index e82c06b..1839d21 100644 --- a/lib/utf8_heavy.pl +++ b/lib/utf8_heavy.pl @@ -6,37 +6,35 @@ 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 +## + 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 ## 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+//; @@ -44,122 +42,60 @@ sub SWASHNEW { print "type = $type\n" if DEBUG; - my $file; - ## Figure out what file to load to get the data.... GETFILE: { ## - ## First, see if it's an "Is" name (the 'Is' is optional) + ## 'Is' is always optional, so if it's there, remove it. + ## Same with 'Category=' and 'Script='. ## - ## Because we check "Is" names first, they have precidence over - ## "In" names. For example, "Greek" is both a script and a - ## block. "IsGreek" always gets the script, while "InGreek" - ## always gets the block. "Greek" gets the script because we - ## check "Is" names first. + ## 'Block=' is replaced by 'In'. ## - if ($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; - } + $type =~ s/^Is(?:\s+|[-_])?//i + or + $type =~ s/^Category\s*=\s*//i + or + $type =~ s/^Script\s*=\s*//i + or + $type =~ s/^Block\s*=\s*/In/i; - ## - ## 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 (exists $utf8::IsPat{$prefix}) - { - while (my ($pat, $name) = each %{$utf8::IsPat{$prefix}}) - { - 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"; - last GETFILE; - } - } - } + ## + ## See if it's in the direct mapping table. + ## + require "unicore/Exact.pl"; + if (my $base = $utf8::Exact{$type}) { + $file = "unicore/lib/$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; - } + my $canonical = lc $type; + $canonical =~ s/(?<=[a-z\d])(?:\s+|[-_])(?=[a-z\d])//g; + print "canonical = $canonical\n" if DEBUG; - my $prefix = substr(lc($intype), 0, 2); - if (exists $utf8::InPat{$prefix}) - { - print "inprefix = $prefix, In = $intype\n" if DEBUG; - while (my ($pat, $name) = each %{$utf8::InPat{$prefix}}) - { - 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; - last GETFILE; - } + require "unicore/Canonical.pl"; + if (my $base = $utf8::Canonical{$canonical}) { + $file = "unicore/lib/$base.pl"; + last GETFILE; + } + + ## + ## It could be a user-defined property. + ## + + if ($type =~ /^I[ns](\w+)$/) { + my @caller = caller(1); + + if (defined $caller[0]) { + my $prop = $caller[0] . "::" . $type; + + if (exists &{$prop}) { + no strict 'refs'; + + $list = &{$prop}; + last GETFILE; } } } @@ -178,19 +114,36 @@ 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), 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 $ORIG = $list; if ($list) { my @tmp = split(/^/m, $list); my %seen; @@ -231,7 +184,14 @@ sub SWASHNEW { print STDERR "$1 => $2\n" if DEBUG; if ($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 = $c->SWASHNEW($t, "", 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}; } @@ -240,8 +200,7 @@ 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, @@ -249,6 +208,12 @@ 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. @@ -306,7 +271,7 @@ 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) { my $min = hex $1; my $max = (defined $2 ? hex $2 : $min); next if $max < $start;