1 ## !!!!!!!!!!!!!! IF YOU MODIFY THIS FILE !!!!!!!!!!!!!!!!!!!!!!!!!
2 ## Any files created or read by this program should be listed in 'mktables.lst'
5 require 5.008; # Needs pack "U". Probably safest to run on 5.8.x
11 ## mktables -- create the runtime Perl Unicode files (lib/unicore/**/*.pl)
12 ## from the Unicode database files (lib/unicore/*.txt).
15 ## "Fuzzy" means this section in Unicode TR18:
17 ## The recommended names for UCD properties and property values are in
18 ## PropertyAliases.txt [Prop] and PropertyValueAliases.txt
19 ## [PropValue]. There are both abbreviated names and longer, more
20 ## descriptive names. It is strongly recommended that both names be
21 ## recognized, and that loose matching of property names be used,
22 ## whereby the case distinctions, whitespace, hyphens, and underbar
25 ## Base names already used in lib/gc_sc (for avoiding 8.3 conflicts)
32 my $MakeTestScript = 0;
35 my $FileList = "$0.lst";
40 my $arg = shift @ARGV;
43 } elsif ($arg eq '-q') {
45 } elsif ($arg eq '-w') {
46 $AlwaysWrite = 1; # update the files even if they havent changed
48 } elsif ($arg eq '-check') {
49 my $this = shift @ARGV;
52 print "Skipping as check params are not the same.\n";
55 } elsif ($arg eq '-maketest') {
57 } elsif ($arg eq '-makelist') {
59 } elsif ($arg eq '-C' && defined ($UseDir = shift)) {
60 -d $UseDir or die "Unknown directory '$UseDir'";
61 } elsif ($arg eq '-L' && defined ($FileList = shift)) {
62 -e $FileList or die "Filelist '$FileList' doesn't appear to exist!";
64 die "usage: $0 [-v|-q|-w|-C dir|-L filelist] [-maketest] [-makelist]\n",
65 " -v : Verbose Mode\n",
67 " -w : Write files regardless\n",
68 " -maketest : Make test script\n",
69 " -makelist : Rewrite the file list based on current setup\n",
70 " -L filelist : Use this file list, (defaults to $0)\n",
71 " -C dir : Change to this directory before proceeding\n",
72 " -check A B : Executes only if A and B are the same\n";
77 print "Reading file list '$FileList'\n"
79 open my $fh,"<",$FileList or die "Failed to read '$FileList':$!";
82 for my $list ( \@input, \@output ) {
85 next if /^ \s* (?: \# .* )? $/x;
87 my ( $file ) = split /\t/, $_;
91 @$list = grep !$dupe{ $_ }++, @$list;
94 die "No input or output files in '$FileList'!"
95 if !@input or !@output;
97 foreach my $file (@output) {
102 print "Expecting ".scalar( @input )." input files. ",
103 "Checking ".scalar( @output )." output files.\n";
105 # we set maxtime to be the youngest input file, including $0 itself.
106 my $maxtime = -M $0; # do this before the chdir!
108 chdir $UseDir or die "Failed to chdir to '$UseDir':$!";
110 foreach my $in (@input) {
112 die "Missing input file '$in'" unless defined $time;
113 $maxtime = $time if $maxtime < $time;
116 # now we check to see if any output files are older than maxtime, if
117 # they are we need to continue on, otherwise we can presumably bail.
119 foreach my $out (@output) {
121 print "'$out' is missing.\n"
126 if ( -M $out > $maxtime ) {
127 print "'$out' is too old.\n"
134 print "Files seem to be ok, not bothering to rebuild.\n";
137 print "Must rebuild tables.\n"
141 print "Not checking filelist.\n";
144 chdir $UseDir or die "Failed to chdir to '$UseDir':$!";
148 foreach my $lib ('To', 'lib',
149 map {File::Spec->catdir("lib",$_)}
150 qw(gc_sc dt bc hst ea jt lb nt ccc)) {
152 mkdir $lib, 0755 or die "mkdir '$lib': $!";
155 my $LastUnicodeCodepoint = 0x10FFFF; # As of Unicode 3.1.1.
158 # !!!!!!! DO NOT EDIT THIS FILE !!!!!!!
159 # This file is built by $0 from e.g. UnicodeData.txt.
160 # Any changes made here will be lost!
165 my $filename = shift;
166 return unless -e $filename;
167 return if CORE::unlink($filename);
168 # We might need write permission
169 chmod 0777, $filename;
170 CORE::unlink($filename) or die "Couldn't unlink $filename: $!\n";
174 ## Given a filename and a reference to an array of lines,
175 ## write the lines to the file only if the contents have not changed.
176 ## Filename can be given as an arrayref of directory names
178 sub WriteIfChanged($\@)
183 $file = File::Spec->catfile(@$file) if ref $file;
185 my $TextToWrite = join '', @$lines;
186 if (open IN, $file) {
188 my $PreviousText = <IN>;
190 if ($PreviousText eq $TextToWrite) {
191 print "$file unchanged.\n" if $Verbose;
192 return unless $AlwaysWrite;
195 force_unlink ($file);
196 if (not open OUT, ">$file") {
197 die "$0: can't open $file for output: $!\n";
199 print "$file written.\n" if $Verbose;
201 print OUT $TextToWrite;
206 ## The main datastructure (a "Table") represents a set of code points that
207 ## are part of a particular quality (that are part of \pL, \p{InGreek},
208 ## etc.). They are kept as ranges of code points (starting and ending of
211 ## For example, a range ASCII LETTERS would be represented as:
212 ## [ [ 0x41 => 0x5A, 'UPPER' ],
213 ## [ 0x61 => 0x7A, 'LOWER, ] ]
215 sub RANGE_START() { 0 } ## index into range element
216 sub RANGE_END() { 1 } ## index into range element
217 sub RANGE_NAME() { 2 } ## index into range element
219 ## Conceptually, these should really be folded into the 'Table' objects
227 ## Turn something like
236 $name =~ s/(?<![a-z])(\w)/\u$1/g;
237 $name =~ s/[-_\s]+//g;
239 $CanonicalToOrig{$name} = $orig if not $CanonicalToOrig{$name};
245 ## Store the alias definitions for later use.
256 ## Most of the work with aliases doesn't occur here,
257 ## but rather in utf8_heavy.pl, which uses PVA.pl,
259 # Placate the warnings about used only once. (They are used again, but
260 # via a typeglob lookup)
261 %utf8::PropertyAlias = ();
262 %utf8::PA_reverse = ();
263 %utf8::PropValueAlias = ();
264 %utf8::PVA_reverse = ();
265 %utf8::PVA_abbr_map = ();
267 open PA, "< PropertyAliases.txt"
268 or confess "Can't open PropertyAliases.txt: $!";
274 my ($abbrev, $name) = split /\s*;\s*/;
275 next if $abbrev eq "n/a";
276 $PropertyAlias{$abbrev} = $name;
277 $PA_reverse{$name} = $abbrev;
279 # The %utf8::... versions use japhy's code originally from utf8_pva.pl
280 # However, it's moved here so that we build the tables at runtime.
281 tr/ _-//d for $abbrev, $name;
282 $utf8::PropertyAlias{lc $abbrev} = $name;
283 $utf8::PA_reverse{lc $name} = $abbrev;
287 open PVA, "< PropValueAliases.txt"
288 or confess "Can't open PropValueAliases.txt: $!";
294 my ($prop, @data) = split /\s*;\s*/;
296 if ($prop eq 'ccc') {
297 $PropValueAlias{$prop}{$data[1]} = [ @data[0,2] ];
298 $PVA_reverse{$prop}{$data[2]} = [ @data[0,1] ];
301 next if $data[0] eq "n/a";
302 $PropValueAlias{$prop}{$data[0]} = $data[1];
303 $PVA_reverse{$prop}{$data[1]} = $data[0];
306 shift @data if $prop eq 'ccc';
307 next if $data[0] eq "n/a";
309 $data[1] =~ tr/ _-//d;
310 $utf8::PropValueAlias{$prop}{lc $data[0]} = $data[1];
311 $utf8::PVA_reverse{$prop}{lc $data[1]} = $data[0];
313 my $abbr_class = ($prop eq 'gc' or $prop eq 'sc') ? 'gc_sc' : $prop;
314 $utf8::PVA_abbr_map{$abbr_class}{lc $data[0]} = $data[0];
318 # backwards compatibility for L& -> LC
319 $utf8::PropValueAlias{gc}{'l&'} = $utf8::PropValueAlias{gc}{lc};
320 $utf8::PVA_abbr_map{gc_sc}{'l&'} = $utf8::PVA_abbr_map{gc_sc}{lc};
326 ## Associates a property ("Greek", "Lu", "Assigned",...) with a Table.
329 ## New_Prop(In => 'Greek', $Table, Desc => 'Greek Block', Fuzzy => 1);
331 ## Normally, these parameters are set when the Table is created (when the
332 ## Table->New constructor is called), but there are times when it needs to
333 ## be done after-the-fact...)
337 my $Type = shift; ## "Is" or "In";
341 ## remaining args are optional key/val
344 my $Fuzzy = delete $Args{Fuzzy};
345 my $Desc = delete $Args{Desc}; # description
347 $Name = CanonicalName($Name) if $Fuzzy;
349 ## sanity check a few args
350 if (%Args or ($Type ne 'Is' and $Type ne 'In') or not ref $Table) {
351 confess "$0: bad args to New_Prop"
354 if (not $TableInfo{$Type}->{$Name})
356 $TableInfo{$Type}->{$Name} = $Table;
357 $TableDesc{$Type}->{$Name} = $Desc;
359 $FuzzyNames{$Type}->{$Name} = $Name;
366 ## Creates a new Table object.
368 ## Args are key/value pairs:
369 ## In => Name -- Name of "In" property to be associated with
370 ## Is => Name -- Name of "Is" property to be associated with
371 ## Fuzzy => Boolean -- True if name can be accessed "fuzzily"
372 ## Desc => String -- Description of the property
374 ## No args are required.
381 my $Table = bless [], $class;
383 my $Fuzzy = delete $Args{Fuzzy};
384 my $Desc = delete $Args{Desc};
386 for my $Type ('Is', 'In')
388 if (my $Name = delete $Args{$Type}) {
389 New_Prop($Type => $Name, $Table, Desc => $Desc, Fuzzy => $Fuzzy);
393 ## shouldn't have any left over
395 confess "$0: bad args to Table->New"
403 ## Returns the maximum code point currently in the table.
407 my $last = $_[0]->[-1]; ## last code point
408 confess "oops" unless $last; ## must have code points to have a max
409 return $last->[RANGE_END];
413 ## Replaces the codepoints in the Table with those in the Table given
414 ## as an arg. (NOTE: this is not a "deep copy").
416 sub Table::Replace($$)
418 my $Table = shift; #self
425 ## Given a new code point, make the last range of the Table extend to
426 ## include the new (and all intervening) code points.
428 ## Takes the time to make sure that the extension is valid.
432 my $Table = shift; #self
433 my $codepoint = shift;
435 my $PrevMax = $Table->Max;
437 confess "oops ($codepoint <= $PrevMax)" if $codepoint <= $PrevMax;
439 $Table->ExtendNoCheck($codepoint);
444 ## Given a new code point, make the last range of the Table extend to
445 ## include the new (and all intervening) code points.
447 ## Does NOT check that the extension is valid. Assumes that the caller
448 ## has already made this check.
450 sub Table::ExtendNoCheck
452 ## Optmized adding: Assumes $Table and $codepoint as parms
453 $_[0]->[-1]->[RANGE_END] = $_[1];
457 ## Given a code point range start and end (and optional name), blindly
458 ## append them to the list of ranges for the Table.
460 ## NOTE: Code points must be added in strictly ascending numeric order.
462 sub Table::RawAppendRange
464 my $Table = shift; #self
468 $name = "" if not defined $name; ## warning: $name can be "0"
470 push @$Table, [ $start, # RANGE_START
472 $name ]; # RANGE_NAME
476 ## Given a code point (and optional name), add it to the Table.
478 ## NOTE: Code points must be added in strictly ascending numeric order.
482 my $Table = shift; #self
483 my $codepoint = shift;
485 $name = "" if not defined $name; ## warning: $name can be "0"
488 ## If we've already got a range working, and this code point is the next
489 ## one in line, and if the name is the same, just extend the current range.
491 my $last = $Table->[-1];
494 $last->[RANGE_END] == $codepoint - 1
496 $last->[RANGE_NAME] eq $name)
498 $Table->ExtendNoCheck($codepoint);
502 $Table->RawAppendRange($codepoint, $codepoint, $name);
507 ## Given a code point range starting value and ending value (and name),
508 ## Add the range to teh Table.
510 ## NOTE: Code points must be added in strictly ascending numeric order.
512 sub Table::AppendRange
514 my $Table = shift; #self
518 $name = "" if not defined $name; ## warning: $name can be "0"
520 $Table->Append($start, $name);
521 $Table->Extend($end) if $end > $start;
525 ## Return a new Table that represents all code points not in the Table.
529 my $Table = shift; #self
531 my $New = Table->New();
533 for my $range (@$Table)
535 my $start = $range->[RANGE_START];
536 my $end = $range->[RANGE_END];
537 if ($start-1 >= $max+1) {
538 $New->AppendRange($max+1, $start-1, "");
542 if ($max+1 < $LastUnicodeCodepoint) {
543 $New->AppendRange($max+1, $LastUnicodeCodepoint);
549 ## Merges any number of other tables with $self, returning the new table.
550 ## (existing tables are not modified)
553 ## Args may be Tables, or individual code points (as integers).
555 ## Can be called as either a constructor or a method.
559 shift(@_) if not ref $_[0]; ## if called as a constructor, lose the class
562 ## Accumulate all records from all tables
564 for my $Arg (@Tables)
567 ## arg is a table -- get its ranges
568 push @Records, @$Arg;
570 ## arg is a codepoint, make a range
571 push @Records, [ $Arg, $Arg ]
575 ## sort by range start, with longer ranges coming first.
576 my ($first, @Rest) = sort {
577 ($a->[RANGE_START] <=> $b->[RANGE_START])
579 ($b->[RANGE_END] <=> $b->[RANGE_END])
582 my $New = Table->New();
584 ## Ensuring the first range is there makes the subsequent loop easier
585 $New->AppendRange($first->[RANGE_START],
586 $first->[RANGE_END]);
588 ## Fold in records so long as they add new information.
591 my $start = $set->[RANGE_START];
592 my $end = $set->[RANGE_END];
593 if ($start > $New->Max) {
594 $New->AppendRange($start, $end);
595 } elsif ($end > $New->Max) {
596 $New->ExtendNoCheck($end);
604 ## Given a filename, write a representation of the Table to a file.
605 ## May have an optional comment as a 2nd arg.
606 ## Filename may actually be an arrayref of directories
610 my $Table = shift; #self
611 my $filename = shift;
615 if (defined $comment) {
616 $comment =~ s/\s+\Z//;
617 $comment =~ s/^/# /gm;
618 push @OUT, "#\n$comment\n#\n";
620 push @OUT, "return <<'END';\n";
622 for my $set (@$Table)
624 my $start = $set->[RANGE_START];
625 my $end = $set->[RANGE_END];
626 my $name = $set->[RANGE_NAME];
628 if ($start == $end) {
629 push @OUT, sprintf "%04X\t\t%s\n", $start, $name;
631 push @OUT, sprintf "%04X\t%04X\t%s\n", $start, $end, $name;
637 WriteIfChanged($filename, @OUT);
640 ## This used only for making the test script.
645 return 0 if $code <= 0x0000; ## don't use null
646 return 0 if $code >= $LastUnicodeCodepoint; ## keep in range
647 return 0 if ($code >= 0xD800 and $code <= 0xDFFF); ## no surrogates
648 return 0 if ($code >= 0xFDD0 and $code <= 0xFDEF); ## utf8.c says no good
649 return 0 if (($code & 0xFFFF) == 0xFFFE); ## utf8.c says no good
650 return 0 if (($code & 0xFFFF) == 0xFFFF); ## utf8.c says no good
654 ## Return a code point that's part of the table.
655 ## Returns nothing if the table is empty (or covers only surrogates).
656 ## This used only for making the test script.
659 my $Table = shift; #self
660 for my $set (@$Table) {
661 return $set->[RANGE_END] if IsUsable($set->[RANGE_END]);
666 ## Return a code point that's not part of the table
667 ## Returns nothing if the table covers all code points.
668 ## This used only for making the test script.
669 sub Table::InvalidCode
671 my $Table = shift; #self
673 return 0x1234 if not @$Table;
675 for my $set (@$Table)
677 if (IsUsable($set->[RANGE_END] + 1))
679 return $set->[RANGE_END] + 1;
682 if (IsUsable($set->[RANGE_START] - 1))
684 return $set->[RANGE_START] - 1;
690 ###########################################################################
691 ###########################################################################
692 ###########################################################################
697 ## New_Alias(Is => 'All', SameAs => 'Any', Fuzzy => 1);
699 ## The args must be in that order, although the Fuzzy pair may be omitted.
701 ## This creates 'IsAll' as an alias for 'IsAny'
705 my $Type = shift; ## "Is" or "In"
707 my $SameAs = shift; # expecting "SameAs" -- just ignored
710 ## remaining args are optional key/val
713 my $Fuzzy = delete $Args{Fuzzy};
715 ## sanity check a few args
716 if (%Args or ($Type ne 'Is' and $Type ne 'In') or $SameAs ne 'SameAs') {
717 confess "$0: bad args to New_Alias"
720 $Alias = CanonicalName($Alias) if $Fuzzy;
722 if (not $TableInfo{$Type}->{$Name})
724 my $CName = CanonicalName($Name);
725 if ($TableInfo{$Type}->{$CName}) {
726 confess "$0: Use canonical form '$CName' instead of '$Name' for alias.";
728 confess "$0: don't have original $Type => $Name to make alias\n";
731 if ($TableInfo{$Alias}) {
732 confess "$0: already have original $Type => $Alias; can't make alias";
734 $AliasInfo{$Type}->{$Name} = $Alias;
736 $FuzzyNames{$Type}->{$Alias} = $Name;
742 ## All assigned code points
743 my $Assigned = Table->New(Is => 'Assigned',
744 Desc => "All assigned code points",
747 my $Name = Table->New(); ## all characters, individually by name
748 my $General = Table->New(); ## all characters, grouped by category
752 ## Simple Data::Dumper alike. Good enough for our needs. We can't use the real
753 ## thing as we have to run under miniperl
759 if (ref $item eq 'ARRAY') {
760 push @lines, "[\n", simple_dumper (@$item), "],\n";
761 } elsif (ref $item eq 'HASH') {
762 push @lines, "{\n", simple_dumper (%$item), "},\n";
764 die "Can't cope with $item";
769 $copy =~ s/([\'\\])/\\$1/gs;
770 push @lines, "'$copy',\n";
772 push @lines, "undef,\n";
780 ## Process UnicodeData.txt (Categories, etc.)
782 # These are the character mappings as defined in the POSIX standard
783 # and in the case of PerlSpace and PerlWord as is defined in the test macros
784 # for binary strings. IOW, PerlWord is [A-Za-z_] and PerlSpace is [\f\r\n\t ]
785 # This differs from Word and the existing SpacePerl (note the prefix/suffix difference)
786 # which is basically the Unicode WhiteSpace without the vertical tab included
788 my %TRUE_POSIX_PERL_CC= (
789 PosixAlnum => { map { $_ => 1 } ( 0x0030..0x0039, 0x0041..0x005a, 0x0061..0x007a )},
790 PosixAlpha => { map { $_ => 1 } ( 0x0041..0x005a, 0x0061..0x007a )},
791 # Not Needed: Ascii => { map { $_ => 1 } ( 0x0000..0x007f )},
792 PosixBlank => { map { $_ => 1 } ( 0x0009, 0x0020 )},
793 PosixCntrl => { map { $_ => 1 } ( 0x0000..0x001f, 0x007f )},
794 PosixGraph => { map { $_ => 1 } ( 0x0021..0x007e )},
795 PosixLower => { map { $_ => 1 } ( 0x0061..0x007a )},
796 PosixPrint => { map { $_ => 1 } ( 0x0020..0x007e )},
797 PosixPunct => { map { $_ => 1 } ( 0x0021..0x002f, 0x003a..0x0040, 0x005b..0x0060, 0x007b..0x007e )},
798 PosixSpace => { map { $_ => 1 } ( 0x0009..0x000d, 0x0020 )},
799 PosixUpper => { map { $_ => 1 } ( 0x0041..0x005a )},
800 # Not needed: PosixXdigit => { map { $_ => 1 } ( 0x0030..0x0039, 0x0041..0x0046, 0x0061..0x0066 )},
801 PosixDigit => { map { $_ => 1 } ( 0x0030..0x0039 )},
803 PerlSpace => { map { $_ => 1 } ( 0x0009..0x000a, 0x000c..0x000d, 0x0020 )},
804 PerlWord => { map { $_ => 1 } ( 0x0030..0x0039, 0x0041..0x005a, 0x005f, 0x0061..0x007a )},
807 sub UnicodeData_Txt()
809 my $Bidi = Table->New();
810 my $Deco = Table->New();
811 my $Comb = Table->New();
812 my $Number = Table->New();
813 my $Mirrored = Table->New();#Is => 'Mirrored',
814 #Desc => "Mirrored in bidirectional text",
820 $DC{can} = Table->New();
821 $DC{com} = Table->New();
823 ## Initialize Broken Perl-generated categories
824 ## (Categories from UnicodeData.txt are auto-initialized in gencat)
826 Table->New(Is => 'Alnum', Desc => "[[:Alnum:]]", Fuzzy => 0);
828 Table->New(Is => 'Alpha', Desc => "[[:Alpha:]]", Fuzzy => 0);
830 Table->New(Is => 'ASCII', Desc => "[[:ASCII:]]", Fuzzy => 0);
832 Table->New(Is => 'Blank', Desc => "[[:Blank:]]", Fuzzy => 0);
834 Table->New(Is => 'Cntrl', Desc => "[[:Cntrl:]]", Fuzzy => 0);
836 Table->New(Is => 'Digit', Desc => "[[:Digit:]]", Fuzzy => 0);
838 Table->New(Is => 'Graph', Desc => "[[:Graph:]]", Fuzzy => 0);
840 Table->New(Is => 'Lower', Desc => "[[:Lower:]]", Fuzzy => 0);
842 Table->New(Is => 'Print', Desc => "[[:Print:]]", Fuzzy => 0);
844 Table->New(Is => 'Punct', Desc => "[[:Punct:]]", Fuzzy => 0);
846 Table->New(Is => 'Space', Desc => "[[:Space:]]", Fuzzy => 0);
848 Table->New(Is => 'Title', Desc => "[[:Title:]]", Fuzzy => 0);
850 Table->New(Is => 'Upper', Desc => "[[:Upper:]]", Fuzzy => 0);
852 Table->New(Is => 'XDigit', Desc => "[[:XDigit:]]", Fuzzy => 0);
854 Table->New(Is => 'Word', Desc => "[[:Word:]]", Fuzzy => 0);
856 Table->New(Is => 'SpacePerl', Desc => '\s', Fuzzy => 0);
858 Table->New(Is => 'VertSpace', Desc => '\v', Fuzzy => 0);
860 Table->New(Is => 'HorizSpace', Desc => '\h', Fuzzy => 0);
862 $To{Upper} = Table->New();
863 $To{Lower} = Table->New();
864 $To{Title} = Table->New();
865 $To{Digit} = Table->New();
867 foreach my $cat (keys %TRUE_POSIX_PERL_CC) {
868 $Cat{$cat} = Table->New(Is=>$cat, Fuzzy => 0);
873 my ($name, ## Name ("LATIN CAPITAL LETTER A")
874 $cat, ## Category ("Lu", "Zp", "Nd", etc.)
875 $code, ## Code point (as an integer)
878 my $MajorCat = substr($cat, 0, 1); ## L, M, Z, S, etc
880 $Assigned->$op($code);
881 $Name->$op($code, $name);
882 $General->$op($code, $cat);
884 ## add to the sub category (e.g. "Lu", "Nd", "Cf", ..)
885 $Cat{$cat} ||= Table->New(Is => $cat,
886 Desc => "General Category '$cat'",
888 $Cat{$cat}->$op($code);
890 ## add to the major category (e.g. "L", "N", "C", ...)
891 $Cat{$MajorCat} ||= Table->New(Is => $MajorCat,
892 Desc => "Major Category '$MajorCat'",
894 $Cat{$MajorCat}->$op($code);
896 ($General{$name} ||= Table->New)->$op($code, $name);
898 # 005F: SPACING UNDERSCORE
899 $Cat{Word}->$op($code) if $cat =~ /^[LMN]|Pc/;
900 $Cat{Alnum}->$op($code) if $cat =~ /^[LM]|Nd/;
901 $Cat{Alpha}->$op($code) if $cat =~ /^[LM]/;
904 ($cat =~ /Zs|Zl|Zp/ &&
905 $code != 0x200B) # 200B is ZWSP which is for line break control
906 # and therefore it is not part of "space" even while it is "Zs".
907 || $code == 0x0009 # 0009: HORIZONTAL TAB
908 || $code == 0x000A # 000A: LINE FEED
909 || $code == 0x000B # 000B: VERTICAL TAB
910 || $code == 0x000C # 000C: FORM FEED
911 || $code == 0x000D # 000D: CARRIAGE RETURN
912 || $code == 0x0085 # 0085: NEL
916 $Cat{Space}->$op($code) if $isspace;
918 $Cat{SpacePerl}->$op($code) if $isspace
919 && $code != 0x000B; # Backward compat.
921 $Cat{VertSpace}->$op($code) if grep {$code == $_}
922 ( 0x0A..0x0D,0x85,0x2028,0x2029 );
924 $Cat{HorizSpace}->$op($code) if grep {$code == $_} (
925 0x09, 0x20, 0xa0, 0x1680, 0x180e, 0x2000, 0x2001, 0x2002,
926 0x2003, 0x2004, 0x2005, 0x2006, 0x2007, 0x2008, 0x2009, 0x200a,
927 0x202f, 0x205f, 0x3000
930 $Cat{Blank}->$op($code) if $isspace
931 && !($code == 0x000A ||
938 $Cat{Digit}->$op($code) if $cat eq "Nd";
939 $Cat{Upper}->$op($code) if $cat eq "Lu";
940 $Cat{Lower}->$op($code) if $cat eq "Ll";
941 $Cat{Title}->$op($code) if $cat eq "Lt";
942 $Cat{ASCII}->$op($code) if $code <= 0x007F;
943 $Cat{Cntrl}->$op($code) if $cat =~ /^C/;
944 my $isgraph = !$isspace && $cat !~ /Cc|Cs|Cn/;
945 $Cat{Graph}->$op($code) if $isgraph;
946 $Cat{Print}->$op($code) if $isgraph || $isspace;
947 $Cat{Punct}->$op($code) if $cat =~ /^P/;
949 $Cat{XDigit}->$op($code) if ($code >= 0x30 && $code <= 0x39) ## 0..9
950 || ($code >= 0x41 && $code <= 0x46) ## A..F
951 || ($code >= 0x61 && $code <= 0x66); ## a..f
953 foreach my $cat (keys %TRUE_POSIX_PERL_CC) {
954 if ($TRUE_POSIX_PERL_CC{$cat}{$code}) {
955 $Cat{$cat}->$op($code);
961 ## open ane read file.....
962 if (not open IN, "UnicodeData.txt") {
963 die "$0: UnicodeData.txt: $!\n";
967 ## For building \p{_CombAbove} and \p{_CanonDCIJ}
969 my %_Above_HexCodes; ## Hexcodes for chars with $comb == 230 ("ABOVE")
971 my %CodeToDeco; ## Maps code to decomp. list for chars with first
972 ## decomp. char an "i" or "j" (for \p{_CanonDCIJ})
974 ## This is filled in as we go....
975 my $CombAbove = Table->New(Is => '_CombAbove',
976 Desc => '(for internal casefolding use)',
981 next unless /^[0-9A-Fa-f]+;/;
984 my ($hexcode, ## code point in hex (e.g. "0041")
985 $name, ## character name (e.g. "LATIN CAPITAL LETTER A")
986 $cat, ## category (e.g. "Lu")
987 $comb, ## Canonical combining class (e.t. "230")
988 $bidi, ## directional category (e.g. "L")
989 $deco, ## decomposition mapping
990 $decimal, ## decimal digit value
991 $digit, ## digit value
992 $number, ## numeric value
993 $mirrored, ## mirrored
994 $unicode10, ## name in Unicode 1.0
995 $comment, ## comment field
996 $upper, ## uppercase mapping
997 $lower, ## lowercase mapping
998 $title, ## titlecase mapping
999 ) = split(/\s*;\s*/);
1001 # Note that in Unicode 3.2 there will be names like
1002 # LINE FEED (LF), which probably means that \N{} needs
1003 # to cope also with LINE FEED and LF.
1004 $name = $unicode10 if $name eq '<control>' && $unicode10 ne '';
1006 my $code = hex($hexcode);
1008 if ($comb and $comb == 230) {
1009 $CombAbove->Append($code);
1010 $_Above_HexCodes{$hexcode} = 1;
1013 ## Used in building \p{_CanonDCIJ}
1014 if ($deco and $deco =~ m/^006[9A]\b/) {
1015 $CodeToDeco{$code} = $deco;
1019 ## There are a few pairs of lines like:
1020 ## AC00;<Hangul Syllable, First>;Lo;0;L;;;;;N;;;;;
1021 ## D7A3;<Hangul Syllable, Last>;Lo;0;L;;;;;N;;;;;
1022 ## that define ranges.
1024 if ($name =~ /^<(.+), (First|Last)>$/)
1027 gencat($name, $cat, $code, $2 eq 'First' ? 'Append' : 'Extend');
1028 #New_Prop(In => $name, $General{$name}, Fuzzy => 1);
1032 ## normal (single-character) lines
1033 gencat($name, $cat, $code, 'Append');
1035 # No Append() here since since several codes may map into one.
1036 $To{Upper}->RawAppendRange($code, $code, $upper) if $upper;
1037 $To{Lower}->RawAppendRange($code, $code, $lower) if $lower;
1038 $To{Title}->RawAppendRange($code, $code, $title) if $title;
1039 $To{Digit}->Append($code, $decimal) if length $decimal;
1041 $Bidi->Append($code, $bidi);
1042 $Comb->Append($code, $comb) if $comb;
1043 $Number->Append($code, $number) if length $number;
1045 length($decimal) and ($Number{De} ||= Table->New())->Append($code)
1047 length($digit) and ($Number{Di} ||= Table->New())->Append($code)
1049 length($number) and ($Number{Nu} ||= Table->New())->Append($code);
1051 $Mirrored->Append($code) if $mirrored eq "Y";
1053 $Bidi{$bidi} ||= Table->New();#Is => "bt/$bidi",
1054 #Desc => "Bi-directional category '$bidi'",
1056 $Bidi{$bidi}->Append($code);
1060 $Deco->Append($code, $deco);
1061 if ($deco =~/^<(\w+)>/)
1063 my $dshort = $PVA_reverse{dt}{ucfirst lc $1};
1064 $DC{com}->Append($code);
1066 $DC{$dshort} ||= Table->New();
1067 $DC{$dshort}->Append($code);
1071 $DC{can}->Append($code);
1079 ## Tidy up a few special cases....
1082 $Cat{Cn} = $Assigned->Invert; ## Cn is everything that doesn't exist
1083 New_Prop(Is => 'Cn',
1085 Desc => "General Category 'Cn' [not functional in Perl]",
1088 ## Unassigned is the same as 'Cn'
1089 New_Alias(Is => 'Unassigned', SameAs => 'Cn', Fuzzy => 0);
1091 $Cat{C}->Replace($Cat{C}->Merge($Cat{Cn})); ## Now merge in Cn into C
1094 # LC is Ll, Lu, and Lt.
1095 # (used to be L& or L_, but PropValueAliases.txt defines it as LC)
1096 New_Prop(Is => 'LC',
1097 Table->Merge(@Cat{qw[Ll Lu Lt]}),
1098 Desc => '[\p{Ll}\p{Lu}\p{Lt}]',
1101 ## Any and All are all code points.
1102 my $Any = Table->New(Is => 'Any',
1103 Desc => sprintf("[\\x{0000}-\\x{%X}]",
1104 $LastUnicodeCodepoint),
1106 $Any->RawAppendRange(0, $LastUnicodeCodepoint);
1108 New_Alias(Is => 'All', SameAs => 'Any', Fuzzy => 0);
1111 ## Build special properties for Perl's internal case-folding needs:
1112 ## \p{_CaseIgnorable}
1115 ## _CombAbove was built above. Others are built here....
1118 ## \p{_CaseIgnorable} is [\p{Mn}\0x00AD\x2010]
1119 New_Prop(Is => '_CaseIgnorable',
1120 Table->Merge($Cat{Mn},
1121 0x00AD, #SOFT HYPHEN
1123 Desc => '(for internal casefolding use)',
1127 ## \p{_CanonDCIJ} is fairly complex...
1128 my $CanonCDIJ = Table->New(Is => '_CanonDCIJ',
1129 Desc => '(for internal casefolding use)',
1131 ## It contains the ASCII 'i' and 'j'....
1132 $CanonCDIJ->Append(0x0069); # ASCII ord("i")
1133 $CanonCDIJ->Append(0x006A); # ASCII ord("j")
1134 ## ...and any character with a decomposition that starts with either of
1135 ## those code points, but only if the decomposition does not have any
1136 ## combining character with the "ABOVE" canonical combining class.
1137 for my $code (sort { $a <=> $b} keys %CodeToDeco)
1139 ## Need to ensure that all decomposition characters do not have
1140 ## a %HexCodeToComb in %AboveCombClasses.
1142 for my $deco_hexcode (split / /, $CodeToDeco{$code})
1144 if (exists $_Above_HexCodes{$deco_hexcode}) {
1145 ## one of the decmposition chars has an ABOVE combination
1146 ## class, so we're not interested in this one
1152 $CanonCDIJ->Append($code);
1159 ## Now dump the files.
1161 $Name->Write("Name.pl");
1165 foreach my $name (qw (PropertyAlias PA_reverse PropValueAlias
1166 PVA_reverse PVA_abbr_map)) {
1167 # Should I really jump through typeglob hoops just to avoid a
1168 # symbolic reference? (%{"utf8::$name})
1169 push @PVA, "\n", "\%utf8::$name = (\n",
1170 simple_dumper (%{$utf8::{$name}}), ");\n";
1173 WriteIfChanged("PVA.pl", @PVA);
1176 # $Bidi->Write("Bidirectional.pl");
1179 ["lib","bc","$_.pl"],
1180 "BidiClass category '$PropValueAlias{bc}{$_}'"
1184 $Comb->Write("CombiningClass.pl");
1185 for (keys %{ $PropValueAlias{ccc} }) {
1186 my ($code, $name) = @{ $PropValueAlias{ccc}{$_} };
1187 (my $c = Table->New())->Append($code);
1189 ["lib","ccc","$_.pl"],
1190 "CombiningClass category '$name'"
1194 $Deco->Write("Decomposition.pl");
1197 ["lib","dt","$_.pl"],
1198 "DecompositionType category '$PropValueAlias{dt}{$_}'"
1202 # $Number->Write("Number.pl");
1203 for (keys %Number) {
1205 ["lib","nt","$_.pl"],
1206 "NumericType category '$PropValueAlias{nt}{$_}'"
1210 # $General->Write("Category.pl");
1212 for my $to (sort keys %To) {
1213 $To{$to}->Write(["To","$to.pl"]);
1216 for (keys %{ $PropValueAlias{gc} }) {
1217 New_Alias(Is => $PropValueAlias{gc}{$_}, SameAs => $_, Fuzzy => 1);
1222 ## Process LineBreak.txt
1226 if (not open IN, "LineBreak.txt") {
1227 die "$0: LineBreak.txt: $!\n";
1230 my $Lbrk = Table->New();
1235 next unless /^([0-9A-Fa-f]+)(?:\.\.([0-9A-Fa-f]+))?\s*;\s*(\w+)/;
1237 my ($first, $last, $lbrk) = (hex($1), hex($2||""), $3);
1239 $Lbrk->Append($first, $lbrk);
1241 $Lbrk{$lbrk} ||= Table->New();
1242 $Lbrk{$lbrk}->Append($first);
1245 $Lbrk->Extend($last);
1246 $Lbrk{$lbrk}->Extend($last);
1251 # $Lbrk->Write("Lbrk.pl");
1256 ["lib","lb","$_.pl"],
1257 "Linebreak category '$PropValueAlias{lb}{$_}'"
1263 ## Process ArabicShaping.txt.
1265 sub ArabicShaping_txt()
1267 if (not open IN, "ArabicShaping.txt") {
1268 die "$0: ArabicShaping.txt: $!\n";
1271 my $ArabLink = Table->New();
1272 my $ArabLinkGroup = Table->New();
1278 next unless /^[0-9A-Fa-f]+;/;
1281 my ($hexcode, $name, $link, $linkgroup) = split(/\s*;\s*/);
1282 my $code = hex($hexcode);
1283 $ArabLink->Append($code, $link);
1284 $ArabLinkGroup->Append($code, $linkgroup);
1286 $JoinType{$link} ||= Table->New(Is => "JoinType$link");
1287 $JoinType{$link}->Append($code);
1291 # $ArabLink->Write("ArabLink.pl");
1292 # $ArabLinkGroup->Write("ArabLnkGrp.pl");
1295 for (keys %JoinType) {
1296 $JoinType{$_}->Write(
1297 ["lib","jt","$_.pl"],
1298 "JoiningType category '$PropValueAlias{jt}{$_}'"
1304 ## Process EastAsianWidth.txt.
1306 sub EastAsianWidth_txt()
1308 if (not open IN, "EastAsianWidth.txt") {
1309 die "$0: EastAsianWidth.txt: $!\n";
1316 next unless /^[0-9A-Fa-f]+(\.\.[0-9A-Fa-f]+)?;/;
1320 my ($hexcodes, $pv) = split(/\s*;\s*/);
1321 $EAW{$pv} ||= Table->New(Is => "EastAsianWidth$pv");
1322 my ($start, $end) = split(/\.\./, $hexcodes);
1324 $EAW{$pv}->AppendRange(hex($start), hex($end));
1326 $EAW{$pv}->Append(hex($start));
1334 ["lib","ea","$_.pl"],
1335 "EastAsianWidth category '$PropValueAlias{ea}{$_}'"
1341 ## Process HangulSyllableType.txt.
1343 sub HangulSyllableType_txt()
1345 if (not open IN, "HangulSyllableType.txt") {
1346 die "$0: HangulSyllableType.txt: $!\n";
1353 next unless /^([0-9A-Fa-f]+)(?:\.\.([0-9A-Fa-f]+))?\s*;\s*(\w+)/;
1354 my ($first, $last, $pv) = (hex($1), hex($2||""), $3);
1356 $HST{$pv} ||= Table->New(Is => "HangulSyllableType$pv");
1357 $HST{$pv}->Append($first);
1359 if ($last) { $HST{$pv}->Extend($last) }
1365 ["lib","hst","$_.pl"],
1366 "HangulSyllableType category '$PropValueAlias{hst}{$_}'"
1372 ## Process Jamo.txt.
1376 if (not open IN, "Jamo.txt") {
1377 die "$0: Jamo.txt: $!\n";
1379 my $Short = Table->New();
1383 next unless /^([0-9A-Fa-f]+)\s*;\s*(\w*)/;
1384 my ($code, $short) = (hex($1), $2);
1386 $Short->Append($code, $short);
1389 # $Short->Write("JamoShort.pl");
1393 ## Process Scripts.txt.
1399 if (not open(IN, "Scripts.txt")) {
1400 die "$0: Scripts.txt: $!\n";
1403 next unless /^([0-9A-Fa-f]+)(?:\.\.([0-9A-Fa-f]+))?\s*;\s*(.+?)\s*\#/;
1405 # Wait until all the scripts have been read since
1406 # they are not listed in numeric order.
1407 push @ScriptInfo, [ hex($1), hex($2||""), $3 ];
1411 # Now append the scripts properties in their code point order.
1414 my $Scripts = Table->New();
1416 for my $script (sort { $a->[0] <=> $b->[0] } @ScriptInfo)
1418 my ($first, $last, $name) = @$script;
1419 $Scripts->Append($first, $name);
1421 $Script{$name} ||= Table->New(Is => $name,
1422 Desc => "Script '$name'",
1424 $Script{$name}->Append($first, $name);
1427 $Scripts->Extend($last);
1428 $Script{$name}->Extend($last);
1432 # $Scripts->Write("Scripts.pl");
1434 ## Common is everything not explicitly assigned to a Script
1436 ## ***shouldn't this be intersected with \p{Assigned}? ******
1438 New_Prop(Is => 'Common',
1440 Desc => 'Pseudo-Script of codepoints not in other Unicode scripts',
1445 ## Given a name like "Close Punctuation", return a regex (that when applied
1446 ## with /i) matches any valid form of that name (e.g. "ClosePunctuation",
1447 ## "Close-Punctuation", etc.)
1449 ## Accept any space, dash, or underbar where in the official name there is
1450 ## space or a dash (or underbar, but there never is).
1456 $Name =~ s/[- _]/(?:[-_]|\\s+)?/g;
1461 ## Process Blocks.txt.
1465 my $Blocks = Table->New();
1468 if (not open IN, "Blocks.txt") {
1469 die "$0: Blocks.txt: $!\n";
1474 #next if not /Private Use$/;
1475 next if not /^([0-9A-Fa-f]+)\.\.([0-9A-Fa-f]+)\s*;\s*(.+?)\s*$/;
1477 my ($first, $last, $name) = (hex($1), hex($2), $3);
1479 $Blocks->Append($first, $name);
1481 $Blocks{$name} ||= Table->New(In => $name,
1482 Desc => "Block '$name'",
1484 $Blocks{$name}->Append($first, $name);
1486 if ($last and $last != $first) {
1487 $Blocks->Extend($last);
1488 $Blocks{$name}->Extend($last);
1493 # $Blocks->Write("Blocks.pl");
1497 ## Read in the PropList.txt. It contains extended properties not
1498 ## listed in the UnicodeData.txt, such as 'Other_Alphabetic':
1499 ## alphabetic but not of the general category L; many modifiers
1500 ## belong to this extended property category: while they are not
1501 ## alphabets, they are alphabetic in nature.
1507 if (not open IN, "PropList.txt") {
1508 die "$0: PropList.txt: $!\n";
1513 next unless /^([0-9A-Fa-f]+)(?:\.\.([0-9A-Fa-f]+))?\s*;\s*(.+?)\s*\#/;
1515 # Wait until all the extended properties have been read since
1516 # they are not listed in numeric order.
1517 push @PropInfo, [ hex($1), hex($2||""), $3 ];
1521 # Now append the extended properties in their code point order.
1522 my $Props = Table->New();
1525 for my $prop (sort { $a->[0] <=> $b->[0] } @PropInfo)
1527 my ($first, $last, $name) = @$prop;
1528 $Props->Append($first, $name);
1530 $Prop{$name} ||= Table->New(Is => $name,
1531 Desc => "Extended property '$name'",
1533 $Prop{$name}->Append($first, $name);
1536 $Props->Extend($last);
1537 $Prop{$name}->Extend($last);
1542 (my $file = $PA_reverse{$_}) =~ tr/_//d;
1543 # XXX I'm assuming that the names from %Prop don't suffer 8.3 clashes.
1544 $BaseNames{lc $file}++;
1546 ["lib","gc_sc","$file.pl"],
1547 "Binary property '$_'"
1551 # Alphabetic is L, Nl, and Other_Alphabetic.
1552 New_Prop(Is => 'Alphabetic',
1553 Table->Merge($Cat{L}, $Cat{Nl}, $Prop{Other_Alphabetic}),
1554 Desc => '[\p{L}\p{Nl}\p{OtherAlphabetic}]', # canonical names
1557 # Lowercase is Ll and Other_Lowercase.
1558 New_Prop(Is => 'Lowercase',
1559 Table->Merge($Cat{Ll}, $Prop{Other_Lowercase}),
1560 Desc => '[\p{Ll}\p{OtherLowercase}]', # canonical names
1563 # Uppercase is Lu and Other_Uppercase.
1564 New_Prop(Is => 'Uppercase',
1565 Table->Merge($Cat{Lu}, $Prop{Other_Uppercase}),
1566 Desc => '[\p{Lu}\p{OtherUppercase}]', # canonical names
1569 # Math is Sm and Other_Math.
1570 New_Prop(Is => 'Math',
1571 Table->Merge($Cat{Sm}, $Prop{Other_Math}),
1572 Desc => '[\p{Sm}\p{OtherMath}]', # canonical names
1575 # ID_Start is Ll, Lu, Lt, Lm, Lo, Nl, and Other_ID_Start.
1576 New_Prop(Is => 'ID_Start',
1577 Table->Merge(@Cat{qw[Ll Lu Lt Lm Lo Nl]}, $Prop{Other_ID_Start}),
1578 Desc => '[\p{Ll}\p{Lu}\p{Lt}\p{Lm}\p{Lo}\p{Nl}\p{OtherIDStart}]',
1581 # ID_Continue is ID_Start, Mn, Mc, Nd, Pc, and Other_ID_Continue.
1582 New_Prop(Is => 'ID_Continue',
1583 Table->Merge(@Cat{qw[Ll Lu Lt Lm Lo Nl Mn Mc Nd Pc ]},
1584 @Prop{qw[Other_ID_Start Other_ID_Continue]}),
1585 Desc => '[\p{ID_Start}\p{Mn}\p{Mc}\p{Nd}\p{Pc}\p{OtherIDContinue}]',
1588 # Default_Ignorable_Code_Point = Other_Default_Ignorable_Code_Point
1589 # + Cf + Cc + Cs + Noncharacter + Variation_Selector
1590 # - WhiteSpace - FFF9..FFFB (Annotation Characters)
1592 my $Annotation = Table->New();
1593 $Annotation->RawAppendRange(0xFFF9, 0xFFFB);
1595 New_Prop(Is => 'Default_Ignorable_Code_Point',
1596 Table->Merge(@Cat{qw[Cf Cc Cs]},
1597 $Prop{Noncharacter_Code_Point},
1598 $Prop{Variation_Selector},
1599 $Prop{Other_Default_Ignorable_Code_Point})
1601 ->Merge($Prop{White_Space}, $Annotation)
1603 Desc => '(?![\p{WhiteSpace}\x{FFF9}-\x{FFFB}])[\p{Cf}\p{Cc}'.
1604 '\p{Cs}\p{NoncharacterCodePoint}\p{VariationSelector}'.
1605 '\p{OtherDefaultIgnorableCodePoint}]',
1612 ## These are used in:
1613 ## MakePropTestScript()
1614 ## WriteAllMappings()
1615 ## for making the test script.
1617 my %FuzzyNameToTest;
1618 my %ExactNameToTest;
1621 ## This used only for making the test script
1626 my $MatchCode = shift;
1627 my $FailCode = shift;
1629 if (defined $MatchCode) {
1630 printf $FH qq/Expect(1, "\\x{%04X}", '\\p{$Prop}' );\n/, $MatchCode;
1631 printf $FH qq/Expect(0, "\\x{%04X}", '\\p{^$Prop}');\n/, $MatchCode;
1632 printf $FH qq/Expect(0, "\\x{%04X}", '\\P{$Prop}' );\n/, $MatchCode;
1633 printf $FH qq/Expect(1, "\\x{%04X}", '\\P{^$Prop}');\n/, $MatchCode;
1635 if (defined $FailCode) {
1636 printf $FH qq/Expect(0, "\\x{%04X}", '\\p{$Prop}' );\n/, $FailCode;
1637 printf $FH qq/Expect(1, "\\x{%04X}", '\\p{^$Prop}');\n/, $FailCode;
1638 printf $FH qq/Expect(1, "\\x{%04X}", '\\P{$Prop}' );\n/, $FailCode;
1639 printf $FH qq/Expect(0, "\\x{%04X}", '\\P{^$Prop}');\n/, $FailCode;
1643 ## This used only for making the test script
1649 print $FH qq/Error('\\p{$prop}');\n/;
1650 print $FH qq/Error('\\P{$prop}');\n/;
1653 ## This used only for making the test script
1669 ## This used only for making the test script
1670 sub RandomlyFuzzifyName($;$)
1673 my $WantError = shift; ## if true, make an error
1676 for my $part (split /[-\s_]+/, $Name)
1679 if ($WantError and rand() < 0.3) {
1680 push @parts, $BadSeps[rand(@BadSeps)];
1683 push @parts, $GoodSeps[rand(@GoodSeps)];
1686 my $switch = int rand(4);
1688 push @parts, uc $part;
1689 } elsif ($switch == 1) {
1690 push @parts, lc $part;
1691 } elsif ($switch == 2) {
1692 push @parts, ucfirst $part;
1697 my $new = join('', @parts);
1700 if (rand() >= 0.5) {
1701 $new .= $BadSeps[rand(@BadSeps)];
1703 $new = $BadSeps[rand(@BadSeps)] . $new;
1709 ## This used only for making the test script
1710 sub MakePropTestScript()
1712 ## this written directly -- it's huge.
1713 force_unlink ("TestProp.pl");
1714 if (not open OUT, ">TestProp.pl") {
1715 die "$0: TestProp.pl: $!\n";
1719 while (my ($Name, $Table) = each %ExactNameToTest)
1721 GenTests(*OUT, $Name, $Table->ValidCode, $Table->InvalidCode);
1722 ExpectError(*OUT, uc $Name) if uc $Name ne $Name;
1723 ExpectError(*OUT, lc $Name) if lc $Name ne $Name;
1727 while (my ($Name, $Table) = each %FuzzyNameToTest)
1729 my $Orig = $CanonicalToOrig{$Name};
1733 RandomlyFuzzifyName($Orig) => 1
1736 for my $N (keys %Names) {
1737 GenTests(*OUT, $N, $Table->ValidCode, $Table->InvalidCode);
1740 ExpectError(*OUT, RandomlyFuzzifyName($Orig, 'ERROR'));
1743 print OUT "Finished();\n";
1749 ## These are used only in:
1750 ## RegisterFileForName()
1751 ## WriteAllMappings()
1753 my %Exact; ## will become %utf8::Exact;
1754 my %Canonical; ## will become %utf8::Canonical;
1755 my %CaComment; ## Comment for %Canonical entry of same key
1758 ## Given info about a name and a datafile that it should be associated with,
1759 ## register that assocation in %Exact and %Canonical.
1760 sub RegisterFileForName($$$$)
1764 my $IsFuzzy = shift;
1765 my $filename = shift;
1768 ## Now in details for the mapping. $Type eq 'Is' has the
1769 ## Is removed, as it will be removed in utf8_heavy when this
1770 ## data is being checked. In keeps its "In", but a second
1771 ## sans-In record is written if it doesn't conflict with
1772 ## anything already there.
1776 if ($Type eq 'Is') {
1777 die "oops[$Name]" if $Exact{$Name};
1778 $Exact{$Name} = $filename;
1780 die "oops[$Type$Name]" if $Exact{"$Type$Name"};
1781 $Exact{"$Type$Name"} = $filename;
1782 $Exact{$Name} = $filename if not $Exact{$Name};
1787 my $CName = lc $Name;
1788 if ($Type eq 'Is') {
1789 die "oops[$CName]" if $Canonical{$CName};
1790 $Canonical{$CName} = $filename;
1791 $CaComment{$CName} = $Name if $Name =~ tr/A-Z// >= 2;
1793 die "oops[$Type$CName]" if $Canonical{lc "$Type$CName"};
1794 $Canonical{lc "$Type$CName"} = $filename;
1795 $CaComment{lc "$Type$CName"} = "$Type$Name";
1796 if (not $Canonical{$CName}) {
1797 $Canonical{$CName} = $filename;
1798 $CaComment{$CName} = "$Type$Name";
1805 ## Writes the info accumulated in
1812 sub WriteAllMappings()
1816 ## 'Is' *MUST* come first, so its names have precidence over 'In's
1817 for my $Type ('Is', 'In')
1819 my %RawNameToFile; ## a per-$Type cache
1821 for my $Name (sort {length $a <=> length $b} keys %{$TableInfo{$Type}})
1823 ## Note: $Name is already canonical
1824 my $Table = $TableInfo{$Type}->{$Name};
1825 my $IsFuzzy = $FuzzyNames{$Type}->{$Name};
1827 ## Need an 8.3 safe filename (which means "an 8 safe" $filename)
1830 ## 'Is' items lose 'Is' from the basename.
1831 $filename = $Type eq 'Is' ?
1832 ($PVA_reverse{sc}{$Name} || $Name) :
1835 $filename =~ s/[^\w_]+/_/g; # "L&" -> "L_"
1836 substr($filename, 8) = '' if length($filename) > 8;
1839 ## Make sure the basename doesn't conflict with something we
1840 ## might have already written. If we have, say,
1847 while (my $num = $BaseNames{lc $filename}++)
1849 $num++; ## so basenames with numbers start with '2', which
1850 ## just looks more natural.
1851 ## Want to append $num, but if it'll make the basename longer
1852 ## than 8 characters, pre-truncate $filename so that the result
1854 my $delta = length($filename) + length($num) - 8;
1856 substr($filename, -$delta) = $num;
1864 ## Construct a nice comment to add to the file, and build data
1865 ## for the "./Properties" file along the way.
1869 my $Desc = $TableDesc{$Type}->{$Name} || "";
1870 ## get list of names this table is reference by
1871 my @Supported = $Name;
1872 while (my ($Orig, $Alias) = each %{ $AliasInfo{$Type} })
1874 if ($Orig eq $Name) {
1875 push @Supported, $Alias;
1879 my $TypeToShow = $Type eq 'Is' ? "" : $Type;
1882 $Comment = "This file supports:\n";
1883 for my $N (@Supported)
1885 my $IsFuzzy = $FuzzyNames{$Type}->{$N};
1886 my $Prop = "\\p{$TypeToShow$Name}";
1887 $OrigProp = $Prop if not $OrigProp; #cache for aliases
1889 $Comment .= "\t$Prop (and fuzzy permutations)\n";
1891 $Comment .= "\t$Prop\n";
1893 my $MyDesc = ($N eq $Name) ? $Desc : "Alias for $OrigProp ($Desc)";
1895 push @MAP, sprintf("%s %-42s %s\n",
1896 $IsFuzzy ? '*' : ' ', $Prop, $MyDesc);
1899 $Comment .= "\nMeaning: $Desc\n";
1904 ## Okay, write the file...
1906 $Table->Write(["lib","gc_sc","$filename.pl"], $Comment);
1909 $RawNameToFile{$Name} = $filename;
1910 RegisterFileForName($Type => $Name, $IsFuzzy, $filename);
1914 my $CName = CanonicalName($Type . '_'. $Name);
1915 $FuzzyNameToTest{$Name} = $Table if !$FuzzyNameToTest{$Name};
1916 $FuzzyNameToTest{$CName} = $Table if !$FuzzyNameToTest{$CName};
1918 $ExactNameToTest{$Name} = $Table;
1923 ## Register aliase info
1924 for my $Name (sort {length $a <=> length $b} keys %{$AliasInfo{$Type}})
1926 my $Alias = $AliasInfo{$Type}->{$Name};
1927 my $IsFuzzy = $FuzzyNames{$Type}->{$Alias};
1928 my $filename = $RawNameToFile{$Name};
1929 die "oops [$Alias]->[$Name]" if not $filename;
1930 RegisterFileForName($Type => $Alias, $IsFuzzy, $filename);
1932 my $Table = $TableInfo{$Type}->{$Name};
1933 die "oops" if not $Table;
1936 my $CName = CanonicalName($Type .'_'. $Alias);
1937 $FuzzyNameToTest{$Alias} = $Table if !$FuzzyNameToTest{$Alias};
1938 $FuzzyNameToTest{$CName} = $Table if !$FuzzyNameToTest{$CName};
1940 $ExactNameToTest{$Alias} = $Table;
1946 ## Write out the property list
1951 "## This file created by $0\n",
1952 "## List of built-in \\p{...}/\\P{...} properties.\n",
1954 "## '*' means name may be 'fuzzy'\n",
1956 sort { substr($a,2) cmp substr($b, 2) } @MAP,
1958 WriteIfChanged('Properties', @OUT);
1961 use Text::Tabs (); ## using this makes the files about half the size
1968 "## Data in this file used by ../utf8_heavy.pl\n",
1970 "## Mapping from name to filename in ./lib/gc_sc\n",
1971 "%utf8::Exact = (\n",
1974 $Exact{InGreek} = 'InGreekA'; # this is evil kludge
1975 for my $Name (sort keys %Exact)
1977 my $File = $Exact{$Name};
1978 $Name = $Name =~ m/\W/ ? qq/'$Name'/ : " $Name ";
1979 my $Text = sprintf("%-15s => %s,\n", $Name, qq/'$File'/);
1980 push @OUT, Text::Tabs::unexpand($Text);
1982 push @OUT, ");\n1;\n";
1984 WriteIfChanged('Exact.pl', @OUT);
1987 ## Write Canonical.pl
1992 "## Data in this file used by ../utf8_heavy.pl\n",
1994 "## Mapping from lc(canonical name) to filename in ./lib\n",
1995 "%utf8::Canonical = (\n",
1997 my $Trail = ""; ## used just to keep the spacing pretty
1998 for my $Name (sort keys %Canonical)
2000 my $File = $Canonical{$Name};
2001 if ($CaComment{$Name}) {
2002 push @OUT, "\n" if not $Trail;
2003 push @OUT, " # $CaComment{$Name}\n";
2008 $Name = $Name =~ m/\W/ ? qq/'$Name'/ : " $Name ";
2009 my $Text = sprintf(" %-41s => %s,\n$Trail", $Name, qq/'$File'/);
2010 push @OUT, Text::Tabs::unexpand($Text);
2012 push @OUT, ");\n1\n";
2013 WriteIfChanged('Canonical.pl', @OUT);
2016 MakePropTestScript() if $MakeTestScript;
2020 sub SpecialCasing_txt()
2023 # Read in the special cases.
2028 if (not open IN, "SpecialCasing.txt") {
2029 die "$0: SpecialCasing.txt: $!\n";
2032 next unless /^[0-9A-Fa-f]+;/;
2036 my ($code, $lower, $title, $upper, $condition) = split(/\s*;\s*/);
2038 if ($condition) { # not implemented yet
2039 print "# SKIPPING $_\n" if $Verbose;
2043 # Wait until all the special cases have been read since
2044 # they are not listed in numeric order.
2045 my $ix = hex($code);
2046 push @{$CaseInfo{Lower}}, [ $ix, $code, $lower ]
2047 unless $code eq $lower;
2048 push @{$CaseInfo{Title}}, [ $ix, $code, $title ]
2049 unless $code eq $title;
2050 push @{$CaseInfo{Upper}}, [ $ix, $code, $upper ]
2051 unless $code eq $upper;
2055 # Now write out the special cases properties in their code point order.
2056 # Prepend them to the To/{Upper,Lower,Title}.pl.
2058 for my $case (qw(Lower Title Upper))
2060 my $NormalCase = do "To/$case.pl" || die "$0: $@\n";
2065 "# The key UTF-8 _bytes_, the value UTF-8 (speed hack)\n",
2066 "%utf8::ToSpec$case =\n(\n",
2069 for my $prop (sort { $a->[0] <=> $b->[0] } @{$CaseInfo{$case}}) {
2070 my ($ix, $code, $to) = @$prop;
2072 join "", map { sprintf "\\x{%s}", $_ } split ' ', $to;
2073 push @OUT, sprintf qq["%s" => "$tostr",\n], join("", map { sprintf "\\x%02X", $_ } unpack("U0C*", pack("U", $ix)));
2074 # Remove any single-character mappings for
2075 # the same character since we are going for
2076 # the special casing rules.
2077 $NormalCase =~ s/^$code\t\t\w+\n//m;
2081 "return <<'END';\n",
2085 WriteIfChanged(["To","$case.pl"], @OUT);
2090 # Read in the case foldings.
2092 # We will do full case folding, C + F + I (see CaseFolding.txt).
2094 sub CaseFolding_txt()
2096 if (not open IN, "CaseFolding.txt") {
2097 die "$0: CaseFolding.txt: $!\n";
2100 my $Fold = Table->New();
2104 # Skip status 'S', simple case folding
2105 next unless /^([0-9A-Fa-f]+)\s*;\s*([CFI])\s*;\s*([0-9A-Fa-f]+(?: [0-9A-Fa-f]+)*)\s*;/;
2107 my ($code, $status, $fold) = (hex($1), $2, $3);
2109 if ($status eq 'C') { # Common: one-to-one folding
2110 # No append() since several codes may fold into one.
2111 $Fold->RawAppendRange($code, $code, $fold);
2112 } else { # F: full, or I: dotted uppercase I -> dotless lowercase I
2113 $Fold{$code} = $fold;
2118 $Fold->Write("To/Fold.pl");
2121 # Prepend the special foldings to the common foldings.
2123 my $CommonFold = do "To/Fold.pl" || die "$0: To/Fold.pl: $!\n";
2128 "# The ke UTF-8 _bytes_, the value UTF-8 (speed hack)\n",
2129 "%utf8::ToSpecFold =\n(\n",
2131 for my $code (sort { $a <=> $b } keys %Fold) {
2133 join "", map { sprintf "\\x{%s}", $_ } split ' ', $Fold{$code};
2134 push @OUT, sprintf qq["%s" => "$foldstr",\n], join("", map { sprintf "\\x%02X", $_ } unpack("U0C*", pack("U", $code)));
2138 "return <<'END';\n",
2143 WriteIfChanged(["To","Fold.pl"], @OUT);
2158 ArabicShaping_txt();
2159 EastAsianWidth_txt();
2160 HangulSyllableType_txt();
2162 SpecialCasing_txt();
2165 if ( $FileList and $MakeList ) {
2167 print "Updating '$FileList'\n"
2170 open my $ofh,">",$FileList
2171 or die "Can't write to '$FileList':$!";
2172 print $ofh <<"EOFHEADER";
2174 # mktables.lst -- File list for mktables.
2176 # Autogenerated on @{[scalar localtime]}
2178 # - First section is input files
2179 # (mktables itself is automatically included)
2180 # - Section seperator is /^=+\$/
2181 # - Second section is a list of output files.
2182 # - Lines matching /^\\s*#/ are treated as comments
2183 # which along with blank lines are ignored.
2189 my @input=("version",glob('*.txt'));
2190 print $ofh "$_\n" for
2192 "\n=================================\n",
2193 "# Output files:\n",
2211 print $ofh "\n# ",scalar(@input)," input files\n",
2212 "# ",scalar($count+1)," output files\n\n",
2215 or warn "Failed to close $ofh: $!";
2217 print "Filelist has ",scalar(@input)," input files and ",
2218 scalar($count+1)," output files\n"
2221 print "All done\n" if $Verbose;
2224 ## TRAILING CODE IS USED BY MakePropTestScript()
2237 my $Line = (caller)[2];
2242 $RegObj = qr/$Regex/;
2243 $String =~ $RegObj ? 1 : 0
2246 if (not defined $result) {
2247 print "couldn't compile /$Regex/ on $0 line $Line: $@\n";
2249 } elsif ($result ^ $Expect) {
2250 print "bad result (expected $Expect) on $0 line $Line: $@\n";
2259 if (eval { 'x' =~ qr/$Regex/; 1 }) {
2261 my $Line = (caller)[2];
2262 print "expected error for /$Regex/ on $0 line $Line: $@\n";
2269 print "All $Tests tests passed.\n";
2272 print "$Tests tests, $Fails failed!\n";