6 ## mktables -- create the runtime Perl Unicode files (lib/unicore/**/*.pl)
7 ## from the Unicode database files (lib/unicore/*.txt).
17 my $MakeTestScript = 0;
21 my $arg = shift @ARGV;
24 } elsif ($arg eq '-q') {
26 } elsif ($arg eq '-maketest') {
29 die "usage: $0 [-v|-q] [-maketest]";
33 my $LastUnicodeCodepoint = 0x10FFFF; # As of Unicode 3.1.1.
36 # !!!!!!! DO NOT EDIT THIS FILE !!!!!!!
37 # This file is built by $0 from e.g. UnicodeData.txt.
38 # Any changes made here will be lost!
44 ## Given a filename and a reference to an array of lines,
45 ## write the lines to the file only if the contents have not changed.
47 sub WriteIfChanged($\@)
52 my $TextToWrite = join '', @$lines;
55 my $PreviousText = <IN>;
57 if ($PreviousText eq $TextToWrite) {
58 print "$file unchanged.\n" if $Verbose;
62 if (not open OUT, ">$file") {
63 die "$0: can't open $file for output: $!\n";
65 print "$file written.\n" if $Verbose;
67 print OUT $TextToWrite;
72 ## The main datastructure (a "Table") represents a set of code points that
73 ## are part of a particular quality (that are part of \pL, \p{InGreek},
74 ## etc.). They are kept as ranges of code points (starting and ending of
77 ## For example, a range ASCII LETTERS would be represented as:
78 ## [ [ 0x41 => 0x5A, 'UPPER' ],
79 ## [ 0x61 => 0x7A, 'LOWER, ] ]
81 sub RANGE_START() { 0 } ## index into range element
82 sub RANGE_END() { 1 } ## index into range element
83 sub RANGE_NAME() { 2 } ## index into range element
85 ## Conceptually, these should really be folded into the 'Table' objects
93 ## Turn something like
102 $name =~ s/(?<![a-z])(\w)/\u$1/g;
103 $name =~ s/[-_\s]+//g;
105 $CanonicalToOrig{$name} = $orig if not $CanonicalToOrig{$name};
110 ## Associates a property ("Greek", "Lu", "Assigned",...) with a Table.
113 ## New_Prop(In => 'Greek', $Table, Desc => 'Greek Block', Fuzzy => 1);
115 ## Normally, these parameters are set when the Table is created (when the
116 ## Table->New constructor is called), but there are times when it needs to
117 ## be done after-the-fact...)
121 my $Type = shift; ## "Is" or "In";
125 ## remaining args are optional key/val
128 my $Fuzzy = delete $Args{Fuzzy};
129 my $Desc = delete $Args{Desc}; # description
131 $Name = CanonicalName($Name) if $Fuzzy;
133 ## sanity check a few args
134 if (%Args or ($Type ne 'Is' and $Type ne 'In') or not ref $Table) {
135 confess "$0: bad args to New_Prop"
138 if (not $TableInfo{$Type}->{$Name})
140 $TableInfo{$Type}->{$Name} = $Table;
141 $TableDesc{$Type}->{$Name} = $Desc;
143 $FuzzyNames{$Type}->{$Name} = $Name;
150 ## Creates a new Table object.
152 ## Args are key/value pairs:
153 ## In => Name -- Name of "In" property to be associated with
154 ## Is => Name -- Name of "Is" property to be associated with
155 ## Fuzzy => Boolean -- True if name can be accessed "fuzzily"
156 ## Desc => String -- Description of the property
158 ## No args are required.
165 my $Table = bless [], $class;
167 my $Fuzzy = delete $Args{Fuzzy};
168 my $Desc = delete $Args{Desc};
170 for my $Type ('Is', 'In')
172 if (my $Name = delete $Args{$Type}) {
173 New_Prop($Type => $Name, $Table, Desc => $Desc, Fuzzy => $Fuzzy);
177 ## shouldn't have any left over
179 confess "$0: bad args to Table->New"
186 ## Returns true if the Table has no code points
190 my $Table = shift; #self
195 ## Returns true if the Table has code points
199 my $Table = shift; #self
204 ## Returns the maximum code point currently in the table.
208 my $Table = shift; #self
209 confess "oops" if $Table->IsEmpty; ## must have code points to have a max
210 return $Table->[-1]->[RANGE_END];
214 ## Replaces the codepoints in the Table with those in the Table given
215 ## as an arg. (NOTE: this is not a "deep copy").
217 sub Table::Replace($$)
219 my $Table = shift; #self
226 ## Given a new code point, make the last range of the Table extend to
227 ## include the new (and all intervening) code points.
231 my $Table = shift; #self
232 my $codepoint = shift;
234 my $PrevMax = $Table->Max;
236 confess "oops ($codepoint <= $PrevMax)" if $codepoint <= $PrevMax;
238 $Table->[-1]->[RANGE_END] = $codepoint;
242 ## Given a code point range start and end (and optional name), blindly
243 ## append them to the list of ranges for the Table.
245 ## NOTE: Code points must be added in strictly ascending numeric order.
247 sub Table::RawAppendRange
249 my $Table = shift; #self
253 $name = "" if not defined $name; ## warning: $name can be "0"
255 push @$Table, [ $start, # RANGE_START
257 $name ]; # RANGE_NAME
261 ## Given a code point (and optional name), add it to the Table.
263 ## NOTE: Code points must be added in strictly ascending numeric order.
267 my $Table = shift; #self
268 my $codepoint = shift;
270 $name = "" if not defined $name; ## warning: $name can be "0"
273 ## If we've already got a range working, and this code point is the next
274 ## one in line, and if the name is the same, just extend the current range.
278 $Table->Max == $codepoint - 1
280 $Table->[-1]->[RANGE_NAME] eq $name)
282 $Table->Extend($codepoint);
286 $Table->RawAppendRange($codepoint, $codepoint, $name);
291 ## Given a code point range starting value and ending value (and name),
292 ## Add the range to teh Table.
294 ## NOTE: Code points must be added in strictly ascending numeric order.
296 sub Table::AppendRange
298 my $Table = shift; #self
302 $name = "" if not defined $name; ## warning: $name can be "0"
304 $Table->Append($start, $name);
305 $Table->Extend($end) if $end > $start;
309 ## Return a new Table that represents all code points not in the Table.
313 my $Table = shift; #self
315 my $New = Table->New();
317 for my $range (@$Table)
319 my $start = $range->[RANGE_START];
320 my $end = $range->[RANGE_END];
321 if ($start-1 >= $max+1) {
322 $New->AppendRange($max+1, $start-1, "");
326 if ($max+1 < $LastUnicodeCodepoint) {
327 $New->AppendRange($max+1, $LastUnicodeCodepoint);
333 ## Merges any number of other tables with $self, returning the new table.
334 ## (existing tables are not modified)
337 ## Args may be Tables, or individual code points (as integers).
339 ## Can be called as either a constructor or a method.
343 shift(@_) if not ref $_[0]; ## if called as a constructor, lose the class
346 ## Accumulate all records from all tables
348 for my $Arg (@Tables)
351 ## arg is a table -- get its ranges
352 push @Records, @$Arg;
354 ## arg is a codepoint, make a range
355 push @Records, [ $Arg, $Arg ]
359 ## sort by range start, with longer ranges coming first.
360 my ($first, @Rest) = sort {
361 ($a->[RANGE_START] <=> $b->[RANGE_START])
363 ($b->[RANGE_END] <=> $b->[RANGE_END])
366 my $New = Table->New();
368 ## Ensuring the first range is there makes the subsequent loop easier
369 $New->AppendRange($first->[RANGE_START],
370 $first->[RANGE_END]);
372 ## Fold in records so long as they add new information.
375 my $start = $set->[RANGE_START];
376 my $end = $set->[RANGE_END];
377 if ($start > $New->Max) {
378 $New->AppendRange($start, $end);
379 } elsif ($end > $New->Max) {
388 ## Given a filename, write a representation of the Table to a file.
389 ## May have an optional comment as a 2nd arg.
393 my $Table = shift; #self
394 my $filename = shift;
398 if (defined $comment) {
399 $comment =~ s/\s+\Z//;
400 $comment =~ s/^/# /gm;
401 push @OUT, "#\n$comment\n#\n";
403 push @OUT, "return <<'END';\n";
405 for my $set (@$Table)
407 my $start = $set->[RANGE_START];
408 my $end = $set->[RANGE_END];
409 my $name = $set->[RANGE_NAME];
411 if ($start == $end) {
412 push @OUT, sprintf "%04X\t\t%s\n", $start, $name;
414 push @OUT, sprintf "%04X\t%04X\t%s\n", $start, $end, $name;
420 WriteIfChanged($filename, @OUT);
423 ## This used only for making the test script.
428 return 0 if $code <= 0x0000; ## don't use null
429 return 0 if $code >= $LastUnicodeCodepoint; ## keep in range
430 return 0 if ($code >= 0xD800 and $code <= 0xDFFF); ## no surrogates
431 return 0 if ($code >= 0xFDD0 and $code <= 0xFDEF); ## utf8.c says no good
432 return 0 if (($code & 0xFFFF) == 0xFFFE); ## utf8.c says no good
433 return 0 if (($code & 0xFFFF) == 0xFFFF); ## utf8.c says no good
437 ## Return a code point that's part of the table.
438 ## Returns nothing if the table is empty (or covers only surrogates).
439 ## This used only for making the test script.
442 my $Table = shift; #self
443 for my $set (@$Table) {
444 return $set->[RANGE_END] if IsUsable($set->[RANGE_END]);
449 ## Return a code point that's not part of the table
450 ## Returns nothing if the table covers all code points.
451 ## This used only for making the test script.
452 sub Table::InvalidCode
454 my $Table = shift; #self
456 return 0x1234 if $Table->IsEmpty();
458 for my $set (@$Table)
460 if (IsUsable($set->[RANGE_END] + 1))
462 return $set->[RANGE_END] + 1;
465 if (IsUsable($set->[RANGE_START] - 1))
467 return $set->[RANGE_START] - 1;
473 ###########################################################################
474 ###########################################################################
475 ###########################################################################
480 ## New_Alias(Is => 'All', SameAs => 'Any', Fuzzy => 1);
482 ## The args must be in that order, although the Fuzzy pair may be omitted.
484 ## This creates 'IsAll' as an alias for 'IsAny'
488 my $Type = shift; ## "Is" or "In"
490 my $SameAs = shift; # expecting "SameAs" -- just ignored
493 ## remaining args are optional key/val
496 my $Fuzzy = delete $Args{Fuzzy};
498 ## sanity check a few args
499 if (%Args or ($Type ne 'Is' and $Type ne 'In') or $SameAs ne 'SameAs') {
500 confess "$0: bad args to New_Alias"
503 $Alias = CanonicalName($Alias) if $Fuzzy;
505 if (not $TableInfo{$Type}->{$Name})
507 my $CName = CanonicalName($Name);
508 if ($TableInfo{$Type}->{$CName}) {
509 confess "$0: Use canonical form '$CName' instead of '$Name' for alias.";
511 confess "$0: don't have orignial $Type => $Name to make alias";
514 if ($TableInfo{$Alias}) {
515 confess "$0: already have original $Type => $Alias; can't make alias";
517 $AliasInfo{$Type}->{$Name} = $Alias;
519 $FuzzyNames{$Type}->{$Alias} = $Name;
525 ## All assigned code points
526 my $Assigned = Table->New(Is => 'Assigned',
527 Desc => "All assigned code points",
530 my $Name = Table->New(); ## all characters, individually by name
531 my $General = Table->New(); ## all characters, grouped by category
536 ## Process UnicodeData.txt (Categories, etc.)
538 sub UnicodeData_Txt()
540 my $Bidi = Table->New();
541 my $Deco = Table->New();
542 my $Comb = Table->New();
543 my $Number = Table->New();
544 my $Mirrored = Table->New(Is => 'Mirrored',
545 Desc => "Mirrored in bidirectional text",
551 $Deco{Canon} = Table->New(Is => 'Canon',
552 Desc => 'Decomposes to multiple characters',
554 $Deco{Compat} = Table->New(Is => 'Compat',
555 Desc => 'Compatible with a more-basic character',
558 ## Initialize Perl-generated categories
559 ## (Categories from UnicodeData.txt are auto-initialized in gencat)
561 Table->New(Is => 'Alnum', Desc => "[[:Alnum:]]", Fuzzy => 0);
563 Table->New(Is => 'Alpha', Desc => "[[:Alpha:]]", Fuzzy => 0);
565 Table->New(Is => 'ASCII', Desc => "[[:ASCII:]]", Fuzzy => 0);
567 Table->New(Is => 'Blank', Desc => "[[:Blank:]]", Fuzzy => 0);
569 Table->New(Is => 'Cntrl', Desc => "[[:Cntrl:]]", Fuzzy => 0);
571 Table->New(Is => 'Digit', Desc => "[[:Digit:]]", Fuzzy => 0);
573 Table->New(Is => 'Graph', Desc => "[[:Graph:]]", Fuzzy => 0);
575 Table->New(Is => 'Lower', Desc => "[[:Lower:]]", Fuzzy => 0);
577 Table->New(Is => 'Print', Desc => "[[:Print:]]", Fuzzy => 0);
579 Table->New(Is => 'Punct', Desc => "[[:Punct:]]", Fuzzy => 0);
581 Table->New(Is => 'Space', Desc => "[[:Space:]]", Fuzzy => 0);
583 Table->New(Is => 'Title', Desc => "[[:Title:]]", Fuzzy => 0);
585 Table->New(Is => 'Upper', Desc => "[[:Upper:]]", Fuzzy => 0);
587 Table->New(Is => 'XDigit', Desc => "[[:XDigit:]]", Fuzzy => 0);
589 Table->New(Is => 'Word', Desc => "[[:Word:]]", Fuzzy => 0);
591 Table->New(Is => 'SpacePerl', Desc => '\s', Fuzzy => 0);
594 $To{Upper} = Table->New();
595 $To{Lower} = Table->New();
596 $To{Title} = Table->New();
597 $To{Digit} = Table->New();
601 my ($name, ## Name ("LATIN CAPITAL LETTER A")
602 $cat, ## Category ("Lu", "Zp", "Nd", etc.)
603 $code, ## Code point (as an integer)
606 my $MajorCat = substr($cat, 0, 1); ## L, M, Z, S, etc
608 $Assigned->$op($code);
609 $Name->$op($code, $name);
610 $General->$op($code, $cat);
612 ## add to the sub category (e.g. "Lu", "Nd", "Cf", ..)
613 $Cat{$cat} ||= Table->New(Is => $cat,
614 Desc => "General Category '$cat'",
616 $Cat{$cat}->$op($code);
618 ## add to the major category (e.g. "L", "N", "C", ...)
619 $Cat{$MajorCat} ||= Table->New(Is => $MajorCat,
620 Desc => "Major Category '$MajorCat'",
622 $Cat{$MajorCat}->$op($code);
624 ($General{$name} ||= Table->New)->$op($code, $name);
626 # 005F: SPACING UNDERSCORE
627 $Cat{Word}->$op($code) if $cat =~ /^[LMN]/ || $code == 0x005F;
628 $Cat{Alnum}->$op($code) if $cat =~ /^[LMN]/;
629 $Cat{Alpha}->$op($code) if $cat =~ /^[LM]/;
633 $Cat{Space}->$op($code) if $cat =~ /^Z/
634 || $code == 0x0009 # 0009: HORIZONTAL TAB
635 || $code == 0x000A # 000A: LINE FEED
636 || $code == 0x000B # 000B: VERTICAL TAB
637 || $code == 0x000C # 000C: FORM FEED
638 || $code == 0x000D; # 000D: CARRIAGE RETURN
641 $Cat{SpacePerl}->$op($code) if $cat =~ /^Z/
642 || $code == 0x0009 # 0009: HORIZONTAL TAB
643 || $code == 0x000A # 000A: LINE FEED
644 || $code == 0x000C # 000C: FORM FEED
645 || $code == 0x000D # 000D: CARRIAGE RETURN
646 || $code == 0x0085 # 0085: <NEXT LINE>
647 || $code == 0x2028 # 2028: LINE SEPARATOR
648 || $code == 0x2029;# 2029: PARAGRAPH SEP.
650 $Cat{Blank}->$op($code) if $cat =~ /^Z[^lp]$/
651 || $code == 0x0009 # 0009: HORIZONTAL TAB
652 || $code == 0x0020; # 0020: SPACE
654 $Cat{Digit}->$op($code) if $cat eq "Nd";
655 $Cat{Upper}->$op($code) if $cat eq "Lu";
656 $Cat{Lower}->$op($code) if $cat eq "Ll";
657 $Cat{Title}->$op($code) if $cat eq "Lt";
658 $Cat{ASCII}->$op($code) if $code <= 0x007F;
659 $Cat{Cntrl}->$op($code) if $cat =~ /^C/;
660 $Cat{Graph}->$op($code) if $cat =~ /^([LMNPS]|Co)/;
661 $Cat{Print}->$op($code) if $cat =~ /^([LMNPS]|Co|Zs)/;
662 $Cat{Punct}->$op($code) if $cat =~ /^P/;
664 $Cat{XDigit}->$op($code) if ($code >= 0x30 && $code <= 0x39) ## 0..9
665 || ($code >= 0x41 && $code <= 0x46) ## A..F
666 || ($code >= 0x61 && $code <= 0x66); ## a..f
669 ## open ane read file.....
670 if (not open IN, "UnicodeData.txt") {
671 die "$0: UnicodeData.txt: $!\n";
675 ## For building \p{_CombAbove} and \p{_CanonDCIJ}
677 my %_Above_HexCodes; ## Hexcodes for chars with $comb == 230 ("ABOVE")
679 my %CodeToDeco; ## Maps code to decomp. list for chars with first
680 ## decomp. char an "i" or "j" (for \p{_CanonDCIJ})
682 ## This is filled in as we go....
683 my $CombAbove = Table->New(Is => '_CombAbove',
684 Desc => '(for internal casefolding use)',
689 next unless /^[0-9A-Fa-f]+;/;
692 my ($hexcode, ## code point in hex (e.g. "0041")
693 $name, ## character name (e.g. "LATIN CAPITAL LETTER A")
694 $cat, ## category (e.g. "Lu")
695 $comb, ## Canonical combining class (e.t. "230")
696 $bidi, ## directional category (e.g. "L")
697 $deco, ## decomposition mapping
698 $decimal, ## decimal digit value
699 $digit, ## digit value
700 $number, ## numeric value
701 $mirrored, ## mirrored
702 $unicode10, ## name in Unicode 1.0
703 $comment, ## comment field
704 $upper, ## uppercase mapping
705 $lower, ## lowercase mapping
706 $title, ## titlecase mapping
707 ) = split(/\s*;\s*/);
709 # Note that in Unicode 3.2 there will be names like
710 # LINE FEED (LF), which probably means that \N{} needs
711 # to cope also with LINE FEED and LF.
712 $name = $unicode10 if $name eq '<control>' && $unicode10 ne '';
714 my $code = hex($hexcode);
716 if ($comb and $comb == 230) {
717 $CombAbove->Append($code);
718 $_Above_HexCodes{$hexcode} = 1;
721 ## Used in building \p{_CanonDCIJ}
722 if ($deco and $deco =~ m/^006[9A]\b/) {
723 $CodeToDeco{$code} = $deco;
727 ## There are a few pairs of lines like:
728 ## AC00;<Hangul Syllable, First>;Lo;0;L;;;;;N;;;;;
729 ## D7A3;<Hangul Syllable, Last>;Lo;0;L;;;;;N;;;;;
730 ## that define ranges.
732 if ($name =~ /^<(.+), (First|Last)>$/)
735 gencat($name, $cat, $code, $2 eq 'First' ? 'Append' : 'Extend');
736 #New_Prop(In => $name, $General{$name}, Fuzzy => 1);
740 ## normal (single-character) lines
741 gencat($name, $cat, $code, 'Append');
743 # No Append() here since since several codes may map into one.
744 $To{Upper}->RawAppendRange($code, $code, $upper) if $upper;
745 $To{Lower}->RawAppendRange($code, $code, $lower) if $lower;
746 $To{Title}->RawAppendRange($code, $code, $title) if $title;
747 $To{Digit}->Append($code, $decimal) if length $decimal;
749 $Bidi->Append($code, $bidi);
750 $Comb->Append($code, $comb) if $comb;
751 $Number->Append($code, $number) if length $number;
753 $Mirrored->Append($code) if $mirrored eq "Y";
755 $Bidi{$bidi} ||= Table->New(Is => "Bidi$bidi",
756 Desc => "Bi-directional category '$bidi'",
758 $Bidi{$bidi}->Append($code);
762 $Deco->Append($code, $deco);
763 if ($deco =~/^<(\w+)>/)
765 $Deco{Compat}->Append($code);
767 $DC{$1} ||= Table->New(Is => "DC$1",
768 Desc => "Compatible with '$1'",
770 $DC{$1}->Append($code);
774 $Deco{Canon}->Append($code);
782 ## Tidy up a few special cases....
785 $Cat{Cn} = $Assigned->Invert; ## Cn is everything that doesn't exist
788 Desc => "General Category 'Cn' [not functional in Perl]",
791 ## Unassigned is the same as 'Cn'
792 New_Alias(Is => 'Unassigned', SameAs => 'Cn', Fuzzy => 0);
794 $Cat{C}->Replace($Cat{C}->Merge($Cat{Cn})); ## Now merge in Cn into C
797 # L& is Ll, Lu, and Lt.
799 Table->Merge(@Cat{qw[Ll Lu Lt]}),
800 Desc => '[\p{Ll}\p{Lu}\p{Lt}]',
803 ## Any and All are all code points.
804 my $Any = Table->New(Is => 'Any',
805 Desc => sprintf("[\\x{0000}-\\x{%X}]",
806 $LastUnicodeCodepoint),
808 $Any->RawAppendRange(0, $LastUnicodeCodepoint);
810 New_Alias(Is => 'All', SameAs => 'Any', Fuzzy => 0);
813 ## Build special properties for Perl's internal case-folding needs:
814 ## \p{_CaseIgnorable}
817 ## _CombAbove was built above. Others are built here....
820 ## \p{_CaseIgnorable} is [\p{Mn}\0x00AD\x2010]
821 New_Prop(Is => '_CaseIgnorable',
822 Table->Merge($Cat{Mn},
825 Desc => '(for internal casefolding use)',
829 ## \p{_CanonDCIJ} is fairly complex...
830 my $CanonCDIJ = Table->New(Is => '_CanonDCIJ',
831 Desc => '(for internal casefolding use)',
833 ## It contains the ASCII 'i' and 'j'....
834 $CanonCDIJ->Append(0x0069); # ASCII ord("i")
835 $CanonCDIJ->Append(0x006A); # ASCII ord("j")
836 ## ...and any character with a decomposition that starts with either of
837 ## those code points, but only if the decomposition does not have any
838 ## combining character with the "ABOVE" canonical combining class.
839 for my $code (sort { $a <=> $b} keys %CodeToDeco)
841 ## Need to ensure that all decomposition characters do not have
842 ## a %HexCodeToComb in %AboveCombClasses.
844 for my $deco_hexcode (split / /, $CodeToDeco{$code})
846 if (exists $_Above_HexCodes{$deco_hexcode}) {
847 ## one of the decmposition chars has an ABOVE combination
848 ## class, so we're not interested in this one
854 $CanonCDIJ->Append($code);
861 ## Now dump the files.
863 $Name->Write("Name.pl");
864 $Bidi->Write("Bidirectional.pl");
865 $Comb->Write("CombiningClass.pl");
866 $Deco->Write("Decomposition.pl");
867 $Number->Write("Number.pl");
868 $General->Write("Category.pl");
870 for my $to (sort keys %To) {
871 $To{$to}->Write("To/$to.pl");
876 ## Process LineBreak.txt
880 if (not open IN, "LineBreak.txt") {
881 die "$0: LineBreak.txt: $!\n";
884 my $Lbrk = Table->New();
889 next unless /^([0-9A-Fa-f]+)(?:\.\.([0-9A-Fa-f]+))?\s*;\s*(\w+)/;
891 my ($first, $last, $lbrk) = (hex($1), hex($2||""), $3);
893 $Lbrk->Append($first, $lbrk);
895 $Lbrk{$lbrk} ||= Table->New(Is => "Lbrk$lbrk",
896 Desc => "Linebreak category '$lbrk'",
898 $Lbrk{$lbrk}->Append($first);
901 $Lbrk->Extend($last);
902 $Lbrk{$lbrk}->Extend($last);
907 $Lbrk->Write("Lbrk.pl");
911 ## Process ArabicShaping.txt.
913 sub ArabicShaping_txt()
915 if (not open IN, "ArabicShaping.txt") {
916 die "$0: ArabicShaping.txt: $!\n";
919 my $ArabLink = Table->New();
920 my $ArabLinkGroup = Table->New();
924 next unless /^[0-9A-Fa-f]+;/;
927 my ($hexcode, $name, $link, $linkgroup) = split(/\s*;\s*/);
928 my $code = hex($hexcode);
929 $ArabLink->Append($code, $link);
930 $ArabLinkGroup->Append($code, $linkgroup);
934 $ArabLink->Write("ArabLink.pl");
935 $ArabLinkGroup->Write("ArabLnkGrp.pl");
943 if (not open IN, "Jamo.txt") {
944 die "$0: Jamo.txt: $!\n";
946 my $Short = Table->New();
950 next unless /^([0-9A-Fa-f]+)\s*;\s*(\w*)/;
951 my ($code, $short) = (hex($1), $2);
953 $Short->Append($code, $short);
956 $Short->Write("JamoShort.pl");
960 ## Process Scripts.txt.
966 if (not open(IN, "Scripts.txt")) {
967 die "$0: Scripts.txt: $!\n";
970 next unless /^([0-9A-Fa-f]+)(?:\.\.([0-9A-Fa-f]+))?\s*;\s*(.+?)\s*\#/;
972 # Wait until all the scripts have been read since
973 # they are not listed in numeric order.
974 push @ScriptInfo, [ hex($1), hex($2||""), $3 ];
978 # Now append the scripts properties in their code point order.
981 my $Scripts = Table->New();
983 for my $script (sort { $a->[0] <=> $b->[0] } @ScriptInfo)
985 my ($first, $last, $name) = @$script;
986 $Scripts->Append($first, $name);
988 $Script{$name} ||= Table->New(Is => $name,
989 Desc => "Script '$name'",
991 $Script{$name}->Append($first, $name);
994 $Scripts->Extend($last);
995 $Script{$name}->Extend($last);
999 $Scripts->Write("Scripts.pl");
1001 ## Common is everything not explicitly assigned to a Script
1003 ## ***shouldn't this be intersected with \p{Assigned}? ******
1005 New_Prop(Is => 'Common',
1007 Desc => 'Pseudo-Script of codepoints not in other Unicode scripts',
1012 ## Given a name like "Close Punctuation", return a regex (that when applied
1013 ## with /i) matches any valid form of that name (e.g. "ClosePunctuation",
1014 ## "Close-Punctuation", etc.)
1016 ## Accept any space, dash, or underbar where in the official name there is
1017 ## space or a dash (or underbar, but there never is).
1023 $Name =~ s/[- _]/(?:[-_]|\\s+)?/g;
1028 ## Process Blocks.txt.
1032 my $Blocks = Table->New();
1035 if (not open IN, "Blocks.txt") {
1036 die "$0: Blocks.txt: $!\n";
1041 #next if not /Private Use$/;
1042 next if not /^([0-9A-Fa-f]+)\.\.([0-9A-Fa-f]+)\s*;\s*(.+?)\s*$/;
1044 my ($first, $last, $name) = (hex($1), hex($2), $3);
1046 $Blocks->Append($first, $name);
1048 $Blocks{$name} ||= Table->New(In => $name,
1049 Desc => "Block '$name'",
1051 $Blocks{$name}->Append($first, $name);
1053 if ($last and $last != $first) {
1054 $Blocks->Extend($last);
1055 $Blocks{$name}->Extend($last);
1060 $Blocks->Write("Blocks.pl");
1064 ## Read in the PropList.txt. It contains extended properties not
1065 ## listed in the UnicodeData.txt, such as 'Other_Alphabetic':
1066 ## alphabetic but not of the general category L; many modifiers
1067 ## belong to this extended property category: while they are not
1068 ## alphabets, they are alphabetic in nature.
1074 if (not open IN, "PropList.txt") {
1075 die "$0: PropList.txt: $!\n";
1080 next unless /^([0-9A-Fa-f]+)(?:\.\.([0-9A-Fa-f]+))?\s*;\s*(.+?)\s*\#/;
1082 # Wait until all the extended properties have been read since
1083 # they are not listed in numeric order.
1084 push @PropInfo, [ hex($1), hex($2||""), $3 ];
1088 # Now append the extended properties in their code point order.
1089 my $Props = Table->New();
1092 for my $prop (sort { $a->[0] <=> $b->[0] } @PropInfo)
1094 my ($first, $last, $name) = @$prop;
1095 $Props->Append($first, $name);
1097 $Prop{$name} ||= Table->New(Is => $name,
1098 Desc => "Extended property '$name'",
1100 $Prop{$name}->Append($first, $name);
1103 $Props->Extend($last);
1104 $Prop{$name}->Extend($last);
1108 # Alphabetic is L and Other_Alphabetic.
1109 New_Prop(Is => 'Alphabetic',
1110 Table->Merge($Cat{L}, $Prop{Other_Alphabetic}),
1111 Desc => '[\p{L}\p{OtherAlphabetic}]', # use canonical names here
1114 # Lowercase is Ll and Other_Lowercase.
1115 New_Prop(Is => 'Lowercase',
1116 Table->Merge($Cat{Ll}, $Prop{Other_Lowercase}),
1117 Desc => '[\p{Ll}\p{OtherLowercase}]', # use canonical names here
1120 # Uppercase is Lu and Other_Uppercase.
1121 New_Prop(Is => 'Uppercase',
1122 Table->Merge($Cat{Lu}, $Prop{Other_Uppercase}),
1123 Desc => '[\p{Lu}\p{Other_Uppercase}]', # use canonical names here
1126 # Math is Sm and Other_Math.
1127 New_Prop(Is => 'Math',
1128 Table->Merge($Cat{Sm}, $Prop{Other_Math}),
1129 Desc => '[\p{Sm}\p{OtherMath}]', # use canonical names here
1132 # ID_Start is Ll, Lu, Lt, Lm, Lo, and Nl.
1133 New_Prop(Is => 'ID_Start',
1134 Table->Merge(@Cat{qw[Ll Lu Lt Lm Lo Nl]}),
1135 Desc => '[\p{Ll}\p{Lu}\p{Lt}\p{Lm}\p{Lo}\p{Nl}]',
1138 # ID_Continue is ID_Start, Mn, Mc, Nd, and Pc.
1139 New_Prop(Is => 'ID_Continue',
1140 Table->Merge(@Cat{qw[Ll Lu Lt Lm Lo Nl Mn Mc Nd Pc ]}),
1141 Desc => '[\p{ID_Start}\p{Mn}\p{Mc}\p{Nd}\p{Pc}]',
1145 sub Make_GC_Aliases()
1148 ## The mapping from General Category long forms to short forms is
1149 ## currently hardwired here since no simple data file in the UCD
1150 ## seems to do that. Unicode 3.2 will assumedly correct this.
1154 'Uppercase_Letter' => 'Lu',
1155 'Lowercase_Letter' => 'Ll',
1156 'Titlecase_Letter' => 'Lt',
1157 'Modifier_Letter' => 'Lm',
1158 'Other_Letter' => 'Lo',
1161 'Non_Spacing_Mark' => 'Mn',
1162 'Spacing_Mark' => 'Mc',
1163 'Enclosing_Mark' => 'Me',
1166 'Space_Separator' => 'Zs',
1167 'Line_Separator' => 'Zl',
1168 'Paragraph_Separator' => 'Zp',
1171 'Decimal_Number' => 'Nd',
1172 'Letter_Number' => 'Nl',
1173 'Other_Number' => 'No',
1175 'Punctuation' => 'P',
1176 'Connector_Punctuation' => 'Pc',
1177 'Dash_Punctuation' => 'Pd',
1178 'Open_Punctuation' => 'Ps',
1179 'Close_Punctuation' => 'Pe',
1180 'Initial_Punctuation' => 'Pi',
1181 'Final_Punctuation' => 'Pf',
1182 'Other_Punctuation' => 'Po',
1185 'Math_Symbol' => 'Sm',
1186 'Currency_Symbol' => 'Sc',
1187 'Modifier_Symbol' => 'Sk',
1188 'Other_Symbol' => 'So',
1193 'Surrogate' => 'Cs',
1194 'Private Use' => 'Co',
1195 'Unassigned' => 'Cn',
1198 ## make the aliases....
1199 while (my ($Alias, $Name) = each %Is) {
1200 New_Alias(Is => $Alias, SameAs => $Name, Fuzzy => 1);
1206 ## These are used in:
1207 ## MakePropTestScript()
1208 ## WriteAllMappings()
1209 ## for making the test script.
1211 my %FuzzyNameToTest;
1212 my %ExactNameToTest;
1215 ## This used only for making the test script
1220 my $MatchCode = shift;
1221 my $FailCode = shift;
1223 if (defined $MatchCode) {
1224 printf $FH qq/Expect(1, "\\x{%04X}", '\\p{$Prop}' );\n/, $MatchCode;
1225 printf $FH qq/Expect(0, "\\x{%04X}", '\\p{^$Prop}');\n/, $MatchCode;
1226 printf $FH qq/Expect(0, "\\x{%04X}", '\\P{$Prop}' );\n/, $MatchCode;
1227 printf $FH qq/Expect(1, "\\x{%04X}", '\\P{^$Prop}');\n/, $MatchCode;
1229 if (defined $FailCode) {
1230 printf $FH qq/Expect(0, "\\x{%04X}", '\\p{$Prop}' );\n/, $FailCode;
1231 printf $FH qq/Expect(1, "\\x{%04X}", '\\p{^$Prop}');\n/, $FailCode;
1232 printf $FH qq/Expect(1, "\\x{%04X}", '\\P{$Prop}' );\n/, $FailCode;
1233 printf $FH qq/Expect(0, "\\x{%04X}", '\\P{^$Prop}');\n/, $FailCode;
1237 ## This used only for making the test script
1243 print $FH qq/Error('\\p{$prop}');\n/;
1244 print $FH qq/Error('\\P{$prop}');\n/;
1247 ## This used only for making the test script
1263 ## This used only for making the test script
1264 sub RandomlyFuzzifyName($;$)
1267 my $WantError = shift; ## if true, make an error
1270 for my $part (split /[-\s_]+/, $Name)
1273 if ($WantError and rand() < 0.3) {
1274 push @parts, $BadSeps[rand(@BadSeps)];
1277 push @parts, $GoodSeps[rand(@GoodSeps)];
1280 my $switch = int rand(4);
1282 push @parts, uc $part;
1283 } elsif ($switch == 1) {
1284 push @parts, lc $part;
1285 } elsif ($switch == 2) {
1286 push @parts, ucfirst $part;
1291 my $new = join('', @parts);
1294 if (rand() >= 0.5) {
1295 $new .= $BadSeps[rand(@BadSeps)];
1297 $new = $BadSeps[rand(@BadSeps)] . $new;
1303 ## This used only for making the test script
1304 sub MakePropTestScript()
1306 ## this written directly -- it's huge.
1307 if (not open OUT, ">TestProp.pl") {
1308 die "$0: TestProp.pl: $!\n";
1312 while (my ($Name, $Table) = each %ExactNameToTest)
1314 GenTests(*OUT, $Name, $Table->ValidCode, $Table->InvalidCode);
1315 ExpectError(*OUT, uc $Name) if uc $Name ne $Name;
1316 ExpectError(*OUT, lc $Name) if lc $Name ne $Name;
1320 while (my ($Name, $Table) = each %FuzzyNameToTest)
1322 my $Orig = $CanonicalToOrig{$Name};
1326 RandomlyFuzzifyName($Orig) => 1
1329 for my $N (keys %Names) {
1330 GenTests(*OUT, $N, $Table->ValidCode, $Table->InvalidCode);
1333 ExpectError(*OUT, RandomlyFuzzifyName($Orig, 'ERROR'));
1336 print OUT "Finished();\n";
1342 ## These are used only in:
1343 ## RegisterFileForName()
1344 ## WriteAllMappings()
1346 my %Exact; ## will become %utf8::Exact;
1347 my %Canonical; ## will become %utf8::Canonical;
1348 my %CaComment; ## Comment for %Canonical entry of same key
1351 ## Given info about a name and a datafile that it should be associated with,
1352 ## register that assocation in %Exact and %Canonical.
1353 sub RegisterFileForName($$$$)
1357 my $IsFuzzy = shift;
1358 my $filename = shift;
1361 ## Now in details for the mapping. $Type eq 'Is' has the
1362 ## Is removed, as it will be removed in utf8_heavy when this
1363 ## data is being checked. In keeps its "In", but a second
1364 ## sans-In record is written if it doesn't conflict with
1365 ## anything already there.
1369 if ($Type eq 'Is') {
1370 die "oops[$Name]" if $Exact{$Name};
1371 $Exact{$Name} = $filename;
1373 die "oops[$Type$Name]" if $Exact{"$Type$Name"};
1374 $Exact{"$Type$Name"} = $filename;
1375 $Exact{$Name} = $filename if not $Exact{$Name};
1380 my $CName = lc $Name;
1381 if ($Type eq 'Is') {
1382 die "oops[$CName]" if $Canonical{$CName};
1383 $Canonical{$CName} = $filename;
1384 $CaComment{$CName} = $Name if $Name =~ tr/A-Z// >= 2;
1386 die "oops[$Type$CName]" if $Canonical{lc "$Type$CName"};
1387 $Canonical{lc "$Type$CName"} = $filename;
1388 $CaComment{lc "$Type$CName"} = "$Type$Name";
1389 if (not $Canonical{$CName}) {
1390 $Canonical{$CName} = $filename;
1391 $CaComment{$CName} = "$Type$Name";
1398 ## Writes the info accumulated in
1405 sub WriteAllMappings()
1409 my %BaseNames; ## Base names already used (for avoiding 8.3 conflicts)
1411 ## 'Is' *MUST* come first, so its names have precidence over 'In's
1412 for my $Type ('Is', 'In')
1414 my %RawNameToFile; ## a per-$Type cache
1416 for my $Name (sort {length $a <=> length $b} keys %{$TableInfo{$Type}})
1418 ## Note: $Name is already canonical
1419 my $Table = $TableInfo{$Type}->{$Name};
1420 my $IsFuzzy = $FuzzyNames{$Type}->{$Name};
1422 ## Need an 8.3 safe filename (which means "an 8 safe" $filename)
1425 ## 'Is' items lose 'Is' from the basename.
1426 $filename = $Type eq 'Is' ? $Name : "$Type$Name";
1428 $filename =~ s/[^\w_]+/_/g; # "L&" -> "L_"
1429 substr($filename, 8) = '' if length($filename) > 8;
1432 ## Make sure the basename doesn't conflict with something we
1433 ## might have already written. If we have, say,
1440 while (my $num = $BaseNames{lc $filename}++)
1442 $num++; ## so basenames with numbers start with '2', which
1443 ## just looks more natural.
1444 ## Want to append $num, but if it'll make the basename longer
1445 ## than 8 characters, pre-truncate $filename so that the result
1447 my $delta = length($filename) + length($num) - 8;
1449 substr($filename, -$delta) = $num;
1457 ## Construct a nice comment to add to the file, and build data
1458 ## for the "./Properties" file along the way.
1462 my $Desc = $TableDesc{$Type}->{$Name} || "";
1463 ## get list of names this table is reference by
1464 my @Supported = $Name;
1465 while (my ($Orig, $Alias) = each %{ $AliasInfo{$Type} })
1467 if ($Orig eq $Name) {
1468 push @Supported, $Alias;
1472 my $TypeToShow = $Type eq 'Is' ? "" : $Type;
1475 $Comment = "This file supports:\n";
1476 for my $N (@Supported)
1478 my $IsFuzzy = $FuzzyNames{$Type}->{$N};
1479 my $Prop = "\\p{$TypeToShow$Name}";
1480 $OrigProp = $Prop if not $OrigProp; #cache for aliases
1482 $Comment .= "\t$Prop (and fuzzy permutations)\n";
1484 $Comment .= "\t$Prop\n";
1486 my $MyDesc = ($N eq $Name) ? $Desc : "Alias for $OrigProp ($Desc)";
1488 push @MAP, sprintf("%s %-42s %s\n",
1489 $IsFuzzy ? '*' : ' ', $Prop, $MyDesc);
1492 $Comment .= "\nMeaning: $Desc\n";
1497 ## Okay, write the file...
1499 $Table->Write("lib/$filename.pl", $Comment);
1502 $RawNameToFile{$Name} = $filename;
1503 RegisterFileForName($Type => $Name, $IsFuzzy, $filename);
1507 my $CName = CanonicalName($Type . '_'. $Name);
1508 $FuzzyNameToTest{$Name} = $Table if !$FuzzyNameToTest{$Name};
1509 $FuzzyNameToTest{$CName} = $Table if !$FuzzyNameToTest{$CName};
1511 $ExactNameToTest{$Name} = $Table;
1516 ## Register aliase info
1517 for my $Name (sort {length $a <=> length $b} keys %{$AliasInfo{$Type}})
1519 my $Alias = $AliasInfo{$Type}->{$Name};
1520 my $IsFuzzy = $FuzzyNames{$Type}->{$Alias};
1521 my $filename = $RawNameToFile{$Name};
1522 die "oops [$Alias]->[$Name]" if not $filename;
1523 RegisterFileForName($Type => $Alias, $IsFuzzy, $filename);
1525 my $Table = $TableInfo{$Type}->{$Name};
1526 die "oops" if not $Table;
1529 my $CName = CanonicalName($Type .'_'. $Alias);
1530 $FuzzyNameToTest{$Alias} = $Table if !$FuzzyNameToTest{$Alias};
1531 $FuzzyNameToTest{$CName} = $Table if !$FuzzyNameToTest{$CName};
1533 $ExactNameToTest{$Alias} = $Table;
1539 ## Write out the property list
1544 "## This file created by $0\n",
1545 "## List of built-in \\p{...}/\\P{...} properties.\n",
1547 "## '*' means name may be 'fuzzy'\n",
1549 sort { substr($a,2) cmp substr($b, 2) } @MAP,
1551 WriteIfChanged('Properties', @OUT);
1554 use Text::Tabs (); ## using this makes the files about half the size
1561 "## Data in this file used by ../utf8_heavy.pl\n",
1563 "## Mapping from name to filename in ./lib\n",
1564 "%utf8::Exact = (\n",
1567 for my $Name (sort keys %Exact)
1569 my $File = $Exact{$Name};
1570 $Name = $Name =~ m/\W/ ? qq/'$Name'/ : " $Name ";
1571 my $Text = sprintf("%-15s => %s,\n", $Name, qq/'$File'/);
1572 push @OUT, Text::Tabs::unexpand($Text);
1574 push @OUT, ");\n1;\n";
1576 WriteIfChanged('Exact.pl', @OUT);
1579 ## Write Canonical.pl
1584 "## Data in this file used by ../utf8_heavy.pl\n",
1586 "## Mapping from lc(canonical name) to filename in ./lib\n",
1587 "%utf8::Canonical = (\n",
1589 my $Trail = ""; ## used just to keep the spacing pretty
1590 for my $Name (sort keys %Canonical)
1592 my $File = $Canonical{$Name};
1593 if ($CaComment{$Name}) {
1594 push @OUT, "\n" if not $Trail;
1595 push @OUT, " # $CaComment{$Name}\n";
1600 $Name = $Name =~ m/\W/ ? qq/'$Name'/ : " $Name ";
1601 my $Text = sprintf(" %-41s => %s,\n$Trail", $Name, qq/'$File'/);
1602 push @OUT, Text::Tabs::unexpand($Text);
1604 push @OUT, ");\n1\n";
1605 WriteIfChanged('Canonical.pl', @OUT);
1608 MakePropTestScript() if $MakeTestScript;
1612 sub SpecialCasing_txt()
1615 # Read in the special cases.
1620 if (not open IN, "SpecialCasing.txt") {
1621 die "$0: SpecialCasing.txt: $!\n";
1624 next unless /^[0-9A-Fa-f]+;/;
1628 my ($code, $lower, $title, $upper, $condition) = split(/\s*;\s*/);
1630 if ($condition) { # not implemented yet
1631 print "# SKIPPING $_\n" if $Verbose;
1635 # Wait until all the special cases have been read since
1636 # they are not listed in numeric order.
1637 my $ix = hex($code);
1638 push @{$CaseInfo{Lower}}, [ $ix, $code, $lower ];
1639 push @{$CaseInfo{Title}}, [ $ix, $code, $title ];
1640 push @{$CaseInfo{Upper}}, [ $ix, $code, $upper ];
1644 # Now write out the special cases properties in their code point order.
1645 # Prepend them to the To/{Upper,Lower,Title}.pl.
1647 for my $case (qw(Lower Title Upper))
1649 my $NormalCase = do "To/$case.pl" || die "$0: $@\n";
1653 "%utf8::ToSpec$case =\n(\n",
1656 for my $prop (sort { $a->[0] <=> $b->[0] } @{$CaseInfo{$case}}) {
1657 my ($ix, $code, $to) = @$prop;
1659 join "", map { sprintf "\\x{%s}", $_ } split ' ', $to;
1660 push @OUT, sprintf qq['%04X' => "$tostr",\n], $ix;
1664 "return <<'END';\n",
1668 WriteIfChanged("To/$case.pl", @OUT);
1673 # Read in the case foldings.
1675 # We will do full case folding, C + F + I (see CaseFolding.txt).
1677 sub CaseFolding_txt()
1679 if (not open IN, "CaseFolding.txt") {
1680 die "$0: CaseFolding.txt: $!\n";
1683 my $Fold = Table->New();
1687 # Skip status 'S', simple case folding
1688 next unless /^([0-9A-Fa-f]+)\s*;\s*([CFI])\s*;\s*([0-9A-Fa-f]+(?: [0-9A-Fa-f]+)*)\s*;/;
1690 my ($code, $status, $fold) = (hex($1), $2, $3);
1692 if ($status eq 'C') { # Common: one-to-one folding
1693 # No append() since several codes may fold into one.
1694 $Fold->RawAppendRange($code, $code, $fold);
1695 } else { # F: full, or I: dotted uppercase I -> dotless lowercase I
1696 $Fold{$code} = $fold;
1701 $Fold->Write("To/Fold.pl");
1704 # Prepend the special foldings to the common foldings.
1706 my $CommonFold = do "To/Fold.pl" || die "$0: To/Fold.pl: $!\n";
1710 "%utf8::ToSpecFold =\n(\n",
1712 for my $code (sort { $a <=> $b } keys %Fold) {
1714 join "", map { sprintf "\\x{%s}", $_ } split ' ', $Fold{$code};
1715 push @OUT, sprintf qq['%04X' => "$foldstr",\n], $code;
1719 "return <<'END';\n",
1724 WriteIfChanged("To/Fold.pl", @OUT);
1739 ArabicShaping_txt();
1741 SpecialCasing_txt();
1746 ## TRAILING CODE IS USED BY MakePropTestScript()
1759 my $Line = (caller)[2];
1764 $RegObj = qr/$Regex/;
1765 $String =~ $RegObj ? 1 : 0
1768 if (not defined $result) {
1769 print "couldn't compile /$Regex/ on $0 line $Line: $@\n";
1771 } elsif ($result ^ $Expect) {
1772 print "bad result (expected $Expect) on $0 line $Line: $@\n";
1781 if (eval { 'x' =~ qr/$Regex/; 1 }) {
1783 my $Line = (caller)[2];
1784 print "expected error for /$Regex/ on $0 line $Line: $@\n";
1791 print "All $Tests tests passed.\n";
1794 print "$Tests tests, $Fails failed!\n";