Add the special casing mappings (from SpecCase.txt)

(except for the hyper special case mappings that have
the condition list); the special casing mappings are
unused for now.  Small tidying up on mktables.

p4raw-id: //depot/perl@12465

4 files changed, 376 insertions, 28 deletions

diff --git a/lib/unicore/To/SpecLower.pl b/lib/unicore/To/SpecLower.pl
new file mode 100644
index 0000000000..18c073b98b
--- /dev/null
+++ b/lib/unicore/To/SpecLower.pl
@@ -0,0 +1,107 @@
+# !!!!!!!   DO NOT EDIT THIS FILE   !!!!!!! 
+# This file is built by mktables from e.g. Unicode.txt.
+# Any changes made here will be lost!
+return <<'END';
+00DF		00DF
+0149		0149
+01F0		01F0
+0390		0390
+03B0		03B0
+0587		0587
+1E96		1E96
+1E97		1E97
+1E98		1E98
+1E99		1E99
+1E9A		1E9A
+1F50		1F50
+1F52		1F52
+1F54		1F54
+1F56		1F56
+1F80		1F80
+1F81		1F81
+1F82		1F82
+1F83		1F83
+1F84		1F84
+1F85		1F85
+1F86		1F86
+1F87		1F87
+1F88		1F80
+1F89		1F81
+1F8A		1F82
+1F8B		1F83
+1F8C		1F84
+1F8D		1F85
+1F8E		1F86
+1F8F		1F87
+1F90		1F90
+1F91		1F91
+1F92		1F92
+1F93		1F93
+1F94		1F94
+1F95		1F95
+1F96		1F96
+1F97		1F97
+1F98		1F90
+1F99		1F91
+1F9A		1F92
+1F9B		1F93
+1F9C		1F94
+1F9D		1F95
+1F9E		1F96
+1F9F		1F97
+1FA0		1FA0
+1FA1		1FA1
+1FA2		1FA2
+1FA3		1FA3
+1FA4		1FA4
+1FA5		1FA5
+1FA6		1FA6
+1FA7		1FA7
+1FA8		1FA0
+1FA9		1FA1
+1FAA		1FA2
+1FAB		1FA3
+1FAC		1FA4
+1FAD		1FA5
+1FAE		1FA6
+1FAF		1FA7
+1FB2		1FB2
+1FB3		1FB3
+1FB4		1FB4
+1FB6		1FB6
+1FB7		1FB7
+1FBC		1FB3
+1FC2		1FC2
+1FC3		1FC3
+1FC4		1FC4
+1FC6		1FC6
+1FC7		1FC7
+1FCC		1FC3
+1FD2		1FD2
+1FD3		1FD3
+1FD6		1FD6
+1FD7		1FD7
+1FE2		1FE2
+1FE3		1FE3
+1FE4		1FE4
+1FE6		1FE6
+1FE7		1FE7
+1FF2		1FF2
+1FF3		1FF3
+1FF4		1FF4
+1FF6		1FF6
+1FF7		1FF7
+1FFC		1FF3
+FB00		FB00
+FB01		FB01
+FB02		FB02
+FB03		FB03
+FB04		FB04
+FB05		FB05
+FB06		FB06
+FB13		FB13
+FB14		FB14
+FB15		FB15
+FB16		FB16
+FB17		FB17
+END
diff --git a/lib/unicore/To/SpecTitle.pl b/lib/unicore/To/SpecTitle.pl
new file mode 100644
index 0000000000..c3e1911408
--- /dev/null
+++ b/lib/unicore/To/SpecTitle.pl
@@ -0,0 +1,106 @@
+# !!!!!!!   DO NOT EDIT THIS FILE   !!!!!!! 
+# This file is built by mktables from e.g. Unicode.txt.
+# Any changes made here will be lost!
+return <<'END';
+00DF		0053 0073
+0149		02BC 004E
+01F0		004A 030C
+0390		0399 0308 0301
+03B0		03A5 0308 0301
+0587		0535 0582
+1E96		0048 0331
+1E97		0054 0308
+1E98		0057 030A
+1E99		0059 030A
+1E9A		0041 02BE
+1F50		03A5 0313
+1F52		03A5 0313 0300
+1F54		03A5 0313 0301
+1F56		03A5 0313 0342
+1F80		1F88
+1F81		1F89
+1F82		1F8A
+1F83		1F8B
+1F84		1F8C
+1F85		1F8D
+1F86		1F8E
+1F87		1F8F
+1F88		1F88
+1F89		1F89
+1F8A		1F8A
+1F8B		1F8B
+1F8C		1F8C
+1F8D		1F8D
+1F8E		1F8E
+1F8F		1F8F
+1F90		1F98
+1F91		1F99
+1F92		1F9A
+1F93		1F9B
+1F94		1F9C
+1F95		1F9D
+1F96		1F9E
+1F97		1F9F
+1F98		1F98
+1F99		1F99
+1F9A		1F9A
+1F9B		1F9B
+1F9C		1F9C
+1F9D		1F9D
+1F9E		1F9E
+1F9F		1F9F
+1FA0		1FA8
+1FA1		1FA9
+1FA2		1FAA
+1FA3		1FAB
+1FA4		1FAC
+1FA5		1FAD
+1FA6		1FAE
+1FA7		1FAF
+1FA8		1FA8
+1FA9		1FA9
+1FAA		1FAA
+1FAB		1FAB
+1FAC		1FAC
+1FAD		1FAD
+1FAE		1FAE
+1FAF		1FAF
+1FB2		1FBA 0345
+1FB3		1FBC
+1FB4		0386 0345
+1FB6		0391 0342
+1FB7		0391 0342 0345
+1FBC		1FBC
+1FC2		1FCA 0345
+1FC3		1FCC
+1FC4		0389 0345
+1FC6		0397 0342
+1FC7		0397 0342 0345
+1FCC		1FCC
+1FD2		0399 0308 0300
+1FD3		0399 0308 0301
+1FD6		0399 0342
+1FD7		0399 0308 0342
+1FE2		03A5 0308 0300
+1FE3		03A5 0308 0301
+1FE4		03A1 0313
+1FE6		03A5 0342
+1FE7		03A5 0308 0342
+1FF2		1FFA 0345
+1FF3		1FFC
+1FF4		038F 0345
+1FF6		03A9 0342
+1FF7		03A9 0342 0345
+1FFC		1FFC
+FB00		0046 0066
+FB01		0046 0069
+FB02		0046 006C
+FB03		0046 0066 0069
+FB04		0046 0066 006C
+FB05	FB06	0053 0074
+FB13		0544 0576
+FB14		0544 0565
+FB15		0544 056B
+FB16		054E 0576
+FB17		0544 056D
+END
diff --git a/lib/unicore/To/SpecUpper.pl b/lib/unicore/To/SpecUpper.pl
new file mode 100644
index 0000000000..e5af4b1089
--- /dev/null
+++ b/lib/unicore/To/SpecUpper.pl
@@ -0,0 +1,106 @@
+# !!!!!!!   DO NOT EDIT THIS FILE   !!!!!!! 
+# This file is built by mktables from e.g. Unicode.txt.
+# Any changes made here will be lost!
+return <<'END';
+00DF		0053 0053
+0149		02BC 004E
+01F0		004A 030C
+0390		0399 0308 0301
+03B0		03A5 0308 0301
+0587		0535 0552
+1E96		0048 0331
+1E97		0054 0308
+1E98		0057 030A
+1E99		0059 030A
+1E9A		0041 02BE
+1F50		03A5 0313
+1F52		03A5 0313 0300
+1F54		03A5 0313 0301
+1F56		03A5 0313 0342
+1F80		1F08 0399
+1F81		1F09 0399
+1F82		1F0A 0399
+1F83		1F0B 0399
+1F84		1F0C 0399
+1F85		1F0D 0399
+1F86		1F0E 0399
+1F87		1F0F 0399
+1F88		1F08 0399
+1F89		1F09 0399
+1F8A		1F0A 0399
+1F8B		1F0B 0399
+1F8C		1F0C 0399
+1F8D		1F0D 0399
+1F8E		1F0E 0399
+1F8F		1F0F 0399
+1F90		1F28 0399
+1F91		1F29 0399
+1F92		1F2A 0399
+1F93		1F2B 0399
+1F94		1F2C 0399
+1F95		1F2D 0399
+1F96		1F2E 0399
+1F97		1F2F 0399
+1F98		1F28 0399
+1F99		1F29 0399
+1F9A		1F2A 0399
+1F9B		1F2B 0399
+1F9C		1F2C 0399
+1F9D		1F2D 0399
+1F9E		1F2E 0399
+1F9F		1F2F 0399
+1FA0		1F68 0399
+1FA1		1F69 0399
+1FA2		1F6A 0399
+1FA3		1F6B 0399
+1FA4		1F6C 0399
+1FA5		1F6D 0399
+1FA6		1F6E 0399
+1FA7		1F6F 0399
+1FA8		1F68 0399
+1FA9		1F69 0399
+1FAA		1F6A 0399
+1FAB		1F6B 0399
+1FAC		1F6C 0399
+1FAD		1F6D 0399
+1FAE		1F6E 0399
+1FAF		1F6F 0399
+1FB2		1FBA 0399
+1FB3		0391 0399
+1FB4		0386 0399
+1FB6		0391 0342
+1FB7		0391 0342 0399
+1FBC		0391 0399
+1FC2		1FCA 0399
+1FC3		0397 0399
+1FC4		0389 0399
+1FC6		0397 0342
+1FC7		0397 0342 0399
+1FCC		0397 0399
+1FD2		0399 0308 0300
+1FD3		0399 0308 0301
+1FD6		0399 0342
+1FD7		0399 0308 0342
+1FE2		03A5 0308 0300
+1FE3		03A5 0308 0301
+1FE4		03A1 0313
+1FE6		03A5 0342
+1FE7		03A5 0308 0342
+1FF2		1FFA 0399
+1FF3		03A9 0399
+1FF4		038F 0399
+1FF6		03A9 0342
+1FF7		03A9 0342 0399
+1FFC		03A9 0399
+FB00		0046 0046
+FB01		0046 0049
+FB02		0046 004C
+FB03		0046 0046 0049
+FB04		0046 0046 004C
+FB05	FB06	0053 0054
+FB13		0544 0546
+FB14		0544 0535
+FB15		0544 053B
+FB16		054E 0546
+FB17		0544 053D
+END
diff --git a/lib/unicore/mktables b/lib/unicore/mktables
index 5b2d7869f5..f851302d34 100644
--- a/lib/unicore/mktables
+++ b/lib/unicore/mktables
@@ -5,10 +5,10 @@
 # from the Unicode database files (lib/unicore/*.txt).
 #
 
-my $LastUnicodeCodepoint = 0x10FFFF; # As of Unicode 3.1.1.
-
 use strict;
 
+my $LastUnicodeCodepoint = 0x10FFFF; # As of Unicode 3.1.1.
+
 mkdir("In", 0755);
 mkdir("Is", 0755);
 mkdir("To", 0755);
@@ -123,9 +123,9 @@ if (open(my $Unicode, "Unicode.txt")) {
     my @Mirrored;
     my %To;
     while (<$Unicode>) {
-	next if /^\#/ || /^\s*$/;
-	next unless /^[0-9a-f]+\s*;/i;
+	next unless /^[0-9A-Fa-f]+;/;
 	s/\s+$//;
+
 	my ($code, $name, $cat, $comb, $bidi, $deco,
 	    $decimal, $digit, $number,
 	    $mirrored, $unicode10, $comment,
@@ -259,10 +259,7 @@ if (open(my $LineBrk, "LineBrk.txt")) {
     my %Lbrk;
 
     while (<$LineBrk>) {
-	next if /^\#/ || /^\s*$/;
-	s/\s+$//;
-	s/\s*\#.*//;
-	next unless /^([0-9a-f]+)(?:\.\.([0-9a-f]+))?\s*;\s*(.+)$/i;
+        next unless /^([0-9A-Fa-f]+)(?:\.\.([0-9A-Fa-f]+))?\s*;\s*(\w+)/;
 
 	my ($first, $last, $lbrk) = ($1, $2, $3);
 
@@ -291,9 +288,9 @@ if (open(my $ArabShap, "ArabShap.txt")) {
     my @ArabLinkGroup;
 
     while (<$ArabShap>) {
-	next if /^\#/ || /^\s*$/;
-	next unless /^[0-9a-f]+\s*;/i;
+	next unless /^[0-9A-Fa-f]+;/;
 	s/\s+$//;
+
 	my ($code, $name, $link, $linkgroup) = split(/\s*;\s*/);
 
 	append(\@ArabLink,      $code, $link);
@@ -314,11 +311,9 @@ if (open(my $Jamo, "Jamo.txt")) {
     my @Short;
 
     while (<$Jamo>) {
-	next if /^\#/ || /^\s*$/;
-	next unless /^[0-9a-f]+\s*;/i;
-	s/\s*\#.*//;
-	s/\s+$//;
-	my ($code, $short) = split(/\s*;\s*/);
+	next unless /^([0-9A-Fa-f]+)\s*;\s*(\w*)/;
+
+	my ($code, $short) = ($1, $2);
 
 	append(\@Short, $code, $short);
     }
@@ -336,10 +331,7 @@ my @Scripts;
 
 if (open(my $Scripts, "Scripts.txt")) {
     while (<$Scripts>) {
-	next if /^\#/ || /^\s*$/;
-	s/\s*\#.*//;
-	s/\s+$//;
-	next unless /^([0-9a-f]+)(?:\.\.([0-9a-f]+))?\s*;\s*(.+)$/i;
+	next unless /^([0-9A-Fa-f]+)(?:\.\.([0-9A-Fa-f]+))?\s*;\s*(.+?)\s*\#/;
 
 	# Wait until all the scripts have been read since
 	# they are not listed in numeric order.
@@ -387,11 +379,8 @@ my %Blocks;
 
 if (open(my $Blocks, "Blocks.txt")) {
     while (<$Blocks>) {
-	next if /^\#/ || /^\s*$/;
-	s/\s*\#.*//;
-	s/\s+$//;
-	next unless /^([0-9a-f]+)\.\.([0-9a-f]+)\s*;\s*(.+)$/i;
-
+	next unless /^([0-9A-Fa-f]+)\.\.([0-9A-Fa-f]+)\s*;\s*(.+?)\s*$/;
+	
 	my ($first, $last, $name) = ($1, $2, $3);
 
 	# If there's a naming conflict (the script names are
@@ -430,10 +419,7 @@ my @Props;
 
 if (open(my $Props, "PropList.txt")) {
     while (<$Props>) {
-	next if /^\#/ || /^\s*$/;
-	s/\s*\#.*//;
-	s/\s+$//;
-	next unless /^([0-9a-f]+)(?:\.\.([0-9a-f]+))?\s*;\s*(\w+)/i;
+	next unless /^([0-9A-Fa-f]+)(?:\.\.([0-9A-Fa-f]+))?\s*;\s*(.+?)\s*\#/;
 
 	# Wait until all the extended properties have been read since
 	# they are not listed in numeric order.
@@ -682,5 +668,48 @@ my %Is = (
 
 mapping(\%Is, "Is");
 
+#
+# Read in the special cases.
+#
+
+my %Case;
+
+if (open(my $SpecCase, "SpecCase.txt")) {
+    while (<$SpecCase>) {
+	next unless /^[0-9A-Fa-f]+;/;
+	s/\#.*//;
+	s/\s+$//;
+
+	my ($code, $lower, $title, $upper, $condition) = split(/\s*;\s*/);
+
+	if ($condition) { # not implemented yet
+	    print "# SKIPPING $_\n";
+	    next;
+	}
+
+	# Wait until all the special cases have been read since
+	# they are not listed in numeric order.
+	my $ix = hex($code);
+	push @{$Case{Lower}}, [ $ix, $code, $lower ];
+	push @{$Case{Title}}, [ $ix, $code, $title ];
+	push @{$Case{Upper}}, [ $ix, $code, $upper ];
+    }
+} else {
+    die "$0: SpecCase.txt: $!\n";
+}
+
+# Now write out the special cases properties in their code point order.
+# The To/Spec{Lower,Title,Upper}.pl are unused for now since the swash
+# routines do not do returning multiple characters.
+
+for my $case (qw(Lower Title Upper)) {
+    my @case;
+    for my $prop (sort { $a->[0] <=> $b->[0] } @{$Case{$case}}) {
+        my ($ix, $code, $to) = @$prop;
+        append(\@case, $code, $to);
+    }
+    flush(\@case, "To/Spec$case.pl");
+}
+
 # That's all, folks!