On closer reading the proposed UTS#18 update required

even more changes.

p4raw-id: //depot/perl@19686

diff --git a/lib/unicore/lib/Blank.pl b/lib/unicore/lib/Blank.pl
index 5de7d7d..639ccf4 100644 (file)
--- a/lib/unicore/lib/Blank.pl
+++ b/lib/unicore/lib/Blank.pl
@@ -14,7 +14,7 @@ return <<'END';
 00A0           
 1680           
 180E           
-2000   200B    
+2000   200A    
 202F           
 205F           
 3000           

diff --git a/lib/unicore/lib/Graph.pl b/lib/unicore/lib/Graph.pl
index 943056c..79ef3db 100644 (file)
--- a/lib/unicore/lib/Graph.pl
+++ b/lib/unicore/lib/Graph.pl
@@ -278,7 +278,8 @@ return <<'END';
 1FDD   1FEF    
 1FF2   1FF4    
 1FF6   1FFE    
-200C   202E    
+200B   2027    
+202A   202E    
 2030   2054    
 2057           
 2060   2063    

diff --git a/lib/unicore/lib/Print.pl b/lib/unicore/lib/Print.pl
index 54954e7..a5bae46 100644 (file)
--- a/lib/unicore/lib/Print.pl
+++ b/lib/unicore/lib/Print.pl
@@ -9,9 +9,10 @@
 # Meaning: [[:Print:]]
 #
 return <<'END';
+0009   000D    
 0020   007E    
-00A0   00AC    
-00AE   0236    
+0085           
+00A0   0236    
 0250   0357    
 035D   036F    
 0374   0375    
@@ -36,14 +37,14 @@ return <<'END';
 05BB   05C4    
 05D0   05EA    
 05F0   05F4    
+0600   0603    
 060C   0615    
 061B           
 061F           
 0621   063A    
 0640   0658    
-0660   06DC    
-06DE   070D    
-0710   074A    
+0660   070D    
+070F   074A    
 074D   074F    
 0780   07B1    
 0901   0939    
@@ -246,8 +247,7 @@ return <<'END';
 1760   176C    
 176E   1770    
 1772   1773    
-1780   17B3    
-17B6   17DD    
+1780   17DD    
 17E0   17E9    
 17F0   17F9    
 1800   180E    
@@ -280,12 +280,10 @@ return <<'END';
 1FDD   1FEF    
 1FF2   1FF4    
 1FF6   1FFE    
-2000   200B    
-2010   2027    
-202F   2054    
+2000   2054    
 2057           
-205F           
-2070   2071    
+205F   2063    
+206A   2071    
 2074   208E    
 20A0   20B1    
 20D0   20EA    
@@ -331,7 +329,7 @@ return <<'END';
 A000   A48C    
 A490   A4C6    
 AC00   D7A3    
-F900   FA2D    
+E000   FA2D    
 FA30   FA6A    
 FB00   FB06    
 FB13   FB17    
@@ -352,6 +350,7 @@ FE54        FE66
 FE68   FE6B    
 FE70   FE74    
 FE76   FEFC    
+FEFF           
 FF01   FFBE    
 FFC2   FFC7    
 FFCA   FFCF    
@@ -359,7 +358,7 @@ FFD2        FFD7
 FFDA   FFDC    
 FFE0   FFE6    
 FFE8   FFEE    
-FFFC   FFFD    
+FFF9   FFFD    
 10000  1000B   
 1000D  10026   
 10028  1003A   
@@ -385,8 +384,7 @@ FFFC        FFFD
 1083F          
 1D000  1D0F5   
 1D100  1D126   
-1D12A  1D172   
-1D17B  1D1DD   
+1D12A  1D1DD   
 1D300  1D356   
 1D400  1D454   
 1D456  1D49C   
@@ -411,5 +409,9 @@ FFFC        FFFD
 1D7CE  1D7FF   
 20000  2A6D6   
 2F800  2FA1D   
+E0001          
+E0020  E007F   
 E0100  E01EF   
+F0000  FFFFD   
+100000 10FFFD  
 END

diff --git a/lib/unicore/lib/Space.pl b/lib/unicore/lib/Space.pl
index 9aa12c8..6c1cc2e 100644 (file)
--- a/lib/unicore/lib/Space.pl
+++ b/lib/unicore/lib/Space.pl
@@ -15,7 +15,7 @@ return <<'END';
 00A0           
 1680           
 180E           
-2000   200B    
+2000   200A    
 2028   2029    
 202F           
 205F           

diff --git a/lib/unicore/lib/SpacePer.pl b/lib/unicore/lib/SpacePer.pl
index 18911cf..cc12e07 100644 (file)
--- a/lib/unicore/lib/SpacePer.pl
+++ b/lib/unicore/lib/SpacePer.pl
@@ -16,7 +16,7 @@ return <<'END';
 00A0           
 1680           
 180E           
-2000   200B    
+2000   200A    
 2028   2029    
 202F           
 205F           

diff --git a/lib/unicore/lib/Word.pl b/lib/unicore/lib/Word.pl
index 0bf2ea4..20936b1 100644 (file)
--- a/lib/unicore/lib/Word.pl
+++ b/lib/unicore/lib/Word.pl
@@ -312,6 +312,8 @@ return <<'END';
 1FE0   1FEC    
 1FF2   1FF4    
 1FF6   1FFC    
+203F   2040    
+2054           
 2070   2071    
 2074   2079    
 207F   2089    
@@ -340,8 +342,7 @@ return <<'END';
 3041   3096    
 3099   309A    
 309D   309F    
-30A1   30FA    
-30FC   30FF    
+30A1   30FF    
 3105   312C    
 3131   318E    
 3192   3195    
@@ -372,12 +373,15 @@ FD92      FDC7
 FDF0   FDFB    
 FE00   FE0F    
 FE20   FE23    
+FE33   FE34    
+FE4D   FE4F    
 FE70   FE74    
 FE76   FEFC    
 FF10   FF19    
 FF21   FF3A    
+FF3F           
 FF41   FF5A    
-FF66   FFBE    
+FF65   FFBE    
 FFC2   FFC7    
 FFCA   FFCF    
 FFD2   FFD7    

diff --git a/lib/unicore/mktables b/lib/unicore/mktables
index 083bc58..5fdac52 100644 (file)
--- a/lib/unicore/mktables
+++ b/lib/unicore/mktables
@@ -627,32 +627,35 @@ sub UnicodeData_Txt()
         ($General{$name} ||= Table->New)->$op($code, $name);
 
         # 005F: SPACING UNDERSCORE
-        $Cat{Word}->$op($code)  if $cat =~ /^[LMN]/ || $code == 0x005F;
+        $Cat{Word}->$op($code)  if $cat =~ /^[LMN]|Pc/;
         $Cat{Alnum}->$op($code) if $cat =~ /^[LM]|Nd/;
         $Cat{Alpha}->$op($code) if $cat =~ /^[LM]/;
 
-
-
-        $Cat{Space}->$op($code) if $cat  =~ /^Z/
+       my $isspace = 
+           ($cat =~ /Zs|Zl|Zp/ &&
+            $code != 0x200B) # 200B is ZWSP which is for line break control
+            # and therefore it is not part of "space" even while it is "Zs".
                                 || $code == 0x0009  # 0009: HORIZONTAL TAB
                                 || $code == 0x000A  # 000A: LINE FEED
                                 || $code == 0x000B  # 000B: VERTICAL TAB
                                 || $code == 0x000C  # 000C: FORM FEED
                                 || $code == 0x000D  # 000D: CARRIAGE RETURN
-                                || $code == 0x0085; # 0085: NEL
+                                || $code == 0x0085  # 0085: NEL
+
+           ;
 
+        $Cat{Space}->$op($code) if $isspace;
 
-        $Cat{SpacePerl}->$op($code) if $cat =~ /^Z/
-                                    || $code == 0x0009 # 0009: HORIZONTAL TAB
-                                    || $code == 0x000A # 000A: LINE FEED
-                                    || $code == 0x000C # 000C: FORM FEED
-                                    || $code == 0x000D # 000D: CARRIAGE RETURN
-                                    || $code == 0x0085 # 0085: <NEXT LINE>
-                                    || $code == 0x2028 # 2028: LINE SEPARATOR
-                                    || $code == 0x2029;# 2029: PARAGRAPH SEP.
+        $Cat{SpacePerl}->$op($code) if $isspace
+                                      && $code != 0x000B; # Backward compat.
 
-        $Cat{Blank}->$op($code) if $cat eq "Zs"
-                                || $code == 0x0009; # 0009: HORIZONTAL TAB
+        $Cat{Blank}->$op($code) if $isspace
+                                && !($code == 0x000A ||
+                                    $code == 0x000B ||
+                                    $code == 0x000C ||
+                                    $code == 0x000D ||
+                                    $code == 0x0085 ||
+                                    $cat =~ /^Z[lp]/);
 
         $Cat{Digit}->$op($code) if $cat eq "Nd";
         $Cat{Upper}->$op($code) if $cat eq "Lu";
@@ -660,9 +663,9 @@ sub UnicodeData_Txt()
         $Cat{Title}->$op($code) if $cat eq "Lt";
         $Cat{ASCII}->$op($code) if $code <= 0x007F;
         $Cat{Cntrl}->$op($code) if $cat =~ /^C/;
-        $Cat{Graph}->$op($code) if $cat !~ /Zs|Cc|Cs|Cn/;
-        $Cat{Print}->$op($code) if $cat =~ /^[LMNPS]/
-                                || $cat eq "Zs";
+       my $isgraph = !$isspace && $cat !~ /Cc|Cs|Cn/;
+        $Cat{Graph}->$op($code) if $isgraph;
+        $Cat{Print}->$op($code) if $isgraph || $isspace;
         $Cat{Punct}->$op($code) if $cat =~ /^P/;
 
         $Cat{XDigit}->$op($code) if ($code >= 0x30 && $code <= 0x39)  ## 0..9