From 6f68543692a7588daa581d00c475715395036b15 Mon Sep 17 00:00:00 2001
From: Alexandre Julliard <julliard@winehq.org>
Date: Thu, 11 Mar 2021 12:46:04 +0100
Subject: [PATCH] make_unicode: Use codepage 949 to add default mappings to
 codepage 20949.

Signed-off-by: Alexandre Julliard <julliard@winehq.org>
---
 dlls/kernelbase/kernelbase.rgs |  86 ++++++++++++++++-----------------
 nls/c_20949.nls                | Bin 177698 -> 177698 bytes
 tools/make_unicode             |  28 +++++++++--
 3 files changed, 66 insertions(+), 48 deletions(-)

diff --git a/dlls/kernelbase/kernelbase.rgs b/dlls/kernelbase/kernelbase.rgs
index d35eb0a71a8..0d9dbd29c43 100644
--- a/dlls/kernelbase/kernelbase.rgs
+++ b/dlls/kernelbase/kernelbase.rgs
@@ -10,6 +10,49 @@ HKLM
                 {
                     Codepage
                     {
+                        val '10000' = s 'c_10000.nls'
+                        val '10001' = s 'c_10001.nls'
+                        val '10002' = s 'c_10002.nls'
+                        val '10003' = s 'c_10003.nls'
+                        val '10004' = s 'c_10004.nls'
+                        val '10005' = s 'c_10005.nls'
+                        val '10006' = s 'c_10006.nls'
+                        val '10007' = s 'c_10007.nls'
+                        val '10008' = s 'c_10008.nls'
+                        val '10010' = s 'c_10010.nls'
+                        val '10017' = s 'c_10017.nls'
+                        val '10021' = s 'c_10021.nls'
+                        val '10029' = s 'c_10029.nls'
+                        val '10079' = s 'c_10079.nls'
+                        val '10081' = s 'c_10081.nls'
+                        val '10082' = s 'c_10082.nls'
+                        val '1026' = s 'c_1026.nls'
+                        val '1250' = s 'c_1250.nls'
+                        val '1251' = s 'c_1251.nls'
+                        val '1252' = s 'c_1252.nls'
+                        val '1253' = s 'c_1253.nls'
+                        val '1254' = s 'c_1254.nls'
+                        val '1255' = s 'c_1255.nls'
+                        val '1256' = s 'c_1256.nls'
+                        val '1257' = s 'c_1257.nls'
+                        val '1258' = s 'c_1258.nls'
+                        val '1361' = s 'c_1361.nls'
+                        val '20127' = s 'c_20127.nls'
+                        val '20866' = s 'c_20866.nls'
+                        val '20932' = s 'c_20932.nls'
+                        val '20949' = s 'c_20949.nls'
+                        val '21866' = s 'c_21866.nls'
+                        val '28591' = s 'c_28591.nls'
+                        val '28592' = s 'c_28592.nls'
+                        val '28593' = s 'c_28593.nls'
+                        val '28594' = s 'c_28594.nls'
+                        val '28595' = s 'c_28595.nls'
+                        val '28596' = s 'c_28596.nls'
+                        val '28597' = s 'c_28597.nls'
+                        val '28598' = s 'c_28598.nls'
+                        val '28599' = s 'c_28599.nls'
+                        val '28603' = s 'c_28603.nls'
+                        val '28605' = s 'c_28605.nls'
                         val '37' = s 'c_037.nls'
                         val '437' = s 'c_437.nls'
                         val '500' = s 'c_500.nls'
@@ -34,49 +77,6 @@ HKLM
                         val '936' = s 'c_936.nls'
                         val '949' = s 'c_949.nls'
                         val '950' = s 'c_950.nls'
-                        val '1026' = s 'c_1026.nls'
-                        val '1250' = s 'c_1250.nls'
-                        val '1251' = s 'c_1251.nls'
-                        val '1252' = s 'c_1252.nls'
-                        val '1253' = s 'c_1253.nls'
-                        val '1254' = s 'c_1254.nls'
-                        val '1255' = s 'c_1255.nls'
-                        val '1256' = s 'c_1256.nls'
-                        val '1257' = s 'c_1257.nls'
-                        val '1258' = s 'c_1258.nls'
-                        val '1361' = s 'c_1361.nls'
-                        val '10000' = s 'c_10000.nls'
-                        val '10001' = s 'c_10001.nls'
-                        val '10002' = s 'c_10002.nls'
-                        val '10003' = s 'c_10003.nls'
-                        val '10004' = s 'c_10004.nls'
-                        val '10005' = s 'c_10005.nls'
-                        val '10006' = s 'c_10006.nls'
-                        val '10007' = s 'c_10007.nls'
-                        val '10008' = s 'c_10008.nls'
-                        val '10010' = s 'c_10010.nls'
-                        val '10017' = s 'c_10017.nls'
-                        val '10021' = s 'c_10021.nls'
-                        val '10029' = s 'c_10029.nls'
-                        val '10079' = s 'c_10079.nls'
-                        val '10081' = s 'c_10081.nls'
-                        val '10082' = s 'c_10082.nls'
-                        val '20127' = s 'c_20127.nls'
-                        val '20866' = s 'c_20866.nls'
-                        val '21866' = s 'c_21866.nls'
-                        val '28591' = s 'c_28591.nls'
-                        val '28592' = s 'c_28592.nls'
-                        val '28593' = s 'c_28593.nls'
-                        val '28594' = s 'c_28594.nls'
-                        val '28595' = s 'c_28595.nls'
-                        val '28596' = s 'c_28596.nls'
-                        val '28597' = s 'c_28597.nls'
-                        val '28598' = s 'c_28598.nls'
-                        val '28599' = s 'c_28599.nls'
-                        val '28603' = s 'c_28603.nls'
-                        val '28605' = s 'c_28605.nls'
-                        val '20932' = s 'c_20932.nls'
-                        val '20949' = s 'c_20949.nls'
                     }
                     Normalization
                     {
diff --git a/nls/c_20949.nls b/nls/c_20949.nls
index 1ee157af00355df9e5539da91cda687394d43399..2ff92aec922ae54b03d377b3e7ea99229fdada0d 100644
GIT binary patch
delta 655
zcmb8r&1(}u7zf~)G+5eyz-p{C!GqouJO;Po7hV(u4}u4yD0ry{@lX&%DCosg4tt*2
zshurJTHHtism&T~)nxM>tksy<SQ|H;G+Enxlh!0#+p|0{Gw<-@eTTtp5VOHq*5BtC
zad}Re;6o_g9x%F|gph*VIE)^HB{YSiRG<uPBN62>2Nuyhs)==R60^w!Cdh%SqJtJu
zgH~yi&fv3PAEM{Lg<Y3{4$~{(>K;Furq_Y)25?e`XahDY=qx=*yaVp;x({@MI0hc<
zk<qmH2<XOvQyjOOjpjZqV!P%&A&q$taQ!y{*RjT%P0%SFuixvp2X;<Nft`s%M^q-T
zS6L-9pH($qLkRs?pG}8K-Yxtm&g>&bnYSE<qnG}r)VzNwQN{IUgsv_W+AT2`T#m=1
zwIcIrn}yYku>7kwv$0lDK7UKjwStX$un?1r<)*8ww0%V{|5S!1`$^+bKG*4L`~R-Z
z@2$9$C~U;k6r0n8rqYgOR!aKYr?jMs*=BivD&<$M!GoSr6MPPzS%#s<GHe)nao+vA
z`x_jPD;&sgbx-SU4k(NRg`cVHm-Kc?E9F!@BbB>*iMdmH-S}G1yorPxo?8|8a=>%U
zZ8!rBDZ)<-L911d_=!y3k^p%}-jnXl0$C&<$Vc*t1bcOeC?rfGBuZkvI!+QKN%-_@
G!^OXAw`ynr

delta 300
zcmZ4VhHKFqt_^S4w?B|&EMwf9#L2R*-k!mpVap;Ad41vXh4u`KSAf_nfh>E5V+;2!
z{J!$X%B4X0ON*8)T(jug$~8dwpDVtt__bm^SQkhg2rdNcgYh8rA{4Rx3s)iW*REKz
z5~zv`svT&+<iZ`ovWr22aCl(hDkT2e$sc#9GA;s=nw#@>u-Q*GVA--+<3}3%<WD<H
zCind}+5GK)faGK~mOImxq!=BhdoVEy0oCM9e+1$=F)>PP{~*iwg0WpmmT|k1ER#g<
c^rjd_A&}q=&h08&nCuvV${VIX+REer0J86DZU6uP

diff --git a/tools/make_unicode b/tools/make_unicode
index 053adb92602..1e0dbe419e1 100755
--- a/tools/make_unicode
+++ b/tools/make_unicode
@@ -945,10 +945,9 @@ sub dump_eucjp_codepage()
 
 ################################################################
 # build Korean Wansung table from the KSX1001 file
-# FIXME: compare to Windows we should add extra mappings for
-# characters added from Unicode 1.1 to enhance the completeness.
-sub dump_krwansung_codepage()
+sub dump_krwansung_codepage(@)
 {
+    my @cp949 = @_;
     @cp2uni = ();
     @glyph2uni = ();
     @lead_bytes = ();
@@ -980,6 +979,24 @@ sub dump_krwansung_codepage()
     }
     close $INPUT;
 
+    # get some extra mappings from cp 949
+    my @defined_lb;
+    map { $defined_lb[$_] = 1; } @lead_bytes;
+    foreach my $i (0x0000 .. 0xffff)
+    {
+        next if ($i >= 0x1100 && $i <= 0x11ff);  # range not used in 20949
+        next unless defined $cp949[$i];
+        if ($cp949[$i] >= 0xff)
+        {
+            # only add chars for lead bytes that exist in 20949
+            my $hi = $cp949[$i] >> 8;
+            my $lo = $cp949[$i] & 0xff;
+            next unless $defined_lb[$hi];
+            next unless $lo >= 0xa1 && $lo <= 0xfe;
+        }
+        add_mapping( $cp949[$i], $i );
+    }
+
     output_codepage_file( 20949 );
 }
 
@@ -2341,6 +2358,8 @@ sub dump_msdata_codepage($)
     close $INPUT;
 
     output_codepage_file( $codepage );
+
+    if ($codepage == 949) { dump_krwansung_codepage( @uni2cp ); }
 }
 
 ################################################################
@@ -2758,7 +2777,7 @@ sub dump_registry_script($%)
                 $i == $#subkeys && $def ? " = s '$def'" : "", 4 * $indent, "";
             $indent++;
         }
-        foreach my $v (@vals) { printf OUTPUT "%*sval $v\n", 4 * $indent, ""; }
+        foreach my $v (sort @vals) { printf OUTPUT "%*sval $v\n", 4 * $indent, ""; }
         for (my $i = 0; $i < @subkeys; $i++) { printf OUTPUT "%*s}\n", 4 * --$indent, ""; }
     }
     while ($indent) { printf OUTPUT "%*s}\n", 4 * --$indent, ""; }
@@ -2813,7 +2832,6 @@ dump_norm_table( "nls/normidna.nls" );
 dump_sortkey_table( "nls/sortdefault.nls", "Windows 10 Sorting Weight Table.txt" );
 foreach my $file (@allfiles) { dump_msdata_codepage( $file ); }
 dump_eucjp_codepage();
-dump_krwansung_codepage();
 dump_registry_script( "dlls/kernelbase/kernelbase.rgs", %registry_keys );
 
 exit 0;