[uptex] set latin_ucs_flag 0x800000 for latin_ucs node (#170)

source/texk/web2c/uptexdir/kanji.c

@@ -8,9 +8,9 @@
 #define CS_TOKEN_FLAG  0x1FFFFFFF
 #define IVS_CHAR_LIMIT  0x4400000
 #define CJK_CHAR_LIMIT  0x1000000
-#define UCS_CHAR_LIMIT   0x120000
+#define LATIN_CHAR_LIMIT   0x2E80
+#define LATIN_UCS_FLAG   0x800000
 #define CJK_TOKEN_FLAG   0xFFFFFF
-#define CAT_LEFT_BRACE  1
 #define CAT_DELIM_NUM  15
 #define KCAT_KANJI     16
 #define KCAT_MODIFIER  20
@@ -25,9 +25,9 @@ boolean check_kanji (integer c)
 
     c0 = c & CJK_TOKEN_FLAG;
     c1 = XXHi(c);
-    if (c1>=CAT_LEFT_BRACE && c1<=CAT_DELIM_NUM &&
-            c0 < UCS_CHAR_LIMIT) {
-        return is_char_kanji(c0);
+    if (c1<=CAT_DELIM_NUM && (c0 & LATIN_UCS_FLAG) &&
+           (c0 - LATIN_UCS_FLAG) < LATIN_CHAR_LIMIT) { /* kcatcode latin_ucs */
+        return is_char_kanji(c0 - LATIN_UCS_FLAG);
     }
     else if (c1>=KCAT_KANJI && c1<=KCAT_MODIFIER) {
         return is_char_kanji(c0);
@@ -45,7 +45,7 @@ boolean is_char_ascii(integer c)
 
 boolean is_char_kanji(integer c)
 {
-    if (is_internalUPTEX()) 
+    if (is_internalUPTEX())
         return ((c >= 0)&&(c<IVS_CHAR_LIMIT));
     else
         return iskanji1(Hi(c)) && iskanji2(Lo(c));
@@ -86,8 +86,11 @@ integer ktoken_to_chr(integer c)
 {
     if (c > KCAT_KANJI_IVS * CJK_CHAR_LIMIT)
         return (c - KCAT_KANJI_IVS * CJK_CHAR_LIMIT);
-    else
+    else {
+        if (XXHi(c)<=CAT_DELIM_NUM && (c & LATIN_UCS_FLAG))
+            c = c - LATIN_UCS_FLAG;
         return (c % CJK_CHAR_LIMIT);
+    }
 }
 
 /* Ref. http://www.unicode.org/Public/UNIDATA/Blocks.txt */

source/texk/web2c/uptexdir/uptex-m.ch

@@ -169,6 +169,7 @@ else if (kcode_pos=1)or((kcode_pos>=@'11)and(kcode_pos<=@'12))
 @d max_ivs_val=@"4400000 {to separate wchar with ivs and kcatcode}
 @d max_ucs_val=@"110000 {largest Unicode Scalar Value}
 @d max_latin_val=@"2E80
+@d latin_ucs_flag=@"800000
 @z
 
 @x
@@ -638,7 +639,7 @@ if cat=other_kchar then k:=k-multilenbuffchar(cur_chr)+1; {now |k| points to fir
     else
       cur_tok:=(cur_cmd*max_cjk_val)+cur_chr
   else if (cur_cmd=latin_ucs) then
-      cur_tok:=(cat_code(cur_chr)*max_cjk_val)+cur_chr
+      cur_tok:=(cat_code(cur_chr)*max_cjk_val)+cur_chr+latin_ucs_flag
   else cur_tok:=(cur_cmd*max_char_val)+cur_chr
 @z
 
@@ -673,7 +674,7 @@ if cat=other_kchar then k:=k-multilenbuffchar(cur_chr)+1; {now |k| points to fir
     else
       cur_tok:=(cur_cmd*max_cjk_val)+cur_chr
   else if (cur_cmd=latin_ucs) then
-      cur_tok:=(cat_code(cur_chr)*max_cjk_val)+cur_chr
+      cur_tok:=(cat_code(cur_chr)*max_cjk_val)+cur_chr+latin_ucs_flag
   else cur_tok:=(cur_cmd*max_char_val)+cur_chr
 @z
 
@@ -688,7 +689,7 @@ if cat=other_kchar then k:=k-multilenbuffchar(cur_chr)+1; {now |k| points to fir
     else
       cur_tok:=(cur_cmd*max_cjk_val)+cur_chr
   else if (cur_cmd=latin_ucs) then
-      cur_tok:=(cat_code(cur_chr)*max_cjk_val)+cur_chr
+      cur_tok:=(cat_code(cur_chr)*max_cjk_val)+cur_chr+latin_ucs_flag
   else cur_tok:=(cur_cmd*max_char_val)+cur_chr
 @z
 
@@ -807,7 +808,7 @@ while k<pool_ptr do
   begin  t:=so(str_pool[k]);
   if t>=@"180 then { there is no |wchar_token| whose code is 0--127. }
     begin t:=fromBUFFshort(str_pool, pool_ptr, k); cc:=kcat_code(kcatcodekey(t));
-    if (cc=latin_ucs) then cc:=other_char;
+    if (cc=latin_ucs) then begin cc:=other_char; t:=t+latin_ucs_flag; end;
     if (cc=not_cjk) then cc:=other_kchar;
     if (cc=kanji)and(t>=max_cjk_val) then cc:=kanji_ivs;
     t:=t+cc*max_cjk_val;
@@ -2121,7 +2122,7 @@ begin if is_char_node(link(p)) then
     else if cur_cmd=not_cjk then
       info(main_p):=KANJI(cur_chr)+other_kchar*max_cjk_val
     else if cur_cmd=latin_ucs then
-      info(main_p):=KANJI(cur_chr)+cat_code(cur_chr)*max_cjk_val
+      info(main_p):=KANJI(cur_chr)+cat_code(cur_chr)*max_cjk_val+latin_ucs_flag
     else { Does this case occur? }
       info(main_p):=KANJI(cur_chr)+kcat_code(kcatcodekey(KANJI(cur_chr)))*max_cjk_val;
 @z
@@ -2216,7 +2217,7 @@ end;
 
 function check_kcat_code(@!ct:integer;@!cx:integer):integer;
 begin
-if (((ct>=kanji)or((ct=latin_ucs)and(cx<max_ucs_val)))and(enable_cjk_token=0))or(enable_cjk_token=2)then
+if (((ct>=kanji)or((ct=latin_ucs)and(cx<max_latin_val)))and(enable_cjk_token=0))or(enable_cjk_token=2)then
   check_kcat_code:=1
 else check_kcat_code:=0;
 end;