[uptex] revert latin_ucs_flag (#170)

texjporg · Jun 26, 2024 · 0fa0c33 · 0fa0c33
1 parent e2e161c
commit 0fa0c33
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 15 deletions.
diff --git a/source/texk/web2c/euptexdir/eptex.ech b/source/texk/web2c/euptexdir/eptex.ech
@@ -730,7 +730,7 @@ if j=1 then
          check_kcat_code(cur_tok,cur_chr) then
       begin if (cur_tok=not_cjk) then cur_tok:=other_kchar;
         if (cur_tok=latin_ucs) then cur_tok:=other_token;
-        cur_tok:=cur_chr+cur_tok*max_cjk_val+latin_ucs_flag;
+        cur_tok:=cur_chr+cur_tok*max_cjk_val;
         loc:=loc+multistrlen(ustringcast(buffer), limit+1,loc);
       end
     else begin cur_chr:=buffer[loc]; incr(loc);

diff --git a/source/texk/web2c/uptexdir/kanji.c b/source/texk/web2c/uptexdir/kanji.c
@@ -9,8 +9,8 @@
 #define IVS_CHAR_LIMIT  0x4400000
 #define CJK_CHAR_LIMIT  0x1000000
 #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_DELIM_NUM && (c0 & LATIN_UCS_FLAG) &&
-           (c0 - LATIN_UCS_FLAG) < LATIN_CHAR_LIMIT) { /* kcatcode latin_ucs */
-        return is_char_kanji(c0 - LATIN_UCS_FLAG);
+    if (c1>=CAT_LEFT_BRACE && c1<=CAT_DELIM_NUM &&
+            c0 < LATIN_CHAR_LIMIT) { /* kcatcode latin_ucs */
+        return is_char_kanji(c0);
     }
     else if (c1>=KCAT_KANJI && c1<=KCAT_MODIFIER) {
         return is_char_kanji(c0);
@@ -86,11 +86,8 @@ integer ktoken_to_chr(integer c)
 {
     if (c > KCAT_KANJI_IVS * CJK_CHAR_LIMIT)
         return (c - KCAT_KANJI_IVS * CJK_CHAR_LIMIT);
-    else {
-        if (XXHi(c)<=CAT_DELIM_NUM && (c & LATIN_UCS_FLAG))
-            c = c - LATIN_UCS_FLAG;
+    else
         return (c % CJK_CHAR_LIMIT);
-    }
 }
 
 /* Ref. http://www.unicode.org/Public/UNIDATA/Blocks.txt */

diff --git a/source/texk/web2c/uptexdir/uptex-m.ch b/source/texk/web2c/uptexdir/uptex-m.ch
@@ -169,7 +169,6 @@ 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
@@ -639,7 +638,7 @@ 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+latin_ucs_flag
+      cur_tok:=(cat_code(cur_chr)*max_cjk_val)+cur_chr
   else cur_tok:=(cur_cmd*max_char_val)+cur_chr
 @z
 
@@ -674,7 +673,7 @@ 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+latin_ucs_flag
+      cur_tok:=(cat_code(cur_chr)*max_cjk_val)+cur_chr
   else cur_tok:=(cur_cmd*max_char_val)+cur_chr
 @z
 
@@ -689,7 +688,7 @@ 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+latin_ucs_flag
+      cur_tok:=(cat_code(cur_chr)*max_cjk_val)+cur_chr
   else cur_tok:=(cur_cmd*max_char_val)+cur_chr
 @z
 
@@ -808,7 +807,7 @@ 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 begin cc:=other_char; t:=t+latin_ucs_flag; end;
+    if (cc=latin_ucs) then cc:=other_char;
     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;
@@ -2126,7 +2125,7 @@ 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+latin_ucs_flag
+      info(main_p):=KANJI(cur_chr)+cat_code(cur_chr)*max_cjk_val
     else { Does this case occur? }
       info(main_p):=KANJI(cur_chr)+kcat_code(kcatcodekey(KANJI(cur_chr)))*max_cjk_val;
 @z