[euptex] improve \Uchar, \Ucharcat (#170)

texjporg · Jun 29, 2024 · 254f9cf · 254f9cf
1 parent 664e90c
commit 254f9cf
Show file tree

Hide file tree

Showing 6 changed files with 55 additions and 11 deletions.
diff --git a/source/texk/web2c/euptexdir/pdfutils.ch b/source/texk/web2c/euptexdir/pdfutils.ch
@@ -1201,9 +1201,9 @@ pdf_strcmp_code: print_int(cur_val);
 uniform_deviate_code:     print_int(unif_rand(cur_val));
 normal_deviate_code:      print_int(norm_rand);
 Uchar_convert_code:
-if is_char_ascii(cur_val) then print_char(cur_val) else print_kanji(cur_val);
+if check_echar_range(cur_val)>1 then print_char(cur_val) else print_kanji(cur_val);
 Ucharcat_convert_code:
-if cat<kanji then print_char(cur_val) else print_kanji(cur_val);
+if (cat<kanji)and(check_echar_range(cur_val)>1) then print_char(cur_val) else print_kanji(cur_val);
 @z
 
 @x e-pTeX: if primitives - leave room for \ifincsname

diff --git a/source/texk/web2c/euptexdir/tests/eup-ofm.log b/source/texk/web2c/euptexdir/tests/eup-ofm.log
@@ -12,5 +12,15 @@ entering extended mode
 l.43 \showbox0
 
 
+> \box0=
+\hbox(0.0+0.0)x10.49997, yoko direction
+.\displace 0.0
+.\uctt ^^c6
+.\uctt ^^a1
+
+! OK.
+l.54 \showbox0
+
+
 We have "2010. We do not have "2014. [1] )
-Output written on eup-ofm.dvi (1 page, 528 bytes).
+Output written on eup-ofm.dvi (1 page, 560 bytes).
diff --git a/source/texk/web2c/euptexdir/tests/eup-ofm.tex b/source/texk/web2c/euptexdir/tests/eup-ofm.tex
@@ -35,33 +35,45 @@
 \catcode"C6=11  \catcode"D8=11  \catcode"DF=11
 \catcode"E6=11  \catcode"F8=11
 
-\char"C6\char"D8\char"DF^^e6^^^^00f8
+\char"C6\Uchar"D8\Ucharcat"DF"B^^e6^^^^00f8
 
 ÆØßæø
 
 \setbox0\hbox{【Æ】}
 \showbox0
 
+\kcatcode"A1=15 % not_cjk: not cjk in legacy encoding
+\kcatcode"1A1=15
+
+\char"C6\Uchar"A1^^c6\Ucharcat"BF"B
+
+ơ% U+01A1 :: UTF-8 0xC6 0xA1
+ƿ% U+01BF :: UTF-8 0xC6 0xBF
+
+\setbox0\hbox{【ơ】}
+\showbox0
+
+\kcatcode"A1=14
 \kcatcode"123=14
 \catcode"123=11  \catcode"131=11  \catcode"152=11
 \catcode"153=11
 
-\char"123\char"131\char"152\char"153
+\char"123\char"131\Uchar"152^^^^0153
 
 ģıŒœ
 
 \kcatcode`Γ=14 % Γ: U+0393
 \catcode"393=11  \catcode"394=11  \catcode"398=11
 \catcode"39B=11  \catcode"39E=11
 
-\char"393\char"394\char`Θ\char"39B^^^^039e
+\char"393\Uchar"394\char`Θ\Ucharcat"39B"B^^^^039e
 
 ΓΔΘΛΞ
 
 \kcatcode"2010=14
 \catcode"2010=11  \catcode"2018=11  \catcode"2019=11
 
-\char"2010\char"2018^^^^2019
+\char"2010\Uchar"2018^^^^2019
 
 ‐‘’
 

diff --git a/source/texk/web2c/uptexdir/tests/up-ofm.log b/source/texk/web2c/uptexdir/tests/up-ofm.log
@@ -11,5 +11,15 @@
 l.43 \showbox0
 
 
+> \box0=
+\hbox(0.0+0.0)x10.49997, yoko direction
+.\displace 0.0
+.\uctt ^^c6
+.\uctt ^^a1
+
+! OK.
+l.54 \showbox0
+
+
 [1] )
-Output written on up-ofm.dvi (1 page, 528 bytes).
+Output written on up-ofm.dvi (1 page, 560 bytes).
diff --git a/source/texk/web2c/uptexdir/tests/up-ofm.tex b/source/texk/web2c/uptexdir/tests/up-ofm.tex
@@ -42,6 +42,18 @@
 \setbox0\hbox{【Æ】}
 \showbox0
 
+\kcatcode"A1=15 % not_cjk: not cjk in legacy encoding
+\kcatcode"1A1=15
+
+\char"C6\char"A1^^c6\char"BF
+
+ơ% U+01A1 :: UTF-8 0xC6 0xA1
+ƿ% U+01BF :: UTF-8 0xC6 0xBF
+
+\setbox0\hbox{【ơ】}
+\showbox0
+
+\kcatcode"A1=14
 \kcatcode"123=14
 \catcode"123=11  \catcode"131=11  \catcode"152=11
 \catcode"153=11

diff --git a/source/texk/web2c/uptexdir/uptex-m.ch b/source/texk/web2c/uptexdir/uptex-m.ch
@@ -404,7 +404,7 @@ left_brace,right_brace,math_shift,tab_mark,sup_mark,sub_mark,spacer,
 case m of
 kanji,kana,other_kchar,hangul,modifier: print_kanji(KANJI(c));
 left_brace,right_brace,math_shift,tab_mark,sup_mark,sub_mark,spacer,
-  letter,other_char: if (check_echar_range(c)=2) then print_kanji(KANJI(c)) else print(c);
+  letter,other_char: if (check_echar_range(c)=1) then print_kanji(KANJI(c)) else print(c);
 @z
 
 @x
@@ -2254,8 +2254,8 @@ end;
 function check_echar_range(@!c:integer):integer;
 begin
 if (c>127)and(c<max_latin_val)and(kcat_code(kcatcodekey(c))=latin_ucs)then
-  check_echar_range:=2
-else if (c>=0)and(c<256)then
   check_echar_range:=1
+else if (c>=0)and(c<256)then
+  check_echar_range:=2
 else check_echar_range:=0;
 @z