treatment of filenames (ticket 125)

 * string which represents filename does not expect to have
   "Japanese character" flag 0x100,
 * slow_print_filename and (rewrited) print_quoted both prints
   a string which represents filename.
    - slow_print_filename assumes that no kanji code conversion is required
    - print_quoted both prints assumes that code conversion should be performed

source/texk/ptexenc/ptexenc.c

@@ -281,6 +281,24 @@ int multistrlen ## SUFF(TYPE *s, int len, int pos) \
 DEFINE_MULTISTRLEN(,unsigned char);
 DEFINE_MULTISTRLEN(short,unsigned short);
 
+/* for outputting filename (*s) to the terminal */
+int multistrlenfilename(unsigned short *s, int len, int pos)
+{
+    s += pos; len -= pos;
+    if (terminal_enc == ENC_UTF8) {
+        int ret = UTF8Slengthshort(s, len);
+        if (ret < 0) return 1;
+        return ret;
+    }
+    if (len < 2) return 1;
+    if (terminal_enc == ENC_SJIS) {
+        if (isSJISkanji1(s[0]) && isSJISkanji2(s[1])) return 2;
+    } else { /* EUC */
+        if (isEUCkanji1(s[0])  && isEUCkanji2(s[1]))  return 2;
+    }
+    return 1;
+}
+
 /* with not so strict range check */
 int multibytelen (int first_byte)
 {

source/texk/ptexenc/ptexenc/ptexenc.h

@@ -65,6 +65,9 @@ extern PTENCDLL long fromBUFF(unsigned char *s, int len, int pos);
 extern PTENCDLL long fromBUFFshort(unsigned short *s, int len, int pos);
 extern PTENCDLL long toBUFF(long inter);
 
+/* for outputting filename (*s) to the terminal */
+extern PTENCDLL int multistrlenfilename(unsigned short *s, int len, int pos);
+
 /* internal (EUC/SJIS/UPTEX) from/to DVI (JIS/UCS) code conversion */
 extern PTENCDLL long toDVI (long kcode);
 extern PTENCDLL long fromDVI (long kcode);

source/texk/web2c/eptexdir/eptex.defines

@@ -12,6 +12,7 @@
 @define function iskanji1 ();
 @define function multistrlen ();
 @define function multistrlenshort ();
+@define function multistrlenfilename ();
 @define function fromBUFF ();
 @define function fromBUFFshort ();
 @define function toBUFF ();

source/texk/web2c/euptexdir/euptex.defines

@@ -17,6 +17,7 @@
 
 @define function multistrlen ();
 @define function multistrlenshort ();
+@define function multistrlenfilename ();
 @define function fromBUFF ();
 @define function fromBUFFshort ();
 @define function toBUFF ();

source/texk/web2c/lib/texmfmp.c

@@ -3040,7 +3040,7 @@ maketexstring(const_string s)
   }
 #else /* ! XeTeX */
   while (len-- > 0)
-    strpool[poolptr++] = *s++;
+    strpool[poolptr++] = 0xFF&(*s++);
 #endif /* ! XeTeX */
 
   return makestring();
@@ -3058,9 +3058,19 @@ makefullnamestring(void)
 strnumber
 getjobname(strnumber name)
 {
-    strnumber ret = name;
+    strnumber ret = name; int i, l, p;
     if (c_job_name != NULL)
       ret = maketexstring(c_job_name);
+#ifdef IS_pTeX
+    i = strstart[ret]; l = strstart[ret+1];
+    while (i<l)
+     {
+        p = multistrlenshort(strpool, l, i);
+        if (p>1) 
+             for (int j=i+p; i<j; i++) strpool[i] = (0xFF&strpool[i])+0x100;
+        else i++;
+     }
+#endif /* IS_pTeX */
     return ret;
 }
 #endif

source/texk/web2c/ptexdir/ptex-base.ch

@@ -376,6 +376,40 @@ else begin j:=str_start[s];
     end;
   end;
 end;
+
+procedure slow_print_filename(@!s:integer);
+  {prints string |s| which represents filename, without code conversion}
+var i,j,l:pool_pointer; p:integer;
+begin if (s>=str_ptr) or (s<256) then print(s)
+else begin i:=str_start[s]; l:=str_start[s+1];
+  while i<l do begin
+    p:=multistrlenfilename(str_pool, l, i);
+    if p<>1 then
+      begin for j:=i to i+p-1 do print_char(so(str_pool[j]));
+      i:=i+p; end
+    else begin print(so(str_pool[i])); incr(i); end;
+    end;
+  end;
+end;
+
+procedure print_quoted(@!s:integer);
+  {prints string |s| which represents filename, 
+  omitting quotes and with code covversion}
+var i,l:pool_pointer; j,p:integer;
+begin if s<>0 then begin
+  i:=str_start[s]; l:=str_start[s+1];
+  while i<l do begin
+    p:=multistrlenshort(str_pool, l, i);
+    if p<>1 then begin
+      for j:=i to i+p-1 do print_char(@"100+so(str_pool[j]));
+      i:=i+p; end
+    else begin
+      if so(str_pool[i])<>"""" then print(so(str_pool[i]));
+      incr(i); end;
+    end;
+  end;
+end;
+
 @z
 
 @x [5.61] l.1656 - pTeX:
@@ -2808,30 +2842,22 @@ else  begin str_room(1); append_char(c); {contribute |c| to the current string}
 end;
 @z
 
-@x [29.518] - print_quoted in pTeX: file name is always printed with code conversion
+@x [29.518] - print_quoted in pTeX is already defined
 @d print_quoted(#) == {print string |#|, omitting quotes}
 if #<>0 then
   for j:=str_start[#] to str_start[#+1]-1 do
     if so(str_pool[j])<>"""" then
       print(so(str_pool[j]))
 
 @y
-@d print_quoted(#) == {print string |#|, omitting quotes}
-if #<>0 then
-  for j:=str_start[#] to str_start[#+1]-1 do
-    if so(str_pool[j])<>"""" then
-      begin 
-      if so(str_pool[j])>=@"100 then print_char(so(str_pool[j]))
-      else print(so(str_pool[j])); end
-
 @z
 
 @x
 @d append_to_name(#)==begin c:=#; if not (c="""") then begin incr(k);
   if k<=file_name_size then name_of_file[k]:=xchr[c];
   end end
 @y
-@d append_to_name(#)==begin if (#)>=@"100 then c:=(#)-256 else c:=#;
+@d append_to_name(#)==begin if (#)>=@"100 then c:=(#)-@"100 else c:=#;
   { Since the type of |c| is |ASCII_code|, above if-statement might not be needed }
   if not (c="""") then begin incr(k);
   if k<=file_name_size then name_of_file[k]:=xchr[c];
@@ -2901,6 +2927,14 @@ else
   wlog(')');
 @z
 
+@x [29.???] pTeX - start_input
+print_char("("); incr(open_parens);
+slow_print(full_source_filename_stack[in_open]); update_terminal;
+@y
+print_char("("); incr(open_parens);
+slow_print_filename(full_source_filename_stack[in_open]); update_terminal;
+@z
+
 @x [30.560] l.10968 - pTeX:
 This is called BigEndian order.
 @!@^BigEndian order@>
@@ -7824,7 +7858,7 @@ are part of Japaense characters when the procedure is called.
 @<Basic printing...@>=
 procedure print_unread_buffer_with_ptenc(@!f, @!l: integer);
 { print |buffer[f..l-1]| with code conversion }
-var @!i,@!j,@!p: integer;
+var @!i,@!j: pool_pointer; @!p: integer;
 begin
   i:=f;
   while i<l do begin

source/texk/web2c/ptexdir/ptex.defines

@@ -8,6 +8,7 @@
 @define function iskanji1 ();
 @define function multistrlen ();
 @define function multistrlenshort ();
+@define function multistrlenfilename ();
 @define function fromBUFF ();
 @define function fromBUFFshort ();
 @define function toBUFF ();

source/texk/web2c/ptexdir/tests/filename_test.sh

@@ -0,0 +1,40 @@
+#!/bin/bash
+cat <<'EOF' > filename_日本語.tex
+% UTF-8 encoding
+\def\TEST#1{\ifx#1\relax\else\immediate\write16{[\meaning#1]}\expandafter\TEST\fi}
+\expandafter\TEST\jobname\relax
+
+\immediate\openout0=\jobname.txt
+\immediate\write0{日本語}
+\immediate\closeout0
+
+\openin0=\jobname.txt
+\ifeof0
+  \immediate\write16{File `\jobname.txt' not found.}
+\else
+  \read0 to \TEXT
+  \immediate\write16{TEXT=\TEXT}
+\fi
+\closein0
+
+\input \jobname.txt
+
+\end
+
+EOF
+
+ln -s filename_日本語.tex "filename_日本a.tex"
+eptex "\input {filename_日本a}"
+euptex "\input {filename_日本a}"
+ls "filename_日本a".*
+rm "filename_日本a".*
+ln -s filename_日本語.tex filename_日本ßſa.tex
+euptex filename_日本ßſa
+ls filename_日本ßſa.*
+rm filename_日本ßſa.*
+
+eptex "-jobname=漢ßſa" filename_日本語
+ls 漢*; rm 漢*
+euptex "-jobname=漢ßſa" filename_日本語
+ls 漢*; rm 漢*
+rm filename_日本語.tex

source/texk/web2c/uptexdir/uptex.defines

@@ -18,6 +18,7 @@
 
 @define function multistrlen ();
 @define function multistrlenshort ();
+@define function multistrlenfilename ();
 @define function multibytelen ();
 @define function fromBUFF ();
 @define function fromBUFFshort ();