patch (1/2) for cpplib and Fortran [committed]

[Zack Weinberg <zack at wolery dot cumb dot org>]

This patch is most of the fix for cpplib's troubles with Fortran (see the
discussion on gcc).  It also corrects some problems that crop up when
processing assembly - e.g. arch/i386/kernel/head.S from the Linux kernel.

The rest of the fix touches files outside cpplib, and will come separately.

zw

2000-01-11 16:24 -0800  Zack Weinberg  <zack@wolery.cumb.org>

	* cpplib.h (struct cpp_options): Change lang_asm to char.
	Add lang_fortran.
	* cppinit.c (builtin_array): Take out __STDC_VERSION__, it's
	done in cpp_handle_option now.
	(initialize_builtins): Take out special case code used only by
	__STDC_VERSION__.
	(cpp_handle_option): Turn off trigraphs and trigraph warnings
	if -traditional.  Recognize -lang-fortran and set
	lang_fortran, also turn off cplusplus_comments.
	(print_help): Document -lang-fortran.
	* cpplib.c (handle_directive): Ignore `# 123 "file"' if
	lang_asm. Ignore all directives other than `# 123 "file"' if
	CPP_PREPROCESSED.
	(cpp_get_token): If -traditional, don't recognize directives
	unless the # is in column 1.
	(parse_string): If lang_fortran or lang_asm, silently
	terminate strings ('' or "") at end of line.
	Remove unnecessary braces.

===================================================================
Index: cpplib.h
--- cpplib.h	1999/10/29 04:31:14	1.44
+++ cpplib.h	2000/01/12 00:24:30
@@ -346,10 +346,18 @@ struct cpp_options {
 
   char objc;
 
-  /* Nonzero means this is an assembly file, and allow
-     unknown directives, which could be comments.  */
+  /* Nonzero means this is an assembly file, so ignore unrecognized
+     directives and the "# 33" form of #line, both of which are
+     probably comments.  Also, permit unbalanced ' strings (again,
+     likely to be in comments).  */
 
-  int lang_asm;
+  char lang_asm;
+
+  /* Nonzero means this is Fortran, and we don't know where the
+     comments are, so permit unbalanced ' strings.  Unlike lang_asm,
+     this does not ignore unrecognized directives.  */
+
+  char lang_fortran;
 
   /* Nonzero means turn NOTREACHED into #pragma NOTREACHED etc */
 
===================================================================
Index: cppinit.c
--- cppinit.c	2000/01/07 00:16:50	1.29
+++ cppinit.c	2000/01/12 00:24:30
@@ -496,8 +496,6 @@ cpp_cleanup (pfile)
    VALUE (if any).  FLAGS tweaks the behavior a little:
    DUMP		write debug info for this macro
    STDC		define only if not -traditional
-   C89		define only if -lang-c89
-   C9X		define only if -lang-c9x
    ULP		value is the global user_label_prefix (which can't be
 		put directly into the table).
  */
@@ -511,8 +509,6 @@ struct builtin
 };
 #define DUMP 0x01
 #define STDC 0x02
-#define C89  0x04
-#define C9X  0x08
 #define ULP  0x10
 
 static const struct builtin builtin_array[] =
@@ -536,8 +532,6 @@ static const struct builtin builtin_arra
   { "__PTRDIFF_TYPE__",		PTRDIFF_TYPE,	 T_CONST, DUMP },
 #endif
   { "__WCHAR_TYPE__",		WCHAR_TYPE,	 T_CONST, DUMP },
-  { "__STDC_VERSION__",		"199409L",	 T_CONST, DUMP|STDC|C89 },
-  { "__STDC_VERSION__",		"199909L",	 T_CONST, DUMP|STDC|C9X },
   { 0, 0, 0, 0 }
 };
 
@@ -555,10 +549,6 @@ initialize_builtins (pfile)
     {
       if ((b->flags & STDC) && CPP_TRADITIONAL (pfile))
 	continue;
-      if ((b->flags & C89) && CPP_OPTIONS (pfile)->c9x)
-	continue;
-      if ((b->flags & C9X) && !CPP_OPTIONS (pfile)->c9x)
-	continue;
 
       val = (b->flags & ULP) ? user_label_prefix : b->value;
       len = strlen (b->name);
@@ -571,8 +561,6 @@ initialize_builtins (pfile)
 }
 #undef DUMP
 #undef STDC
-#undef C89
-#undef C9X
 #undef ULP
 
 /* Another subroutine of cpp_start_read.  This one sets up to do
@@ -1312,6 +1300,8 @@ cpp_handle_option (pfile, argc, argv)
 	  {
 	    opts->traditional = 1;
 	    opts->cplusplus_comments = 0;
+	    opts->trigraphs = 0;
+	    opts->warn_trigraphs = 0;
 	  }
 	else if (!strcmp (argv[i], "-trigraphs"))
 	  opts->trigraphs = 1;
@@ -1339,6 +1329,8 @@ cpp_handle_option (pfile, argc, argv)
 	    opts->c9x = 0, opts->objc = 1;
 	if (! strcmp (argv[i], "-lang-asm"))
 	  opts->lang_asm = 1;
+	if (! strcmp (argv[i], "-lang-fortran"))
+	  opts->lang_fortran = 1, opts->cplusplus_comments = 0;
 	if (! strcmp (argv[i], "-lint"))
 	  opts->for_lint = 1;
 	if (! strcmp (argv[i], "-lang-chill"))
@@ -1722,6 +1714,7 @@ print_help ()
   -lang-objc                Assume that the input sources are in ObjectiveC\n\
   -lang-objc++              Assume that the input sources are in ObjectiveC++\n\
   -lang-asm                 Assume that the input sources are in assembler\n\
+  -lang-fortran		    Assume that the input sources are in Fortran\n\
   -lang-chill               Assume that the input sources are in Chill\n\
   -std=<std name>           Specify the conformance standard; one of:\n\
                             gnu89, gnu9x, c89, c9x, iso9899:1990,\n\
===================================================================
Index: cpplib.c
--- cpplib.c	1999/10/29 04:31:13	1.96
+++ cpplib.c	2000/01/12 00:24:31
@@ -527,11 +527,18 @@ handle_directive (pfile)
   cpp_skip_hspace (pfile);
 
   c = PEEKC ();
+  /* # followed by a number is equivalent to #line.  Do not recognize
+     this form in assembly language source files.  Complain about this
+     form if we're being pedantic, but not if this is regurgitated
+     input (preprocessed or fed back in by the C++ frontend).  */
   if (c >= '0' && c <= '9')
     {
-      /* Handle # followed by a line number.  Complain about using that
-         form if we're being pedantic, but not if this is regurgitated
-         input (preprocessed or fed back in by the C++ frontend).  */
+      if (CPP_OPTIONS (pfile)->lang_asm)
+	{
+	  skip_rest_of_line (pfile);
+	  return 1;
+	}
+
       if (CPP_PEDANTIC (pfile)
 	  && ! CPP_PREPROCESSED (pfile)
 	  && ! CPP_BUFFER (pfile)->manual_pop)
@@ -540,6 +547,11 @@ handle_directive (pfile)
       return 1;
     }
 
+  /* If we are rescanning preprocessed input, don't obey any directives
+     other than # nnn.  */
+  if (CPP_PREPROCESSED (pfile))
+    return 0;
+
   /* Now find the directive name.  */
   CPP_PUTC (pfile, '#');
   parse_name (pfile, GETC());
@@ -2388,6 +2400,12 @@ cpp_get_token (pfile)
 
 	  if (!pfile->only_seen_white)
 	    goto randomchar;
+	  /* -traditional directives are recognized only with the # in
+	     column 1.
+	     XXX Layering violation.  */
+	  if (CPP_TRADITIONAL (pfile)
+	      && CPP_BUFFER (pfile)->cur - CPP_BUFFER (pfile)->line_base != 1)
+	    goto randomchar;
 	  if (handle_directive (pfile))
 	    return CPP_DIRECTIVE;
 	  pfile->only_seen_white = 0;
@@ -2872,9 +2890,17 @@ parse_string (pfile, c)
 	case '\n':
 	  CPP_BUMP_LINE (pfile);
 	  pfile->lineno++;
+
+	  /* In Fortran and assembly language, silently terminate
+	     strings of either variety at end of line.  This is a
+	     kludge around not knowing where comments are in these
+	     languages.  */
+	  if (CPP_OPTIONS (pfile)->lang_fortran
+	      || CPP_OPTIONS (pfile)->lang_asm)
+	    return;
 	  /* Character constants may not extend over multiple lines.
-	     In ANSI, neither may strings.  We accept multiline strings
-	     as an extension.  */
+	     In Standard C, neither may strings.  We accept multiline
+	     strings as an extension.  */
 	  if (c == '\'')
 	    {
 	      cpp_error_with_line (pfile, start_line, start_column,
@@ -2882,10 +2908,8 @@ parse_string (pfile, c)
 	      return;
 	    }
 	  if (CPP_PEDANTIC (pfile) && pfile->multiline_string_line == 0)
-	    {
-	      cpp_pedwarn_with_line (pfile, start_line, start_column,
-				     "string constant runs past end of line");
-	    }
+	    cpp_pedwarn_with_line (pfile, start_line, start_column,
+				   "string constant runs past end of line");
 	  if (pfile->multiline_string_line == 0)
 	    pfile->multiline_string_line = start_line;
 	  break;