Re: Add rules for ObjC files

[Klee Dienes <klee@mit.edu>]

[-- Attachment #1: Type: text/plain, Size: 975 bytes --]

I'm not sure if this is something you want to hassle with at the  
moment, or if you'd prefer to let it wait as a modification once a  
working version of the Objective-C patches are successfully committed.

But just as a heads-up, I should mention that in more recent versions  
of our GDB, we've folded in all of the Objective-C support directly  
into c-exp.y, and removed objc-exp.y as a file entirely.  Most of  
objc-exp.y was always just an out-of-date copy of c-exp.y, anyway;  
moving the (rather small) changes into c-exp.y directly not only  
allowed us to track changes to the C parser much more smoothly, it  
should also allow for much better Objective-C++ support.

You'll probably want to look at the most recent version of our GDB tree  
(http://www.opensource.apple.com/projects/darwin/6.0/source/other/gdb- 
20020918.tar.gz; no registration or anything similar required), and  
check out our changes to c-exp.y; I've attached them for your  
convenience.


[-- Attachment #2: objc.diffs --]
[-- Type: application/octet-stream, Size: 9129 bytes --]

diff -u -r1.1.1.8 c-exp.y
--- c-exp.y	2002/09/26 20:56:42	1.1.1.8
+++ c-exp.y	2002/10/04 08:31:39
@@ -46,9 +46,11 @@
 #include "parser-defs.h"
 #include "language.h"
 #include "c-lang.h"
+#include "objc-lang.h" /* For Obj-C language constructs. */
 #include "bfd.h" /* Required by objfiles.h.  */
 #include "symfile.h" /* Required by objfiles.h.  */
 #include "objfiles.h" /* For have_full_symbols and have_partial_symbols */
+#include "completer.h" /* For skip_quoted. */
 #include "charset.h"
 
 /* Flag indicating we're dealing with HP-compiled objects */ 
@@ -140,6 +142,7 @@
     struct block *bval;
     enum exp_opcode opcode;
     struct internalvar *ivar;
+    struct objc_class_str class;
 
     struct type **tvec;
     int *ivec;
@@ -173,8 +176,11 @@
    nonterminal "name", which matches either NAME or TYPENAME.  */
 
 %token <sval> STRING
-%token <ssym> NAME /* BLOCKNAME defined below to give it higher precedence. */
+%token <sval> NSSTRING		/* Obj-C Foundation "NSString" literal */ 
+%token <sval> SELECTOR		/* Obj-C "@selector" pseudo-operator   */ 
+%token <ssym> NAME		/* BLOCKNAME defined below to give it higher precedence. */
 %token <tsym> TYPENAME
+%token <class> CLASSNAME	/* Obj-C Class name */ 
 %type <sval> name
 %type <ssym> name_not_typename
 %type <tsym> typename
@@ -324,6 +330,71 @@
 			{ write_exp_elt_opcode (BINOP_SUBSCRIPT); }
 	;
 
+/* The rules below parse Obj-C message calls of the form:
+   '[' target selector {':' argument}* ']' */
+
+exp	: 	'[' TYPENAME
+			{
+			  CORE_ADDR class;
+
+			  class = lookup_objc_class (copy_name ($2.stoken));
+			  if (class == 0)
+			    error ("%s is not an ObjC Class", 
+				   copy_name ($2.stoken));
+			  write_exp_elt_opcode (OP_LONG);
+			  write_exp_elt_type (builtin_type_int);
+			  write_exp_elt_longcst ((LONGEST) class);
+			  write_exp_elt_opcode (OP_LONG);
+			  start_msglist();
+			}
+		msglist ']'
+			{ write_exp_elt_opcode (OP_MSGCALL);
+			  end_msglist();
+			  write_exp_elt_opcode (OP_MSGCALL); 
+			}
+	;
+
+exp	:	'[' CLASSNAME
+			{
+			  write_exp_elt_opcode (OP_LONG);
+			  write_exp_elt_type (builtin_type_int);
+			  write_exp_elt_longcst ((LONGEST) $2.class);
+			  write_exp_elt_opcode (OP_LONG);
+			  start_msglist();
+			}
+		msglist ']'
+			{ write_exp_elt_opcode (OP_MSGCALL);
+			  end_msglist();
+			  write_exp_elt_opcode (OP_MSGCALL); 
+			}
+	;
+
+exp	:	'[' exp
+			{ start_msglist(); }
+		msglist ']'
+			{ write_exp_elt_opcode (OP_MSGCALL);
+			  end_msglist();
+			  write_exp_elt_opcode (OP_MSGCALL); 
+			}
+	;
+
+msglist :	name
+			{ add_msglist(&$1, 0); }
+	|	msgarglist
+	;
+
+msgarglist :	msgarg
+	|	msgarglist msgarg
+	;
+
+msgarg	:	name ':' exp
+			{ add_msglist(&$1, 1); }
+	|	':' exp	/* Unnamed argument. */
+			{ add_msglist(0, 1);   }
+	|	',' exp	/* Variable number of arguments. */
+			{ add_msglist(0, 0);   }
+	;
+ 
 exp	:	exp '(' 
 			/* This is to save the value of arglist_len
 			   being accumulated by an outer function call.  */
@@ -499,6 +570,13 @@
 			/* Already written by write_dollar_variable. */
 	;
 
+exp	:	SELECTOR 
+			{
+			  write_exp_elt_opcode (OP_SELECTOR);
+			  write_exp_string ($1);
+			  write_exp_elt_opcode (OP_SELECTOR); }
+
+
 exp	:	SIZEOF '(' type ')'	%prec UNARY
 			{ write_exp_elt_opcode (OP_LONG);
 			  write_exp_elt_type (builtin_type_int);
@@ -531,6 +609,14 @@
 			  write_exp_elt_opcode (OP_ARRAY); }
 	;
 
+exp     :	NSSTRING	/* ObjC NextStep NSString constant
+				 * of the form '@' '"' string '"'.
+				 */
+			{ write_exp_elt_opcode (OP_NSSTRING);
+			  write_exp_string ($1);
+			  write_exp_elt_opcode (OP_NSSTRING); }
+	;
+
 /* C++.  */
 exp	:	THIS
 			{ write_exp_elt_opcode (OP_THIS);
@@ -804,6 +899,14 @@
 typebase  /* Implements (approximately): (type-qualifier)* type-specifier */
 	:	TYPENAME
 			{ $$ = $1.type; }
+	|	CLASSNAME
+			{
+			  if ($1.type == NULL)
+			    error ("No symbol \"%s\" in current context.", 
+				   copy_name($1.stoken));
+			  else
+			    $$ = $1.type;
+			}
 	|	INT_KEYWORD
 			{ $$ = builtin_type_int; }
 	|	LONG
@@ -951,7 +1054,8 @@
 name	:	NAME { $$ = $1.stoken; }
 	|	BLOCKNAME { $$ = $1.stoken; }
 	|	TYPENAME { $$ = $1.stoken; }
-	|	NAME_OR_INT  { $$ = $1.stoken; }
+	|	CLASSNAME { $$ = $1.stoken; }
+	|	NAME_OR_INT { $$ = $1.stoken; }
 	;
 
 name_not_typename :	NAME
@@ -1239,6 +1343,7 @@
   int c;
   int namelen;
   unsigned int i;
+  int tokchr;
   char *tokstart;
   char *tokptr;
   int tempbufindex;
@@ -1284,7 +1389,7 @@
 	return tokentab2[i].token;
       }
 
-  switch (c = *tokstart)
+  switch (tokchr = *tokstart)
     {
     case 0:
       /* If we were just scanning the result of a macro expansion,
@@ -1331,7 +1436,8 @@
       c = *lexptr++;
       if (c != '\'')
 	{
-	  namelen = skip_quoted (tokstart) - tokstart;
+	  namelen = skip_quoted (tokstart, get_gdb_completer_word_break_characters ())
+	    - tokstart;
 	  if (namelen > 2)
 	    {
 	      lexptr = tokstart + namelen;
@@ -1349,14 +1455,14 @@
     case '(':
       paren_depth++;
       lexptr++;
-      return c;
+      return '(';
 
     case ')':
       if (paren_depth == 0)
 	return 0;
       paren_depth--;
       lexptr++;
-      return c;
+      return ')';
 
     case ',':
       if (comma_terminates
@@ -1364,7 +1470,7 @@
           && ! scanning_macro_expansion ())
 	return 0;
       lexptr++;
-      return c;
+      return ',';
 
     case '.':
       /* Might be a floating point number.  */
@@ -1445,7 +1572,6 @@
     case '^':
     case '~':
     case '!':
-    case '@':
     case '<':
     case '>':
     case '[':
@@ -1457,8 +1583,42 @@
     case '}':
     symbol:
       lexptr++;
-      return c;
+      return tokchr;
 
+    case '@':
+      if (strncmp (tokstart, "@selector", 9) == 0)
+	{
+	  tokptr = strchr (tokstart, '(');
+	  if (tokptr == NULL)
+	    error ("Missing '(' in @selector(...)");
+	  tempbufindex = 0;
+	  /* skip the '(' */
+	  tokptr++;
+	  do {
+	    /* Grow the static temp buffer if necessary, including allocating
+	       the first one on demand. */
+	    if (tempbufindex + 1 >= tempbufsize)
+	      {
+		tempbuf = (char *) realloc (tempbuf, tempbufsize += 64);
+	      }
+	    tempbuf[tempbufindex++] = *tokptr++;
+	  } while ((*tokptr != ')') && (*tokptr != '\0'));
+	  if (*tokptr++ != ')')
+	    error ("Missing ')' in @selector(...)");
+	  tempbuf[tempbufindex] = '\0';
+	  yylval.sval.ptr = tempbuf;
+	  yylval.sval.length = tempbufindex;
+	  lexptr = tokptr;
+	  return SELECTOR;
+	}
+      if (tokstart[1] != '"')
+        {
+          lexptr++;
+          return tokchr;
+        }
+      /* Obj-C NSString constant: fall through and parse like STRING. */
+      tokstart++;
+
     case '"':
 
       /* Build the gdb internal form of the input string in tempbuf,
@@ -1522,13 +1682,13 @@
       yylval.sval.ptr = tempbuf;
       yylval.sval.length = tempbufindex;
       lexptr = tokptr;
-      return (STRING);
+      return (tokchr == '@' ? NSSTRING : STRING);
     }
 
-  if (!(c == '_' || c == '$'
-	|| (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')))
+  if (!(tokchr == '_' || tokchr == '$' || 
+       (tokchr >= 'a' && tokchr <= 'z') || (tokchr >= 'A' && tokchr <= 'Z')))
     /* We must have come across a bad character (e.g. ';').  */
-    error ("Invalid character '%c' in expression.", c);
+    error ("Invalid character '%c' in expression.", tokchr);
 
   /* It's a name.  See how long it is.  */
   namelen = 0;
@@ -1672,16 +1832,22 @@
     char *tmp = copy_name (yylval.sval);
     struct symbol *sym;
     int is_a_field_of_this = 0;
+    int *need_this;
     int hextype;
 
+    if (current_language->la_language == language_cplus
+	|| current_language->la_language == language_objc
+	|| current_language->la_language == language_objcplus)
+      need_this = &is_a_field_of_this;
+    else
+      need_this = (int *) NULL;
+
     sym = lookup_symbol (tmp, expression_context_block,
-			 VAR_NAMESPACE,
-			 current_language->la_language == language_cplus
-			 ? &is_a_field_of_this : (int *) NULL,
+			 VAR_NAMESPACE, need_this,
 			 (struct symtab **) NULL);
     /* Call lookup_symtab, not lookup_partial_symtab, in case there are
        no psymtabs (coff, xcoff, or some future change to blow away the
-       psymtabs once once symbols are read).  */
+       psymtabs once symbols are read).  */
     if (sym && SYMBOL_CLASS (sym) == LOC_BLOCK)
       {
 	yylval.ssym.sym = sym;
@@ -1794,6 +1960,23 @@
     if ((yylval.tsym.type = lookup_primitive_typename (tmp)) != 0)
       return TYPENAME;
 
+    /* See if it's an Obj-C classname. */
+    if (!sym && should_lookup_objc_class ())  
+      {
+	extern struct symbol *lookup_struct_typedef ();
+	sym = lookup_struct_typedef (tmp, expression_context_block, 1);
+	if (sym)
+	  {
+	    CORE_ADDR Class = lookup_objc_class (tmp);
+	    if (Class)
+	      {
+		yylval.class.class = Class;
+		yylval.class.type = SYMBOL_TYPE (sym);
+		return CLASSNAME;
+	      }
+	  }
+      }
+
     /* Input names that aren't symbols but ARE valid hex numbers,
        when the input radix permits them, can be names or numbers
        depending on the parse.  Note we support radixes > 16 here.  */

[-- Attachment #3: Type: text/plain, Size: 482 bytes --]



One downside to having the changes directly in c-exp.y is that I don't 
know of any good way to conditionalize them; though I am by no means a 
bison guru.

The following diffs are unrelated to Objective-C; I'm attaching them 
just for your reference (the ones in generic.diffs are likely to be 
removed in our sources soon as well; the ones in number-parsing.diffs 
remove a memory leak caused by calling parse_number on invalid, but are 
unrelated to the Objective-C changes).


[-- Attachment #4: number-parsing.diffs --]
[-- Type: application/octet-stream, Size: 3114 bytes --]

Index: c-exp.y
===================================================================
RCS file: /cvs/Darwin/src/live/cygnus/src/gdb/c-exp.y,v
retrieving revision 1.1.1.8
diff -u -r1.1.1.8 c-exp.y
--- c-exp.y	2002/09/26 20:56:42	1.1.1.8
+++ c-exp.y	2002/10/04 08:33:41
@@ -1384,19 +1490,23 @@
     case '9':
       {
 	/* It's a number.  */
-	int got_dot = 0, got_e = 0, toktype;
+	/* Initialize toktype to anything other than ERROR. */
+	int got_dot = 0, got_e = 0, toktype = FLOAT;
 	register char *p = tokstart;
 	int hex = input_radix > 10;
+	int local_radix = input_radix;
 
-	if (c == '0' && (p[1] == 'x' || p[1] == 'X'))
+	if (tokchr == '0' && (p[1] == 'x' || p[1] == 'X'))
 	  {
 	    p += 2;
 	    hex = 1;
+	    local_radix = 16;
 	  }
-	else if (c == '0' && (p[1]=='t' || p[1]=='T' || p[1]=='d' || p[1]=='D'))
+	else if (tokchr == '0' && (p[1]=='t' || p[1]=='T' || p[1]=='d' || p[1]=='D'))
 	  {
 	    p += 2;
 	    hex = 0;
+	    local_radix = 10;
 	  }
 
 	for (;; ++p)
@@ -1404,25 +1514,42 @@
 	    /* This test includes !hex because 'e' is a valid hex digit
 	       and thus does not indicate a floating point number when
 	       the radix is hex.  */
-	    if (!hex && !got_e && (*p == 'e' || *p == 'E'))
-	      got_dot = got_e = 1;
+
+	    if (!hex && (*p == 'e' || *p == 'E'))
+	      if (got_e)
+		toktype = ERROR;	/* only one 'e' in a float */
+	      else
+		got_e = 1;
 	    /* This test does not include !hex, because a '.' always indicates
 	       a decimal floating point number regardless of the radix.  */
-	    else if (!got_dot && *p == '.')
-	      got_dot = 1;
-	    else if (got_e && (p[-1] == 'e' || p[-1] == 'E')
-		     && (*p == '-' || *p == '+'))
+	    else if (*p == '.')
+	      if (got_dot)
+		toktype = ERROR;	/* only one '.' in a float */
+	      else
+		got_dot = 1;
+	    else if (got_e && (p[-1] == 'e' || p[-1] == 'E') &&
+		    (*p == '-' || *p == '+'))
 	      /* This is the sign of the exponent, not the end of the
 		 number.  */
 	      continue;
-	    /* We will take any letters or digits.  parse_number will
-	       complain if past the radix, or if L or U are not final.  */
-	    else if ((*p < '0' || *p > '9')
-		     && ((*p < 'a' || *p > 'z')
-				  && (*p < 'A' || *p > 'Z')))
-	      break;
+	    /* Always take decimal digits; parse_number handles radix error */
+	    else if (*p >= '0' && *p <= '9')
+	      continue;
+	    /* We will take letters only if hex is true, and only 
+	       up to what the input radix would permit.  FSF was content
+	       to rely on parse_number to validate; but it leaks. */
+	    else if (*p >= 'a' && *p <= 'z') {
+	      if (!hex || *p >= ('a' + local_radix - 10))
+		toktype = ERROR;
+	    }
+	    else if (*p >= 'A' && *p <= 'Z') {
+	      if (!hex || *p >= ('A' + local_radix - 10))
+		toktype = ERROR;
+	    }
+	    else break;
 	  }
-	toktype = parse_number (tokstart, p - tokstart, got_dot|got_e, &yylval);
+	if (toktype != ERROR)
+	  toktype = parse_number (tokstart, p - tokstart, got_dot|got_e, &yylval);
         if (toktype == ERROR)
 	  {
 	    char *err_copy = (char *) alloca (p - tokstart + 1);

[-- Attachment #5: Type: text/plain, Size: 2 bytes --]




[-- Attachment #6: generic.diffs --]
[-- Type: application/octet-stream, Size: 1216 bytes --]

diff -u -r1.1.1.8 c-exp.y
--- c-exp.y	2002/09/26 20:56:42	1.1.1.8
+++ c-exp.y	2002/10/04 08:31:39
@@ -659,7 +745,7 @@
 						 builtin_type_int);
 			    }
 			  else
-			    if (!have_full_symbols () && !have_partial_symbols ())
+			    if (!have_full_symbols () && !have_partial_symbols () && !have_minimal_symbols ())
 			      error ("No symbol table is loaded.  Use the \"file\" command.");
 			    else
 			      error ("No symbol \"%s\" in current context.", name);
@@ -714,7 +809,7 @@
 						     lookup_function_type (builtin_type_int),
 						     builtin_type_int);
 				}
-			      else if (!have_full_symbols () && !have_partial_symbols ())
+			      else if (!have_full_symbols () && !have_partial_symbols () && !have_minimal_symbols ())
 				error ("No symbol table is loaded.  Use the \"file\" command.");
 			      else
 				error ("No symbol \"%s\" in current context.",
@@ -1825,5 +2008,8 @@
   if (prev_lexptr)
     lexptr = prev_lexptr;
 
-  error ("A %s in expression, near `%s'.", (msg ? msg : "error"), lexptr);
+  if (*lexptr == '\0')
+    error("A %s near end of expression.",  (msg ? msg : "error"));
+  else
+    error ("A %s in expression, near `%s'.", (msg ? msg : "error"), lexptr);
 }

[-- Attachment #7: Type: text/plain, Size: 409 bytes --]




On Thursday, October 3, 2002, at 09:54 PM, Adam Fedor wrote:

> This patch adds rules for the ObjC files but does not enable them.
>
> 2002-10-03  Adam Fedor  <fedor@gnu.org>
>
> 	* Makefile.in (SFILES): Add objc-exp.y objc-lang.c.
> 	(objc_lang_h): Define.
> 	(YYFILES): Add objc-exp.tab.c.
> 	(local-maintainer-clean): Remove objc-exp.tab.c.
> 	(objc-exp.tab.c, objc-exp.tab.o, objc-lang.o): New target.