Stricter treatment of '$' in identifiers.

- Reject identifiers consisting of a single '$' sign.
They are rejected by some assemblers.

- Warn on identifiers containing '$' signs
This is an extension, not part of the ISO C standards.
The warning is off by default.

Author: xavierleroy

cparser/Diagnostics.ml

@@ -105,6 +105,7 @@ type warning_type =
   | Reduced_alignment
   | Non_linear_cond_expr
   | Invalid_UTF8
+  | Dollar_in_identifier
 
 (* List of all warnings with default status.
    "true" means the warning is active by default.
@@ -142,6 +143,7 @@ let all_warnings =
     (Reduced_alignment, false);
     (Non_linear_cond_expr, false);
     (Invalid_UTF8, true);
+    (Dollar_in_identifier, false)
   ]
 
 (* List of active warnings *)
@@ -185,6 +187,7 @@ let string_of_warning = function
   | Reduced_alignment -> "reduced-alignment"
   | Non_linear_cond_expr -> "non-linear-cond-expr"
   | Invalid_UTF8 -> "invalid-utf8"
+  | Dollar_in_identifier -> "dollar-in-identifier-extension"
 
 (* Activate the given warning *)
 let activate_warning w () =

cparser/Diagnostics.mli

@@ -58,6 +58,7 @@ type warning_type =
   | Reduced_alignment              (** alignment reduction *)
   | Non_linear_cond_expr           (** condition that cannot be linearized *)
   | Invalid_UTF8                   (** invalid UTF-8 encoding *)
+  | Dollar_in_identifier           (** '$' sign in identifier *)
 
 val warning  : (string * int) -> warning_type -> ('a, Format.formatter, unit, unit, unit, unit) format6 -> 'a
 (** [warning (f,c) w fmt arg1 ... argN] formats the arguments [arg1] to [argN] as warining according to

cparser/Lexer.mll

@@ -150,6 +150,21 @@ let warning lb kind fmt =
   Diagnostics.warning
       (lb.lex_curr_p.pos_fname,lb.lex_curr_p.pos_lnum) kind fmt
 
+(* Identifiers or keywords *)
+
+let ident_or_keyword lb id =
+  try
+    let f = Hashtbl.find lexicon id in
+    f (currentLoc lb)
+  with Not_found ->
+    if String.contains id '$' then begin
+      if id = "$" then
+        error lb "not supported: identifier consisting of a single '$' sign"
+      else
+        warning lb Diagnostics.Dollar_in_identifier "'$' in identifier";
+    end;
+    PRE_NAME id
+
 (* Simple character escapes *)
 
 let convert_escape = function
@@ -416,12 +431,9 @@ rule initial = parse
   | ";"                           { SEMICOLON(currentLoc lexbuf) }
   | ","                           { COMMA(currentLoc lexbuf) }
   | "."                           { DOT(currentLoc lexbuf) }
-  | identifier as id              {
-    if SSet.mem id !ignored_keywords then
-      initial lexbuf
-    else
-      try Hashtbl.find lexicon id (currentLoc lexbuf)
-      with Not_found -> PRE_NAME id }
+  | identifier as id              { if SSet.mem id !ignored_keywords
+                                    then initial lexbuf
+                                    else ident_or_keyword lexbuf id }
   | eof                           { EOF }
   | _ as c                        { fatal_error lexbuf "invalid symbol %C" c }