Implement the FIXME that ends an user action with a semicolon

[bison.git] / src / scan-code.l

diff --git a/src/scan-code.l b/src/scan-code.l
index e883e714dc4767cd35198efbb3e3a9264c3609e4..7a655fb875dec0f805bde448cb2b7c1ffe2adcbd 100644 (file)
--- a/src/scan-code.l
+++ b/src/scan-code.l
@@ -81,6 +81,19 @@ splice        (\\[ \f\t\v]*\n)*
   /* Nesting level of the current code in braces.  */
   int braces_level = 0;
 
+  /* Whether a semicolon is probably needed.
+     The heuristic is that a semicolon is not needed after `{', `}', `;',
+     or a C preprocessor directive, and that whitespaces and comments
+     do not affect this flag.
+     Note that `{' does not need a semicolon because of `{}'.
+     A semicolon may be needed before a cpp direcive, but don't bother.  */
+  bool need_semicolon = false;
+
+  /* Whether in a C preprocessor directive.  Don't use a start condition
+     for this because, at the end of strings and comments, we still need
+     to know whether we're in a directive.  */
+  bool in_cpp = false;
+
   /* This scanner is special: it is invoked only once, henceforth
      is expected to return only once.  This initialization is
      therefore done once per action to translate. */
@@ -135,10 +148,12 @@ splice     (\\[ \f\t\v]*\n)*
   "'" {
     STRING_GROW;
     BEGIN SC_CHARACTER;
+    need_semicolon = true;
   }
   "\"" {
     STRING_GROW;
     BEGIN SC_STRING;
+    need_semicolon = true;
   }
   "/"{splice}"*" {
     STRING_GROW;
@@ -154,46 +169,63 @@ splice     (\\[ \f\t\v]*\n)*
 {
   "$"("<"{tag}">")?(-?[0-9]+|"$")  {
     handle_action_dollar (self->rule, yytext, *loc);
+    need_semicolon = true;
   }
   "@"(-?[0-9]+|"$") {
     handle_action_at (self->rule, yytext, *loc);
+    need_semicolon = true;
   }
 
   "$"  {
     warn_at (*loc, _("stray `$'"));
     obstack_sgrow (&obstack_for_string, "$][");
+    need_semicolon = true;
   }
   "@"  {
     warn_at (*loc, _("stray `@'"));
     obstack_sgrow (&obstack_for_string, "@@");
+    need_semicolon = true;
+  }
+  "["  {
+    obstack_sgrow (&obstack_for_string, "@{");
+    need_semicolon = true;
+  }
+  "]"  {
+    obstack_sgrow (&obstack_for_string, "@}");
+    need_semicolon = true;
   }
 
-  "{"  STRING_GROW; ++braces_level;
+  ";"  STRING_GROW;                 need_semicolon = false;
+  "{"  STRING_GROW; ++braces_level; need_semicolon = false;
   "}"  {
     bool outer_brace = --braces_level == 0;
 
     /* As an undocumented Bison extension, append `;' before the last
        brace in braced code, so that the user code can omit trailing
        `;'.  But do not append `;' if emulating Yacc, since Yacc does
-       not append one.  Also, some output languages (like Java) do not
-       accept an extra semicolon, so don't append if the user specified
-       a skeleton or language.
-
-       FIXME: Bison should warn if a semicolon seems to be necessary
-       here, and should omit the semicolon if it seems unnecessary
-       (e.g., after ';', '{', or '}', each followed by comments or
-       white space).  Such a warning shouldn't depend on --yacc; it
-       should depend on a new --pedantic option, which would cause
-       Bison to warn if it detects an extension to POSIX.  --pedantic
-       should also diagnose other Bison extensions like %yacc.
-       Perhaps there should also be a GCC-style --pedantic-errors
-       option, so that such warnings are diagnosed as errors.  */
+       not append one.  */
     if (outer_brace && !yacc_flag && language_prio == default_prio
-        && skeleton_prio == default_prio)
-      obstack_1grow (&obstack_for_string, ';');
+        && skeleton_prio == default_prio && need_semicolon && ! in_cpp)
+      {
+       warn_at (*loc, _("a `;' might be needed at the end of action code"));
+       warn_at (*loc, _("future versions of Bison will not add the `;'"));
+       obstack_1grow (&obstack_for_string, ';');
+      }
 
     STRING_GROW;
+    need_semicolon = false;
   }
+
+  /* Preprocessing directives should only be recognized at the beginning
+     of lines, allowing whitespace including comments, but in C/C++,
+     `#' can only be the start of preprocessor directives or within
+     `#define' directives anyway, so don't bother with begin of line.  */
+  "#"       STRING_GROW; in_cpp = true;
+
+  {splice}  STRING_GROW;
+  [\n\r]    STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false; 
+  [ \t\f]   STRING_GROW;
+  .         STRING_GROW; need_semicolon = true;
 }
 
 <SC_SYMBOL_ACTION>