Re: Dubious features

[Akim Demaille]

>>> "Akim" == Akim Demaille <address@hidden> writes:

 > But it gives an easy mean to implement this bug: /%union[^{]*/.  I
 > suppose that I can accumulate the various values if there are several
 > unions, after all GIGO is a sane principle :)

Here is my proposal.

2006-06-08  Akim Demaille  <address@hidden>

        * src/scan-gram.l: No longer "parse" things after `%union' until
        `{'.  Rather, implement the simpler behavior of Solaris Yacc:
        match a single token from '%union[^{]*'.
        (SC_PRE_CODE, token_type): Remove.
        * src/parse-gram.y: Adjust.

Index: src/parse-gram.y
===================================================================
RCS file: /cvsroot/bison/bison/src/parse-gram.y,v
retrieving revision 1.75
diff -u -u -r1.75 parse-gram.y
--- src/parse-gram.y 6 Jun 2006 16:40:06 -0000 1.75
+++ src/parse-gram.y 8 Jun 2006 12:12:39 -0000
@@ -113,7 +113,7 @@
 %token PERCENT_DESTRUCTOR  "%destructor"
 %token PERCENT_PRINTER     "%printer"

-%token PERCENT_UNION       "%union {...}"
+%token PERCENT_UNION       "%union"

 %token PERCENT_LEFT        "%left"
 %token PERCENT_RIGHT       "%right"
@@ -170,13 +170,13 @@

 %type <chars> STRING string_content
              "{...}"
-             "%union {...}"
+             "%union"
              PROLOGUE EPILOGUE
 %printer { fprintf (stderr, "\"%s\"", $$); }
              STRING string_content
 %printer { fprintf (stderr, "{\n%s\n}", $$); }
              "{...}"
-             "%union {...}"
+             "%union"
              PROLOGUE EPILOGUE
 %type <uniqstr> TYPE
 %printer { fprintf (stderr, "<%s>", $$); } TYPE
@@ -252,9 +252,9 @@
     {
       grammar_start_symbol_set ($2, @2);
     }
-| "%union {...}"
+| "%union" "{...}"
     {
-      char const *body = $1;
+      char const *body = $2;

       if (typed)
        {
@@ -266,7 +266,8 @@
        }

       typed = true;
-      muscle_code_grow ("stype", body, @1);
+      muscle_code_grow ("stype", $1,   @1);
+      muscle_code_grow ("stype", body, @2);
     }
 | "%destructor" "{...}" symbols.1
     {
Index: src/scan-gram.l
===================================================================
RCS file: /cvsroot/bison/bison/src/scan-gram.l,v
retrieving revision 1.90
diff -u -u -r1.90 scan-gram.l
--- src/scan-gram.l 7 Jun 2006 21:17:35 -0000 1.90
+++ src/scan-gram.l 8 Jun 2006 12:12:39 -0000
@@ -88,9 +88,6 @@
  /* A identifier was just read in directives/rules.  Special state
     to capture the sequence `identifier :'. */
 %x SC_AFTER_IDENTIFIER
- /* A keyword that should be followed by some code was read (e.g.
-    %printer). */
-%x SC_PRE_CODE

  /* Three types of user code:
     - prologue (code between `%{' `%}' in the first section, before %%);
@@ -124,9 +121,6 @@
   /* Parent context state, when applicable.  */
   int context_state IF_LINT (= 0);

-  /* Token type to return, when applicable.  */
-  int token_type IF_LINT (= 0);
-
   /* Location of most recent identifier, when applicable.  */
   location id_loc IF_LINT (= empty_location);

@@ -145,7 +139,7 @@
   | Scanning white space.  |
   `-----------------------*/

-<INITIAL,SC_AFTER_IDENTIFIER,SC_PRE_CODE>
+<INITIAL,SC_AFTER_IDENTIFIER>
 {
   /* Comments and white space.  */
   ","         warn_at (*loc, _("stray `,' treated as white space"));
@@ -175,17 +169,17 @@
   "%default"[-_]"prec"    return PERCENT_DEFAULT_PREC;
   "%define"               return PERCENT_DEFINE;
   "%defines"              return PERCENT_DEFINES;
-  "%destructor"                  /* FIXME: Remove once %union handled 
differently.  */ token_type = BRACED_CODE; return PERCENT_DESTRUCTOR;
+  "%destructor"                  return PERCENT_DESTRUCTOR;
   "%dprec"               return PERCENT_DPREC;
   "%error"[-_]"verbose"   return PERCENT_ERROR_VERBOSE;
   "%expect"               return PERCENT_EXPECT;
   "%expect"[-_]"rr"      return PERCENT_EXPECT_RR;
   "%file-prefix"          return PERCENT_FILE_PREFIX;
   "%fixed"[-_]"output"[-_]"files"   return PERCENT_YACC;
-  "%initial-action"       /* FIXME: Remove once %union handled differently.  
*/ token_type = BRACED_CODE; return PERCENT_INITIAL_ACTION;
+  "%initial-action"       return PERCENT_INITIAL_ACTION;
   "%glr-parser"           return PERCENT_GLR_PARSER;
   "%left"                 return PERCENT_LEFT;
-  "%lex-param"           /* FIXME: Remove once %union handled differently.  */ 
token_type = BRACED_CODE; return PERCENT_LEX_PARAM;
+  "%lex-param"           return PERCENT_LEX_PARAM;
   "%locations"            return PERCENT_LOCATIONS;
   "%merge"               return PERCENT_MERGE;
   "%name"[-_]"prefix"     return PERCENT_NAME_PREFIX;
@@ -195,9 +189,9 @@
   "%nondeterministic-parser"   return PERCENT_NONDETERMINISTIC_PARSER;
   "%nterm"                return PERCENT_NTERM;
   "%output"               return PERCENT_OUTPUT;
-  "%parse-param"         /* FIXME: Remove once %union handled differently.  */ 
token_type = BRACED_CODE; return PERCENT_PARSE_PARAM;
+  "%parse-param"         return PERCENT_PARSE_PARAM;
   "%prec"                 return PERCENT_PREC;
-  "%printer"              /* FIXME: Remove once %union handled differently.  
*/ token_type = BRACED_CODE; return PERCENT_PRINTER;
+  "%printer"              return PERCENT_PRINTER;
   "%pure"[-_]"parser"     return PERCENT_PURE_PARSER;
   "%require"              return PERCENT_REQUIRE;
   "%right"                return PERCENT_RIGHT;
@@ -207,7 +201,13 @@
   "%token"                return PERCENT_TOKEN;
   "%token"[-_]"table"     return PERCENT_TOKEN_TABLE;
   "%type"                 return PERCENT_TYPE;
-  "%union"               token_type = PERCENT_UNION; BEGIN SC_PRE_CODE;
+  "%union"[^{]*                  {
+                            obstack_grow (&obstack_for_string,
+                                         yytext + 6, yyleng - 6);
+                           STRING_FINISH;
+                           val->chars = last_string;
+                           return PERCENT_UNION;
+                          }
   "%verbose"              return PERCENT_VERBOSE;
   "%yacc"                 return PERCENT_YACC;

@@ -248,7 +248,6 @@
     if (current_rule && current_rule->action)
       grammar_midrule_action ();
     STRING_GROW;
-    token_type = BRACED_CODE;
     braces_level = 0;
     code_start = loc->start;
     BEGIN SC_BRACED_CODE;
@@ -501,45 +500,11 @@
 }


-  /*---------------------------------------------------------------.
-  | Scanning after %union etc., possibly followed by white space.  |
-  | For %union only, allow arbitrary C code to appear before the   |
-  | following brace, as an extension to POSIX.                    |
-  `---------------------------------------------------------------*/
-
-<SC_PRE_CODE>
-{
-  . {
-    bool valid = yytext[0] == '{' || token_type == PERCENT_UNION;
-    scanner_cursor.column -= mbsnwidth (yytext, yyleng, 0);
-    yyless (0);

-    if (valid)
-      {
-       braces_level = -1;
-       code_start = loc->start;
-       BEGIN SC_BRACED_CODE;
-      }
-    else
-      {
-       complain_at (*loc, _("missing `{' in %s"),
-                    token_name (token_type));
-       obstack_sgrow (&obstack_for_string, "{}");
-       STRING_FINISH;
-       val->chars = last_string;
-       BEGIN INITIAL;
-       return token_type;
-      }
-  }
-
-  <<EOF>>  unexpected_eof (scanner_cursor, "{}"); BEGIN INITIAL;
-}
-
-
-  /*---------------------------------------------------------------.
-  | Scanning some code in braces (%union and actions). The initial |
-  | "{" is already eaten.                                          |
-  `---------------------------------------------------------------*/
+  /*-----------------------------------------------------------.
+  | Scanning some code in braces (actions). The initial "{" is |
+  | already eaten.                                             |
+  `-----------------------------------------------------------*/

 <SC_BRACED_CODE>
 {
@@ -556,7 +521,7 @@
        val->chars = last_string;
        gram_last_braced_code_loc = *loc;
        BEGIN INITIAL;
-       return token_type;
+       return BRACED_CODE;
       }
   }