01-fyi-bad-defines.patch

[Akim Demaille]

Index: ChangeLog
from  Akim Demaille  <address@hidden>
        
        * src/symtab.h (SALIAS, SUNDEF): Rename as...
        (USER_NUMBER_ALIAS, USER_NUMBER_UNDEFINED): these.
        Adjust dependencies.
        * src/output.c (token_definitions_output): Be sure not to output a
        `#define 'a'' when fed with `%token 'a' "a"'.
        * tests/regression.at (Token definitions): New.
        
Index: NEWS
--- NEWS Thu, 02 May 2002 19:30:29 +0200 akim
+++ NEWS Sat, 04 May 2002 10:43:39 +0200 akim
@@ -64,6 +64,9 @@
 * New tranlations
   Croatian, thanks to Denis Lackovic.
 
+* Token definitions
+  When fed with `%token 'a' "A"', Bison used to output `#define 'a' 65'.
+
 
 Changes in version 1.35, 2002-03-25:
 
Index: TODO
--- TODO Fri, 03 May 2002 23:39:21 +0200 akim
+++ TODO Sat, 04 May 2002 10:42:45 +0200 akim
@@ -276,38 +276,6 @@ grammar.y:1234: foo: bar(0x123456) baz(0
 Rewrite the reader in Flex/Bison.  There will be delicate parts, in
 particular, expect the scanner to be hard to write.
 
-* Problems with aliases
-From: "Baum, Nathan I" <address@hidden>
-Subject: Token Alias Bug
-To: "'address@hidden'" <address@hidden>
-
-I've noticed a bug in bison. Sadly, our eternally wise sysadmins won't let
-us use CVS, so I can't find out if it's been fixed already...
-
-Basically, I made a program (in flex) that went through a .y file looking
-for "..."-tokens, and then outputed a %token
-line for it. For single-character ""-tokens, I reasoned, I could just use
-[%token 'A' "A"]. However, this causes Bison to output a [#define 'A' 65],
-which cppp chokes on, not unreasonably. (And even if cppp didn't choke, I
-obviously wouldn't want (char)'A' to be replaced with (int)65 throughout my
-code.
-
-Bison normally forgoes outputing a #define for a character token. However,
-it always outputs an aliased token -- even if the token is an alias for a
-character token. We don't want that. The problem is in /output.c/, as I
-recall. When it outputs the token definitions, it checks for a character
-token, and then checks for an alias token. If the character token check is
-placed after the alias check, then it works correctly.
-
-Alias tokens seem to be something of a kludge. What about an [%alias "..."]
-command...
-
-       %alias T_IF "IF"
-
-Hmm. I can't help thinking... What about a --generate-lex option that
-creates an .l file for the alias tokens used... (Or an option to make a
-gperf file, etc...)
-
 * Presentation of the report file
 From: "Baum, Nathan I" <address@hidden>
 Subject: Token Alias Bug
Index: src/lex.c
--- src/lex.c Wed, 10 Apr 2002 23:20:06 +0200 akim
+++ src/lex.c Sat, 04 May 2002 10:29:09 +0200 akim
@@ -368,7 +368,7 @@
          {
            symval->number = ntokens++;
            symval->class = token_sym;
-           if (symval->user_token_number == SUNDEF)
+           if (symval->user_token_number == USER_NUMBER_UNDEFINED)
              symval->user_token_number = code;
          }
        return tok_identifier;
Index: src/output.c
--- src/output.c Fri, 03 May 2002 07:58:20 +0200 akim
+++ src/output.c Sat, 04 May 2002 10:36:07 +0200 akim
@@ -582,21 +582,25 @@
       symbol_t *symbol = symbols[i];
       int number = symbol->user_token_number;
 
-      if (number == SALIAS)
-       continue;
+      /* At this stage, if there are literal aliases, they are part of
+        SYMBOLS, so we should not find symbols which are the aliases
+        here.  */
+      assert (number != USER_NUMBER_ALIAS);
+
       /* Skip error token.  */
       if (symbol == errtoken)
        continue;
-      if (symbol->tag[0] == '\'')
-       continue;               /* skip literal character */
-      if (symbol->tag[0] == '\"')
-       {
-         /* use literal string only if given a symbol with an alias */
-         if (symbol->alias)
-           symbol = symbol->alias;
-         else
-           continue;
-       }
+
+      /* If this string has an alias, then it is necessarily the alias
+        which is to be output.  */
+      if (symbol->alias)
+       symbol = symbol->alias;
+
+      /* Don't output literal chars or strings (when defined only as a
+        string).  Note that must be done after the alias resolution:
+        think about `%token 'f' "f"'.  */
+      if (symbol->tag[0] == '\'' || symbol->tag[0] == '\"')
+       continue;
 
       /* Don't #define nonliteral tokens whose names contain periods
         or '$' (as does the default value of the EOF token).  */
@@ -605,6 +609,7 @@
 
       fprintf (out, "%s[[[%s]], [%d]]",
               first ? "" : ",\n", symbol->tag, number);
+
       first = 0;
     }
 }
Index: src/reader.c
--- src/reader.c Thu, 02 May 2002 19:30:29 +0200 akim
+++ src/reader.c Sat, 04 May 2002 10:41:47 +0200 akim
@@ -126,7 +126,7 @@
       symval->class = token_sym;
       symval->type_name = typename;
       symval->user_token_number = symbol->user_token_number;
-      symbol->user_token_number = SALIAS;
+      symbol->user_token_number = USER_NUMBER_ALIAS;
       symval->alias = symbol;
       symbol->alias = symval;
       /* symbol and symval combined are only one symbol */
@@ -149,7 +149,7 @@
 symbol_check_alias_consistence (symbol_t *this)
 {
   /* Check only those who _are_ the aliases. */
-  if (this->alias && this->user_token_number == SALIAS)
+  if (this->alias && this->user_token_number == USER_NUMBER_ALIAS)
     {
       if (this->prec != this->alias->prec)
        {
@@ -204,8 +204,8 @@
              this->number = this->alias->number;
            }
        }
-      /* Do not do processing below for SALIASs.  */
-      if (this->user_token_number == SALIAS)
+      /* Do not do processing below for USER_NUMBER_ALIASs.  */
+      if (this->user_token_number == USER_NUMBER_ALIAS)
        return TRUE;
     }
   else /* this->class == token_sym */
@@ -229,7 +229,7 @@
 {
   /* Non-terminal? */
   if (this->class == token_sym
-      && this->user_token_number != SALIAS)
+      && this->user_token_number != USER_NUMBER_ALIAS)
     {
       /* A token which translation has already been set? */
       if (token_translations[this->user_token_number] != undeftoken->number)
@@ -998,10 +998,10 @@
 |                                                                    |
 | Two symbols are entered in the table, one for the token symbol and |
 | one for the literal.  Both are given the <type>, if any, from the  |
-| declaration.  The ->user_token_number of the first is SALIAS and   |
-| the ->user_token_number of the second is set to the number, if     |
-| any, from the declaration.  The two symbols are linked via         |
-| pointers in their ->alias fields.                                  |
+| declaration.  The ->user_token_number of the first is              |
+| USER_NUMBER_ALIAS and the ->user_token_number of the second is set |
+| to the number, if any, from the declaration.  The two symbols are  |
+| linked via pointers in their ->alias fields.                       |
 |                                                                    |
 | During OUTPUT_DEFINES_TABLE, the symbol is reported thereafter,    |
 | only the literal string is retained it is the literal string that  |
@@ -1014,7 +1014,7 @@
   token_t token;
   symbol_t *symbol;
   char *typename = 0;
-  int usrtoknum = SUNDEF;
+  int usrtoknum = USER_NUMBER_UNDEFINED;
 
   token = lex ();              /* fetch typename or first token */
   if (token == tok_typename)
@@ -1035,7 +1035,7 @@
     }
   symval->class = token_sym;
   symval->type_name = typename;
-  symval->user_token_number = SALIAS;
+  symval->user_token_number = USER_NUMBER_ALIAS;
   symbol = symval;
 
   token = lex ();              /* get number or literal string */
@@ -1666,7 +1666,7 @@
   for (i = 0; i < ntokens; ++i)
     {
       symbol_t *this = symbols[i];
-      if (this->user_token_number != SUNDEF)
+      if (this->user_token_number != USER_NUMBER_UNDEFINED)
        {
          if (this->user_token_number > max_user_token_number)
            max_user_token_number = this->user_token_number;
@@ -1676,7 +1676,8 @@
     }
 
   /* If 256 is not used, assign it to error, to follow POSIX.  */
-  if (num_256_available_p && errtoken->user_token_number == SUNDEF)
+  if (num_256_available_p
+      && errtoken->user_token_number == USER_NUMBER_UNDEFINED)
     errtoken->user_token_number = 256;
 
   /* Set the missing user numbers. */
@@ -1686,7 +1687,7 @@
   for (i = 0; i < ntokens; ++i)
     {
       symbol_t *this = symbols[i];
-      if (this->user_token_number == SUNDEF)
+      if (this->user_token_number == USER_NUMBER_UNDEFINED)
        this->user_token_number = ++max_user_token_number;
       if (this->user_token_number > max_user_token_number)
        max_user_token_number = this->user_token_number;
Index: src/symtab.c
--- src/symtab.c Wed, 10 Apr 2002 23:20:06 +0200 akim
+++ src/symtab.c Sat, 04 May 2002 10:29:43 +0200 akim
@@ -38,7 +38,7 @@
   res->number = NUMBER_UNDEFINED;
   res->prec = 0;
   res->assoc = right_assoc;
-  res->user_token_number = SUNDEF;
+  res->user_token_number = USER_NUMBER_UNDEFINED;
   res->alias = NULL;
   res->class = unknown_sym;
 
Index: src/symtab.h
--- src/symtab.h Wed, 10 Apr 2002 23:20:06 +0200 akim
+++ src/symtab.h Sat, 04 May 2002 10:23:37 +0200 akim
@@ -41,14 +41,12 @@
   nterm_sym            /* non-terminal */
 } symbol_class;
 
+
 /* Internal token numbers. */
 typedef short token_number_t;
 
-#define SUNDEF  -1              /* For undefined user number. */
-#define SALIAS -9991           /* for symbol generated with an alias */
-
-#define NUMBER_UNDEFINED ((token_number_t) -1)
 
+typedef struct symbol_s symbol_t;
 struct symbol_s
 {
   /* The key, name of the symbol. */
@@ -61,14 +59,25 @@
   associativity assoc;
   int user_token_number;
 
-  /* Points to the other in the identifier-symbol pair for an
-     alias. Special value SALIAS in the identifier half of the
+  /* Points to the other in the identifier-symbol pair for an alias.
+     Special value USER_NUMBER_ALIAS in the identifier half of the
      identifier-symbol pair for an alias.  */
-  struct symbol_s *alias;
+  symbol_t *alias;
   symbol_class class;
 };
 
-typedef struct symbol_s symbol_t;
+/* Undefined user number.  */
+#define USER_NUMBER_UNDEFINED -1
+
+/* `symbol->user_token_number == USER_NUMBER_ALIAS' means this symbol
+   *has* (not is) a string literal alias.  For instance, `%token foo
+   "foo"' has `"foo"' numbered regularly, and `foo' numbered as
+   USER_NUMBER_ALIAS.  */
+#define USER_NUMBER_ALIAS -9991
+
+/* Undefined internal token number.  */
+#define NUMBER_UNDEFINED ((token_number_t) -1)
+
 
 /* A function to apply to each symbol. */
 typedef bool (*symbol_processor) PARAMS ((symbol_t *));
Index: tests/regression.at
--- tests/regression.at Thu, 18 Apr 2002 21:09:12 +0200 akim
+++ tests/regression.at Sat, 04 May 2002 10:39:40 +0200 akim
@@ -287,6 +287,32 @@ input.y:2: fatal error: no input grammar
 
 
 
+## ------------------- ##
+## Token definitions.  ##
+## ------------------- ##
+
+
+AT_SETUP([Token definitions])
+
+# Bison managed, when fed with `%token 'f' "f"' to #define 'f'!
+AT_DATA([input.y],
+[[%token "end of file"
+%token 'a' "a"
+%token "b" 'b'
+%token "c" c
+%token d "d"
+%token e 'e'
+%token 'f' e
+%%
+exp: "a";
+]])
+
+AT_CHECK([bison input.y -o input.c])
+AT_CHECK([$CC $CFLAGS $CPPFLAGS input.c -c])
+AT_CLEANUP
+
+
+
 ## -------------- ##
 ## Web2c Report.  ##
 ## -------------- ##