]> git.saurik.com Git - bison.git/blobdiff - data/lalr1.cc
Support i18n of the parse error messages.
[bison.git] / data / lalr1.cc
index 52aae209b88a3270f798ee134ce7c33f05b03c8d..bacdf37ce42c2743e6d5810677c4f8ce3261189c 100644 (file)
@@ -53,6 +53,12 @@ b4_variant_if([
 ]) # b4_variant_if
 
 
+# b4_lex_symbol_if([IF-YYLEX-RETURNS-A-COMPLETE-SYMBOL], [IF-NOT])
+# ----------------------------------------------------------------
+m4_define([b4_lex_symbol_if],
+[b4_percent_define_ifdef([[lex_symbol]], [$1], [$2])])
+
+
 # b4_assert_if([IF-ASSERTIONS-ARE-USED], [IF-NOT])
 # ------------------------------------------------
 m4_define([b4_assert_if],
@@ -113,6 +119,17 @@ m4_define([b4_symbol],
 [m4_indir([b4_symbol($1, $2)])])
 
 
+# b4_symbol_if(NUM, FIELD, IF-TRUE, IF-FALSE)
+# -------------------------------------------
+# If FIELD about symbol #NUM is 1 expand IF-TRUE, if is 0, expand IF-FALSE.
+# Otherwise an error.
+m4_define([b4_symbol_if],
+[m4_case(b4_symbol([$1], [$2]),
+         [1], [$3],
+         [0], [$4],
+         [m4_fatal([$0: field $2 of $1 is not a Boolean:] b4_symbol([$1], [$2]))])])
+
+
 # b4_symbol_actions(FILENAME, LINENO,
 #                   SYMBOL-TAG, SYMBOL-NUM,
 #                   SYMBOL-ACTION, SYMBOL-TYPENAME)
@@ -143,15 +160,97 @@ m4_define([b4_symbol_case_],
 # b4_type_action_(NUMS)
 # ---------------------
 # Run actions for the symbol NUMS that all have the same type-name.
+# Skip NUMS that have no type-name.
 m4_define([b4_type_action_],
+[b4_symbol_if([$1], [has_type_name],
 [m4_map([b4_symbol_case_], [$@])[]dnl
         b4_dollar_dollar([b4_symbol([$1], [number])],
                          [b4_symbol([$1], [tag])],
                          [b4_symbol([$1], [type_name])]);
        break;
 
+])])
+
+
+# b4_symbol_constructor_declaration_(SYMBOL-NUMBERS)
+# ----------------------------------------------------
+# Declare the overloaded version of make_symbol for the (common) type of
+# these SYMBOL-NUMBERS.  Use at class-level.
+m4_define([b4_symbol_constructor_declaration_],
+[    template <token_type>
+    static inline symbol_type
+    make_symbol (b4_symbol_if([$1], [has_type_name],
+                              [const b4_symbol([$1], [type_name])& v, ])dnl
+const location_type& l);
+
 ])
 
+# b4_symbol_constructor_declarations
+# ----------------------------------
+# Declare the overloaded versions of make_symbol for all the value types.
+# Use at class-level.
+m4_define([b4_symbol_constructor_declarations],
+[b4_variant_if([
+    // Declaration of make_symbol for each value type.
+m4_map([b4_symbol_constructor_declaration_], m4_defn([b4_type_names]))])])
+
+
+
+# b4_symbol_constructor_specialization_(SYMBOL-NUMBER)
+# ----------------------------------------------------
+# Declare the specialization of make_symbol for this each SYMBOL-NUMBER.
+# Specializations cannot be declared at class-level, this must be done
+# at namespace-level.
+m4_define([b4_symbol_constructor_specialization_],
+[b4_symbol_if([$1], [is_token], [b4_symbol_if([$1], [tag_is_id],
+[  template <>
+  inline
+  b4_parser_class_name::symbol_type
+  b4_parser_class_name::make_symbol <b4_parser_class_name::token::b4_symbol([$1], [tag])> (dnl
+b4_symbol_if([$1], [has_type_name],
+             [const b4_symbol([$1], [type_name])& v, ])dnl
+const b4_parser_class_name::location_type& l);
+])])])
+
+# b4_symbol_constructor_specializations
+# -------------------------------------
+# Declare specializations of make_symbol.
+m4_define([b4_symbol_constructor_specializations],
+[b4_variant_if([
+  // Specializations of make_symbol for each symbol type.
+m4_map([b4_symbol_constructor_specialization_],
+       m4_defn([b4_symbol_numbers]))])dnl
+])
+
+
+
+# b4_symbol_constructor_definition_(SYMBOL-NUMBER)
+# ------------------------------------------------
+# Define make_symbol for this SYMBOL-NUMBER.
+m4_define([b4_symbol_constructor_definition_],
+[b4_symbol_if([$1], [is_token], [b4_symbol_if([$1], [tag_is_id],
+[  template <>
+  b4_parser_class_name::symbol_type
+  b4_parser_class_name::make_symbol <b4_parser_class_name::token::b4_symbol([$1], [tag])> (dnl
+b4_symbol_if([$1], [has_type_name],
+             [const b4_symbol([$1], [type_name])& v, ])dnl
+const location_type& l)
+  {
+    return symbol_type (yytranslate_ (token::b4_symbol([$1], [tag])),dnl
+ b4_symbol_if([$1], [has_type_name], [v, ])l);
+  }
+
+])])])
+
+
+# b4_symbol_constructor_declarations
+# ----------------------------------
+# Define the overloaded versions of make_symbol for all the value types.
+m4_define([b4_symbol_constructor_definitions],
+[b4_variant_if(
+[  // Implementation of make_symbol for each symbol type.
+m4_map([b4_symbol_constructor_definition_], m4_defn([b4_symbol_numbers]))])])
+
 
 # b4_symbol_variant(YYTYPE, YYVAL, ACTION, [ARGS])
 # ------------------------------------------------
@@ -163,8 +262,8 @@ m4_define([b4_symbol_variant],
   switch ($1)
     {
 m4_map([b4_type_action_], m4_defn([b4_type_names]))[]dnl
-     default:
-       break;
+      default:
+        break;
     }
 m4_popdef([b4_dollar_dollar])dnl
 ])
@@ -249,11 +348,12 @@ dnl FIXME: This is wrong, we want computed header guards.
   {]b4_assert_if([
     /// Whether something is contained.
     bool built;
-
-    /// Initially uninitialized.
-    variant ()
-      : built(false)
-    {}])[
+])[
+    /// Empty construction.
+    inline
+    variant ()]b4_assert_if([
+      : built(false)])[
+    {}
 
     /// Instantiate a \a T in here.
     template <typename T>
@@ -275,6 +375,15 @@ dnl FIXME: This is wrong, we want computed header guards.
       return *new (buffer) T(t);
     }
 
+    /// Construct and fill.
+    template <typename T>
+    inline
+    variant (const T& t)]b4_assert_if([
+      : built(true)])[
+    {
+      new (buffer) T(t);
+    }
+
     /// Accessor to a built \a T.
     template <typename T>
     inline T&
@@ -497,7 +606,7 @@ m4_ifdef([b4_stype],
 #endif
 
     /// Convert a scanner token number \a t to a symbol number.
-    token_number_type yytranslate_ (int t);
+    static inline token_number_type yytranslate_ (int t);
 
     /// A complete symbol, with its type.
     template <typename Exact>
@@ -507,6 +616,7 @@ m4_ifdef([b4_stype],
       inline symbol_base_type ();
 
       /// Constructor.
+      inline symbol_base_type (const location_type& l);
       inline symbol_base_type (const semantic_type& v, const location_type& l);
 
       /// Return this with its exact type.
@@ -540,6 +650,7 @@ m4_ifdef([b4_stype],
     inline void yy_destroy_ (const char* yymsg,
                              symbol_base_type<Exact>& yysym) const;
 
+  public:
     /// Element of the stack: a state and its attributes.
     struct symbol_type : symbol_base_type<symbol_type>
     {
@@ -553,6 +664,9 @@ m4_ifdef([b4_stype],
       inline symbol_type (int t,
                           const semantic_type& v, const location_type& l);
 
+      inline symbol_type (int t,
+                          const location_type& l);
+
       /// The symbol type.
       int type;
 
@@ -560,6 +674,9 @@ m4_ifdef([b4_stype],
       inline int type_get_ () const;
     };
 
+]b4_symbol_constructor_declarations[
+
+  private:
     /// Element of the stack: a state and its attributes.
     struct stack_symbol_type : symbol_base_type<stack_symbol_type>
     {
@@ -605,20 +722,21 @@ m4_ifdef([b4_stype],
     inline void yypop_ (unsigned int n = 1);
 
     /* Constants.  */
-    static const int yyeof_;
-    /* LAST_ -- Last index in TABLE_.  */
-    static const int yylast_;
-    static const int yynnts_;
-    static const int yyempty_;
-    static const int yyfinal_;
-    static const int yyterror_;
-    static const int yyerrcode_;
-    static const int yyntokens_;
-    static const unsigned int yyuser_token_number_max_;
-    static const token_number_type yyundef_token_;
+    enum
+    {
+      yyeof_ = 0,
+      yylast_ = ]b4_last[,           //< Last index in yytable_.
+      yynnts_ = ]b4_nterms_number[,  //< Number of nonterminal symbols.
+      yyempty_ = -2,
+      yyfinal_ = ]b4_final_state_number[, //< Termination state number.
+      yyterror_ = 1,
+      yyerrcode_ = 256,
+      yyntokens_ = ]b4_tokens_number[,   //< Number of tokens.
+    };
+
 ]b4_parse_param_vars[
   };
-
+]b4_symbol_constructor_specializations[
 ]b4_namespace_close[
 
 ]b4_percent_define_flag_if([[global_tokens_and_yystype]],
@@ -703,12 +821,12 @@ b4_percent_code_get[]dnl
 
 #endif /* !YYDEBUG */
 
-#define yyerrok                (yyerrstatus_ = 0)
-#define yyclearin      (yychar = yyempty_)
+#define yyerrok         (yyerrstatus_ = 0)
+#define yyclearin       (yyempty = true)
 
-#define YYACCEPT       goto yyacceptlab
-#define YYABORT                goto yyabortlab
-#define YYERROR                goto yyerrorlab
+#define YYACCEPT        goto yyacceptlab
+#define YYABORT         goto yyabortlab
+#define YYERROR         goto yyerrorlab
 #define YYRECOVERING()  (!!yyerrstatus_)
 
 ]b4_namespace_open[
@@ -780,6 +898,13 @@ b4_percent_code_get[]dnl
   {
   }
 
+  template <typename Exact>
+  ]b4_parser_class_name[::symbol_base_type<Exact>::symbol_base_type (const location_type& l)
+    : value()
+    , location(l)
+  {
+  }
+
   template <typename Exact>
   ]b4_parser_class_name[::symbol_base_type<Exact>::symbol_base_type (const semantic_type& v, const location_type& l)
     : value(v)
@@ -815,6 +940,13 @@ b4_percent_code_get[]dnl
   {
   }
 
+  ]b4_parser_class_name[::symbol_type::symbol_type (int t,
+                           const location_type& l)
+    : super_type (l)
+    , type (t)
+  {
+  }
+
   ]b4_parser_class_name[::symbol_type::symbol_type (int t,
                            const semantic_type& v, const location_type& l)
     : super_type (v, l)
@@ -828,6 +960,8 @@ b4_percent_code_get[]dnl
     return type;
   }
 
+]b4_symbol_constructor_definitions[
+
   // stack_symbol_type.
   ]b4_parser_class_name[::stack_symbol_type::stack_symbol_type ()
     : super_type ()
@@ -883,8 +1017,8 @@ b4_percent_code_get[]dnl
         << yysym.location << ": ";
     switch (yytype)
       {
-]m4_map([b4_symbol_actions], m4_defn([b4_symbol_printers]))dnl
-[       default:
+]m4_map([b4_symbol_actions], m4_defn([b4_symbol_printers]))[
+       default:
          break;
       }
     yyo << ')';
@@ -952,8 +1086,8 @@ b4_percent_code_get[]dnl
   int
   ]b4_parser_class_name[::parse ()
   {
-    /// Coded type of the lookahead.
-    int yychar = yyempty_;
+    /// Whether yyla contains a lookahead.
+    bool yyempty = true;
 
     /* State.  */
     int yyn;
@@ -1014,27 +1148,18 @@ m4_popdef([b4_at_dollar])])dnl
       goto yydefault;
 
     /* Read a lookahead token.  */
-    if (yychar == yyempty_)
+    if (yyempty)
       {
-       YYCDEBUG << "Reading a token: ";
-       yychar = ]b4_c_function_call([yylex], [int],
+        YYCDEBUG << "Reading a token: ";
+]b4_lex_symbol_if(
+[        yyla = yylex();],
+[        yyla.type = yytranslate_ (b4_c_function_call([yylex], [int],
                                     [[YYSTYPE*], [&yyla.value]][]dnl
 b4_locations_if([, [[location*], [&yyla.location]]])dnl
-m4_ifdef([b4_lex_param], [, ]b4_lex_param))[;
-      }
-
-
-    /* Convert token to internal form.  */
-    if (yychar <= yyeof_)
-      {
-       yychar = yyla.type = yyeof_;
-       YYCDEBUG << "Now at end of input." << std::endl;
-      }
-    else
-      {
-       yyla.type = yytranslate_ (yychar);
-       YY_SYMBOL_PRINT ("Next token is", yyla);
+m4_ifdef([b4_lex_param], [, ]b4_lex_param)));])[
+        yyempty = false;
       }
+    YY_SYMBOL_PRINT ("Next token is", yyla);
 
     /* If the proper action on seeing token YYLA.TYPE is to reduce or
        to detect an error, take that action.  */
@@ -1053,7 +1178,7 @@ m4_ifdef([b4_lex_param], [, ]b4_lex_param))[;
       }
 
     /* Discard the token being shifted.  */
-    yychar = yyempty_;
+    yyempty = true;
 
     /* Count tokens shifted since error; after three, turn off error
        status.  */
@@ -1104,7 +1229,7 @@ m4_ifdef([b4_lex_param], [, ]b4_lex_param))[;
     YY_REDUCE_PRINT (yyn);
     switch (yyn)
       {
-       ]b4_user_actions[
+]b4_user_actions[
        default:
           break;
       }
@@ -1154,18 +1279,15 @@ m4_ifdef([b4_lex_param], [, ]b4_lex_param))[;
     if (yyerrstatus_ == 3)
       {
        /* If just tried and failed to reuse lookahead token after an
-        error, discard it.  */
-
-       if (yychar <= yyeof_)
-         {
-            /* Return failure if at end of input.  */
-            if (yychar == yyeof_)
-              YYABORT;
-         }
-       else
-         {
-           yy_destroy_ ("Error: discarding", yyla);
-           yychar = yyempty_;
+           error, discard it.  */
+
+        /* Return failure if at end of input.  */
+        if (yyla.type == yyeof_)
+          YYABORT;
+        else
+          {
+            yy_destroy_ ("Error: discarding", yyla);
+            yyempty = true;
          }
       }
 
@@ -1245,7 +1367,7 @@ m4_ifdef([b4_lex_param], [, ]b4_lex_param))[;
     goto yyreturn;
 
   yyreturn:
-    if (yychar != yyempty_)
+    if (!yyempty)
       yy_destroy_ ("Cleanup: discarding lookahead", yyla);
 
     /* Do not reclaim the symbols of the rule which action triggered
@@ -1263,9 +1385,9 @@ m4_ifdef([b4_lex_param], [, ]b4_lex_param))[;
   // Generate an error message.
   std::string
   ]b4_parser_class_name[::yysyntax_error_ (int yystate, int]dnl
-b4_error_verbose_if([ tok])[)
+b4_error_verbose_if([ yytoken])[)
   {
-    std::string res;
+    std::string yyres;
     YYUSE (yystate);
 #if YYERROR_VERBOSE
     int yyn = yypact_[yystate];
@@ -1278,36 +1400,55 @@ b4_error_verbose_if([ tok])[)
        /* Stay within bounds of both yycheck and yytname.  */
        int yychecklim = yylast_ - yyn + 1;
        int yyxend = yychecklim < yyntokens_ ? yychecklim : yyntokens_;
-       int count = 0;
-       for (int x = yyxbegin; x < yyxend; ++x)
-         if (yycheck_[x + yyn] == x && x != yyterror_)
-           ++count;
-
-       // FIXME: This method of building the message is not compatible
-       // with internationalization.  It should work like yacc.c does it.
-       // That is, first build a string that looks like this:
-       // "syntax error, unexpected %s or %s or %s"
-       // Then, invoke YY_ on this string.
-       // Finally, use the string as a format to output
-       // yytname_[tok], etc.
-       // Until this gets fixed, this message appears in English only.
-       res = "syntax error, unexpected ";
-       res += yytnamerr_ (yytname_[tok]);
-       if (count < 5)
-         {
-           count = 0;
-           for (int x = yyxbegin; x < yyxend; ++x)
-             if (yycheck_[x + yyn] == x && x != yyterror_)
-               {
-                 res += (!count++) ? ", expecting " : " or ";
-                 res += yytnamerr_ (yytname_[x]);
-               }
-         }
+
+        // Number of "expected" tokens.
+       size_t yycount = 0;
+        // Its maximum.
+        enum { YYERROR_VERBOSE_ARGS_MAXIMUM = 5 };
+        // Arguments of yyformat.
+        char const *yyarg[YYERROR_VERBOSE_ARGS_MAXIMUM];
+        yyarg[yycount++] = yytname_[yytoken];
+       for (int yyx = yyxbegin; yyx < yyxend; ++yyx)
+         if (yycheck_[yyx + yyn] == yyx && yyx != yyterror_)
+          {
+            if (yycount == YYERROR_VERBOSE_ARGS_MAXIMUM)
+            {
+              yycount = 1;
+              break;
+            }
+            else
+              yyarg[yycount++] = yytname_[yyx];
+          }
+
+        char const* yyformat = 0;
+        switch (yycount)
+        {
+#define YYCASE_(N, S)                           \
+          case N:                               \
+            yyformat = S;                       \
+          break
+          YYCASE_(1, YY_("syntax error, unexpected %s"));
+          YYCASE_(2, YY_("syntax error, unexpected %s, expecting %s"));
+          YYCASE_(3, YY_("syntax error, unexpected %s, expecting %s or %s"));
+          YYCASE_(4, YY_("syntax error, unexpected %s, expecting %s or %s or %s"));
+          YYCASE_(5, YY_("syntax error, unexpected %s, expecting %s or %s or %s or %s"));
+#undef YYCASE_
+        }
+        // Argument number.
+        size_t yyi = 0;
+        for (char const* yyp = yyformat; *yyp; ++yyp)
+          if (yyp[0] == '%' && yyp[1] == 's' && yyi < yycount)
+          {
+            yyres += yytnamerr_ (yyarg[yyi++]);
+            ++yyp;
+          }
+          else
+            yyres += *yyp;
       }
     else
 #endif
-      res = YY_("syntax error");
-    return res;
+      yyres = YY_("syntax error");
+    return yyres;
   }
 
 
@@ -1403,26 +1544,19 @@ b4_error_verbose_if([ tok])[)
     const token_number_type
     translate_table[] =
     {
-  ]b4_translate[
+]b4_translate[
     };
-    if ((unsigned int) t <= yyuser_token_number_max_)
+    const unsigned int user_token_number_max_ = ]b4_user_token_number_max[;
+    const token_number_type undef_token_ = ]b4_undef_token_number[;
+
+    if (t <= yyeof_)
+      return yyeof_;
+    else if (static_cast<unsigned int> (t) <= user_token_number_max_)
       return translate_table[t];
     else
-      return yyundef_token_;
+      return undef_token_;
   }
 
-  const int ]b4_parser_class_name[::yyeof_ = 0;
-  const int ]b4_parser_class_name[::yylast_ = ]b4_last[;
-  const int ]b4_parser_class_name[::yynnts_ = ]b4_nterms_number[;
-  const int ]b4_parser_class_name[::yyempty_ = -2;
-  const int ]b4_parser_class_name[::yyfinal_ = ]b4_final_state_number[;
-  const int ]b4_parser_class_name[::yyterror_ = 1;
-  const int ]b4_parser_class_name[::yyerrcode_ = 256;
-  const int ]b4_parser_class_name[::yyntokens_ = ]b4_tokens_number[;
-
-  const unsigned int ]b4_parser_class_name[::yyuser_token_number_max_ = ]b4_user_token_number_max[;
-  const ]b4_parser_class_name[::token_number_type ]b4_parser_class_name[::yyundef_token_ = ]b4_undef_token_number[;
-
 ]b4_namespace_close[
 
 ]b4_epilogue[]dnl