factor the handling of m4 escaping

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

diff --git a/src/scan-code.l b/src/scan-code.l
index f1c979083e4e6255f55bbebe59817af3bd888e55..31befe9172f9bc7e8d6c99f543e4f8c5649ff2af 100644 (file)
--- a/src/scan-code.l
+++ b/src/scan-code.l
@@ -46,6 +46,9 @@ YY_DECL;
 
 #define YY_USER_ACTION  location_compute (loc, &loc->end, yytext, yyleng);
 
 
 #define YY_USER_ACTION  location_compute (loc, &loc->end, yytext, yyleng);
 

+static char *fetch_type_name (char *cp, char const **type_name,
+                              location dollar_loc);
+

 static void handle_action_dollar (symbol_list *rule, char *cp,
                                  location dollar_loc);
 static void handle_action_at (symbol_list *rule, char *cp, location at_loc);
 static void handle_action_dollar (symbol_list *rule, char *cp,
                                  location dollar_loc);
 static void handle_action_at (symbol_list *rule, char *cp, location at_loc);


@@ -95,11 +98,15 @@ ref      -?[0-9]+|{id}|"["{id}"]"|"$"
   int braces_level = 0;
 
   /* Whether a semicolon is probably needed.
   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.  */
+
+     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 directive, but don't bother.
+
+     While it is maintained in several start-conditions (factoring
+     opportunities), it is meaningful only for SC_RULE_ACTION. */

   bool need_semicolon = false;
 
   /* Whether in a C preprocessor directive.  Don't use a start condition
   bool need_semicolon = false;
 
   /* Whether in a C preprocessor directive.  Don't use a start condition


@@ -117,7 +124,7 @@ ref      -?[0-9]+|{id}|"["{id}"]"|"$"
 %}
 
   /*------------------------------------------------------------.
 %}
 
   /*------------------------------------------------------------.

-  | Scanning a C comment.  The initial `/ *' is already eaten.  |
+  | Scanning a C comment.  The initial '/ *' is already eaten.  |

   `------------------------------------------------------------*/
 
 <SC_COMMENT>
   `------------------------------------------------------------*/
 
 <SC_COMMENT>


@@ -127,7 +134,7 @@ ref      -?[0-9]+|{id}|"["{id}"]"|"$"
 
 
   /*--------------------------------------------------------------.
 
 
   /*--------------------------------------------------------------.

-  | Scanning a line comment.  The initial `//' is already eaten.  |
+  | Scanning a line comment.  The initial '//' is already eaten.  |

   `--------------------------------------------------------------*/
 
 <SC_LINE_COMMENT>
   `--------------------------------------------------------------*/
 
 <SC_LINE_COMMENT>


@@ -157,7 +164,8 @@ ref      -?[0-9]+|{id}|"["{id}"]"|"$"
 }
 
 
 }
 
 

-<SC_RULE_ACTION,SC_SYMBOL_ACTION>{
+<SC_RULE_ACTION,SC_SYMBOL_ACTION>
+{

   "'" {
     STRING_GROW;
     BEGIN SC_CHARACTER;
   "'" {
     STRING_GROW;
     BEGIN SC_CHARACTER;


@@ -176,42 +184,31 @@ ref      -?[0-9]+|{id}|"["{id}"]"|"$"
     STRING_GROW;
     BEGIN SC_LINE_COMMENT;
   }
     STRING_GROW;
     BEGIN SC_LINE_COMMENT;
   }

+  [$@]  {
+    warn_at (*loc, _("stray '%s'"), yytext);
+    obstack_escape (&obstack_for_string, yytext);
+    need_semicolon = true;
+  }
+  [\[\]]  {
+    obstack_escape (&obstack_for_string, yytext);
+    need_semicolon = true;
+  }

 }
 
 <SC_RULE_ACTION>
 {
   "$"("<"{tag}">")?{ref}  {
 }
 
 <SC_RULE_ACTION>
 {
   "$"("<"{tag}">")?{ref}  {

-    ref_tail_fields = 0;
+    ref_tail_fields = NULL;

     handle_action_dollar (self->rule, yytext, *loc);
     handle_action_dollar (self->rule, yytext, *loc);

-    if (ref_tail_fields) {
+    if (ref_tail_fields)

       obstack_sgrow (&obstack_for_string, ref_tail_fields);
       obstack_sgrow (&obstack_for_string, ref_tail_fields);

-    }

     need_semicolon = true;
   }
   "@"{ref} {
     need_semicolon = true;
   }
   "@"{ref} {

-    ref_tail_fields = 0;
+    ref_tail_fields = NULL;

     handle_action_at (self->rule, yytext, *loc);
     handle_action_at (self->rule, yytext, *loc);

-    if (ref_tail_fields) {
+    if (ref_tail_fields)

       obstack_sgrow (&obstack_for_string, ref_tail_fields);
       obstack_sgrow (&obstack_for_string, ref_tail_fields);

-    }
-    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;
   }
 
     need_semicolon = true;
   }
 


@@ -220,16 +217,16 @@ ref      -?[0-9]+|{id}|"["{id}"]"|"$"
   "}"  {
     bool outer_brace = --braces_level == 0;
 
   "}"  {
     bool outer_brace = --braces_level == 0;
 

-    /* As an undocumented Bison extension, append `;' before the last
+    /* As an undocumented Bison extension, append ';' before the last

        brace in braced code, so that the user code can omit trailing
        brace in braced code, so that the user code can omit trailing

-       `;'.  But do not append `;' if emulating Yacc, since Yacc does
+       ';'.  But do not append ';' if emulating Yacc, since Yacc does

        not append one.  */
     if (outer_brace && !yacc_flag && language_prio == default_prio
         && skeleton_prio == default_prio && need_semicolon && ! in_cpp)
       {
        not append one.  */
     if (outer_brace && !yacc_flag && language_prio == default_prio
         && 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, ';');
+        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;
       }
 
     STRING_GROW;


@@ -238,12 +235,12 @@ ref      -?[0-9]+|{id}|"["{id}"]"|"$"
 
   /* Preprocessing directives should only be recognized at the beginning
      of lines, allowing whitespace including comments, but in C/C++,
 
   /* 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.  */
+     '#' 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;
   "#"       STRING_GROW; in_cpp = true;
 
   {splice}  STRING_GROW;

-  [\n\r]    STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false; 
+  [\n\r]    STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false;

   [ \t\f]   STRING_GROW;
 
   /* YYFAIL is undocumented and was formally deprecated in Bison
   [ \t\f]   STRING_GROW;
 
   /* YYFAIL is undocumented and was formally deprecated in Bison


@@ -263,8 +260,12 @@ ref      -?[0-9]+|{id}|"["{id}"]"|"$"
 
 <SC_SYMBOL_ACTION>
 {
 
 <SC_SYMBOL_ACTION>
 {

-  "$$" {
-    obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar[");
+  "$"("<"{tag}">")?"$" {
+    const char *type_name = NULL;
+    fetch_type_name (yytext + 1, &type_name, *loc)[-1] = 0;
+    obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar(");
+    obstack_quote (&obstack_for_string, type_name);
+    obstack_sgrow (&obstack_for_string, ")[");

     self->is_value_used = true;
   }
   "@$" {
     self->is_value_used = true;
   }
   "@$" {


@@ -274,29 +275,17 @@ ref      -?[0-9]+|{id}|"["{id}"]"|"$"
 }
 
 
 }
 
 

-  /*-----------------------------------------.
-  | Escape M4 quoting characters in C code.  |
-  `-----------------------------------------*/
-

 <*>
 {
 <*>
 {

-  \$   obstack_sgrow (&obstack_for_string, "$][");
-  \@   obstack_sgrow (&obstack_for_string, "@@");
-  \[   obstack_sgrow (&obstack_for_string, "@{");
-  \]   obstack_sgrow (&obstack_for_string, "@}");
-}
+  /* Escape M4 quoting characters in C code.  */
+  [$@\[\]]    obstack_escape (&obstack_for_string, yytext);

 
 

-  /*-----------------------------------------------------.
-  | By default, grow the string obstack with the input.  |
-  `-----------------------------------------------------*/
+  /* By default, grow the string obstack with the input.  */
+  .|\n        STRING_GROW;

 
 

-<*>.|\n        STRING_GROW;
-
- /* End of processing. */
-<*><<EOF>>      {
-                   STRING_FINISH;
-                   return last_string;
-                 }
+  /* End of processing. */
+  <<EOF>>     STRING_FINISH; return last_string;
+}

 
 %%
 
 
 %%
 


@@ -345,7 +334,7 @@ typedef struct
    not visible from current midrule. */
 #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2)
 
    not visible from current midrule. */
 #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2)
 

-static variant *variant_table = 0;
+static variant *variant_table = NULL;

 static unsigned variant_table_size = 0;
 static unsigned variant_count = 0;
 
 static unsigned variant_table_size = 0;
 static unsigned variant_count = 0;
 


@@ -367,7 +356,7 @@ static void
 variant_table_free (void)
 {
   free (variant_table);
 variant_table_free (void)
 {
   free (variant_table);

-  variant_table = 0;
+  variant_table = NULL;

   variant_table_size = variant_count = 0;
 }
 
   variant_table_size = variant_count = 0;
 }
 


@@ -410,7 +399,7 @@ variant_add (uniqstr id, location id_loc, unsigned symbol_index,
 }
 
 static const char *
 }
 
 static const char *

-get_at_spec(unsigned symbol_index)
+get_at_spec (unsigned symbol_index)

 {
   static char at_buf[20];
   if (symbol_index == 0)
 {
   static char at_buf[20];
   if (symbol_index == 0)


@@ -626,8 +615,8 @@ parse_ref (char *cp, symbol_list *rule, int rule_length,
             sym_loc.start.column += 1;
             sym_loc.end = sym_loc.start;
             const char *format =
             sym_loc.start.column += 1;
             sym_loc.end = sym_loc.start;
             const char *format =

-              _("syntax error after `%c', expecting integer, letter,"
-                " `_', `[', or `$'");
+              _("syntax error after '%c', expecting integer, letter,"
+                " '_', '[', or '$'");

             complain_at_indent (sym_loc, &indent, format, dollar_or_at);
           }
         else if (midrule_rhs_index)
             complain_at_indent (sym_loc, &indent, format, dollar_or_at);
           }
         else if (midrule_rhs_index)


@@ -688,8 +677,33 @@ parse_ref (char *cp, symbol_list *rule, int rule_length,
 int max_left_semantic_context = 0;
 
 
 int max_left_semantic_context = 0;
 
 

+/* If CP points to a typename (i.e., <.*?>), set TYPE_NAME to its
+   beginning (i.e., after the opening "<", and return the pointer
+   immediately after it.  */
+
+static
+char *
+fetch_type_name (char *cp, char const **type_name,
+                 location dollar_loc)
+{
+  if (*cp == '<')
+    {
+      *type_name = ++cp;
+      while (*cp != '>')
+       ++cp;
+
+      /* The '>' symbol will be later replaced by '\0'. Original
+        'text' is needed for error messages. */
+      ++cp;
+      if (untyped_var_seen)
+       complain_at (dollar_loc, _("explicit type given in untyped grammar"));
+      tag_seen = true;
+    }
+  return cp;
+}
+

 /*------------------------------------------------------------------.
 /*------------------------------------------------------------------.

-| TEXT is pointing to a wannabee semantic value (i.e., a `$').      |
+| TEXT is pointing to a wannabee semantic value (i.e., a '$').      |

 |                                                                   |
 | Possible inputs: $[<TYPENAME>]($|integer)                         |
 |                                                                   |
 |                                                                   |
 | Possible inputs: $[<TYPENAME>]($|integer)                         |
 |                                                                   |


@@ -701,7 +715,6 @@ handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
 {
   char const *type_name = NULL;
   char *cp = text + 1;
 {
   char const *type_name = NULL;
   char *cp = text + 1;

-  char *gt_ptr = 0;

   symbol_list *effective_rule;
   int effective_rule_length;
   int n;
   symbol_list *effective_rule;
   int effective_rule_length;
   int n;


@@ -718,26 +731,14 @@ handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
     }
 
   /* Get the type name if explicit. */
     }
 
   /* Get the type name if explicit. */

-  if (*cp == '<')
-    {
-      type_name = ++cp;
-      while (*cp != '>')
-       ++cp;
-
-      /* The '>' symbol will be later replaced by '\0'. Original
-        'text' is needed for error messages. */
-      gt_ptr = cp;
-      ++cp;
-      if (untyped_var_seen)
-       complain_at (dollar_loc, _("explicit type given in untyped grammar"));
-      tag_seen = true;
-    }
+  cp = fetch_type_name (cp, &type_name, dollar_loc);

 
   n = parse_ref (cp, effective_rule, effective_rule_length,
                 rule->midrule_parent_rhs_index, text, dollar_loc, '$');
 
 
   n = parse_ref (cp, effective_rule, effective_rule_length,
                 rule->midrule_parent_rhs_index, text, dollar_loc, '$');
 

-  if (gt_ptr)
-    *gt_ptr = '\0';
+  /* End type_name. */
+  if (type_name)
+    cp[-1] = '\0';

 
   switch (n)
     {
 
   switch (n)
     {


@@ -749,26 +750,26 @@ handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
        type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0);
 
       if (!type_name)
        type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0);
 
       if (!type_name)

-       {
-         if (union_seen | tag_seen)
-           {
-             if (rule->midrule_parent_rule)
-               complain_at (dollar_loc,
-                            _("$$ for the midrule at $%d of `%s'"
-                              " has no declared type"),
-                            rule->midrule_parent_rhs_index,
-                            effective_rule->content.sym->tag);
-             else
-               complain_at (dollar_loc, _("$$ of `%s' has no declared type"),
-                            rule->content.sym->tag);
-           }
-         else
-           untyped_var_seen = true;
-         type_name = "";
-       }
+        {
+          if (union_seen | tag_seen)
+            {
+              if (rule->midrule_parent_rule)
+                complain_at (dollar_loc,
+                             _("$$ for the midrule at $%d of %s"
+                               " has no declared type"),
+                             rule->midrule_parent_rhs_index,
+                             quote (effective_rule->content.sym->tag));
+              else
+                complain_at (dollar_loc, _("$$ of %s has no declared type"),
+                             quote (rule->content.sym->tag));
+            }
+          else
+            untyped_var_seen = true;
+        }

 
 

-      obstack_fgrow1 (&obstack_for_string,
-                     "]b4_lhs_value([%s])[", type_name);
+      obstack_sgrow (&obstack_for_string, "]b4_lhs_value(");
+      obstack_quote (&obstack_for_string, type_name);
+      obstack_sgrow (&obstack_for_string, ")[");

       rule->action_props.is_value_used = true;
       break;
 
       rule->action_props.is_value_used = true;
       break;
 


@@ -779,18 +780,18 @@ handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
        type_name =
          symbol_list_n_type_name_get (effective_rule, dollar_loc, n);
       if (!type_name)
        type_name =
          symbol_list_n_type_name_get (effective_rule, dollar_loc, n);
       if (!type_name)

-       {
-         if (union_seen | tag_seen)
-           complain_at (dollar_loc, _("$%s of `%s' has no declared type"),
-                        cp, effective_rule->content.sym->tag);
-         else
-           untyped_var_seen = true;
-         type_name = "";
-       }
+        {
+          if (union_seen | tag_seen)
+            complain_at (dollar_loc, _("$%s of %s has no declared type"),
+                         cp, quote (effective_rule->content.sym->tag));
+          else
+            untyped_var_seen = true;
+        }

 
 

-      obstack_fgrow3 (&obstack_for_string,
-                     "]b4_rhs_value(%d, %d, [%s])[",
-                     effective_rule_length, n, type_name);
+      obstack_fgrow2 (&obstack_for_string,
+                     "]b4_rhs_value(%d, %d, ", effective_rule_length, n);
+      obstack_quote (&obstack_for_string, type_name);
+      obstack_sgrow (&obstack_for_string, ")[");

       if (n > 0)
        symbol_list_n_get (effective_rule, n)->action_props.is_value_used =
          true;
       if (n > 0)
        symbol_list_n_get (effective_rule, n)->action_props.is_value_used =
          true;


@@ -800,7 +801,7 @@ handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
 
 
 /*------------------------------------------------------.
 
 
 /*------------------------------------------------------.

-| TEXT is a location token (i.e., a `@...').  Output to |
+| TEXT is a location token (i.e., a '@...').  Output to |

 | OBSTACK_FOR_STRING a reference to this location.      |
 `------------------------------------------------------*/
 
 | OBSTACK_FOR_STRING a reference to this location.      |
 `------------------------------------------------------*/
 


@@ -826,7 +827,7 @@ handle_action_at (symbol_list *rule, char *text, location at_loc)
   locations_flag = true;
 
   n = parse_ref (cp, effective_rule, effective_rule_length,
   locations_flag = true;
 
   n = parse_ref (cp, effective_rule, effective_rule_length,

-                      rule->midrule_parent_rhs_index, text, at_loc, '@');
+                 rule->midrule_parent_rhs_index, text, at_loc, '@');

   switch (n)
     {
     case INVALID_REF:
   switch (n)
     {
     case INVALID_REF: