[PATCH 10/13] parsers: support translatable token aliases

[Akim Demaille]

In addition to

    %token NUM "number"

accept

    %token NUM _("number")

in which case the token will be translated in error messages.
Do not use _() in the output if there are no translatable tokens.

* src/symtab.h, src/symtab.c (symbol): Add a 'translatable' member.
* src/parse-gram.y (TSTRING): New token.
(string_as_id.opt): Replace with...
(alias): this.
Use it.
* src/scan-gram.l (SC_ESCAPED_TSTRING): New start conditions, to match
TSTRINGs.
* src/output.c (prepare_symbols): Define b4_translatable if there are
translatable strings.

* data/skeletons/glr.c, data/skeletons/lalr1.cc,
* data/skeletons/yacc.c (yytnamerr): Receive b4_translatable, and use it.
---
 data/skeletons/yacc.c | 10 ++++++++--
 src/output.c          | 24 ++++++++++++++++++++++++
 src/parse-gram.y      | 26 +++++++++++++++++---------
 src/scan-gram.l       | 25 ++++++++++++++++++++-----
 src/symtab.c          |  3 ++-
 src/symtab.h          |  7 ++++++-
 6 files changed, 77 insertions(+), 18 deletions(-)

diff --git a/data/skeletons/yacc.c b/data/skeletons/yacc.c
index 83777ebb..d5b74ce4 100644
--- a/data/skeletons/yacc.c
+++ b/data/skeletons/yacc.c
@@ -631,8 +631,15 @@ yysymbol_name (int yysymbol)
   static const char *const yy_sname[] =
   {
   ]b4_symbol_names[
+  };]m4_ifdef([b4_translatable], [[
+  /* YYTRANSLATABLE[SYMBOL-NUM] -- Whether YYTNAME[SYMBOL-NUM] is
+     internationalizable.  */
+  static ]b4_int_type_for([b4_translate])[ yytranslatable[] =
+  {
+  ]b4_translatable[
   };
-  return yy_sname[yysymbol];
+  return yytranslatable[yysymbol] ? _(yy_sname[yysymbol]) : 
yy_sname[yysymbol];]], [[
+  return yy_sname[yysymbol];]])[
 }]])[
 #endif
 
@@ -1248,7 +1255,6 @@ yytnamerr (char *yyres, const char *yystr)
     {
       YYPTRDIFF_T yyn = 0;
       char const *yyp = yystr;
-
       for (;;)
         switch (*++yyp)
           {
diff --git a/src/output.c b/src/output.c
index 8de00ed2..6f1e9f45 100644
--- a/src/output.c
+++ b/src/output.c
@@ -50,6 +50,10 @@ static struct obstack format_obstack;
 | result of formatting the FIRST and then TABLE_DATA[BEGIN..END[ (of |
 | TYPE), and to the muscle NAME_max, the max value of the            |
 | TABLE_DATA.                                                        |
+|                                                                    |
+| For the typical case of outputting a complete table from 0, pass   |
+| TABLE[0] as FIRST, and 1 as BEGIN.  For instance                   |
+| muscle_insert_base_table ("pact", base, base[0], 1, nstates);      |
 `-------------------------------------------------------------------*/
 
 
@@ -248,6 +252,26 @@ prepare_symbols (void)
   prepare_symbol_names ("tname");
   prepare_symbol_names ("symbol_names");
 
+  /* translatable -- whether a token is translatable. */
+  {
+    bool translatable = false;
+    for (int i = 0; i < ntokens; ++i)
+      if (symbols[i]->translatable)
+        {
+          translatable = true;
+          break;
+        }
+    if (translatable)
+      {
+        int *values = xnmalloc (nsyms, sizeof *values);
+        for (int i = 0; i < ntokens; ++i)
+          values[i] = symbols[i]->translatable;
+        muscle_insert_int_table ("translatable", values,
+                                 values[0], 1, ntokens);
+        free (values);
+      }
+  }
+
   /* Output YYTOKNUM. */
   {
     int *values = xnmalloc (ntokens, sizeof *values);
diff --git a/src/parse-gram.y b/src/parse-gram.y
index 199dfae8..328d7e62 100644
--- a/src/parse-gram.y
+++ b/src/parse-gram.y
@@ -143,6 +143,7 @@
 /* Define the tokens together with their human representation.  */
 %token GRAM_EOF 0 "end of file"
 %token STRING     "string"
+       TSTRING    "translatable string"
 
 %token PERCENT_TOKEN       "%token"
 %token PERCENT_NTERM       "%nterm"
@@ -216,8 +217,8 @@
 %type <unsigned char> CHAR
 %printer { fputs (char_name ($$), yyo); } <unsigned char>
 
-%type <char*> "{...}" "%?{...}" "%{...%}" EPILOGUE STRING
-%printer { fputs ($$, yyo); }  <char*>
+%type <char*> "{...}" "%?{...}" "%{...%}" EPILOGUE STRING TSTRING
+%printer { fputs ($$, yyo); } <char*>
 
 %type <uniqstr>
   BRACKETED_ID ID ID_COLON
@@ -532,7 +533,7 @@ token_decl.1:
 
 // One symbol declaration for %token or %nterm.
 token_decl:
-  id int.opt[num] string_as_id.opt[alias]
+  id int.opt[num] alias
     {
       $$ = $id;
       symbol_class_set ($id, current_class, @id, true);
@@ -549,6 +550,19 @@ int.opt:
 | INT
 ;
 
+%type <symbol*> alias;
+alias:
+  %empty         { $$ = NULL; }
+| string_as_id   { $$ = $1; }
+| TSTRING
+    {
+      $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
+      symbol_class_set ($$, token_sym, @1, false);
+      $$->translatable = true;
+    }
+;
+
+
 /*-------------------------------------.
 | token_decls_for_prec (%left, etc.).  |
 `-------------------------------------*/
@@ -782,12 +796,6 @@ string_as_id:
     }
 ;
 
-%type <symbol*> string_as_id.opt;
-string_as_id.opt:
-  %empty             { $$ = NULL; }
-| string_as_id
-;
-
 epilogue.opt:
   %empty
 | "%%" EPILOGUE
diff --git a/src/scan-gram.l b/src/scan-gram.l
index 2b5eb7f9..94e8a8ba 100644
--- a/src/scan-gram.l
+++ b/src/scan-gram.l
@@ -110,8 +110,8 @@ static void unexpected_newline (boundary, char const *);
 %}
  /* A C-like comment in directives/rules. */
 %x SC_YACC_COMMENT
- /* Strings and characters in directives/rules. */
-%x SC_ESCAPED_STRING SC_ESCAPED_CHARACTER
+ /* Characters and strings in directives/rules. */
+%x SC_ESCAPED_CHARACTER SC_ESCAPED_STRING SC_ESCAPED_TSTRING
  /* A identifier was just read in directives/rules.  Special state
     to capture the sequence 'identifier :'. */
 %x SC_AFTER_IDENTIFIER
@@ -318,6 +318,7 @@ eqopt    ({sp}=)?
 
   /* Strings. */
   "\""        token_start = loc->start; BEGIN SC_ESCAPED_STRING;
+  "_(\""      token_start = loc->start; BEGIN SC_ESCAPED_TSTRING;
 
   /* Prologue. */
   "%{"        code_start = loc->start; BEGIN SC_PROLOGUE;
@@ -378,7 +379,7 @@ eqopt    ({sp}=)?
   | added value.                                                  |
   `--------------------------------------------------------------*/
 
-<SC_ESCAPED_CHARACTER,SC_ESCAPED_STRING,SC_TAG>
+<SC_ESCAPED_CHARACTER,SC_ESCAPED_STRING,SC_ESCAPED_TSTRING,SC_TAG>
 {
   \0        complain (loc, complaint, _("invalid null character"));
 }
@@ -539,6 +540,20 @@ eqopt    ({sp}=)?
   {eol}     unexpected_newline (token_start, "\"");
 }
 
+<SC_ESCAPED_TSTRING>
+{
+  "\")" {
+    STRING_FINISH;
+    BEGIN INITIAL;
+    loc->start = token_start;
+    complain (loc, Wyacc,
+              _("POSIX Yacc does not support string literals"));
+    RETURN_VALUE (TSTRING, last_string);
+  }
+  <<EOF>>   unexpected_eof (token_start, "\"");
+  "\n"      unexpected_newline (token_start, "\"");
+}
+
   /*----------------------------------------------------------.
   | Scanning a Bison character literal, decoding its escapes. |
   | The initial quote is already eaten.                       |
@@ -601,7 +616,7 @@ eqopt    ({sp}=)?
   | Decode escaped characters.  |
   `----------------------------*/
 
-<SC_ESCAPED_STRING,SC_ESCAPED_CHARACTER>
+<SC_ESCAPED_CHARACTER,SC_ESCAPED_STRING,SC_ESCAPED_TSTRING>
 {
   \\[0-7]{1,3} {
     verify (UCHAR_MAX < ULONG_MAX);
@@ -797,7 +812,7 @@ eqopt    ({sp}=)?
   | By default, grow the string obstack with the input.  |
   `-----------------------------------------------------*/
 
-<SC_COMMENT,SC_LINE_COMMENT,SC_BRACED_CODE,SC_PREDICATE,SC_PROLOGUE,SC_EPILOGUE,SC_STRING,SC_CHARACTER,SC_ESCAPED_STRING,SC_ESCAPED_CHARACTER>
+<SC_COMMENT,SC_LINE_COMMENT,SC_BRACED_CODE,SC_PREDICATE,SC_PROLOGUE,SC_EPILOGUE,SC_STRING,SC_CHARACTER,SC_ESCAPED_CHARACTER,SC_ESCAPED_STRING,SC_ESCAPED_TSTRING>
 {
   /* Accept multibyte characters in one block instead of byte after
      byte, so that add_column_width and mbsnwidth can compute correct
diff --git a/src/symtab.c b/src/symtab.c
index 35c0930e..f4903c86 100644
--- a/src/symtab.c
+++ b/src/symtab.c
@@ -108,6 +108,7 @@ symbol_new (uniqstr tag, location loc)
               _("POSIX Yacc forbids dashes in symbol names: %s"), tag);
 
   res->tag = tag;
+  res->translatable = false;
   res->location = loc;
   res->location_of_lhs = false;
   res->alias = NULL;
@@ -954,7 +955,7 @@ dummy_symbol_get (location loc)
 }
 
 bool
-symbol_is_dummy (const symbol *sym)
+symbol_is_dummy (symbol const *sym)
 {
   return sym->tag[0] == '@' || (sym->tag[0] == '$' && sym->tag[1] == '@');
 }
diff --git a/src/symtab.h b/src/symtab.h
index a92e00a9..66017525 100644
--- a/src/symtab.h
+++ b/src/symtab.h
@@ -97,6 +97,9 @@ struct symbol
   /** The "defining" location.  */
   location location;
 
+  /** Whether this symbol is translatable. */
+  bool translatable;
+
   /** Whether \a location is about the first uses as left-hand side
       symbol of a rule (true), or simply the first occurrence (e.g.,
       in a %type, or as a rhs symbol of a rule).  The former type of
@@ -117,6 +120,8 @@ struct symbol
 
 struct sym_content
 {
+  /** The main symbol that denotes this content (it contains the
+      possible alias). */
   symbol *symbol;
 
   /** Its \c \%type.
@@ -179,7 +184,7 @@ symbol *dummy_symbol_get (location loc);
 void symbol_print (symbol const *s, FILE *f);
 
 /** Is this a dummy nonterminal?  */
-bool symbol_is_dummy (const symbol *sym);
+bool symbol_is_dummy (symbol const *sym);
 
 /** The name of the code_props type: "\%destructor" or "\%printer".  */
 char const *code_props_type_string (code_props_type kind);
-- 
2.24.1