[PATCH 10/10] skeletons: clarify the tag of special tokens

[Akim Demaille]

From

    GRAM_EOF = 0,                  /* $end  */
    GRAM_ERRCODE = 1,              /* error  */
    GRAM_UNDEF = 2,                /* $undefined  */

to

    GRAM_EOF = 0,                  /* "end of file"  */
    GRAM_ERRCODE = 1,              /* error  */
    GRAM_UNDEF = 2,                /* "invalid token"  */

* src/output.c (symbol_tag): New.
Use it to pass the token names and the symbol tags to the skeletons.

* tests/input.at: Adjust.
---
 src/output.c     | 48 ++++++++++++++++++++++++++++--------------------
 src/parse-gram.c |  4 ++--
 src/parse-gram.h |  4 ++--
 tests/input.at   |  4 ++--
 4 files changed, 34 insertions(+), 26 deletions(-)

diff --git a/src/output.c b/src/output.c
index 1a7aa906..01669f5a 100644
--- a/src/output.c
+++ b/src/output.c
@@ -196,6 +196,23 @@ has_translations (void)
   return false;
 }
 
+/* The tag to show in the generated parsers.  Use "end of file" rather
+   than "$end".  But keep "$end" in the reports, it's shorter and more
+   consistent.  Support i18n if the user already uses it.  */
+static const char *
+symbol_tag (const symbol *sym)
+{
+  const bool eof_is_user_defined
+    = !endtoken->alias || STRNEQ (endtoken->alias->tag, "$end");
+
+  if (!eof_is_user_defined && sym->content == endtoken->content)
+    return "\"end of file\"";
+  else if (sym->content == undeftoken->content)
+    return "\"invalid token\"";
+  else
+    return sym->tag;
+}
+
 /* Generate the b4_<MUSCLE_NAME> (e.g., b4_tname) table with the
    symbol names (aka tags). */
 
@@ -209,27 +226,18 @@ prepare_symbol_names (char const *muscle_name)
   const bool with_translations = !quote && has_translations ();
 
   /* We assume that the table will be output starting at column 2. */
-  int j = 2;
+  int col = 2;
   struct quoting_options *qo = clone_quoting_options (0);
   set_quoting_style (qo, c_quoting_style);
   set_quoting_flags (qo, QA_SPLIT_TRIGRAPHS);
   for (int i = 0; i < nsyms; i++)
     {
-      /* Use "end of file" rather than "$end".  But keep "$end" in the
-         reports, it's shorter and more consistent.  Support i18n if
-         the user already uses it.  */
-      const char *tag = symbols[i]->tag;
-      bool translatable = with_translations && symbols[i]->translatable;
-      if (!eof_is_user_defined && symbols[i]->content == endtoken->content)
-        {
-          tag = "\"end of file\"";
-          translatable = with_translations;
-        }
-      else if (symbols[i]->content == undeftoken->content)
-        {
-          tag = "\"invalid token\"";
-          translatable = with_translations;
-        }
+      const char *tag = symbol_tag (symbols[i]);
+      bool translatable =
+        with_translations
+        && (symbols[i]->translatable
+            || (!eof_is_user_defined && symbols[i]->content == 
endtoken->content)
+            || symbols[i]->content == undeftoken->content);
 
       char *cp
         = tag[0] == '"' && !quote
@@ -241,10 +249,10 @@ prepare_symbol_names (char const *muscle_name)
         = strlen (cp) + 2
         + (translatable ? strlen ("N_()") : 0);
 
-      if (j + width > 75)
+      if (col + width > 75)
         {
           obstack_sgrow (&format_obstack, "\n ");
-          j = 1;
+          col = 1;
         }
 
       if (i)
@@ -256,7 +264,7 @@ prepare_symbol_names (char const *muscle_name)
         obstack_sgrow (&format_obstack, "])[");
       free (cp);
       obstack_1grow (&format_obstack, ',');
-      j += width;
+      col += width;
     }
   free (qo);
   obstack_sgrow (&format_obstack, " ]b4_null[");
@@ -570,7 +578,7 @@ prepare_symbol_definitions (void)
 
       /* Its tag.  Typically for documentation purpose.  */
       SET_KEY ("tag");
-      MUSCLE_INSERT_STRING (key, sym->tag);
+      MUSCLE_INSERT_STRING (key, symbol_tag (sym));
 
       SET_KEY ("user_number");
       MUSCLE_INSERT_INT (key, sym->content->user_token_number);
diff --git a/src/parse-gram.c b/src/parse-gram.c
index be7e599d..fb74a582 100644
--- a/src/parse-gram.c
+++ b/src/parse-gram.c
@@ -104,9 +104,9 @@
 enum yysymbol_kind_t
 {
   YYSYMBOL_YYEMPTY = -2,
-  YYSYMBOL_YYEOF = 0,                      /* $end  */
+  YYSYMBOL_YYEOF = 0,                      /* "end of file"  */
   YYSYMBOL_YYERROR = 1,                    /* error  */
-  YYSYMBOL_YYUNDEF = 2,                    /* $undefined  */
+  YYSYMBOL_YYUNDEF = 2,                    /* "invalid token"  */
   YYSYMBOL_STRING = 3,                     /* "string"  */
   YYSYMBOL_TSTRING = 4,                    /* "translatable string"  */
   YYSYMBOL_PERCENT_TOKEN = 5,              /* "%token"  */
diff --git a/src/parse-gram.h b/src/parse-gram.h
index ac501c5f..b40347ba 100644
--- a/src/parse-gram.h
+++ b/src/parse-gram.h
@@ -77,9 +77,9 @@ extern int gram_debug;
 # define GRAM_TOKENTYPE
   enum gram_tokentype
   {
-    GRAM_EOF = 0,                  /* $end  */
+    GRAM_EOF = 0,                  /* "end of file"  */
     GRAM_ERRCODE = 1,              /* error  */
-    GRAM_UNDEF = 2,                /* $undefined  */
+    GRAM_UNDEF = 2,                /* "invalid token"  */
     STRING = 3,                    /* "string"  */
     TSTRING = 4,                   /* "translatable string"  */
     PERCENT_TOKEN = 5,             /* "%token"  */
diff --git a/tests/input.at b/tests/input.at
index 7cf2eccb..f1cab823 100644
--- a/tests/input.at
+++ b/tests/input.at
@@ -431,9 +431,9 @@ exp:;
 AT_BISON_CHECK([-Wno-other -S./dump-symbols.m4 input.y])
 AT_CHECK([cat symbols.csv], [],
 [[number, class, tag, id, user_number, type,
-0, Token, $end, YYEOF, 0, ,
+0, Token, "end of file", YYEOF, 0, ,
 1, Token, error, YYERRCODE, 256, ,
-2, Token, $undefined, YYUNDEF, 257, ,
+2, Token, "invalid token", YYUNDEF, 257, ,
 3, Token, 'a', , 97, ,
 4, Token, "A1", A1, 1, ,
 5, Token, A2, A2, 258, ,
-- 
2.26.0