Implemented long variable names a la spss V12.

[pspp-builds.git] / src / lexer.c

diff --git a/src/lexer.c b/src/lexer.c
index e72f7f1bc6f11ac5e5efde7123751ba5d2c47b54..568a87e316654c45b3b979460d8805bf0c9f8261 100644 (file)
--- a/src/lexer.c
+++ b/src/lexer.c
@@ -44,14 +44,14 @@
 /* Current token. */
 int token;
 
-/* T_NUM: the token's value. */
+/* T_POS_NUM, T_NEG_NUM: the token's value. */
 double tokval;
 
 /* T_ID: the identifier. */
-char tokid[9];
+char tokid[LONG_NAME_LEN + 1];
 
 /* T_ID, T_STRING: token string value.
-   For T_ID, this is not truncated to 8 characters as is tokid. */
+   For T_ID, this is not truncated to SHORT_NAME_LEN characters as is tokid. */
 struct string tokstr;
 \f
 /* Static variables. */
@@ -81,7 +81,6 @@ static struct string put_tokstr;
 static double put_tokval;
 
 static void unexpected_eof (void);
-static inline int check_id (const char *id, size_t len);
 static void convert_numeric_string_to_char_string (int type);
 static int parse_string (int type);
 
@@ -99,6 +98,13 @@ lex_init (void)
   if (!lex_get_line ())
     unexpected_eof ();
 }
+
+void
+lex_done (void)
+{
+  ds_destroy(&put_tokstr);
+}
+
 \f
 /* Common functions. */
 
@@ -110,8 +116,8 @@ restore_token (void)
   assert (put_token != 0);
   token = put_token;
   ds_replace (&tokstr, ds_c_str (&put_tokstr));
-  strncpy (tokid, ds_c_str (&put_tokstr), 8);
-  tokid[8] = 0;
+  strncpy (tokid, ds_c_str (&put_tokstr), SHORT_NAME_LEN);
+  tokid[SHORT_NAME_LEN] = 0;
   tokval = put_tokval;
   put_token = 0;
 }
@@ -131,6 +137,8 @@ save_token (void)
 void
 lex_get (void)
 {
+  int i;
+
   /* If a token was pushed ahead, return it. */
   if (put_token)
     {
@@ -217,8 +225,11 @@ lex_get (void)
                    token = '-';
                    break;
                  }
+                token = T_NEG_NUM;
              }
-
+            else 
+              token = T_POS_NUM;
+                
            /* Parse the number, copying it into tokstr. */
            while (isdigit ((unsigned char) *prog))
              ds_putc (&tokstr, *prog++);
@@ -249,7 +260,6 @@ lex_get (void)
                ds_putc (&tokstr, '0');
              }
 
-           token = T_NUM;
            break;
          }
 
@@ -346,15 +356,19 @@ lex_get (void)
            }
 
          /* Copy id to tokstr. */
-         ds_putc (&tokstr, toupper ((unsigned char) *prog++));
+         ds_putc (&tokstr, *prog++);
          while (CHAR_IS_IDN (*prog))
-           ds_putc (&tokstr, toupper ((unsigned char) *prog++));
+           ds_putc (&tokstr, *prog++);
+
+         /* Copy tokstr to tokid, truncating it to LONG_NAME_LEN characters.*/
+         strncpy (tokid, ds_c_str (&tokstr), LONG_NAME_LEN);
+         tokid[LONG_NAME_LEN] = 0;
 
-         /* Copy tokstr to tokid, truncating it to 8 characters. */
-         strncpy (tokid, ds_c_str (&tokstr), 8);
-         tokid[8] = 0;
+         /* Convert to upper case */
+         for ( i = 0 ; i < ds_length(&tokstr) ; ++i ) 
+                 tokstr.string[i] = toupper(tokstr.string[i]);
 
-         token = check_id (ds_c_str (&tokstr), ds_length (&tokstr));
+         token = lex_id_to_token (ds_c_str (&tokstr), ds_length (&tokstr));
          break;
 
        default:
@@ -418,11 +432,27 @@ lex_end_of_command (void)
 \f
 /* Token testing functions. */
 
-/* Returns nonzero if the current token is an integer. */
-int
-lex_integer_p (void)
+/* Returns true if the current token is a number. */
+bool
+lex_is_number (void) 
 {
-  return (token == T_NUM
+  return token == T_POS_NUM || token == T_NEG_NUM;
+}
+
+/* Returns the value of the current token, which must be a
+   floating point number. */
+double
+lex_number (void)
+{
+  assert (lex_is_number ());
+  return tokval;
+}
+
+/* Returns true iff the current token is an integer. */
+bool
+lex_is_integer (void)
+{
+  return (lex_is_number ()
          && tokval != NOT_LONG
          && tokval >= LONG_MIN
          && tokval <= LONG_MAX
@@ -434,26 +464,9 @@ lex_integer_p (void)
 long
 lex_integer (void)
 {
-  assert (lex_integer_p ());
-  return tokval;
-}
-/* Returns nonzero if the current token is an floating point. */
-int
-lex_double_p (void)
-{
-  return ( token == T_NUM
-          && tokval != NOT_DOUBLE );
-}
-
-/* Returns the value of the current token, which must be a
-   floating point number. */
-double
-lex_double (void)
-{
-  assert (lex_double_p ());
+  assert (lex_is_integer ());
   return tokval;
 }
-
 \f  
 /* Token matching functions. */
 
@@ -491,7 +504,7 @@ lex_match_id (const char *s)
 int
 lex_match_int (int x)
 {
-  if (lex_integer_p () && lex_integer () == x)
+  if (lex_is_integer () && lex_integer () == x)
     {
       lex_get ();
       return 1;
@@ -556,7 +569,7 @@ lex_force_string (void)
 int
 lex_force_int (void)
 {
-  if (lex_integer_p ())
+  if (lex_is_integer ())
     return 1;
   else
     {
@@ -570,7 +583,7 @@ lex_force_int (void)
 int
 lex_force_num (void)
 {
-  if (token == T_NUM)
+  if (lex_is_number ())
     return 1;
   else
     {
@@ -628,6 +641,23 @@ lex_id_match (const char *kw, const char *tok)
 {
   return lex_id_match_len (kw, strlen (kw), tok, strlen (tok));
 }
+
+/* Returns the proper token type, either T_ID or a reserved keyword
+   enum, for ID[], which must contain LEN characters. */
+int
+lex_id_to_token (const char *id, size_t len)
+{
+  const char **kwp;
+
+  if (len < 2 || len > 4)
+    return T_ID;
+  
+  for (kwp = keywords; *kwp; kwp++)
+    if (!strcasecmp (*kwp, id))
+      return T_FIRST_KEYWORD + (kwp - keywords);
+
+  return T_ID;
+}
 \f
 /* Weird token functions. */
 
@@ -690,8 +720,8 @@ lex_put_back_id (const char *id)
   save_token ();
   token = T_ID;
   ds_replace (&tokstr, id);
-  strncpy (tokid, ds_c_str (&tokstr), 8);
-  tokid[8] = 0;
+  strncpy (tokid, ds_c_str (&tokstr), SHORT_NAME_LEN);
+  tokid[SHORT_NAME_LEN] = 0;
 }
 \f
 /* Weird line processing functions. */
@@ -870,7 +900,8 @@ lex_token_representation (void)
   switch (token)
     {
     case T_ID:
-    case T_NUM:
+    case T_POS_NUM:
+    case T_NEG_NUM:
       return xstrdup (ds_c_str (&tokstr));
       break;
 
@@ -945,9 +976,9 @@ lex_token_representation (void)
 void
 lex_negative_to_dash (void)
 {
-  if (token == T_NUM && tokval < 0.0)
+  if (token == T_NEG_NUM)
     {
-      token = T_NUM;
+      token = T_POS_NUM;
       tokval = -tokval;
       ds_replace (&tokstr, ds_c_str (&tokstr) + 1);
       save_token ();
@@ -993,23 +1024,6 @@ unexpected_eof (void)
   msg (FE, _("Unexpected end of file."));
 }
 
-/* Returns the proper token type, either T_ID or a reserved keyword
-   enum, for ID[], which must contain LEN characters. */
-static inline int
-check_id (const char *id, size_t len)
-{
-  const char **kwp;
-
-  if (len < 2 || len > 4)
-    return T_ID;
-  
-  for (kwp = keywords; *kwp; kwp++)
-    if (!strcmp (*kwp, id))
-      return T_FIRST_KEYWORD + (kwp - keywords);
-
-  return T_ID;
-}
-
 /* When invoked, tokstr contains a string of binary, octal, or hex
    digits, for values of TYPE of 0, 1, or 2, respectively.  The string
    is converted to characters having the specified values. */
@@ -1209,7 +1223,8 @@ dump_token (void)
       fprintf (stderr, "ID\t%s\n", tokid);
       break;
 
-    case T_NUM:
+    case T_POS_NUM:
+    case T_NEG_NUM:
       fprintf (stderr, "NUM\t%f\n", tokval);
       break;