gcc/tradcif.y

   1 /* Parse C expressions for CCCP.
   2    Copyright (C) 1987, 2000, 2001 Free Software Foundation.
   3    Adapted from expread.y of GDB by Paul Rubin, July 1986.
   4    Adapted to ANSI C, Richard Stallman, Jan 1987
   5    Dusted off, polished, and adapted for use as traditional
   6    preprocessor only, Zack Weinberg, Jul 2000
   7
   8 This program is free software; you can redistribute it and/or modify it
   9 under the terms of the GNU General Public License as published by the
  10 Free Software Foundation; either version 2, or (at your option) any
  11 later version.
  12
  13 This program is distributed in the hope that it will be useful,
  14 but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16 GNU General Public License for more details.
  17
  18 You should have received a copy of the GNU General Public License
  19 along with this program; if not, write to the Free Software
  20 Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
  21
  22 /* Parse a C expression from text in a string  */
  23
  24 %{
  25 #include "config.h"
  26 #include "system.h"
  27 #include "tradcpp.h"
  28 #include <setjmp.h>
  29
  30   static int yylex PARAMS ((void));
  31   static void yyerror PARAMS ((const char *msgid)) ATTRIBUTE_NORETURN;
  32
  33   static int parse_number PARAMS ((int));
  34   static int parse_escape PARAMS ((const char **));
  35
  36   static int expression_value;
  37   static jmp_buf parse_return_error;
  38
  39   /* During parsing of a C expression, the pointer to the next
  40      character is in this variable.  */
  41
  42   static const char *lexptr;
  43 %}
  44
  45 %union {
  46   struct constant {long value; int unsignedp;} integer;
  47   int voidval;
  48   char *sval;
  49 }
  50
  51 %type <integer> exp exp1 start
  52 %token <integer> INT CHAR
  53 %token <sval> NAME
  54 %token <integer> ERROR
  55
  56 %right '?' ':'
  57 %left ','
  58 %left OR
  59 %left AND
  60 %left '|'
  61 %left '^'
  62 %left '&'
  63 %left EQUAL NOTEQUAL
  64 %left '<' '>' LEQ GEQ
  65 %left LSH RSH
  66 %left '+' '-'
  67 %left '*' '/' '%'
  68 %right UNARY
  69
  70 /* %expect 40 */
  71 \f
  72 %%
  73
  74 start   :       exp1
  75                 { expression_value = $1.value; }
  76         ;
  77
  78 /* Expressions, including the comma operator.  */
  79 exp1    :       exp
  80         |       exp1 ',' exp
  81                         { $$ = $3; }
  82         ;
  83
  84 /* Expressions, not including the comma operator.  */
  85 exp     :       '-' exp    %prec UNARY
  86                         { $$.value = - $2.value;
  87                           $$.unsignedp = $2.unsignedp; }
  88         |       '!' exp    %prec UNARY
  89                         { $$.value = ! $2.value;
  90                           $$.unsignedp = 0; }
  91         |       '+' exp    %prec UNARY
  92                         { $$ = $2; }
  93         |       '~' exp    %prec UNARY
  94                         { $$.value = ~ $2.value;
  95                           $$.unsignedp = $2.unsignedp; }
  96         |       '(' exp1 ')'
  97                         { $$ = $2; }
  98         ;
  99
 100 /* Binary operators in order of decreasing precedence.  */
 101 exp     :       exp '*' exp
 102                         { $$.unsignedp = $1.unsignedp || $3.unsignedp;
 103                           if ($$.unsignedp)
 104                             $$.value = (unsigned) $1.value * $3.value;
 105                           else
 106                             $$.value = $1.value * $3.value; }
 107         |       exp '/' exp
 108                         { if ($3.value == 0)
 109                             {
 110                               error ("division by zero in #if");
 111                               $3.value = 1;
 112                             }
 113                           $$.unsignedp = $1.unsignedp || $3.unsignedp;
 114                           if ($$.unsignedp)
 115                             $$.value = (unsigned) $1.value / $3.value;
 116                           else
 117                             $$.value = $1.value / $3.value; }
 118         |       exp '%' exp
 119                         { if ($3.value == 0)
 120                             {
 121                               error ("division by zero in #if");
 122                               $3.value = 1;
 123                             }
 124                           $$.unsignedp = $1.unsignedp || $3.unsignedp;
 125                           if ($$.unsignedp)
 126                             $$.value = (unsigned) $1.value % $3.value;
 127                           else
 128                             $$.value = $1.value % $3.value; }
 129         |       exp '+' exp
 130                         { $$.value = $1.value + $3.value;
 131                           $$.unsignedp = $1.unsignedp || $3.unsignedp; }
 132         |       exp '-' exp
 133                         { $$.value = $1.value - $3.value;
 134                           $$.unsignedp = $1.unsignedp || $3.unsignedp; }
 135         |       exp LSH exp
 136                         { $$.unsignedp = $1.unsignedp;
 137                           if ($$.unsignedp)
 138                             $$.value = (unsigned) $1.value << $3.value;
 139                           else
 140                             $$.value = $1.value << $3.value; }
 141         |       exp RSH exp
 142                         { $$.unsignedp = $1.unsignedp;
 143                           if ($$.unsignedp)
 144                             $$.value = (unsigned) $1.value >> $3.value;
 145                           else
 146                             $$.value = $1.value >> $3.value; }
 147         |       exp EQUAL exp
 148                         { $$.value = ($1.value == $3.value);
 149                           $$.unsignedp = 0; }
 150         |       exp NOTEQUAL exp
 151                         { $$.value = ($1.value != $3.value);
 152                           $$.unsignedp = 0; }
 153         |       exp LEQ exp
 154                         { $$.unsignedp = 0;
 155                           if ($1.unsignedp || $3.unsignedp)
 156                             $$.value =
 157                               (unsigned) $1.value <= (unsigned) $3.value;
 158                           else
 159                             $$.value = $1.value <= $3.value; }
 160         |       exp GEQ exp
 161                         { $$.unsignedp = 0;
 162                           if ($1.unsignedp || $3.unsignedp)
 163                             $$.value =
 164                               (unsigned) $1.value >= (unsigned) $3.value;
 165                           else
 166                             $$.value = $1.value >= $3.value; }
 167         |       exp '<' exp
 168                         { $$.unsignedp = 0;
 169                           if ($1.unsignedp || $3.unsignedp)
 170                             $$.value =
 171                               (unsigned) $1.value < (unsigned) $3.value;
 172                           else
 173                             $$.value = $1.value < $3.value; }
 174         |       exp '>' exp
 175                         { $$.unsignedp = 0;
 176                           if ($1.unsignedp || $3.unsignedp)
 177                             $$.value =
 178                               (unsigned) $1.value > (unsigned) $3.value;
 179                           else
 180                             $$.value = $1.value > $3.value; }
 181         |       exp '&' exp
 182                         { $$.value = $1.value & $3.value;
 183                           $$.unsignedp = $1.unsignedp || $3.unsignedp; }
 184         |       exp '^' exp
 185                         { $$.value = $1.value ^ $3.value;
 186                           $$.unsignedp = $1.unsignedp || $3.unsignedp; }
 187         |       exp '|' exp
 188                         { $$.value = $1.value | $3.value;
 189                           $$.unsignedp = $1.unsignedp || $3.unsignedp; }
 190         |       exp AND exp
 191                         { $$.value = ($1.value && $3.value);
 192                           $$.unsignedp = 0; }
 193         |       exp OR exp
 194                         { $$.value = ($1.value || $3.value);
 195                           $$.unsignedp = 0; }
 196         |       exp '?' exp ':' exp
 197                         { $$.value = $1.value ? $3.value : $5.value;
 198                           $$.unsignedp = $3.unsignedp || $5.unsignedp; }
 199         |       INT
 200                         { $$ = yylval.integer; }
 201         |       CHAR
 202                         { $$ = yylval.integer; }
 203         |       NAME
 204                         { $$.value = 0;
 205                           $$.unsignedp = 0; }
 206         |       '#'     { $$.value =
 207                             test_assertion ((unsigned char **) &lexptr); }
 208         ;
 209 %%
 210 \f
 211 /* Take care of parsing a number (anything that starts with a digit).
 212    Set yylval and return the token type; update lexptr.
 213    LEN is the number of characters in it.  */
 214
 215 /* maybe needs to actually deal with floating point numbers */
 216
 217 static int
 218 parse_number (olen)
 219      int olen;
 220 {
 221   register const char *p = lexptr;
 222   register long n = 0;
 223   register int c;
 224   register int base = 10;
 225   register int len = olen;
 226
 227   for (c = 0; c < len; c++)
 228     if (p[c] == '.') {
 229       /* It's a float since it contains a point.  */
 230       yyerror ("floating point numbers not allowed in #if expressions");
 231       return ERROR;
 232     }
 233
 234   yylval.integer.unsignedp = 0;
 235
 236   if (len >= 3 && (!strncmp (p, "0x", 2) || !strncmp (p, "0X", 2))) {
 237     p += 2;
 238     base = 16;
 239     len -= 2;
 240   }
 241   else if (*p == '0')
 242     base = 8;
 243
 244   while (len > 0) {
 245     c = *p++;
 246     len--;
 247     if (c >= 'A' && c <= 'Z') c += 'a' - 'A';
 248
 249     if (c >= '0' && c <= '9') {
 250       n *= base;
 251       n += c - '0';
 252     } else if (base == 16 && c >= 'a' && c <= 'f') {
 253       n *= base;
 254       n += c - 'a' + 10;
 255     } else {
 256       /* `l' means long, and `u' means unsigned.  */
 257       while (1) {
 258         if (c == 'l' || c == 'L')
 259           ;
 260         else if (c == 'u' || c == 'U')
 261           yylval.integer.unsignedp = 1;
 262         else
 263           break;
 264
 265         if (len == 0)
 266           break;
 267         c = *p++;
 268         len--;
 269       }
 270       /* Don't look for any more digits after the suffixes.  */
 271       break;
 272     }
 273   }
 274
 275   if (len != 0) {
 276     yyerror ("Invalid number in #if expression");
 277     return ERROR;
 278   }
 279
 280   /* If too big to be signed, consider it unsigned.  */
 281   if (n < 0)
 282     yylval.integer.unsignedp = 1;
 283
 284   lexptr = p;
 285   yylval.integer.value = n;
 286   return INT;
 287 }
 288
 289 struct token {
 290   const char *operator;
 291   int token;
 292 };
 293
 294 #ifndef NULL
 295 #define NULL 0
 296 #endif
 297
 298 static struct token tokentab2[] = {
 299   {"&&", AND},
 300   {"||", OR},
 301   {"<<", LSH},
 302   {">>", RSH},
 303   {"==", EQUAL},
 304   {"!=", NOTEQUAL},
 305   {"<=", LEQ},
 306   {">=", GEQ},
 307   {NULL, ERROR}
 308 };
 309
 310 /* Read one token, getting characters through lexptr.  */
 311
 312 static int
 313 yylex ()
 314 {
 315   register int c;
 316   register int namelen;
 317   register const char *tokstart;
 318   register struct token *toktab;
 319
 320  retry:
 321
 322   tokstart = lexptr;
 323   c = *tokstart;
 324   /* See if it is a special token of length 2.  */
 325   for (toktab = tokentab2; toktab->operator != NULL; toktab++)
 326     if (c == *toktab->operator && tokstart[1] == toktab->operator[1]) {
 327       lexptr += 2;
 328       return toktab->token;
 329     }
 330
 331   switch (c) {
 332   case 0:
 333     return 0;
 334
 335   case ' ':
 336   case '\t':
 337   case '\r':
 338   case '\n':
 339     lexptr++;
 340     goto retry;
 341
 342   case '\'':
 343     lexptr++;
 344     c = *lexptr++;
 345     if (c == '\\')
 346       c = parse_escape (&lexptr);
 347
 348     /* Sign-extend the constant if chars are signed on target machine.  */
 349     {
 350       if (lookup ((const unsigned char *)"__CHAR_UNSIGNED__",
 351                    sizeof ("__CHAR_UNSIGNED__")-1, -1)
 352           || ((c >> (CHAR_TYPE_SIZE - 1)) & 1) == 0)
 353         yylval.integer.value = c & ((1 << CHAR_TYPE_SIZE) - 1);
 354       else
 355         yylval.integer.value = c | ~((1 << CHAR_TYPE_SIZE) - 1);
 356     }
 357
 358     yylval.integer.unsignedp = 0;
 359     c = *lexptr++;
 360     if (c != '\'') {
 361       yyerror ("Invalid character constant in #if");
 362       return ERROR;
 363     }
 364
 365     return CHAR;
 366
 367     /* some of these chars are invalid in constant expressions;
 368        maybe do something about them later */
 369   case '/':
 370   case '+':
 371   case '-':
 372   case '*':
 373   case '%':
 374   case '|':
 375   case '&':
 376   case '^':
 377   case '~':
 378   case '!':
 379   case '@':
 380   case '<':
 381   case '>':
 382   case '(':
 383   case ')':
 384   case '[':
 385   case ']':
 386   case '.':
 387   case '?':
 388   case ':':
 389   case '=':
 390   case '{':
 391   case '}':
 392   case ',':
 393   case '#':
 394     lexptr++;
 395     return c;
 396
 397   case '"':
 398     yyerror ("double quoted strings not allowed in #if expressions");
 399     return ERROR;
 400   }
 401   if (c >= '0' && c <= '9') {
 402     /* It's a number */
 403     for (namelen = 0;
 404          c = tokstart[namelen], is_idchar (c) || c == '.';
 405          namelen++)
 406       ;
 407     return parse_number (namelen);
 408   }
 409
 410   if (!is_idstart (c)) {
 411     yyerror ("Invalid token in expression");
 412     return ERROR;
 413   }
 414
 415   /* It is a name.  See how long it is.  */
 416
 417   for (namelen = 0;
 418        is_idchar (tokstart[namelen]);
 419        namelen++)
 420     ;
 421
 422   lexptr += namelen;
 423   return NAME;
 424 }
 425
 426
 427 /* Parse a C escape sequence.  STRING_PTR points to a variable
 428    containing a pointer to the string to parse.  That pointer
 429    is updated past the characters we use.  The value of the
 430    escape sequence is returned.
 431
 432    A negative value means the sequence \ newline was seen,
 433    which is supposed to be equivalent to nothing at all.
 434
 435    If \ is followed by a null character, we return a negative
 436    value and leave the string pointer pointing at the null character.
 437
 438    If \ is followed by 000, we return 0 and leave the string pointer
 439    after the zeros.  A value of 0 does not mean end of string.  */
 440
 441 static int
 442 parse_escape (string_ptr)
 443      const char **string_ptr;
 444 {
 445   register int c = *(*string_ptr)++;
 446   switch (c)
 447     {
 448     case 'a':
 449       return TARGET_BELL;
 450     case 'b':
 451       return TARGET_BS;
 452     case 'e':
 453       return 033;
 454     case 'f':
 455       return TARGET_FF;
 456     case 'n':
 457       return TARGET_NEWLINE;
 458     case 'r':
 459       return TARGET_CR;
 460     case 't':
 461       return TARGET_TAB;
 462     case 'v':
 463       return TARGET_VT;
 464     case '\n':
 465       return -2;
 466     case 0:
 467       (*string_ptr)--;
 468       return 0;
 469     case '^':
 470       c = *(*string_ptr)++;
 471       if (c == '\\')
 472         c = parse_escape (string_ptr);
 473       if (c == '?')
 474         return 0177;
 475       return (c & 0200) | (c & 037);
 476
 477     case '0':
 478     case '1':
 479     case '2':
 480     case '3':
 481     case '4':
 482     case '5':
 483     case '6':
 484     case '7':
 485       {
 486         register int i = c - '0';
 487         register int count = 0;
 488         while (++count < 3)
 489           {
 490             c = *(*string_ptr)++;
 491             if (c >= '0' && c <= '7')
 492               i = (i << 3) + c - '0';
 493             else
 494               {
 495                 (*string_ptr)--;
 496                 break;
 497               }
 498           }
 499         if ((i & ~((1 << CHAR_TYPE_SIZE) - 1)) != 0)
 500           {
 501             i &= (1 << CHAR_TYPE_SIZE) - 1;
 502             warning ("octal character constant does not fit in a byte");
 503           }
 504         return i;
 505       }
 506     case 'x':
 507       {
 508         register int i = 0;
 509         for (;;)
 510           {
 511             c = *(*string_ptr)++;
 512             if (c >= '0' && c <= '9')
 513               i = (i << 4) + c - '0';
 514             else if (c >= 'a' && c <= 'f')
 515               i = (i << 4) + c - 'a' + 10;
 516             else if (c >= 'A' && c <= 'F')
 517               i = (i << 4) + c - 'A' + 10;
 518             else
 519               {
 520                 (*string_ptr)--;
 521                 break;
 522               }
 523           }
 524         if ((i & ~((1 << BITS_PER_UNIT) - 1)) != 0)
 525           {
 526             i &= (1 << BITS_PER_UNIT) - 1;
 527             warning ("hex character constant does not fit in a byte");
 528           }
 529         return i;
 530       }
 531     default:
 532       return c;
 533     }
 534 }
 535
 536 static void
 537 yyerror (s)
 538      const char *s;
 539 {
 540   error ("%s", s);
 541   longjmp (parse_return_error, 1);
 542 }
 543 \f
 544 /* This page contains the entry point to this file.  */
 545
 546 /* Parse STRING as an expression, and complain if this fails
 547    to use up all of the contents of STRING.  */
 548 /* We do not support C comments.  They should be removed before
 549    this function is called.  */
 550
 551 int
 552 parse_c_expression (string)
 553      const char *string;
 554 {
 555   lexptr = string;
 556
 557   if (lexptr == 0 || *lexptr == 0) {
 558     error ("empty #if expression");
 559     return 0;                   /* don't include the #if group */
 560   }
 561
 562   /* if there is some sort of scanning error, just return 0 and assume
 563      the parsing routine has printed an error message somewhere.
 564      there is surely a better thing to do than this.     */
 565   if (setjmp (parse_return_error))
 566     return 0;
 567
 568   if (yyparse ())
 569     return 0;                   /* actually this is never reached
 570                                    the way things stand. */
 571   if (*lexptr)
 572     error ("Junk after end of expression.");
 573
 574   return expression_value;      /* set by yyparse () */
 575 }