scan.l revision 1.11
1 %{ 2 /* $NetBSD: scan.l,v 1.11 1998/02/22 15:40:40 christos Exp $ */ 3 4 /* 5 * Copyright (c) 1996 Christopher G. Demetriou. All Rights Reserved. 6 * Copyright (c) 1994, 1995 Jochen Pohl 7 * All Rights Reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 3. All advertising materials mentioning features or use of this software 18 * must display the following acknowledgement: 19 * This product includes software developed by Jochen Pohl for 20 * The NetBSD Project. 21 * 4. The name of the author may not be used to endorse or promote products 22 * derived from this software without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 25 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 26 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 27 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 28 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 29 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 30 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 31 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 32 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 33 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 34 */ 35 36 #include <sys/cdefs.h> 37 #ifndef lint 38 __RCSID("$NetBSD: scan.l,v 1.11 1998/02/22 15:40:40 christos Exp $"); 39 #endif 40 41 #include <stdlib.h> 42 #include <string.h> 43 #include <limits.h> 44 #include <float.h> 45 #include <ctype.h> 46 #include <errno.h> 47 #include <math.h> 48 #include <err.h> 49 50 #include "lint1.h" 51 #include "y.tab.h" 52 53 #define CHAR_MASK (~(~0 << CHAR_BIT)) 54 #define YY_NO_UNPUT 55 56 /* XXX declaration of strtouq() is missing in stdlib.h ? */ 57 extern u_quad_t strtouq __P((const char *, char **, int)); 58 59 /* Current position (its also updated when an included file is parsed) */ 60 pos_t curr_pos = { 1, "", 0 }; 61 62 /* 63 * Current position in C source (not updated when an included file is 64 * parsed). 65 */ 66 pos_t csrc_pos = { 1, "", 0 }; 67 68 static void incline __P((void)); 69 static void badchar __P((int)); 70 static sbuf_t *allocsb __P((void)); 71 static void freesb __P((sbuf_t *)); 72 static int inpc __P((void)); 73 static int hash __P((const char *)); 74 static sym_t *search __P((sbuf_t *)); 75 static int name __P((void)); 76 static int keyw __P((sym_t *)); 77 static int icon __P((int)); 78 static int fcon __P((void)); 79 static int operator __P((int, op_t)); 80 static int ccon __P((void)); 81 static int wccon __P((void)); 82 static int getescc __P((int)); 83 static void directive __P((void)); 84 static void comment __P((void)); 85 static int string __P((void)); 86 static int wcstrg __P((void)); 87 88 %} 89 90 L [_A-Za-z] 91 D [0-9] 92 NZD [1-9] 93 OD [0-7] 94 HD [0-9A-Fa-f] 95 EX ([eE][+-]?[0-9]+) 96 97 %% 98 99 {L}({L}|{D})* return (name()); 100 0{OD}*[lLuU]* return (icon(8)); 101 {NZD}{D}*[lLuU]* return (icon(10)); 102 0[xX]{HD}+[lLuU]* return (icon(16)); 103 {D}+\.{D}*{EX}?[fFlL]? | 104 {D}+{EX}[fFlL]? | 105 \.{D}+{EX}?[fFlL]? return (fcon()); 106 "=" return (operator(T_ASSIGN, ASSIGN)); 107 "*=" return (operator(T_OPASS, MULASS)); 108 "/=" return (operator(T_OPASS, DIVASS)); 109 "%=" return (operator(T_OPASS, MODASS)); 110 "+=" return (operator(T_OPASS, ADDASS)); 111 "-=" return (operator(T_OPASS, SUBASS)); 112 "<<=" return (operator(T_OPASS, SHLASS)); 113 ">>=" return (operator(T_OPASS, SHRASS)); 114 "&=" return (operator(T_OPASS, ANDASS)); 115 "^=" return (operator(T_OPASS, XORASS)); 116 "|=" return (operator(T_OPASS, ORASS)); 117 "||" return (operator(T_LOGOR, LOGOR)); 118 "&&" return (operator(T_LOGAND, LOGAND)); 119 "|" return (operator(T_OR, OR)); 120 "&" return (operator(T_AND, AND)); 121 "^" return (operator(T_XOR, XOR)); 122 "==" return (operator(T_EQOP, EQ)); 123 "!=" return (operator(T_EQOP, NE)); 124 "<" return (operator(T_RELOP, LT)); 125 ">" return (operator(T_RELOP, GT)); 126 "<=" return (operator(T_RELOP, LE)); 127 ">=" return (operator(T_RELOP, GE)); 128 "<<" return (operator(T_SHFTOP, SHL)); 129 ">>" return (operator(T_SHFTOP, SHR)); 130 "++" return (operator(T_INCDEC, INC)); 131 "--" return (operator(T_INCDEC, DEC)); 132 "->" return (operator(T_STROP, ARROW)); 133 "." return (operator(T_STROP, POINT)); 134 "+" return (operator(T_ADDOP, PLUS)); 135 "-" return (operator(T_ADDOP, MINUS)); 136 "*" return (operator(T_MULT, MULT)); 137 "/" return (operator(T_DIVOP, DIV)); 138 "%" return (operator(T_DIVOP, MOD)); 139 "!" return (operator(T_UNOP, NOT)); 140 "~" return (operator(T_UNOP, COMPL)); 141 "\"" return (string()); 142 "L\"" return (wcstrg()); 143 ";" return (T_SEMI); 144 "{" return (T_LBRACE); 145 "}" return (T_RBRACE); 146 "," return (T_COMMA); 147 ":" return (T_COLON); 148 "?" return (T_QUEST); 149 "[" return (T_LBRACK); 150 "]" return (T_RBRACK); 151 "(" return (T_LPARN); 152 ")" return (T_RPARN); 153 "..." return (T_ELLIPSE); 154 "'" return (ccon()); 155 "L'" return (wccon()); 156 ^#.*$ directive(); 157 \n incline(); 158 \t|" "|\f|\v ; 159 "/*" comment(); 160 . badchar(yytext[0]); 161 162 %% 163 164 static void 165 incline() 166 { 167 curr_pos.p_line++; 168 curr_pos.p_uniq = 0; 169 if (curr_pos.p_file == csrc_pos.p_file) { 170 csrc_pos.p_line++; 171 csrc_pos.p_uniq = 0; 172 } 173 } 174 175 static void 176 badchar(c) 177 int c; 178 { 179 /* unknown character \%o */ 180 error(250, c); 181 } 182 183 /* 184 * Keywords. 185 * During initialisation they are written to the symbol table. 186 */ 187 static struct kwtab { 188 const char *kw_name; /* keyword */ 189 int kw_token; /* token returned by yylex() */ 190 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 191 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 192 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 193 u_int kw_stdc : 1; /* STDC keyword */ 194 u_int kw_gcc : 1; /* GCC keyword */ 195 } kwtab[] = { 196 { "asm", T_ASM, 0, 0, 0, 0, 1 }, 197 { "__asm", T_ASM, 0, 0, 0, 0, 0 }, 198 { "__asm__", T_ASM, 0, 0, 0, 0, 0 }, 199 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 }, 200 { "break", T_BREAK, 0, 0, 0, 0, 0 }, 201 { "case", T_CASE, 0, 0, 0, 0, 0 }, 202 { "char", T_TYPE, 0, CHAR, 0, 0, 0 }, 203 { "const", T_QUAL, 0, 0, CONST, 1, 0 }, 204 { "__const__", T_QUAL, 0, 0, CONST, 0, 0 }, 205 { "__const", T_QUAL, 0, 0, CONST, 0, 0 }, 206 { "continue", T_CONTINUE, 0, 0, 0, 0, 0 }, 207 { "default", T_DEFAULT, 0, 0, 0, 0, 0 }, 208 { "do", T_DO, 0, 0, 0, 0, 0 }, 209 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 }, 210 { "else", T_ELSE, 0, 0, 0, 0, 0 }, 211 { "enum", T_ENUM, 0, 0, 0, 0, 0 }, 212 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 }, 213 { "float", T_TYPE, 0, FLOAT, 0, 0, 0 }, 214 { "for", T_FOR, 0, 0, 0, 0, 0 }, 215 { "goto", T_GOTO, 0, 0, 0, 0, 0 }, 216 { "if", T_IF, 0, 0, 0, 0, 0 }, 217 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 }, 218 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 }, 219 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 }, 220 { "int", T_TYPE, 0, INT, 0, 0, 0 }, 221 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0 }, 222 { "long", T_TYPE, 0, LONG, 0, 0, 0 }, 223 { "register", T_SCLASS, REG, 0, 0, 0, 0 }, 224 { "return", T_RETURN, 0, 0, 0, 0, 0 }, 225 { "short", T_TYPE, 0, SHORT, 0, 0, 0 }, 226 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 }, 227 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 }, 228 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 }, 229 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 }, 230 { "static", T_SCLASS, STATIC, 0, 0, 0, 0 }, 231 { "struct", T_SOU, 0, STRUCT, 0, 0, 0 }, 232 { "switch", T_SWITCH, 0, 0, 0, 0, 0 }, 233 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 }, 234 { "union", T_SOU, 0, UNION, 0, 0, 0 }, 235 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 }, 236 { "void", T_TYPE, 0, VOID, 0, 0, 0 }, 237 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 }, 238 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 239 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 240 { "while", T_WHILE, 0, 0, 0, 0, 0 }, 241 { NULL, 0, 0, 0, 0, 0, 0 } 242 }; 243 244 /* Symbol table */ 245 static sym_t *symtab[HSHSIZ1]; 246 247 /* bit i of the entry with index i is set */ 248 u_quad_t qbmasks[sizeof(u_quad_t) * CHAR_BIT]; 249 250 /* least significant i bits are set in the entry with index i */ 251 u_quad_t qlmasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 252 253 /* least significant i bits are not set in the entry with index i */ 254 u_quad_t qumasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 255 256 /* free list for sbuf structures */ 257 static sbuf_t *sbfrlst; 258 259 /* Typ of next expected symbol */ 260 symt_t symtyp; 261 262 263 /* 264 * All keywords are written to the symbol table. This saves us looking 265 * in a extra table for each name we found. 266 */ 267 void 268 initscan() 269 { 270 struct kwtab *kw; 271 sym_t *sym; 272 int h, i; 273 u_quad_t uq; 274 275 for (kw = kwtab; kw->kw_name != NULL; kw++) { 276 if (kw->kw_stdc && tflag) 277 continue; 278 if (kw->kw_gcc && !gflag) 279 continue; 280 sym = getblk(sizeof (sym_t)); 281 sym->s_name = kw->kw_name; 282 sym->s_keyw = 1; 283 sym->s_value.v_quad = kw->kw_token; 284 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 285 sym->s_tspec = kw->kw_tspec; 286 } else if (kw->kw_token == T_SCLASS) { 287 sym->s_scl = kw->kw_scl; 288 } else if (kw->kw_token == T_QUAL) { 289 sym->s_tqual = kw->kw_tqual; 290 } 291 h = hash(sym->s_name); 292 if ((sym->s_link = symtab[h]) != NULL) 293 symtab[h]->s_rlink = &sym->s_link; 294 (symtab[h] = sym)->s_rlink = &symtab[h]; 295 } 296 297 /* initialize bit-masks for quads */ 298 for (i = 0; i < sizeof (u_quad_t) * CHAR_BIT; i++) { 299 qbmasks[i] = (u_quad_t)1 << i; 300 uq = ~(u_quad_t)0 << i; 301 qumasks[i] = uq; 302 qlmasks[i] = ~uq; 303 } 304 qumasks[i] = 0; 305 qlmasks[i] = ~(u_quad_t)0; 306 } 307 308 /* 309 * Get a free sbuf structure, if possible from the free list 310 */ 311 static sbuf_t * 312 allocsb() 313 { 314 sbuf_t *sb; 315 316 if ((sb = sbfrlst) != NULL) { 317 sbfrlst = sb->sb_nxt; 318 } else { 319 sb = xmalloc(sizeof (sbuf_t)); 320 } 321 (void)memset(sb, 0, sizeof (sb)); 322 return (sb); 323 } 324 325 /* 326 * Put a sbuf structure to the free list 327 */ 328 static void 329 freesb(sb) 330 sbuf_t *sb; 331 { 332 sb->sb_nxt = sbfrlst; 333 sbfrlst = sb; 334 } 335 336 /* 337 * Read a character and ensure that it is positive (except EOF). 338 * Increment line count(s) if necessary. 339 */ 340 static int 341 inpc() 342 { 343 int c; 344 345 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 346 incline(); 347 return (c); 348 } 349 350 static int 351 hash(s) 352 const char *s; 353 { 354 u_int v; 355 const u_char *us; 356 357 v = 0; 358 for (us = (const u_char *)s; *us != '\0'; us++) { 359 v = (v << sizeof (v)) + *us; 360 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 361 } 362 return (v % HSHSIZ1); 363 } 364 365 /* 366 * Lex has found a letter followed by zero or more letters or digits. 367 * It looks for a symbol in the symbol table with the same name. This 368 * symbol must either be a keyword or a symbol of the type required by 369 * symtyp (label, member, tag, ...). 370 * 371 * If it is a keyword, the token is returned. In some cases it is described 372 * more deeply by data written to yylval. 373 * 374 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 375 * is stored in yylval. This struct contains the name of the symbol, it's 376 * length and hash value. If there is already a symbol of the same name 377 * and type in the symbol table, the sbuf struct also contains a pointer 378 * to the symbol table entry. 379 */ 380 static int 381 name() 382 { 383 char *s; 384 sbuf_t *sb; 385 sym_t *sym; 386 int tok; 387 388 sb = allocsb(); 389 sb->sb_name = yytext; 390 sb->sb_len = yyleng; 391 sb->sb_hash = hash(yytext); 392 393 if ((sym = search(sb)) != NULL && sym->s_keyw) { 394 freesb(sb); 395 return (keyw(sym)); 396 } 397 398 sb->sb_sym = sym; 399 400 if (sym != NULL) { 401 if (blklev < sym->s_blklev) 402 lerror("name() 1"); 403 sb->sb_name = sym->s_name; 404 sb->sb_len = strlen(sym->s_name); 405 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 406 } else { 407 s = getblk(yyleng + 1); 408 (void)memcpy(s, yytext, yyleng + 1); 409 sb->sb_name = s; 410 sb->sb_len = yyleng; 411 tok = T_NAME; 412 } 413 414 yylval.y_sb = sb; 415 return (tok); 416 } 417 418 static sym_t * 419 search(sb) 420 sbuf_t *sb; 421 { 422 sym_t *sym; 423 424 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 425 if (strcmp(sym->s_name, sb->sb_name) == 0) { 426 if (sym->s_keyw || sym->s_kind == symtyp) 427 return (sym); 428 } 429 } 430 431 return (NULL); 432 } 433 434 static int 435 keyw(sym) 436 sym_t *sym; 437 { 438 int t; 439 440 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 441 yylval.y_scl = sym->s_scl; 442 } else if (t == T_TYPE || t == T_SOU) { 443 yylval.y_tspec = sym->s_tspec; 444 } else if (t == T_QUAL) { 445 yylval.y_tqual = sym->s_tqual; 446 } 447 return (t); 448 } 449 450 /* 451 * Convert a string representing an integer into internal representation. 452 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 453 */ 454 static int 455 icon(base) 456 int base; 457 { 458 int l_suffix, u_suffix; 459 int len; 460 const char *cp; 461 char c, *eptr; 462 tspec_t typ; 463 u_long ul = 0; 464 u_quad_t uq = 0; 465 int ansiu; 466 static tspec_t contypes[2][3] = { 467 { INT, LONG, QUAD }, 468 { UINT, ULONG, UQUAD } 469 }; 470 471 cp = yytext; 472 len = yyleng; 473 474 /* skip 0x */ 475 if (base == 16) { 476 cp += 2; 477 len -= 2; 478 } 479 480 /* read suffixes */ 481 l_suffix = u_suffix = 0; 482 for ( ; ; ) { 483 if ((c = cp[len - 1]) == 'l' || c == 'L') { 484 l_suffix++; 485 } else if (c == 'u' || c == 'U') { 486 u_suffix++; 487 } else { 488 break; 489 } 490 len--; 491 } 492 if (l_suffix > 2 || u_suffix > 1) { 493 /* malformed integer constant */ 494 warning(251); 495 if (l_suffix > 2) 496 l_suffix = 2; 497 if (u_suffix > 1) 498 u_suffix = 1; 499 } 500 if (tflag && u_suffix != 0) { 501 /* suffix U is illegal in traditional C */ 502 warning(97); 503 } 504 typ = contypes[u_suffix][l_suffix]; 505 506 errno = 0; 507 if (l_suffix < 2) { 508 ul = strtoul(cp, &eptr, base); 509 } else { 510 uq = strtouq(cp, &eptr, base); 511 } 512 if (eptr != cp + len) 513 lerror("icon() 1"); 514 if (errno != 0) 515 /* integer constant out of range */ 516 warning(252); 517 518 /* 519 * If the value is to big for the current type, we must choose 520 * another type. 521 */ 522 ansiu = 0; 523 switch (typ) { 524 case INT: 525 if (ul <= INT_MAX) { 526 /* ok */ 527 } else if (ul <= (unsigned)UINT_MAX && base != 10) { 528 typ = UINT; 529 } else if (ul <= LONG_MAX) { 530 typ = LONG; 531 } else { 532 typ = ULONG; 533 } 534 if (typ == UINT || typ == ULONG) { 535 if (tflag) { 536 typ = LONG; 537 } else if (!sflag) { 538 /* 539 * Remember that the constant is unsigned 540 * only in ANSI C 541 */ 542 ansiu = 1; 543 } 544 } 545 break; 546 case UINT: 547 if (ul > (u_int)UINT_MAX) 548 typ = ULONG; 549 break; 550 case LONG: 551 if (ul > LONG_MAX && !tflag) { 552 typ = ULONG; 553 if (!sflag) 554 ansiu = 1; 555 } 556 break; 557 case QUAD: 558 if (uq > QUAD_MAX && !tflag) { 559 typ = UQUAD; 560 if (!sflag) 561 ansiu = 1; 562 } 563 break; 564 /* LINTED (enumeration values not handled in switch) */ 565 case STRUCT: 566 case VOID: 567 case LDOUBLE: 568 case FUNC: 569 case ARRAY: 570 case PTR: 571 case ENUM: 572 case UNION: 573 case SIGNED: 574 case NOTSPEC: 575 case DOUBLE: 576 case FLOAT: 577 case UQUAD: 578 case ULONG: 579 case USHORT: 580 case SHORT: 581 case UCHAR: 582 case SCHAR: 583 case CHAR: 584 case UNSIGN: 585 break; 586 } 587 588 if (typ != QUAD && typ != UQUAD) { 589 if (isutyp(typ)) { 590 uq = ul; 591 } else { 592 uq = (quad_t)(long)ul; 593 } 594 } 595 596 uq = (u_quad_t)xsign((quad_t)uq, typ, -1); 597 598 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 599 yylval.y_val->v_ansiu = ansiu; 600 yylval.y_val->v_quad = (quad_t)uq; 601 602 return (T_CON); 603 } 604 605 /* 606 * Returns 1 if t is a signed type and the value is negative. 607 * 608 * len is the number of significant bits. If len is -1, len is set 609 * to the width of type t. 610 */ 611 int 612 sign(q, t, len) 613 quad_t q; 614 tspec_t t; 615 int len; 616 { 617 if (t == PTR || isutyp(t)) 618 return (0); 619 return (msb(q, t, len)); 620 } 621 622 int 623 msb(q, t, len) 624 quad_t q; 625 tspec_t t; 626 int len; 627 { 628 if (len <= 0) 629 len = size(t); 630 return ((q & qbmasks[len - 1]) != 0); 631 } 632 633 /* 634 * Extends the sign of q. 635 */ 636 quad_t 637 xsign(q, t, len) 638 quad_t q; 639 tspec_t t; 640 int len; 641 { 642 if (len <= 0) 643 len = size(t); 644 645 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 646 q &= qlmasks[len]; 647 } else { 648 q |= qumasks[len]; 649 } 650 return (q); 651 } 652 653 /* 654 * Convert a string representing a floating point value into its interal 655 * representation. Type and value are returned in yylval. fcon() 656 * (and yylex()) returns T_CON. 657 * XXX Currently it is not possible to convert constants of type 658 * long double which are greater then DBL_MAX. 659 */ 660 static int 661 fcon() 662 { 663 const char *cp; 664 int len; 665 tspec_t typ; 666 char c, *eptr; 667 double d; 668 float f = 0; 669 670 cp = yytext; 671 len = yyleng; 672 673 if ((c = cp[len - 1]) == 'f' || c == 'F') { 674 typ = FLOAT; 675 len--; 676 } else if (c == 'l' || c == 'L') { 677 typ = LDOUBLE; 678 len--; 679 } else { 680 typ = DOUBLE; 681 } 682 683 if (tflag && typ != DOUBLE) { 684 /* suffixes F and L are illegal in traditional C */ 685 warning(98); 686 } 687 688 errno = 0; 689 d = strtod(cp, &eptr); 690 if (eptr != cp + len) 691 lerror("fcon() 1"); 692 if (errno != 0) 693 /* floating-point constant out of range */ 694 warning(248); 695 696 if (typ == FLOAT) { 697 f = (float)d; 698 if (isinf(f)) { 699 /* floating-point constant out of range */ 700 warning(248); 701 f = f > 0 ? FLT_MAX : -FLT_MAX; 702 } 703 } 704 705 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 706 if (typ == FLOAT) { 707 yylval.y_val->v_ldbl = f; 708 } else { 709 yylval.y_val->v_ldbl = d; 710 } 711 712 return (T_CON); 713 } 714 715 static int 716 operator(t, o) 717 int t; 718 op_t o; 719 { 720 yylval.y_op = o; 721 return (t); 722 } 723 724 /* 725 * Called if lex found a leading \'. 726 */ 727 static int 728 ccon() 729 { 730 int n, val, c; 731 char cv; 732 733 n = 0; 734 val = 0; 735 while ((c = getescc('\'')) >= 0) { 736 val = (val << CHAR_BIT) + c; 737 n++; 738 } 739 if (c == -2) { 740 /* unterminated character constant */ 741 error(253); 742 } else { 743 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 744 /* too many characters in character constant */ 745 error(71); 746 } else if (n > 1) { 747 /* multi-character character constant */ 748 warning(294); 749 } else if (n == 0) { 750 /* empty character constant */ 751 error(73); 752 } 753 } 754 if (n == 1) { 755 cv = (char)val; 756 val = cv; 757 } 758 759 yylval.y_val = xcalloc(1, sizeof (val_t)); 760 yylval.y_val->v_tspec = INT; 761 yylval.y_val->v_quad = val; 762 763 return (T_CON); 764 } 765 766 /* 767 * Called if lex found a leading L\' 768 */ 769 static int 770 wccon() 771 { 772 static char buf[MB_LEN_MAX + 1]; 773 int i, c; 774 wchar_t wc; 775 776 i = 0; 777 while ((c = getescc('\'')) >= 0) { 778 if (i < MB_CUR_MAX) 779 buf[i] = (char)c; 780 i++; 781 } 782 783 wc = 0; 784 785 if (c == -2) { 786 /* unterminated character constant */ 787 error(253); 788 } else if (c == 0) { 789 /* empty character constant */ 790 error(73); 791 } else { 792 if (i > MB_CUR_MAX) { 793 i = MB_CUR_MAX; 794 /* too many characters in character constant */ 795 error(71); 796 } else { 797 buf[i] = '\0'; 798 (void)mbtowc(NULL, NULL, 0); 799 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 800 /* invalid multibyte character */ 801 error(291); 802 } 803 } 804 805 yylval.y_val = xcalloc(1, sizeof (val_t)); 806 yylval.y_val->v_tspec = WCHAR; 807 yylval.y_val->v_quad = wc; 808 809 return (T_CON); 810 } 811 812 /* 813 * Read a character which is part of a character constant or of a string 814 * and handle escapes. 815 * 816 * The Argument is the character which delimits the character constant or 817 * string. 818 * 819 * Returns -1 if the end of the character constant or string is reached, 820 * -2 if the EOF is reached, and the charachter otherwise. 821 */ 822 static int 823 getescc(d) 824 int d; 825 { 826 static int pbc = -1; 827 int n, c, v; 828 829 if (pbc == -1) { 830 c = inpc(); 831 } else { 832 c = pbc; 833 pbc = -1; 834 } 835 if (c == d) 836 return (-1); 837 switch (c) { 838 case '\n': 839 /* newline in string or char constant */ 840 error(254); 841 return (-2); 842 case EOF: 843 return (-2); 844 case '\\': 845 switch (c = inpc()) { 846 case '"': 847 if (tflag && d == '\'') 848 /* \" inside character constant undef. ... */ 849 warning(262); 850 return ('"'); 851 case '\'': 852 return ('\''); 853 case '?': 854 if (tflag) 855 /* \? undefined in traditional C */ 856 warning(263); 857 return ('?'); 858 case '\\': 859 return ('\\'); 860 case 'a': 861 if (tflag) 862 /* \a undefined in traditional C */ 863 warning(81); 864 #ifdef __STDC__ 865 return ('\a'); 866 #else 867 return ('\007'); 868 #endif 869 case 'b': 870 return ('\b'); 871 case 'f': 872 return ('\f'); 873 case 'n': 874 return ('\n'); 875 case 'r': 876 return ('\r'); 877 case 't': 878 return ('\t'); 879 case 'v': 880 if (tflag) 881 /* \v undefined in traditional C */ 882 warning(264); 883 #ifdef __STDC__ 884 return ('\v'); 885 #else 886 return ('\013'); 887 #endif 888 case '8': case '9': 889 /* bad octal digit %c */ 890 warning(77, c); 891 /* FALLTHROUGH */ 892 case '0': case '1': case '2': case '3': 893 case '4': case '5': case '6': case '7': 894 n = 3; 895 v = 0; 896 do { 897 v = (v << 3) + (c - '0'); 898 c = inpc(); 899 } while (--n && isdigit(c) && (tflag || c <= '7')); 900 if (tflag && n > 0 && isdigit(c)) 901 /* bad octal digit %c */ 902 warning(77, c); 903 pbc = c; 904 if (v > UCHAR_MAX) { 905 /* character escape does not fit in char. */ 906 warning(76); 907 v &= CHAR_MASK; 908 } 909 return (v); 910 case 'x': 911 if (tflag) 912 /* \x undefined in traditional C */ 913 warning(82); 914 v = 0; 915 n = 0; 916 while ((c = inpc()) >= 0 && isxdigit(c)) { 917 c = isdigit(c) ? 918 c - '0' : toupper(c) - 'A' + 10; 919 v = (v << 4) + c; 920 if (n >= 0) { 921 if ((v & ~CHAR_MASK) != 0) { 922 /* overflow in hex escape */ 923 warning(75); 924 n = -1; 925 } else { 926 n++; 927 } 928 } 929 } 930 pbc = c; 931 if (n == 0) { 932 /* no hex digits follow \x */ 933 error(74); 934 } if (n == -1) { 935 v &= CHAR_MASK; 936 } 937 return (v); 938 case '\n': 939 return (getescc(d)); 940 case EOF: 941 return (-2); 942 default: 943 if (isprint(c)) { 944 /* dubious escape \%c */ 945 warning(79, c); 946 } else { 947 /* dubious escape \%o */ 948 warning(80, c); 949 } 950 } 951 } 952 return (c); 953 } 954 955 /* 956 * Called for preprocessor directives. Currently implemented are: 957 * # lineno 958 * # lineno "filename" 959 */ 960 static void 961 directive() 962 { 963 const char *cp, *fn; 964 char c, *eptr; 965 size_t fnl; 966 long ln; 967 static int first = 1; 968 969 /* Go to first non-whitespace after # */ 970 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) ; 971 972 if (!isdigit(c)) { 973 error: 974 /* undefined or invalid # directive */ 975 warning(255); 976 return; 977 } 978 ln = strtol(--cp, &eptr, 10); 979 if (cp == eptr) 980 goto error; 981 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 982 goto error; 983 while ((c = *cp++) == ' ' || c == '\t') ; 984 if (c != '\0') { 985 if (c != '"') 986 goto error; 987 fn = cp; 988 while ((c = *cp) != '"' && c != '\0') 989 cp++; 990 if (c != '"') 991 goto error; 992 if ((fnl = cp++ - fn) > PATH_MAX) 993 goto error; 994 while ((c = *cp++) == ' ' || c == '\t') ; 995 #if 0 996 if (c != '\0') 997 warning("extra character(s) after directive"); 998 #endif 999 1000 /* empty string means stdin */ 1001 if (fnl == 0) { 1002 fn = "{standard input}"; 1003 fnl = 16; /* strlen (fn) */ 1004 } 1005 curr_pos.p_file = fnnalloc(fn, fnl); 1006 /* 1007 * If this is the first directive, the name is the name 1008 * of the C source file as specified at the command line. 1009 * It is written to the output file. 1010 */ 1011 if (first) { 1012 csrc_pos.p_file = curr_pos.p_file; 1013 outsrc(curr_pos.p_file); 1014 first = 0; 1015 } 1016 } 1017 curr_pos.p_line = (int)ln - 1; 1018 curr_pos.p_uniq = 0; 1019 if (curr_pos.p_file == csrc_pos.p_file) { 1020 csrc_pos.p_line = (int)ln - 1; 1021 csrc_pos.p_uniq = 0; 1022 } 1023 } 1024 1025 /* 1026 * Handle lint comments. Following comments are currently understood: 1027 * ARGSUSEDn 1028 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 1029 * FALLTHRU FALLTHROUGH 1030 * LINTLIBRARY 1031 * LINTED NOSTRICT 1032 * LONGLONG 1033 * NOTREACHED 1034 * PRINTFLIKEn 1035 * PROTOLIB 1036 * SCANFLIKEn 1037 * VARARGSn 1038 * If one of this comments is recognized, the arguments, if any, are 1039 * parsed and a function which handles this comment is called. 1040 */ 1041 static void 1042 comment() 1043 { 1044 int c, lc; 1045 static struct { 1046 const char *keywd; 1047 int arg; 1048 void (*func) __P((int)); 1049 } keywtab[] = { 1050 { "ARGSUSED", 1, argsused }, 1051 { "CONSTCOND", 0, constcond }, 1052 { "CONSTANTCOND", 0, constcond }, 1053 { "CONSTANTCONDITION", 0, constcond }, 1054 { "FALLTHRU", 0, fallthru }, 1055 { "FALLTHROUGH", 0, fallthru }, 1056 { "LINTLIBRARY", 0, lintlib }, 1057 { "LINTED", 0, linted }, 1058 { "LONGLONG", 0, longlong }, 1059 { "NOSTRICT", 0, linted }, 1060 { "NOTREACHED", 0, notreach }, 1061 { "PRINTFLIKE", 1, printflike }, 1062 { "PROTOLIB", 1, protolib }, 1063 { "SCANFLIKE", 1, scanflike }, 1064 { "VARARGS", 1, varargs }, 1065 }; 1066 char keywd[32]; 1067 char arg[32]; 1068 int l, i, a; 1069 int eoc; 1070 1071 eoc = 0; 1072 1073 /* Skip white spaces after the start of the comment */ 1074 while ((c = inpc()) != EOF && isspace(c)) ; 1075 1076 /* Read the potential keyword to keywd */ 1077 l = 0; 1078 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1079 keywd[l++] = (char)c; 1080 c = inpc(); 1081 } 1082 keywd[l] = '\0'; 1083 1084 /* look for the keyword */ 1085 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1086 if (strcmp(keywtab[i].keywd, keywd) == 0) 1087 break; 1088 } 1089 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1090 goto skip_rest; 1091 1092 /* skip white spaces after the keyword */ 1093 while (c != EOF && isspace(c)) 1094 c = inpc(); 1095 1096 /* read the argument, if the keyword accepts one and there is one */ 1097 l = 0; 1098 if (keywtab[i].arg) { 1099 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1100 arg[l++] = (char)c; 1101 c = inpc(); 1102 } 1103 } 1104 arg[l] = '\0'; 1105 a = l != 0 ? atoi(arg) : -1; 1106 1107 /* skip white spaces after the argument */ 1108 while (c != EOF && isspace(c)) 1109 c = inpc(); 1110 1111 if (c != '*' || (c = inpc()) != '/') { 1112 if (keywtab[i].func != linted) 1113 /* extra characters in lint comment */ 1114 warning(257); 1115 } else { 1116 /* 1117 * remember that we have already found the end of the 1118 * comment 1119 */ 1120 eoc = 1; 1121 } 1122 1123 if (keywtab[i].func != NULL) 1124 (*keywtab[i].func)(a); 1125 1126 skip_rest: 1127 while (!eoc) { 1128 lc = c; 1129 if ((c = inpc()) == EOF) { 1130 /* unterminated comment */ 1131 error(256); 1132 break; 1133 } 1134 if (lc == '*' && c == '/') 1135 eoc = 1; 1136 } 1137 } 1138 1139 /* 1140 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. 1141 * clrwflgs() is called after function definitions and global and 1142 * local declarations and definitions. It is also called between 1143 * the controlling expression and the body of control statements 1144 * (if, switch, for, while). 1145 */ 1146 void 1147 clrwflgs() 1148 { 1149 nowarn = 0; 1150 quadflg = 0; 1151 ccflg = 0; 1152 } 1153 1154 /* 1155 * Strings are stored in a dynamically alloceted buffer and passed 1156 * in yylval.y_xstrg to the parser. The parser or the routines called 1157 * by the parser are responsible for freeing this buffer. 1158 */ 1159 static int 1160 string() 1161 { 1162 u_char *s; 1163 int c; 1164 size_t len, max; 1165 strg_t *strg; 1166 1167 s = xmalloc(max = 64); 1168 1169 len = 0; 1170 while ((c = getescc('"')) >= 0) { 1171 /* +1 to reserve space for a trailing NUL character */ 1172 if (len + 1 == max) 1173 s = xrealloc(s, max *= 2); 1174 s[len++] = (char)c; 1175 } 1176 s[len] = '\0'; 1177 if (c == -2) 1178 /* unterminated string constant */ 1179 error(258); 1180 1181 strg = xcalloc(1, sizeof (strg_t)); 1182 strg->st_tspec = CHAR; 1183 strg->st_len = len; 1184 strg->st_cp = s; 1185 1186 yylval.y_strg = strg; 1187 return (T_STRING); 1188 } 1189 1190 static int 1191 wcstrg() 1192 { 1193 char *s; 1194 int c, i, n, wi; 1195 size_t len, max, wlen; 1196 wchar_t *ws; 1197 strg_t *strg; 1198 1199 s = xmalloc(max = 64); 1200 len = 0; 1201 while ((c = getescc('"')) >= 0) { 1202 /* +1 to save space for a trailing NUL character */ 1203 if (len + 1 >= max) 1204 s = xrealloc(s, max *= 2); 1205 s[len++] = (char)c; 1206 } 1207 s[len] = '\0'; 1208 if (c == -2) 1209 /* unterminated string constant */ 1210 error(258); 1211 1212 /* get length of wide character string */ 1213 (void)mblen(NULL, 0); 1214 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1215 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1216 /* invalid multibyte character */ 1217 error(291); 1218 break; 1219 } 1220 if (n == 0) 1221 n = 1; 1222 } 1223 1224 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1225 1226 /* convert from multibyte to wide char */ 1227 (void)mbtowc(NULL, NULL, 0); 1228 for (i = 0, wi = 0; i < len; i += n, wi++) { 1229 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1230 break; 1231 if (n == 0) 1232 n = 1; 1233 } 1234 ws[wi] = 0; 1235 free(s); 1236 1237 strg = xcalloc(1, sizeof (strg_t)); 1238 strg->st_tspec = WCHAR; 1239 strg->st_len = wlen; 1240 strg->st_wcp = ws; 1241 1242 yylval.y_strg = strg; 1243 return (T_STRING); 1244 } 1245 1246 /* 1247 * As noted above the scanner does not create new symbol table entries 1248 * for symbols it cannot find in the symbol table. This is to avoid 1249 * putting undeclared symbols into the symbol table if a syntax error 1250 * occurs. 1251 * 1252 * getsym() is called as soon as it is probably ok to put the symbol to 1253 * the symbol table. This does not mean that it is not possible that 1254 * symbols are put to the symbol table which are than not completely 1255 * declared due to syntax errors. To avoid too many problems in this 1256 * case symbols get type int in getsym(). 1257 * 1258 * XXX calls to getsym() should be delayed until decl1*() is called 1259 */ 1260 sym_t * 1261 getsym(sb) 1262 sbuf_t *sb; 1263 { 1264 dinfo_t *di; 1265 char *s; 1266 sym_t *sym; 1267 1268 sym = sb->sb_sym; 1269 1270 /* 1271 * During member declaration it is possible that name() looked 1272 * for symbols of type FVFT, although it should have looked for 1273 * symbols of type FTAG. Same can happen for labels. Both cases 1274 * are compensated here. 1275 */ 1276 if (symtyp == FMOS || symtyp == FLAB) { 1277 if (sym == NULL || sym->s_kind == FVFT) 1278 sym = search(sb); 1279 } 1280 1281 if (sym != NULL) { 1282 if (sym->s_kind != symtyp) 1283 lerror("storesym() 1"); 1284 symtyp = FVFT; 1285 freesb(sb); 1286 return (sym); 1287 } 1288 1289 /* create a new symbol table entry */ 1290 1291 /* labels must always be allocated at level 1 (outhermost block) */ 1292 if (symtyp == FLAB) { 1293 sym = getlblk(1, sizeof (sym_t)); 1294 s = getlblk(1, sb->sb_len + 1); 1295 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1296 sym->s_name = s; 1297 sym->s_blklev = 1; 1298 di = dcs; 1299 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1300 di = di->d_nxt; 1301 if (di->d_ctx != AUTO) 1302 lerror("storesym() 2"); 1303 } else { 1304 sym = getblk(sizeof (sym_t)); 1305 sym->s_name = sb->sb_name; 1306 sym->s_blklev = blklev; 1307 di = dcs; 1308 } 1309 1310 UNIQUE_CURR_POS(sym->s_dpos); 1311 if ((sym->s_kind = symtyp) != FLAB) 1312 sym->s_type = gettyp(INT); 1313 1314 symtyp = FVFT; 1315 1316 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1317 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1318 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1319 1320 *di->d_ldlsym = sym; 1321 di->d_ldlsym = &sym->s_dlnxt; 1322 1323 freesb(sb); 1324 return (sym); 1325 } 1326 1327 /* 1328 * Remove a symbol forever from the symbol table. s_blklev 1329 * is set to -1 to avoid that the symbol will later be put 1330 * back to the symbol table. 1331 */ 1332 void 1333 rmsym(sym) 1334 sym_t *sym; 1335 { 1336 if ((*sym->s_rlink = sym->s_link) != NULL) 1337 sym->s_link->s_rlink = sym->s_rlink; 1338 sym->s_blklev = -1; 1339 sym->s_link = NULL; 1340 } 1341 1342 /* 1343 * Remove a list of symbols declared at one level from the symbol 1344 * table. 1345 */ 1346 void 1347 rmsyms(syms) 1348 sym_t *syms; 1349 { 1350 sym_t *sym; 1351 1352 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1353 if (sym->s_blklev != -1) { 1354 if ((*sym->s_rlink = sym->s_link) != NULL) 1355 sym->s_link->s_rlink = sym->s_rlink; 1356 sym->s_link = NULL; 1357 sym->s_rlink = NULL; 1358 } 1359 } 1360 } 1361 1362 /* 1363 * Put a symbol into the symbol table 1364 */ 1365 void 1366 inssym(bl, sym) 1367 int bl; 1368 sym_t *sym; 1369 { 1370 int h; 1371 1372 h = hash(sym->s_name); 1373 if ((sym->s_link = symtab[h]) != NULL) 1374 symtab[h]->s_rlink = &sym->s_link; 1375 (symtab[h] = sym)->s_rlink = &symtab[h]; 1376 sym->s_blklev = bl; 1377 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1378 lerror("inssym()"); 1379 } 1380 1381 /* 1382 * Called at level 0 after syntax errors 1383 * Removes all symbols which are not declared at level 0 from the 1384 * symbol table. Also frees all memory which is not associated with 1385 * level 0. 1386 */ 1387 void 1388 cleanup() 1389 { 1390 sym_t *sym, *nsym; 1391 int i; 1392 1393 for (i = 0; i < HSHSIZ1; i++) { 1394 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1395 nsym = sym->s_link; 1396 if (sym->s_blklev >= 1) { 1397 if ((*sym->s_rlink = nsym) != NULL) 1398 nsym->s_rlink = sym->s_rlink; 1399 } 1400 } 1401 } 1402 1403 for (i = mblklev; i > 0; i--) 1404 freelblk(i); 1405 } 1406 1407 /* 1408 * Create a new symbol with the name of an existing symbol. 1409 */ 1410 sym_t * 1411 pushdown(sym) 1412 sym_t *sym; 1413 { 1414 int h; 1415 sym_t *nsym; 1416 1417 h = hash(sym->s_name); 1418 nsym = getblk(sizeof (sym_t)); 1419 if (sym->s_blklev > blklev) 1420 lerror("pushdown()"); 1421 nsym->s_name = sym->s_name; 1422 UNIQUE_CURR_POS(nsym->s_dpos); 1423 nsym->s_kind = sym->s_kind; 1424 nsym->s_blklev = blklev; 1425 1426 if ((nsym->s_link = symtab[h]) != NULL) 1427 symtab[h]->s_rlink = &nsym->s_link; 1428 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1429 1430 *dcs->d_ldlsym = nsym; 1431 dcs->d_ldlsym = &nsym->s_dlnxt; 1432 1433 return (nsym); 1434 } 1435 1436 /* 1437 * Free any dynamically allocated memory referenced by 1438 * the value stack or yylval. 1439 * The type of information in yylval is described by tok. 1440 */ 1441 void 1442 freeyyv(sp, tok) 1443 void *sp; 1444 int tok; 1445 { 1446 if (tok == T_NAME || tok == T_TYPENAME) { 1447 sbuf_t *sb = *(sbuf_t **)sp; 1448 freesb(sb); 1449 } else if (tok == T_CON) { 1450 val_t *val = *(val_t **)sp; 1451 free(val); 1452 } else if (tok == T_STRING) { 1453 strg_t *strg = *(strg_t **)sp; 1454 if (strg->st_tspec == CHAR) { 1455 free(strg->st_cp); 1456 } else if (strg->st_tspec == WCHAR) { 1457 free(strg->st_wcp); 1458 } else { 1459 lerror("fryylv() 1"); 1460 } 1461 free(strg); 1462 } 1463 } 1464
Indexes created Wed Oct 01 15:09:59 GMT 2025