scan.l revision 1.6
1 %{ 2 /* $NetBSD: scan.l,v 1.6 1995/10/02 17:26:57 jpo Exp $ */ 3 4 /* 5 * Copyright (c) 1994, 1995 Jochen Pohl 6 * All Rights Reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by Jochen Pohl for 19 * The NetBSD Project. 20 * 4. The name of the author may not be used to endorse or promote products 21 * derived from this software without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 24 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 25 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 26 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 27 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 28 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 29 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 30 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 31 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 32 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 33 */ 34 35 #ifndef lint 36 static char rcsid[] = "$NetBSD: scan.l,v 1.6 1995/10/02 17:26:57 jpo Exp $"; 37 #endif 38 39 #include <stdlib.h> 40 #include <string.h> 41 #include <limits.h> 42 #include <float.h> 43 #include <ctype.h> 44 #include <errno.h> 45 #include <math.h> 46 #include <err.h> 47 48 #include "lint1.h" 49 #include "y.tab.h" 50 51 #define CHAR_MASK (~(~0 << CHAR_BIT)) 52 53 /* XXX declaration of strtouq() is missing in stdlib.h ? */ 54 extern u_quad_t strtouq __P((const char *, char **, int)); 55 56 /* Current position (its also updated when an included file is parsed) */ 57 pos_t curr_pos = { 1, "" }; 58 59 /* 60 * Current position in C source (not updated when an included file is 61 * parsed). 62 */ 63 pos_t csrc_pos = { 1, "" }; 64 65 /* Line number in output of cpp */ 66 int isrcline = 1; 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 isrcline++; 169 if (curr_pos.p_file == csrc_pos.p_file) 170 csrc_pos.p_line++; 171 } 172 173 static void 174 badchar(c) 175 int c; 176 { 177 /* unknown character \%o */ 178 error(250, c); 179 } 180 181 /* 182 * Keywords. 183 * During initialisation they are written to the symbol table. 184 */ 185 static struct kwtab { 186 const char *kw_name; /* keyword */ 187 int kw_token; /* token returned by yylex() */ 188 scl_t kw_scl; /* storage class if kw_token T_SCLASS */ 189 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ 190 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ 191 u_int kw_stdc : 1; /* STDC keyword */ 192 u_int kw_gcc : 1; /* GCC keyword */ 193 } kwtab[] = { 194 { "asm", T_ASM, 0, 0, 0, 0, 1 }, 195 { "__asm", T_ASM, 0, 0, 0, 0, 0 }, 196 { "__asm__", T_ASM, 0, 0, 0, 0, 0 }, 197 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 }, 198 { "break", T_BREAK, 0, 0, 0, 0, 0 }, 199 { "case", T_CASE, 0, 0, 0, 0, 0 }, 200 { "char", T_TYPE, 0, CHAR, 0, 0, 0 }, 201 { "const", T_QUAL, 0, 0, CONST, 1, 0 }, 202 { "__const__", T_QUAL, 0, 0, CONST, 0, 0 }, 203 { "__const", T_QUAL, 0, 0, CONST, 0, 0 }, 204 { "continue", T_CONTINUE, 0, 0, 0, 0, 0 }, 205 { "default", T_DEFAULT, 0, 0, 0, 0, 0 }, 206 { "do", T_DO, 0, 0, 0, 0, 0 }, 207 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 }, 208 { "else", T_ELSE, 0, 0, 0, 0, 0 }, 209 { "enum", T_ENUM, 0, 0, 0, 0, 0 }, 210 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 }, 211 { "float", T_TYPE, 0, FLOAT, 0, 0, 0 }, 212 { "for", T_FOR, 0, 0, 0, 0, 0 }, 213 { "goto", T_GOTO, 0, 0, 0, 0, 0 }, 214 { "if", T_IF, 0, 0, 0, 0, 0 }, 215 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 }, 216 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 }, 217 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 }, 218 { "int", T_TYPE, 0, INT, 0, 0, 0 }, 219 { "long", T_TYPE, 0, LONG, 0, 0, 0 }, 220 { "register", T_SCLASS, REG, 0, 0, 0, 0 }, 221 { "return", T_RETURN, 0, 0, 0, 0, 0 }, 222 { "short", T_TYPE, 0, SHORT, 0, 0, 0 }, 223 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 }, 224 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 }, 225 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 }, 226 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 }, 227 { "static", T_SCLASS, STATIC, 0, 0, 0, 0 }, 228 { "struct", T_SOU, 0, STRUCT, 0, 0, 0 }, 229 { "switch", T_SWITCH, 0, 0, 0, 0, 0 }, 230 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 }, 231 { "union", T_SOU, 0, UNION, 0, 0, 0 }, 232 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 }, 233 { "void", T_TYPE, 0, VOID, 0, 0, 0 }, 234 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 }, 235 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 236 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 237 { "while", T_WHILE, 0, 0, 0, 0, 0 }, 238 { NULL, 0, 0, 0, 0, 0, 0 } 239 }; 240 241 /* Symbol table */ 242 static sym_t *symtab[HSHSIZ1]; 243 244 /* bit i of the entry with index i is set */ 245 u_quad_t qbmasks[sizeof(u_quad_t) * CHAR_BIT]; 246 247 /* least significant i bits are set in the entry with index i */ 248 u_quad_t qlmasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 249 250 /* least significant i bits are not set in the entry with index i */ 251 u_quad_t qumasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 252 253 /* free list for sbuf structures */ 254 static sbuf_t *sbfrlst; 255 256 /* Typ of next expected symbol */ 257 symt_t symtyp; 258 259 260 /* 261 * All keywords are written to the symbol table. This saves us looking 262 * in a extra table for each name we found. 263 */ 264 void 265 initscan() 266 { 267 struct kwtab *kw; 268 sym_t *sym; 269 int h, i; 270 u_quad_t uq; 271 272 for (kw = kwtab; kw->kw_name != NULL; kw++) { 273 if (kw->kw_stdc && tflag) 274 continue; 275 if (kw->kw_gcc && !gflag) 276 continue; 277 sym = getblk(sizeof (sym_t)); 278 sym->s_name = kw->kw_name; 279 sym->s_keyw = 1; 280 sym->s_value.v_quad = kw->kw_token; 281 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 282 sym->s_tspec = kw->kw_tspec; 283 } else if (kw->kw_token == T_SCLASS) { 284 sym->s_scl = kw->kw_scl; 285 } else if (kw->kw_token == T_QUAL) { 286 sym->s_tqual = kw->kw_tqual; 287 } 288 h = hash(sym->s_name); 289 if ((sym->s_link = symtab[h]) != NULL) 290 symtab[h]->s_rlink = &sym->s_link; 291 (symtab[h] = sym)->s_rlink = &symtab[h]; 292 } 293 294 /* initialize bit-masks for quads */ 295 for (i = 0; i < sizeof (u_quad_t) * CHAR_BIT; i++) { 296 qbmasks[i] = (u_quad_t)1 << i; 297 uq = ~(u_quad_t)0 << i; 298 qumasks[i] = uq; 299 qlmasks[i] = ~uq; 300 } 301 qumasks[i] = 0; 302 qlmasks[i] = ~(u_quad_t)0; 303 } 304 305 /* 306 * Get a free sbuf structure, if possible from the free list 307 */ 308 static sbuf_t * 309 allocsb() 310 { 311 sbuf_t *sb; 312 313 if ((sb = sbfrlst) != NULL) { 314 sbfrlst = sb->sb_nxt; 315 } else { 316 sb = xmalloc(sizeof (sbuf_t)); 317 } 318 (void)memset(sb, 0, sizeof (sb)); 319 return (sb); 320 } 321 322 /* 323 * Put a sbuf structure to the free list 324 */ 325 static void 326 freesb(sb) 327 sbuf_t *sb; 328 { 329 sb->sb_nxt = sbfrlst; 330 sbfrlst = sb; 331 } 332 333 /* 334 * Read a character and ensure that it is positive (except EOF). 335 * Increment line count(s) if necessary. 336 */ 337 static int 338 inpc() 339 { 340 int c; 341 342 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 343 incline(); 344 return (c); 345 } 346 347 static int 348 hash(s) 349 const char *s; 350 { 351 u_int v; 352 const u_char *us; 353 354 v = 0; 355 for (us = (const u_char *)s; *us != '\0'; us++) { 356 v = (v << sizeof (v)) + *us; 357 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 358 } 359 return (v % HSHSIZ1); 360 } 361 362 /* 363 * Lex has found a letter followed by zero or more letters or digits. 364 * It looks for a symbol in the symbol table with the same name. This 365 * symbol must either be a keyword or a symbol of the type required by 366 * symtyp (label, member, tag, ...). 367 * 368 * If it is a keyword, the token is returned. In some cases it is described 369 * more deeply by data written to yylval. 370 * 371 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 372 * is stored in yylval. This struct contains the name of the symbol, it's 373 * length and hash value. If there is already a symbol of the same name 374 * and type in the symbol table, the sbuf struct also contains a pointer 375 * to the symbol table entry. 376 */ 377 static int 378 name() 379 { 380 char *s; 381 sbuf_t *sb; 382 sym_t *sym; 383 int tok; 384 385 sb = allocsb(); 386 sb->sb_name = yytext; 387 sb->sb_len = yyleng; 388 sb->sb_hash = hash(yytext); 389 390 if ((sym = search(sb)) != NULL && sym->s_keyw) { 391 freesb(sb); 392 return (keyw(sym)); 393 } 394 395 sb->sb_sym = sym; 396 397 if (sym != NULL) { 398 if (blklev < sym->s_blklev) 399 lerror("name() 1"); 400 sb->sb_name = sym->s_name; 401 sb->sb_len = strlen(sym->s_name); 402 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 403 } else { 404 s = getblk(yyleng + 1); 405 (void)memcpy(s, yytext, yyleng + 1); 406 sb->sb_name = s; 407 sb->sb_len = yyleng; 408 tok = T_NAME; 409 } 410 411 yylval.y_sb = sb; 412 return (tok); 413 } 414 415 static sym_t * 416 search(sb) 417 sbuf_t *sb; 418 { 419 sym_t *sym; 420 421 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 422 if (strcmp(sym->s_name, sb->sb_name) == 0) { 423 if (sym->s_keyw || sym->s_kind == symtyp) 424 return (sym); 425 } 426 } 427 428 return (NULL); 429 } 430 431 static int 432 keyw(sym) 433 sym_t *sym; 434 { 435 int t; 436 437 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 438 yylval.y_scl = sym->s_scl; 439 } else if (t == T_TYPE || t == T_SOU) { 440 yylval.y_tspec = sym->s_tspec; 441 } else if (t == T_QUAL) { 442 yylval.y_tqual = sym->s_tqual; 443 } 444 return (t); 445 } 446 447 /* 448 * Convert a string representing an integer into internal representation. 449 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 450 */ 451 static int 452 icon(base) 453 int base; 454 { 455 int l_suffix, u_suffix; 456 int len; 457 const char *cp; 458 char c, *eptr; 459 tspec_t typ; 460 u_long ul; 461 u_quad_t uq; 462 int ansiu; 463 static tspec_t contypes[2][3] = { 464 { INT, LONG, QUAD }, 465 { UINT, ULONG, UQUAD } 466 }; 467 468 cp = yytext; 469 len = yyleng; 470 471 /* skip 0x */ 472 if (base == 16) { 473 cp += 2; 474 len -= 2; 475 } 476 477 /* read suffixes */ 478 l_suffix = u_suffix = 0; 479 for ( ; ; ) { 480 if ((c = cp[len - 1]) == 'l' || c == 'L') { 481 l_suffix++; 482 } else if (c == 'u' || c == 'U') { 483 u_suffix++; 484 } else { 485 break; 486 } 487 len--; 488 } 489 if (l_suffix > 2 || u_suffix > 1) { 490 /* malformed integer constant */ 491 warning(251); 492 if (l_suffix > 2) 493 l_suffix = 2; 494 if (u_suffix > 1) 495 u_suffix = 1; 496 } 497 if (tflag && u_suffix != 0) { 498 /* suffix U is illegal in traditional C */ 499 warning(97); 500 } 501 typ = contypes[u_suffix][l_suffix]; 502 503 errno = 0; 504 if (l_suffix < 2) { 505 ul = strtoul(cp, &eptr, base); 506 } else { 507 uq = strtouq(cp, &eptr, base); 508 } 509 if (eptr != cp + len) 510 lerror("icon() 1"); 511 if (errno != 0) 512 /* integer constant out of range */ 513 warning(252); 514 515 /* 516 * If the value is to big for the current type, we must choose 517 * another type. 518 */ 519 ansiu = 0; 520 switch (typ) { 521 case INT: 522 if (ul <= INT_MAX) { 523 /* ok */ 524 } else if (ul <= (unsigned)UINT_MAX && base != 10) { 525 typ = UINT; 526 } else if (ul <= LONG_MAX) { 527 typ = LONG; 528 } else { 529 typ = ULONG; 530 } 531 if (typ == UINT || typ == ULONG) { 532 if (tflag) { 533 typ = LONG; 534 } else if (!sflag) { 535 /* 536 * Remember that the constant is unsigned 537 * only in ANSI C 538 */ 539 ansiu = 1; 540 } 541 } 542 break; 543 case UINT: 544 if (ul > (u_int)UINT_MAX) 545 typ = ULONG; 546 break; 547 case LONG: 548 if (ul > LONG_MAX && !tflag) { 549 typ = ULONG; 550 if (!sflag) 551 ansiu = 1; 552 } 553 break; 554 case QUAD: 555 if (uq > QUAD_MAX && !tflag) { 556 typ = UQUAD; 557 if (!sflag) 558 ansiu = 1; 559 } 560 break; 561 /* LINTED (enumeration values not handled in switch) */ 562 } 563 564 if (typ != QUAD && typ != UQUAD) { 565 if (isutyp(typ)) { 566 uq = ul; 567 } else { 568 uq = (quad_t)(long)ul; 569 } 570 } 571 572 uq = (u_quad_t)xsign((quad_t)uq, typ, -1); 573 574 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 575 yylval.y_val->v_ansiu = ansiu; 576 yylval.y_val->v_quad = (quad_t)uq; 577 578 return (T_CON); 579 } 580 581 /* 582 * Returns 1 if t is a signed type and the value is negative. 583 * 584 * len is the number of significant bits. If len is -1, len is set 585 * to the width of type t. 586 */ 587 int 588 sign(q, t, len) 589 quad_t q; 590 tspec_t t; 591 int len; 592 { 593 if (t == PTR || isutyp(t)) 594 return (0); 595 return (msb(q, t, len)); 596 } 597 598 int 599 msb(q, t, len) 600 quad_t q; 601 tspec_t t; 602 int len; 603 { 604 if (len <= 0) 605 len = size(t); 606 return ((q & qbmasks[len - 1]) != 0); 607 } 608 609 /* 610 * Extends the sign of q. 611 */ 612 quad_t 613 xsign(q, t, len) 614 quad_t q; 615 tspec_t t; 616 int len; 617 { 618 if (len <= 0) 619 len = size(t); 620 621 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 622 q &= qlmasks[len]; 623 } else { 624 q |= qumasks[len]; 625 } 626 return (q); 627 } 628 629 /* 630 * Convert a string representing a floating point value into its interal 631 * representation. Type and value are returned in yylval. fcon() 632 * (and yylex()) returns T_CON. 633 * XXX Currently it is not possible to convert constants of type 634 * long double which are greater then DBL_MAX. 635 */ 636 static int 637 fcon() 638 { 639 const char *cp; 640 int len; 641 tspec_t typ; 642 char c, *eptr; 643 double d; 644 float f; 645 646 cp = yytext; 647 len = yyleng; 648 649 if ((c = cp[len - 1]) == 'f' || c == 'F') { 650 typ = FLOAT; 651 len--; 652 } else if (c == 'l' || c == 'L') { 653 typ = LDOUBLE; 654 len--; 655 } else { 656 typ = DOUBLE; 657 } 658 659 if (tflag && typ != DOUBLE) { 660 /* suffixes F and L are illegal in traditional C */ 661 warning(98); 662 } 663 664 errno = 0; 665 d = strtod(cp, &eptr); 666 if (eptr != cp + len) 667 lerror("fcon() 1"); 668 if (errno != 0) 669 /* floating-point constant out of range */ 670 warning(248); 671 672 if (typ == FLOAT) { 673 f = (float)d; 674 if (isinf(f)) { 675 /* floating-point constant out of range */ 676 warning(248); 677 f = f > 0 ? FLT_MAX : -FLT_MAX; 678 } 679 } 680 681 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 682 if (typ == FLOAT) { 683 yylval.y_val->v_ldbl = f; 684 } else { 685 yylval.y_val->v_ldbl = d; 686 } 687 688 return (T_CON); 689 } 690 691 static int 692 operator(t, o) 693 int t; 694 op_t o; 695 { 696 yylval.y_op = o; 697 return (t); 698 } 699 700 /* 701 * Called if lex found a leading \'. 702 */ 703 static int 704 ccon() 705 { 706 int n, val, c; 707 char cv; 708 709 n = 0; 710 val = 0; 711 while ((c = getescc('\'')) >= 0) { 712 val = (val << CHAR_BIT) + c; 713 n++; 714 } 715 if (c == -2) { 716 /* unterminated character constant */ 717 error(253); 718 } else { 719 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 720 /* too many characters in character constant */ 721 error(71); 722 } else if (n > 1) { 723 /* multi-character character constant */ 724 warning(294); 725 } else if (n == 0) { 726 /* empty character constant */ 727 error(73); 728 } 729 } 730 if (n == 1) { 731 cv = (char)val; 732 val = cv; 733 } 734 735 yylval.y_val = xcalloc(1, sizeof (val_t)); 736 yylval.y_val->v_tspec = INT; 737 yylval.y_val->v_quad = val; 738 739 return (T_CON); 740 } 741 742 /* 743 * Called if lex found a leading L\' 744 */ 745 static int 746 wccon() 747 { 748 static char buf[MB_CUR_MAX + 1]; 749 int i, c; 750 wchar_t wc; 751 752 i = 0; 753 while ((c = getescc('\'')) >= 0) { 754 if (i < MB_CUR_MAX) 755 buf[i] = (char)c; 756 i++; 757 } 758 759 wc = 0; 760 761 if (c == -2) { 762 /* unterminated character constant */ 763 error(253); 764 } else if (c == 0) { 765 /* empty character constant */ 766 error(73); 767 } else { 768 if (i > MB_CUR_MAX) { 769 i = MB_CUR_MAX; 770 /* too many characters in character constant */ 771 error(71); 772 } else { 773 buf[i] = '\0'; 774 (void)mbtowc(NULL, NULL, 0); 775 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 776 /* invalid multibyte character */ 777 error(291); 778 } 779 } 780 781 yylval.y_val = xcalloc(1, sizeof (val_t)); 782 yylval.y_val->v_tspec = WCHAR; 783 yylval.y_val->v_quad = wc; 784 785 return (T_CON); 786 } 787 788 /* 789 * Read a character which is part of a character constant or of a string 790 * and handle escapes. 791 * 792 * The Argument is the character which delimits the character constant or 793 * string. 794 * 795 * Returns -1 if the end of the character constant or string is reached, 796 * -2 if the EOF is reached, and the charachter otherwise. 797 */ 798 static int 799 getescc(d) 800 int d; 801 { 802 static int pbc = -1; 803 int n, c, v; 804 805 if (pbc == -1) { 806 c = inpc(); 807 } else { 808 c = pbc; 809 pbc = -1; 810 } 811 if (c == d) 812 return (-1); 813 switch (c) { 814 case '\n': 815 /* newline in string or char constant */ 816 error(254); 817 return (-2); 818 case EOF: 819 return (-2); 820 case '\\': 821 switch (c = inpc()) { 822 case '"': 823 if (tflag && d == '\'') 824 /* \" inside character constant undef. ... */ 825 warning(262); 826 return ('"'); 827 case '\'': 828 return ('\''); 829 case '?': 830 if (tflag) 831 /* \? undefined in traditional C */ 832 warning(263); 833 return ('?'); 834 case '\\': 835 return ('\\'); 836 case 'a': 837 if (tflag) 838 /* \a undefined in traditional C */ 839 warning(81); 840 #ifdef __STDC__ 841 return ('\a'); 842 #else 843 return ('\007'); 844 #endif 845 case 'b': 846 return ('\b'); 847 case 'f': 848 return ('\f'); 849 case 'n': 850 return ('\n'); 851 case 'r': 852 return ('\r'); 853 case 't': 854 return ('\t'); 855 case 'v': 856 if (tflag) 857 /* \v undefined in traditional C */ 858 warning(264); 859 #ifdef __STDC__ 860 return ('\v'); 861 #else 862 return ('\013'); 863 #endif 864 case '8': case '9': 865 /* bad octal digit %c */ 866 warning(77, c); 867 /* FALLTHROUGH */ 868 case '0': case '1': case '2': case '3': 869 case '4': case '5': case '6': case '7': 870 n = 3; 871 v = 0; 872 do { 873 v = (v << 3) + (c - '0'); 874 c = inpc(); 875 } while (--n && isdigit(c) && (tflag || c <= '7')); 876 if (tflag && n > 0 && isdigit(c)) 877 /* bad octal digit %c */ 878 warning(77, c); 879 pbc = c; 880 if (v > UCHAR_MAX) { 881 /* character escape does not fit in char. */ 882 warning(76); 883 v &= CHAR_MASK; 884 } 885 return (v); 886 case 'x': 887 if (tflag) 888 /* \x undefined in traditional C */ 889 warning(82); 890 v = 0; 891 n = 0; 892 while ((c = inpc()) >= 0 && isxdigit(c)) { 893 c = isdigit(c) ? 894 c - '0' : toupper(c) - 'A' + 10; 895 v = (v << 4) + c; 896 if (n >= 0) { 897 if ((v & ~CHAR_MASK) != 0) { 898 /* overflow in hex escape */ 899 warning(75); 900 n = -1; 901 } else { 902 n++; 903 } 904 } 905 } 906 pbc = c; 907 if (n == 0) { 908 /* no hex digits follow \x */ 909 error(74); 910 } if (n == -1) { 911 v &= CHAR_MASK; 912 } 913 return (v); 914 case '\n': 915 return (getescc(d)); 916 case EOF: 917 return (-2); 918 default: 919 if (isprint(c)) { 920 /* dubious escape \%c */ 921 warning(79, c); 922 } else { 923 /* dubious escape \%o */ 924 warning(80, c); 925 } 926 } 927 } 928 return (c); 929 } 930 931 /* 932 * Called for preprocessor directives. Currently implemented are: 933 * # lineno 934 * # lineno "filename" 935 */ 936 static void 937 directive() 938 { 939 const char *cp, *fn; 940 char c, *eptr; 941 size_t fnl; 942 long ln; 943 static int first = 1; 944 945 /* Go to first non-whitespace after # */ 946 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) ; 947 948 if (!isdigit(c)) { 949 error: 950 /* undefined or invalid # directive */ 951 warning(255); 952 return; 953 } 954 ln = strtol(--cp, &eptr, 10); 955 if (cp == eptr) 956 goto error; 957 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 958 goto error; 959 while ((c = *cp++) == ' ' || c == '\t') ; 960 if (c != '\0') { 961 if (c != '"') 962 goto error; 963 fn = cp; 964 while ((c = *cp) != '"' && c != '\0') 965 cp++; 966 if (c != '"') 967 goto error; 968 if ((fnl = cp++ - fn) > PATH_MAX) 969 goto error; 970 while ((c = *cp++) == ' ' || c == '\t') ; 971 #if 0 972 if (c != '\0') 973 warning("extra character(s) after directive"); 974 #endif 975 curr_pos.p_file = fnnalloc(fn, fnl); 976 /* 977 * If this is the first directive, the name is the name 978 * of the C source file as specified at the command line. 979 * It is written to the output file. 980 */ 981 if (first) { 982 csrc_pos.p_file = curr_pos.p_file; 983 outsrc(curr_pos.p_file); 984 first = 0; 985 } 986 } 987 curr_pos.p_line = (int)ln - 1; 988 if (curr_pos.p_file == csrc_pos.p_file) 989 csrc_pos.p_line = (int)ln - 1; 990 } 991 992 /* 993 * Handle lint comments. Following comments are currently understood: 994 * ARGSUSEDn 995 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 996 * FALLTHRU FALLTHROUGH 997 * LINTLIBRARY 998 * LINTED NOSTRICT 999 * NOTREACHED 1000 * PRINTFLIKEn 1001 * PROTOLIB 1002 * SCANFLIKEn 1003 * VARARGSn 1004 * If one of this comments is recognized, the arguments, if any, are 1005 * parsed and a function which handles this comment is called. 1006 */ 1007 static void 1008 comment() 1009 { 1010 int c, lc; 1011 static struct { 1012 const char *keywd; 1013 int arg; 1014 void (*func) __P((int)); 1015 } keywtab[] = { 1016 { "ARGSUSED", 1, argsused }, 1017 { "CONSTCOND", 0, constcond }, 1018 { "CONSTANTCOND", 0, constcond }, 1019 { "CONSTANTCONDITION", 0, constcond }, 1020 { "FALLTHRU", 0, fallthru }, 1021 { "FALLTHROUGH", 0, fallthru }, 1022 { "LINTLIBRARY", 0, lintlib }, 1023 { "LINTED", 0, linted }, 1024 { "NOSTRICT", 0, linted }, 1025 { "NOTREACHED", 0, notreach }, 1026 { "PRINTFLIKE", 1, printflike }, 1027 { "PROTOLIB", 1, protolib }, 1028 { "SCANFLIKE", 1, scanflike }, 1029 { "VARARGS", 1, varargs }, 1030 }; 1031 char keywd[32]; 1032 char arg[32]; 1033 int l, i, a; 1034 int eoc; 1035 1036 eoc = 0; 1037 1038 /* Skip white spaces after the start of the comment */ 1039 while ((c = inpc()) != EOF && isspace(c)) ; 1040 1041 /* Read the potential keyword to keywd */ 1042 l = 0; 1043 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1044 keywd[l++] = (char)c; 1045 c = inpc(); 1046 } 1047 keywd[l] = '\0'; 1048 1049 /* look for the keyword */ 1050 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1051 if (strcmp(keywtab[i].keywd, keywd) == 0) 1052 break; 1053 } 1054 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1055 goto skip_rest; 1056 1057 /* skip white spaces after the keyword */ 1058 while (c != EOF && isspace(c)) 1059 c = inpc(); 1060 1061 /* read the argument, if the keyword accepts one and there is one */ 1062 l = 0; 1063 if (keywtab[i].arg) { 1064 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1065 arg[l++] = (char)c; 1066 c = inpc(); 1067 } 1068 } 1069 arg[l] = '\0'; 1070 a = l != 0 ? atoi(arg) : -1; 1071 1072 /* skip white spaces after the argument */ 1073 while (c != EOF && isspace(c)) 1074 c = inpc(); 1075 1076 if (c != '*' || (c = inpc()) != '/') { 1077 if (keywtab[i].func != linted) 1078 /* extra characters in lint comment */ 1079 warning(257); 1080 } else { 1081 /* 1082 * remember that we have already found the end of the 1083 * comment 1084 */ 1085 eoc = 1; 1086 } 1087 1088 if (keywtab[i].func != NULL) 1089 (*keywtab[i].func)(a); 1090 1091 skip_rest: 1092 while (!eoc) { 1093 lc = c; 1094 if ((c = inpc()) == EOF) { 1095 /* unterminated comment */ 1096 error(256); 1097 break; 1098 } 1099 if (lc == '*' && c == '/') 1100 eoc = 1; 1101 } 1102 } 1103 1104 /* 1105 * Strings are stored in a dynamically alloceted buffer and passed 1106 * in yylval.y_xstrg to the parser. The parser or the routines called 1107 * by the parser are responsible for freeing this buffer. 1108 */ 1109 static int 1110 string() 1111 { 1112 u_char *s; 1113 int c; 1114 size_t len, max; 1115 strg_t *strg; 1116 1117 s = xmalloc(max = 64); 1118 1119 len = 0; 1120 while ((c = getescc('"')) >= 0) { 1121 /* +1 to reserve space for a trailing NUL character */ 1122 if (len + 1 == max) 1123 s = xrealloc(s, max *= 2); 1124 s[len++] = (char)c; 1125 } 1126 s[len] = '\0'; 1127 if (c == -2) 1128 /* unterminated string constant */ 1129 error(258); 1130 1131 strg = xcalloc(1, sizeof (strg_t)); 1132 strg->st_tspec = CHAR; 1133 strg->st_len = len; 1134 strg->st_cp = s; 1135 1136 yylval.y_strg = strg; 1137 return (T_STRING); 1138 } 1139 1140 static int 1141 wcstrg() 1142 { 1143 char *s; 1144 int c, i, n, wi; 1145 size_t len, max, wlen; 1146 wchar_t *ws; 1147 strg_t *strg; 1148 1149 s = xmalloc(max = 64); 1150 len = 0; 1151 while ((c = getescc('"')) >= 0) { 1152 /* +1 to save space for a trailing NUL character */ 1153 if (len + 1 >= max) 1154 s = xrealloc(s, max *= 2); 1155 s[len++] = (char)c; 1156 } 1157 s[len] = '\0'; 1158 if (c == -2) 1159 /* unterminated string constant */ 1160 error(258); 1161 1162 /* get length of wide character string */ 1163 (void)mblen(NULL, 0); 1164 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1165 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1166 /* invalid multibyte character */ 1167 error(291); 1168 break; 1169 } 1170 if (n == 0) 1171 n = 1; 1172 } 1173 1174 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1175 1176 /* convert from multibyte to wide char */ 1177 (void)mbtowc(NULL, NULL, 0); 1178 for (i = 0, wi = 0; i < len; i += n, wi++) { 1179 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1180 break; 1181 if (n == 0) 1182 n = 1; 1183 } 1184 ws[wi] = 0; 1185 free(s); 1186 1187 strg = xcalloc(1, sizeof (strg_t)); 1188 strg->st_tspec = WCHAR; 1189 strg->st_len = wlen; 1190 strg->st_wcp = ws; 1191 1192 yylval.y_strg = strg; 1193 return (T_STRING); 1194 } 1195 1196 /* 1197 * As noted above the scanner does not create new symbol table entries 1198 * for symbols it cannot find in the symbol table. This is to avoid 1199 * putting undeclared symbols into the symbol table if a syntax error 1200 * occurs. 1201 * 1202 * getsym() is called as soon as it is probably ok to put the symbol to 1203 * the symbol table. This does not mean that it is not possible that 1204 * symbols are put to the symbol table which are than not completely 1205 * declared due to syntax errors. To avoid too many problems in this 1206 * case symbols get type int in getsym(). 1207 * 1208 * XXX calls to getsym() should be delayed until decl1*() is called 1209 */ 1210 sym_t * 1211 getsym(sb) 1212 sbuf_t *sb; 1213 { 1214 dinfo_t *di; 1215 char *s; 1216 sym_t *sym; 1217 1218 sym = sb->sb_sym; 1219 1220 /* 1221 * During member declaration it is possible that name() looked 1222 * for symbols of type FVFT, although it should have looked for 1223 * symbols of type FTAG. Same can happen for labels. Both cases 1224 * are compensated here. 1225 */ 1226 if (symtyp == FMOS || symtyp == FLAB) { 1227 if (sym == NULL || sym->s_kind == FVFT) 1228 sym = search(sb); 1229 } 1230 1231 if (sym != NULL) { 1232 if (sym->s_kind != symtyp) 1233 lerror("storesym() 1"); 1234 symtyp = FVFT; 1235 freesb(sb); 1236 return (sym); 1237 } 1238 1239 /* create a new symbol table entry */ 1240 1241 /* labels must always be allocated at level 1 (outhermost block) */ 1242 if (symtyp == FLAB) { 1243 sym = getlblk(1, sizeof (sym_t)); 1244 s = getlblk(1, sb->sb_len + 1); 1245 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1246 sym->s_name = s; 1247 sym->s_blklev = 1; 1248 di = dcs; 1249 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1250 di = di->d_nxt; 1251 if (di->d_ctx != AUTO) 1252 lerror("storesym() 2"); 1253 } else { 1254 sym = getblk(sizeof (sym_t)); 1255 sym->s_name = sb->sb_name; 1256 sym->s_blklev = blklev; 1257 di = dcs; 1258 } 1259 1260 STRUCT_ASSIGN(sym->s_dpos, curr_pos); 1261 if ((sym->s_kind = symtyp) != FLAB) 1262 sym->s_type = gettyp(INT); 1263 1264 symtyp = FVFT; 1265 1266 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1267 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1268 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1269 1270 *di->d_ldlsym = sym; 1271 di->d_ldlsym = &sym->s_dlnxt; 1272 1273 freesb(sb); 1274 return (sym); 1275 } 1276 1277 /* 1278 * Remove a symbol forever from the symbol table. s_blklev 1279 * is set to -1 to avoid that the symbol will later be put 1280 * back to the symbol table. 1281 */ 1282 void 1283 rmsym(sym) 1284 sym_t *sym; 1285 { 1286 if ((*sym->s_rlink = sym->s_link) != NULL) 1287 sym->s_link->s_rlink = sym->s_rlink; 1288 sym->s_blklev = -1; 1289 sym->s_link = NULL; 1290 } 1291 1292 /* 1293 * Remove a list of symbols declared at one level from the symbol 1294 * table. 1295 */ 1296 void 1297 rmsyms(syms) 1298 sym_t *syms; 1299 { 1300 sym_t *sym; 1301 1302 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1303 if (sym->s_blklev != -1) { 1304 if ((*sym->s_rlink = sym->s_link) != NULL) 1305 sym->s_link->s_rlink = sym->s_rlink; 1306 sym->s_link = NULL; 1307 sym->s_rlink = NULL; 1308 } 1309 } 1310 } 1311 1312 /* 1313 * Put a symbol into the symbol table 1314 */ 1315 void 1316 inssym(bl, sym) 1317 int bl; 1318 sym_t *sym; 1319 { 1320 int h; 1321 1322 h = hash(sym->s_name); 1323 if ((sym->s_link = symtab[h]) != NULL) 1324 symtab[h]->s_rlink = &sym->s_link; 1325 (symtab[h] = sym)->s_rlink = &symtab[h]; 1326 sym->s_blklev = bl; 1327 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1328 lerror("inssym()"); 1329 } 1330 1331 /* 1332 * Called at level 0 after syntax errors 1333 * Removes all symbols which are not declared at level 0 from the 1334 * symbol table. Also frees all memory which is not associated with 1335 * level 0. 1336 */ 1337 void 1338 cleanup() 1339 { 1340 sym_t *sym, *nsym; 1341 int i; 1342 1343 for (i = 0; i < HSHSIZ1; i++) { 1344 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1345 nsym = sym->s_link; 1346 if (sym->s_blklev >= 1) { 1347 if ((*sym->s_rlink = nsym) != NULL) 1348 nsym->s_rlink = sym->s_rlink; 1349 } 1350 } 1351 } 1352 1353 for (i = mblklev; i > 0; i--) 1354 freelblk(i); 1355 } 1356 1357 /* 1358 * Create a new symbol with the name of an existing symbol. 1359 */ 1360 sym_t * 1361 pushdown(sym) 1362 sym_t *sym; 1363 { 1364 int h; 1365 sym_t *nsym; 1366 1367 h = hash(sym->s_name); 1368 nsym = getblk(sizeof (sym_t)); 1369 if (sym->s_blklev > blklev) 1370 lerror("pushdown()"); 1371 nsym->s_name = sym->s_name; 1372 STRUCT_ASSIGN(nsym->s_dpos, curr_pos); 1373 nsym->s_kind = sym->s_kind; 1374 nsym->s_blklev = blklev; 1375 1376 if ((nsym->s_link = symtab[h]) != NULL) 1377 symtab[h]->s_rlink = &nsym->s_link; 1378 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1379 1380 *dcs->d_ldlsym = nsym; 1381 dcs->d_ldlsym = &nsym->s_dlnxt; 1382 1383 return (nsym); 1384 } 1385 1386 /* 1387 * Free any dynamically allocated memory referenced by 1388 * the value stack or yylval. 1389 * The type of information in yylval is described by tok. 1390 */ 1391 void 1392 freeyyv(sp, tok) 1393 void *sp; 1394 int tok; 1395 { 1396 if (tok == T_NAME || tok == T_TYPENAME) { 1397 sbuf_t *sb = *(sbuf_t **)sp; 1398 freesb(sb); 1399 } else if (tok == T_CON) { 1400 val_t *val = *(val_t **)sp; 1401 free(val); 1402 } else if (tok == T_STRING) { 1403 strg_t *strg = *(strg_t **)sp; 1404 if (strg->st_tspec == CHAR) { 1405 free(strg->st_cp); 1406 } else if (strg->st_tspec == WCHAR) { 1407 free(strg->st_wcp); 1408 } else { 1409 lerror("fryylv() 1"); 1410 } 1411 free(strg); 1412 } 1413 } 1414
Indexes created Tue Sep 30 17:09:57 GMT 2025