scan.l revision 1.5
1 %{ 2 /* $NetBSD: scan.l,v 1.5 1995/10/02 17:19:03 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 1.5 1995/10/02 17:19:03 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 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 }, 195 { "break", T_BREAK, 0, 0, 0, 0, 0 }, 196 { "case", T_CASE, 0, 0, 0, 0, 0 }, 197 { "char", T_TYPE, 0, CHAR, 0, 0, 0 }, 198 { "const", T_QUAL, 0, 0, CONST, 1, 0 }, 199 { "__const__", T_QUAL, 0, 0, CONST, 0, 0 }, 200 { "__const", T_QUAL, 0, 0, CONST, 0, 0 }, 201 { "continue", T_CONTINUE, 0, 0, 0, 0, 0 }, 202 { "default", T_DEFAULT, 0, 0, 0, 0, 0 }, 203 { "do", T_DO, 0, 0, 0, 0, 0 }, 204 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 }, 205 { "else", T_ELSE, 0, 0, 0, 0, 0 }, 206 { "enum", T_ENUM, 0, 0, 0, 0, 0 }, 207 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 }, 208 { "float", T_TYPE, 0, FLOAT, 0, 0, 0 }, 209 { "for", T_FOR, 0, 0, 0, 0, 0 }, 210 { "goto", T_GOTO, 0, 0, 0, 0, 0 }, 211 { "if", T_IF, 0, 0, 0, 0, 0 }, 212 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 }, 213 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 }, 214 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 }, 215 { "int", T_TYPE, 0, INT, 0, 0, 0 }, 216 { "long", T_TYPE, 0, LONG, 0, 0, 0 }, 217 { "register", T_SCLASS, REG, 0, 0, 0, 0 }, 218 { "return", T_RETURN, 0, 0, 0, 0, 0 }, 219 { "short", T_TYPE, 0, SHORT, 0, 0, 0 }, 220 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 }, 221 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 }, 222 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 }, 223 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 }, 224 { "static", T_SCLASS, STATIC, 0, 0, 0, 0 }, 225 { "struct", T_SOU, 0, STRUCT, 0, 0, 0 }, 226 { "switch", T_SWITCH, 0, 0, 0, 0, 0 }, 227 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 }, 228 { "union", T_SOU, 0, UNION, 0, 0, 0 }, 229 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 }, 230 { "void", T_TYPE, 0, VOID, 0, 0, 0 }, 231 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 }, 232 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 233 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 }, 234 { "while", T_WHILE, 0, 0, 0, 0, 0 }, 235 { NULL, 0, 0, 0, 0, 0, 0 } 236 }; 237 238 /* Symbol table */ 239 static sym_t *symtab[HSHSIZ1]; 240 241 /* bit i of the entry with index i is set */ 242 u_quad_t qbmasks[sizeof(u_quad_t) * CHAR_BIT]; 243 244 /* least significant i bits are set in the entry with index i */ 245 u_quad_t qlmasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 246 247 /* least significant i bits are not set in the entry with index i */ 248 u_quad_t qumasks[sizeof(u_quad_t) * CHAR_BIT + 1]; 249 250 /* free list for sbuf structures */ 251 static sbuf_t *sbfrlst; 252 253 /* Typ of next expected symbol */ 254 symt_t symtyp; 255 256 257 /* 258 * All keywords are written to the symbol table. This saves us looking 259 * in a extra table for each name we found. 260 */ 261 void 262 initscan() 263 { 264 struct kwtab *kw; 265 sym_t *sym; 266 int h, i; 267 u_quad_t uq; 268 269 for (kw = kwtab; kw->kw_name != NULL; kw++) { 270 if (kw->kw_stdc && tflag) 271 continue; 272 if (kw->kw_gcc && !gflag) 273 continue; 274 sym = getblk(sizeof (sym_t)); 275 sym->s_name = kw->kw_name; 276 sym->s_keyw = 1; 277 sym->s_value.v_quad = kw->kw_token; 278 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { 279 sym->s_tspec = kw->kw_tspec; 280 } else if (kw->kw_token == T_SCLASS) { 281 sym->s_scl = kw->kw_scl; 282 } else if (kw->kw_token == T_QUAL) { 283 sym->s_tqual = kw->kw_tqual; 284 } 285 h = hash(sym->s_name); 286 if ((sym->s_link = symtab[h]) != NULL) 287 symtab[h]->s_rlink = &sym->s_link; 288 (symtab[h] = sym)->s_rlink = &symtab[h]; 289 } 290 291 /* initialize bit-masks for quads */ 292 for (i = 0; i < sizeof (u_quad_t) * CHAR_BIT; i++) { 293 qbmasks[i] = (u_quad_t)1 << i; 294 uq = ~(u_quad_t)0 << i; 295 qumasks[i] = uq; 296 qlmasks[i] = ~uq; 297 } 298 qumasks[i] = 0; 299 qlmasks[i] = ~(u_quad_t)0; 300 } 301 302 /* 303 * Get a free sbuf structure, if possible from the free list 304 */ 305 static sbuf_t * 306 allocsb() 307 { 308 sbuf_t *sb; 309 310 if ((sb = sbfrlst) != NULL) { 311 sbfrlst = sb->sb_nxt; 312 } else { 313 sb = xmalloc(sizeof (sbuf_t)); 314 } 315 (void)memset(sb, 0, sizeof (sb)); 316 return (sb); 317 } 318 319 /* 320 * Put a sbuf structure to the free list 321 */ 322 static void 323 freesb(sb) 324 sbuf_t *sb; 325 { 326 sb->sb_nxt = sbfrlst; 327 sbfrlst = sb; 328 } 329 330 /* 331 * Read a character and ensure that it is positive (except EOF). 332 * Increment line count(s) if necessary. 333 */ 334 static int 335 inpc() 336 { 337 int c; 338 339 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') 340 incline(); 341 return (c); 342 } 343 344 static int 345 hash(s) 346 const char *s; 347 { 348 u_int v; 349 const u_char *us; 350 351 v = 0; 352 for (us = (const u_char *)s; *us != '\0'; us++) { 353 v = (v << sizeof (v)) + *us; 354 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 355 } 356 return (v % HSHSIZ1); 357 } 358 359 /* 360 * Lex has found a letter followed by zero or more letters or digits. 361 * It looks for a symbol in the symbol table with the same name. This 362 * symbol must either be a keyword or a symbol of the type required by 363 * symtyp (label, member, tag, ...). 364 * 365 * If it is a keyword, the token is returned. In some cases it is described 366 * more deeply by data written to yylval. 367 * 368 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct 369 * is stored in yylval. This struct contains the name of the symbol, it's 370 * length and hash value. If there is already a symbol of the same name 371 * and type in the symbol table, the sbuf struct also contains a pointer 372 * to the symbol table entry. 373 */ 374 static int 375 name() 376 { 377 char *s; 378 sbuf_t *sb; 379 sym_t *sym; 380 int tok; 381 382 sb = allocsb(); 383 sb->sb_name = yytext; 384 sb->sb_len = yyleng; 385 sb->sb_hash = hash(yytext); 386 387 if ((sym = search(sb)) != NULL && sym->s_keyw) { 388 freesb(sb); 389 return (keyw(sym)); 390 } 391 392 sb->sb_sym = sym; 393 394 if (sym != NULL) { 395 if (blklev < sym->s_blklev) 396 lerror("name() 1"); 397 sb->sb_name = sym->s_name; 398 sb->sb_len = strlen(sym->s_name); 399 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; 400 } else { 401 s = getblk(yyleng + 1); 402 (void)memcpy(s, yytext, yyleng + 1); 403 sb->sb_name = s; 404 sb->sb_len = yyleng; 405 tok = T_NAME; 406 } 407 408 yylval.y_sb = sb; 409 return (tok); 410 } 411 412 static sym_t * 413 search(sb) 414 sbuf_t *sb; 415 { 416 sym_t *sym; 417 418 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { 419 if (strcmp(sym->s_name, sb->sb_name) == 0) { 420 if (sym->s_keyw || sym->s_kind == symtyp) 421 return (sym); 422 } 423 } 424 425 return (NULL); 426 } 427 428 static int 429 keyw(sym) 430 sym_t *sym; 431 { 432 int t; 433 434 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { 435 yylval.y_scl = sym->s_scl; 436 } else if (t == T_TYPE || t == T_SOU) { 437 yylval.y_tspec = sym->s_tspec; 438 } else if (t == T_QUAL) { 439 yylval.y_tqual = sym->s_tqual; 440 } 441 return (t); 442 } 443 444 /* 445 * Convert a string representing an integer into internal representation. 446 * The value is returned in yylval. icon() (and yylex()) returns T_CON. 447 */ 448 static int 449 icon(base) 450 int base; 451 { 452 int l_suffix, u_suffix; 453 int len; 454 const char *cp; 455 char c, *eptr; 456 tspec_t typ; 457 u_long ul; 458 u_quad_t uq; 459 int ansiu; 460 static tspec_t contypes[2][3] = { 461 { INT, LONG, QUAD }, 462 { UINT, ULONG, UQUAD } 463 }; 464 465 cp = yytext; 466 len = yyleng; 467 468 /* skip 0x */ 469 if (base == 16) { 470 cp += 2; 471 len -= 2; 472 } 473 474 /* read suffixes */ 475 l_suffix = u_suffix = 0; 476 for ( ; ; ) { 477 if ((c = cp[len - 1]) == 'l' || c == 'L') { 478 l_suffix++; 479 } else if (c == 'u' || c == 'U') { 480 u_suffix++; 481 } else { 482 break; 483 } 484 len--; 485 } 486 if (l_suffix > 2 || u_suffix > 1) { 487 /* malformed integer constant */ 488 warning(251); 489 if (l_suffix > 2) 490 l_suffix = 2; 491 if (u_suffix > 1) 492 u_suffix = 1; 493 } 494 if (tflag && u_suffix != 0) { 495 /* suffix U is illegal in traditional C */ 496 warning(97); 497 } 498 typ = contypes[u_suffix][l_suffix]; 499 500 errno = 0; 501 if (l_suffix < 2) { 502 ul = strtoul(cp, &eptr, base); 503 } else { 504 uq = strtouq(cp, &eptr, base); 505 } 506 if (eptr != cp + len) 507 lerror("icon() 1"); 508 if (errno != 0) 509 /* integer constant out of range */ 510 warning(252); 511 512 /* 513 * If the value is to big for the current type, we must choose 514 * another type. 515 */ 516 ansiu = 0; 517 switch (typ) { 518 case INT: 519 if (ul <= INT_MAX) { 520 /* ok */ 521 } else if (ul <= (unsigned)UINT_MAX && base != 10) { 522 typ = UINT; 523 } else if (ul <= LONG_MAX) { 524 typ = LONG; 525 } else { 526 typ = ULONG; 527 } 528 if (typ == UINT || typ == ULONG) { 529 if (tflag) { 530 typ = LONG; 531 } else if (!sflag) { 532 /* 533 * Remember that the constant is unsigned 534 * only in ANSI C 535 */ 536 ansiu = 1; 537 } 538 } 539 break; 540 case UINT: 541 if (ul > (u_int)UINT_MAX) 542 typ = ULONG; 543 break; 544 case LONG: 545 if (ul > LONG_MAX && !tflag) { 546 typ = ULONG; 547 if (!sflag) 548 ansiu = 1; 549 } 550 break; 551 case QUAD: 552 if (uq > QUAD_MAX && !tflag) { 553 typ = UQUAD; 554 if (!sflag) 555 ansiu = 1; 556 } 557 break; 558 /* LINTED (enumeration values not handled in switch) */ 559 } 560 561 if (typ != QUAD && typ != UQUAD) { 562 if (isutyp(typ)) { 563 uq = ul; 564 } else { 565 uq = (quad_t)(long)ul; 566 } 567 } 568 569 uq = (u_quad_t)xsign((quad_t)uq, typ, -1); 570 571 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 572 yylval.y_val->v_ansiu = ansiu; 573 yylval.y_val->v_quad = (quad_t)uq; 574 575 return (T_CON); 576 } 577 578 /* 579 * Returns 1 if t is a signed type and the value is negative. 580 * 581 * len is the number of significant bits. If len is -1, len is set 582 * to the width of type t. 583 */ 584 int 585 sign(q, t, len) 586 quad_t q; 587 tspec_t t; 588 int len; 589 { 590 if (t == PTR || isutyp(t)) 591 return (0); 592 return (msb(q, t, len)); 593 } 594 595 int 596 msb(q, t, len) 597 quad_t q; 598 tspec_t t; 599 int len; 600 { 601 if (len <= 0) 602 len = size(t); 603 return ((q & qbmasks[len - 1]) != 0); 604 } 605 606 /* 607 * Extends the sign of q. 608 */ 609 quad_t 610 xsign(q, t, len) 611 quad_t q; 612 tspec_t t; 613 int len; 614 { 615 if (len <= 0) 616 len = size(t); 617 618 if (t == PTR || isutyp(t) || !sign(q, t, len)) { 619 q &= qlmasks[len]; 620 } else { 621 q |= qumasks[len]; 622 } 623 return (q); 624 } 625 626 /* 627 * Convert a string representing a floating point value into its interal 628 * representation. Type and value are returned in yylval. fcon() 629 * (and yylex()) returns T_CON. 630 * XXX Currently it is not possible to convert constants of type 631 * long double which are greater then DBL_MAX. 632 */ 633 static int 634 fcon() 635 { 636 const char *cp; 637 int len; 638 tspec_t typ; 639 char c, *eptr; 640 double d; 641 float f; 642 643 cp = yytext; 644 len = yyleng; 645 646 if ((c = cp[len - 1]) == 'f' || c == 'F') { 647 typ = FLOAT; 648 len--; 649 } else if (c == 'l' || c == 'L') { 650 typ = LDOUBLE; 651 len--; 652 } else { 653 typ = DOUBLE; 654 } 655 656 if (tflag && typ != DOUBLE) { 657 /* suffixes F and L are illegal in traditional C */ 658 warning(98); 659 } 660 661 errno = 0; 662 d = strtod(cp, &eptr); 663 if (eptr != cp + len) 664 lerror("fcon() 1"); 665 if (errno != 0) 666 /* floating-point constant out of range */ 667 warning(248); 668 669 if (typ == FLOAT) { 670 f = (float)d; 671 if (isinf(f)) { 672 /* floating-point constant out of range */ 673 warning(248); 674 f = f > 0 ? FLT_MAX : -FLT_MAX; 675 } 676 } 677 678 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; 679 if (typ == FLOAT) { 680 yylval.y_val->v_ldbl = f; 681 } else { 682 yylval.y_val->v_ldbl = d; 683 } 684 685 return (T_CON); 686 } 687 688 static int 689 operator(t, o) 690 int t; 691 op_t o; 692 { 693 yylval.y_op = o; 694 return (t); 695 } 696 697 /* 698 * Called if lex found a leading \'. 699 */ 700 static int 701 ccon() 702 { 703 int n, val, c; 704 char cv; 705 706 n = 0; 707 val = 0; 708 while ((c = getescc('\'')) >= 0) { 709 val = (val << CHAR_BIT) + c; 710 n++; 711 } 712 if (c == -2) { 713 /* unterminated character constant */ 714 error(253); 715 } else { 716 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { 717 /* too many characters in character constant */ 718 error(71); 719 } else if (n > 1) { 720 /* multi-character character constant */ 721 warning(294); 722 } else if (n == 0) { 723 /* empty character constant */ 724 error(73); 725 } 726 } 727 if (n == 1) { 728 cv = (char)val; 729 val = cv; 730 } 731 732 yylval.y_val = xcalloc(1, sizeof (val_t)); 733 yylval.y_val->v_tspec = INT; 734 yylval.y_val->v_quad = val; 735 736 return (T_CON); 737 } 738 739 /* 740 * Called if lex found a leading L\' 741 */ 742 static int 743 wccon() 744 { 745 static char buf[MB_CUR_MAX + 1]; 746 int i, c; 747 wchar_t wc; 748 749 i = 0; 750 while ((c = getescc('\'')) >= 0) { 751 if (i < MB_CUR_MAX) 752 buf[i] = (char)c; 753 i++; 754 } 755 756 wc = 0; 757 758 if (c == -2) { 759 /* unterminated character constant */ 760 error(253); 761 } else if (c == 0) { 762 /* empty character constant */ 763 error(73); 764 } else { 765 if (i > MB_CUR_MAX) { 766 i = MB_CUR_MAX; 767 /* too many characters in character constant */ 768 error(71); 769 } else { 770 buf[i] = '\0'; 771 (void)mbtowc(NULL, NULL, 0); 772 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) 773 /* invalid multibyte character */ 774 error(291); 775 } 776 } 777 778 yylval.y_val = xcalloc(1, sizeof (val_t)); 779 yylval.y_val->v_tspec = WCHAR; 780 yylval.y_val->v_quad = wc; 781 782 return (T_CON); 783 } 784 785 /* 786 * Read a character which is part of a character constant or of a string 787 * and handle escapes. 788 * 789 * The Argument is the character which delimits the character constant or 790 * string. 791 * 792 * Returns -1 if the end of the character constant or string is reached, 793 * -2 if the EOF is reached, and the charachter otherwise. 794 */ 795 static int 796 getescc(d) 797 int d; 798 { 799 static int pbc = -1; 800 int n, c, v; 801 802 if (pbc == -1) { 803 c = inpc(); 804 } else { 805 c = pbc; 806 pbc = -1; 807 } 808 if (c == d) 809 return (-1); 810 switch (c) { 811 case '\n': 812 /* newline in string or char constant */ 813 error(254); 814 return (-2); 815 case EOF: 816 return (-2); 817 case '\\': 818 switch (c = inpc()) { 819 case '"': 820 if (tflag && d == '\'') 821 /* \" inside character constant undef. ... */ 822 warning(262); 823 return ('"'); 824 case '\'': 825 return ('\''); 826 case '?': 827 if (tflag) 828 /* \? undefined in traditional C */ 829 warning(263); 830 return ('?'); 831 case '\\': 832 return ('\\'); 833 case 'a': 834 if (tflag) 835 /* \a undefined in traditional C */ 836 warning(81); 837 #ifdef __STDC__ 838 return ('\a'); 839 #else 840 return ('\007'); 841 #endif 842 case 'b': 843 return ('\b'); 844 case 'f': 845 return ('\f'); 846 case 'n': 847 return ('\n'); 848 case 'r': 849 return ('\r'); 850 case 't': 851 return ('\t'); 852 case 'v': 853 if (tflag) 854 /* \v undefined in traditional C */ 855 warning(264); 856 #ifdef __STDC__ 857 return ('\v'); 858 #else 859 return ('\013'); 860 #endif 861 case '8': case '9': 862 /* bad octal digit %c */ 863 warning(77, c); 864 /* FALLTHROUGH */ 865 case '0': case '1': case '2': case '3': 866 case '4': case '5': case '6': case '7': 867 n = 3; 868 v = 0; 869 do { 870 v = (v << 3) + (c - '0'); 871 c = inpc(); 872 } while (--n && isdigit(c) && (tflag || c <= '7')); 873 if (tflag && n > 0 && isdigit(c)) 874 /* bad octal digit %c */ 875 warning(77, c); 876 pbc = c; 877 if (v > UCHAR_MAX) { 878 /* character escape does not fit in char. */ 879 warning(76); 880 v &= CHAR_MASK; 881 } 882 return (v); 883 case 'x': 884 if (tflag) 885 /* \x undefined in traditional C */ 886 warning(82); 887 v = 0; 888 n = 0; 889 while ((c = inpc()) >= 0 && isxdigit(c)) { 890 c = isdigit(c) ? 891 c - '0' : toupper(c) - 'A' + 10; 892 v = (v << 4) + c; 893 if (n >= 0) { 894 if ((v & ~CHAR_MASK) != 0) { 895 /* overflow in hex escape */ 896 warning(75); 897 n = -1; 898 } else { 899 n++; 900 } 901 } 902 } 903 pbc = c; 904 if (n == 0) { 905 /* no hex digits follow \x */ 906 error(74); 907 } if (n == -1) { 908 v &= CHAR_MASK; 909 } 910 return (v); 911 case '\n': 912 return (getescc(d)); 913 case EOF: 914 return (-2); 915 default: 916 if (isprint(c)) { 917 /* dubious escape \%c */ 918 warning(79, c); 919 } else { 920 /* dubious escape \%o */ 921 warning(80, c); 922 } 923 } 924 } 925 return (c); 926 } 927 928 /* 929 * Called for preprocessor directives. Currently implemented are: 930 * # lineno 931 * # lineno "filename" 932 */ 933 static void 934 directive() 935 { 936 const char *cp, *fn; 937 char c, *eptr; 938 size_t fnl; 939 long ln; 940 static int first = 1; 941 942 /* Go to first non-whitespace after # */ 943 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) ; 944 945 if (!isdigit(c)) { 946 error: 947 /* undefined or invalid # directive */ 948 warning(255); 949 return; 950 } 951 ln = strtol(--cp, &eptr, 10); 952 if (cp == eptr) 953 goto error; 954 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') 955 goto error; 956 while ((c = *cp++) == ' ' || c == '\t') ; 957 if (c != '\0') { 958 if (c != '"') 959 goto error; 960 fn = cp; 961 while ((c = *cp) != '"' && c != '\0') 962 cp++; 963 if (c != '"') 964 goto error; 965 if ((fnl = cp++ - fn) > PATH_MAX) 966 goto error; 967 while ((c = *cp++) == ' ' || c == '\t') ; 968 #if 0 969 if (c != '\0') 970 warning("extra character(s) after directive"); 971 #endif 972 curr_pos.p_file = fnnalloc(fn, fnl); 973 /* 974 * If this is the first directive, the name is the name 975 * of the C source file as specified at the command line. 976 * It is written to the output file. 977 */ 978 if (first) { 979 csrc_pos.p_file = curr_pos.p_file; 980 outsrc(curr_pos.p_file); 981 first = 0; 982 } 983 } 984 curr_pos.p_line = (int)ln - 1; 985 if (curr_pos.p_file == csrc_pos.p_file) 986 csrc_pos.p_line = (int)ln - 1; 987 } 988 989 /* 990 * Handle lint comments. Following comments are currently understood: 991 * ARGSUSEDn 992 * CONSTCOND CONSTANTCOND CONSTANTCONDITION 993 * FALLTHRU FALLTHROUGH 994 * LINTLIBRARY 995 * LINTED NOSTRICT 996 * NOTREACHED 997 * PRINTFLIKEn 998 * PROTOLIB 999 * SCANFLIKEn 1000 * VARARGSn 1001 * If one of this comments is recognized, the arguments, if any, are 1002 * parsed and a function which handles this comment is called. 1003 */ 1004 static void 1005 comment() 1006 { 1007 int c, lc; 1008 static struct { 1009 const char *keywd; 1010 int arg; 1011 void (*func) __P((int)); 1012 } keywtab[] = { 1013 { "ARGSUSED", 1, argsused }, 1014 { "CONSTCOND", 0, constcond }, 1015 { "CONSTANTCOND", 0, constcond }, 1016 { "CONSTANTCONDITION", 0, constcond }, 1017 { "FALLTHRU", 0, fallthru }, 1018 { "FALLTHROUGH", 0, fallthru }, 1019 { "LINTLIBRARY", 0, lintlib }, 1020 { "LINTED", 0, linted }, 1021 { "NOSTRICT", 0, linted }, 1022 { "NOTREACHED", 0, notreach }, 1023 { "PRINTFLIKE", 1, printflike }, 1024 { "PROTOLIB", 1, protolib }, 1025 { "SCANFLIKE", 1, scanflike }, 1026 { "VARARGS", 1, varargs }, 1027 }; 1028 char keywd[32]; 1029 char arg[32]; 1030 int l, i, a; 1031 int eoc; 1032 1033 eoc = 0; 1034 1035 /* Skip white spaces after the start of the comment */ 1036 while ((c = inpc()) != EOF && isspace(c)) ; 1037 1038 /* Read the potential keyword to keywd */ 1039 l = 0; 1040 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { 1041 keywd[l++] = (char)c; 1042 c = inpc(); 1043 } 1044 keywd[l] = '\0'; 1045 1046 /* look for the keyword */ 1047 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { 1048 if (strcmp(keywtab[i].keywd, keywd) == 0) 1049 break; 1050 } 1051 if (i == sizeof (keywtab) / sizeof (keywtab[0])) 1052 goto skip_rest; 1053 1054 /* skip white spaces after the keyword */ 1055 while (c != EOF && isspace(c)) 1056 c = inpc(); 1057 1058 /* read the argument, if the keyword accepts one and there is one */ 1059 l = 0; 1060 if (keywtab[i].arg) { 1061 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { 1062 arg[l++] = (char)c; 1063 c = inpc(); 1064 } 1065 } 1066 arg[l] = '\0'; 1067 a = l != 0 ? atoi(arg) : -1; 1068 1069 /* skip white spaces after the argument */ 1070 while (c != EOF && isspace(c)) 1071 c = inpc(); 1072 1073 if (c != '*' || (c = inpc()) != '/') { 1074 if (keywtab[i].func != linted) 1075 /* extra characters in lint comment */ 1076 warning(257); 1077 } else { 1078 /* 1079 * remember that we have already found the end of the 1080 * comment 1081 */ 1082 eoc = 1; 1083 } 1084 1085 if (keywtab[i].func != NULL) 1086 (*keywtab[i].func)(a); 1087 1088 skip_rest: 1089 while (!eoc) { 1090 lc = c; 1091 if ((c = inpc()) == EOF) { 1092 /* unterminated comment */ 1093 error(256); 1094 break; 1095 } 1096 if (lc == '*' && c == '/') 1097 eoc = 1; 1098 } 1099 } 1100 1101 /* 1102 * Strings are stored in a dynamically alloceted buffer and passed 1103 * in yylval.y_xstrg to the parser. The parser or the routines called 1104 * by the parser are responsible for freeing this buffer. 1105 */ 1106 static int 1107 string() 1108 { 1109 u_char *s; 1110 int c; 1111 size_t len, max; 1112 strg_t *strg; 1113 1114 s = xmalloc(max = 64); 1115 1116 len = 0; 1117 while ((c = getescc('"')) >= 0) { 1118 /* +1 to reserve space for a trailing NUL character */ 1119 if (len + 1 == max) 1120 s = xrealloc(s, max *= 2); 1121 s[len++] = (char)c; 1122 } 1123 s[len] = '\0'; 1124 if (c == -2) 1125 /* unterminated string constant */ 1126 error(258); 1127 1128 strg = xcalloc(1, sizeof (strg_t)); 1129 strg->st_tspec = CHAR; 1130 strg->st_len = len; 1131 strg->st_cp = s; 1132 1133 yylval.y_strg = strg; 1134 return (T_STRING); 1135 } 1136 1137 static int 1138 wcstrg() 1139 { 1140 char *s; 1141 int c, i, n, wi; 1142 size_t len, max, wlen; 1143 wchar_t *ws; 1144 strg_t *strg; 1145 1146 s = xmalloc(max = 64); 1147 len = 0; 1148 while ((c = getescc('"')) >= 0) { 1149 /* +1 to save space for a trailing NUL character */ 1150 if (len + 1 >= max) 1151 s = xrealloc(s, max *= 2); 1152 s[len++] = (char)c; 1153 } 1154 s[len] = '\0'; 1155 if (c == -2) 1156 /* unterminated string constant */ 1157 error(258); 1158 1159 /* get length of wide character string */ 1160 (void)mblen(NULL, 0); 1161 for (i = 0, wlen = 0; i < len; i += n, wlen++) { 1162 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { 1163 /* invalid multibyte character */ 1164 error(291); 1165 break; 1166 } 1167 if (n == 0) 1168 n = 1; 1169 } 1170 1171 ws = xmalloc((wlen + 1) * sizeof (wchar_t)); 1172 1173 /* convert from multibyte to wide char */ 1174 (void)mbtowc(NULL, NULL, 0); 1175 for (i = 0, wi = 0; i < len; i += n, wi++) { 1176 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) 1177 break; 1178 if (n == 0) 1179 n = 1; 1180 } 1181 ws[wi] = 0; 1182 free(s); 1183 1184 strg = xcalloc(1, sizeof (strg_t)); 1185 strg->st_tspec = WCHAR; 1186 strg->st_len = wlen; 1187 strg->st_wcp = ws; 1188 1189 yylval.y_strg = strg; 1190 return (T_STRING); 1191 } 1192 1193 /* 1194 * As noted above the scanner does not create new symbol table entries 1195 * for symbols it cannot find in the symbol table. This is to avoid 1196 * putting undeclared symbols into the symbol table if a syntax error 1197 * occurs. 1198 * 1199 * getsym() is called as soon as it is probably ok to put the symbol to 1200 * the symbol table. This does not mean that it is not possible that 1201 * symbols are put to the symbol table which are than not completely 1202 * declared due to syntax errors. To avoid too many problems in this 1203 * case symbols get type int in getsym(). 1204 * 1205 * XXX calls to getsym() should be delayed until decl1*() is called 1206 */ 1207 sym_t * 1208 getsym(sb) 1209 sbuf_t *sb; 1210 { 1211 dinfo_t *di; 1212 char *s; 1213 sym_t *sym; 1214 1215 sym = sb->sb_sym; 1216 1217 /* 1218 * During member declaration it is possible that name() looked 1219 * for symbols of type FVFT, although it should have looked for 1220 * symbols of type FTAG. Same can happen for labels. Both cases 1221 * are compensated here. 1222 */ 1223 if (symtyp == FMOS || symtyp == FLAB) { 1224 if (sym == NULL || sym->s_kind == FVFT) 1225 sym = search(sb); 1226 } 1227 1228 if (sym != NULL) { 1229 if (sym->s_kind != symtyp) 1230 lerror("storesym() 1"); 1231 symtyp = FVFT; 1232 freesb(sb); 1233 return (sym); 1234 } 1235 1236 /* create a new symbol table entry */ 1237 1238 /* labels must always be allocated at level 1 (outhermost block) */ 1239 if (symtyp == FLAB) { 1240 sym = getlblk(1, sizeof (sym_t)); 1241 s = getlblk(1, sb->sb_len + 1); 1242 (void)memcpy(s, sb->sb_name, sb->sb_len + 1); 1243 sym->s_name = s; 1244 sym->s_blklev = 1; 1245 di = dcs; 1246 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) 1247 di = di->d_nxt; 1248 if (di->d_ctx != AUTO) 1249 lerror("storesym() 2"); 1250 } else { 1251 sym = getblk(sizeof (sym_t)); 1252 sym->s_name = sb->sb_name; 1253 sym->s_blklev = blklev; 1254 di = dcs; 1255 } 1256 1257 STRUCT_ASSIGN(sym->s_dpos, curr_pos); 1258 if ((sym->s_kind = symtyp) != FLAB) 1259 sym->s_type = gettyp(INT); 1260 1261 symtyp = FVFT; 1262 1263 if ((sym->s_link = symtab[sb->sb_hash]) != NULL) 1264 symtab[sb->sb_hash]->s_rlink = &sym->s_link; 1265 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; 1266 1267 *di->d_ldlsym = sym; 1268 di->d_ldlsym = &sym->s_dlnxt; 1269 1270 freesb(sb); 1271 return (sym); 1272 } 1273 1274 /* 1275 * Remove a symbol forever from the symbol table. s_blklev 1276 * is set to -1 to avoid that the symbol will later be put 1277 * back to the symbol table. 1278 */ 1279 void 1280 rmsym(sym) 1281 sym_t *sym; 1282 { 1283 if ((*sym->s_rlink = sym->s_link) != NULL) 1284 sym->s_link->s_rlink = sym->s_rlink; 1285 sym->s_blklev = -1; 1286 sym->s_link = NULL; 1287 } 1288 1289 /* 1290 * Remove a list of symbols declared at one level from the symbol 1291 * table. 1292 */ 1293 void 1294 rmsyms(syms) 1295 sym_t *syms; 1296 { 1297 sym_t *sym; 1298 1299 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { 1300 if (sym->s_blklev != -1) { 1301 if ((*sym->s_rlink = sym->s_link) != NULL) 1302 sym->s_link->s_rlink = sym->s_rlink; 1303 sym->s_link = NULL; 1304 sym->s_rlink = NULL; 1305 } 1306 } 1307 } 1308 1309 /* 1310 * Put a symbol into the symbol table 1311 */ 1312 void 1313 inssym(bl, sym) 1314 int bl; 1315 sym_t *sym; 1316 { 1317 int h; 1318 1319 h = hash(sym->s_name); 1320 if ((sym->s_link = symtab[h]) != NULL) 1321 symtab[h]->s_rlink = &sym->s_link; 1322 (symtab[h] = sym)->s_rlink = &symtab[h]; 1323 sym->s_blklev = bl; 1324 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) 1325 lerror("inssym()"); 1326 } 1327 1328 /* 1329 * Called at level 0 after syntax errors 1330 * Removes all symbols which are not declared at level 0 from the 1331 * symbol table. Also frees all memory which is not associated with 1332 * level 0. 1333 */ 1334 void 1335 cleanup() 1336 { 1337 sym_t *sym, *nsym; 1338 int i; 1339 1340 for (i = 0; i < HSHSIZ1; i++) { 1341 for (sym = symtab[i]; sym != NULL; sym = nsym) { 1342 nsym = sym->s_link; 1343 if (sym->s_blklev >= 1) { 1344 if ((*sym->s_rlink = nsym) != NULL) 1345 nsym->s_rlink = sym->s_rlink; 1346 } 1347 } 1348 } 1349 1350 for (i = mblklev; i > 0; i--) 1351 freelblk(i); 1352 } 1353 1354 /* 1355 * Create a new symbol with the name of an existing symbol. 1356 */ 1357 sym_t * 1358 pushdown(sym) 1359 sym_t *sym; 1360 { 1361 int h; 1362 sym_t *nsym; 1363 1364 h = hash(sym->s_name); 1365 nsym = getblk(sizeof (sym_t)); 1366 if (sym->s_blklev > blklev) 1367 lerror("pushdown()"); 1368 nsym->s_name = sym->s_name; 1369 STRUCT_ASSIGN(nsym->s_dpos, curr_pos); 1370 nsym->s_kind = sym->s_kind; 1371 nsym->s_blklev = blklev; 1372 1373 if ((nsym->s_link = symtab[h]) != NULL) 1374 symtab[h]->s_rlink = &nsym->s_link; 1375 (symtab[h] = nsym)->s_rlink = &symtab[h]; 1376 1377 *dcs->d_ldlsym = nsym; 1378 dcs->d_ldlsym = &nsym->s_dlnxt; 1379 1380 return (nsym); 1381 } 1382
Indexes created Thu Oct 02 01:09:59 GMT 2025