lexi.c revision 1.79
1 /* $NetBSD: lexi.c,v 1.79 2021/10/08 16:20:33 rillig Exp $ */ 2 3 /*- 4 * SPDX-License-Identifier: BSD-4-Clause 5 * 6 * Copyright (c) 1985 Sun Microsystems, Inc. 7 * Copyright (c) 1980, 1993 8 * The Regents of the University of California. All rights reserved. 9 * All rights reserved. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. All advertising materials mentioning features or use of this software 20 * must display the following acknowledgement: 21 * This product includes software developed by the University of 22 * California, Berkeley and its contributors. 23 * 4. Neither the name of the University nor the names of its contributors 24 * may be used to endorse or promote products derived from this software 25 * without specific prior written permission. 26 * 27 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 29 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 30 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 31 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 32 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 33 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 34 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 35 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 36 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 37 * SUCH DAMAGE. 38 */ 39 40 #if 0 41 static char sccsid[] = "@(#)lexi.c 8.1 (Berkeley) 6/6/93"; 42 #endif 43 44 #include <sys/cdefs.h> 45 #if defined(__NetBSD__) 46 __RCSID("$NetBSD: lexi.c,v 1.79 2021/10/08 16:20:33 rillig Exp $"); 47 #elif defined(__FreeBSD__) 48 __FBSDID("$FreeBSD: head/usr.bin/indent/lexi.c 337862 2018-08-15 18:19:45Z pstef $"); 49 #endif 50 51 #include <assert.h> 52 #include <stdio.h> 53 #include <ctype.h> 54 #include <stdlib.h> 55 #include <string.h> 56 #include <sys/param.h> 57 58 #include "indent.h" 59 60 /* must be sorted alphabetically, is used in binary search */ 61 static const struct keyword { 62 const char *name; 63 enum keyword_kind kind; 64 } keywords[] = { 65 {"_Bool", kw_type}, 66 {"_Complex", kw_type}, 67 {"_Imaginary", kw_type}, 68 {"auto", kw_storage_class}, 69 {"bool", kw_type}, 70 {"break", kw_jump}, 71 {"case", kw_case_or_default}, 72 {"char", kw_type}, 73 {"complex", kw_type}, 74 {"const", kw_type}, 75 {"continue", kw_jump}, 76 {"default", kw_case_or_default}, 77 {"do", kw_do_or_else}, 78 {"double", kw_type}, 79 {"else", kw_do_or_else}, 80 {"enum", kw_struct_or_union_or_enum}, 81 {"extern", kw_storage_class}, 82 {"float", kw_type}, 83 {"for", kw_for_or_if_or_while}, 84 {"global", kw_type}, 85 {"goto", kw_jump}, 86 {"if", kw_for_or_if_or_while}, 87 {"imaginary", kw_type}, 88 {"inline", kw_inline_or_restrict}, 89 {"int", kw_type}, 90 {"long", kw_type}, 91 {"offsetof", kw_offsetof}, 92 {"register", kw_storage_class}, 93 {"restrict", kw_inline_or_restrict}, 94 {"return", kw_jump}, 95 {"short", kw_type}, 96 {"signed", kw_type}, 97 {"sizeof", kw_sizeof}, 98 {"static", kw_storage_class}, 99 {"struct", kw_struct_or_union_or_enum}, 100 {"switch", kw_switch}, 101 {"typedef", kw_typedef}, 102 {"union", kw_struct_or_union_or_enum}, 103 {"unsigned", kw_type}, 104 {"void", kw_type}, 105 {"volatile", kw_type}, 106 {"while", kw_for_or_if_or_while} 107 }; 108 109 struct { 110 const char **items; 111 unsigned int len; 112 unsigned int cap; 113 } typenames; 114 115 /* 116 * The transition table below was rewritten by hand from lx's output, given 117 * the following definitions. lx is Katherine Flavel's lexer generator. 118 * 119 * O = /[0-7]/; D = /[0-9]/; NZ = /[1-9]/; 120 * H = /[a-f0-9]/i; B = /[0-1]/; HP = /0x/i; 121 * BP = /0b/i; E = /e[+\-]?/i D+; P = /p[+\-]?/i D+; 122 * FS = /[fl]/i; IS = /u/i /(l|L|ll|LL)/? | /(l|L|ll|LL)/ /u/i?; 123 * 124 * D+ E FS? -> $float; 125 * D* "." D+ E? FS? -> $float; 126 * D+ "." E? FS? -> $float; HP H+ IS? -> $int; 127 * HP H+ P FS? -> $float; NZ D* IS? -> $int; 128 * HP H* "." H+ P FS? -> $float; "0" O* IS? -> $int; 129 * HP H+ "." P FS -> $float; BP B+ IS? -> $int; 130 */ 131 /* INDENT OFF */ 132 static const unsigned char num_lex_state[][26] = { 133 /* examples: 134 00 135 s 0xx 136 t 00xaa 137 a 11 101100xxa.. 138 r 11ee0001101lbuuxx.a.pp 139 t.01.e+008bLuxll0Ll.aa.p+0 140 states: ABCDEFGHIJKLMNOPQRSTUVWXYZ */ 141 [0] = "uuiifuufiuuiiuiiiiiuiuuuuu", 142 [1] = "CEIDEHHHIJQ U Q VUVVZZZ", 143 [2] = "DEIDEHHHIJQ U Q VUVVZZZ", 144 [3] = "DEIDEHHHIJ U VUVVZZZ", 145 [4] = "DEJDEHHHJJ U VUVVZZZ", 146 [5] = " U VUVV ", 147 [6] = " K U VUVV ", 148 [7] = " FFF FF U VUVV ", 149 [8] = " f f U VUVV f", 150 [9] = " LLf fL PR Li L f", 151 [10] = " OOf fO S P O i O f", 152 [11] = " FFX ", 153 [12] = " MM M i iiM M ", 154 [13] = " N ", 155 [14] = " G Y ", 156 [15] = "B EE EE T W ", 157 /* ABCDEFGHIJKLMNOPQRSTUVWXYZ */ 158 }; 159 /* INDENT ON */ 160 161 static const uint8_t num_lex_row[] = { 162 ['0'] = 1, 163 ['1'] = 2, 164 ['2'] = 3, ['3'] = 3, ['4'] = 3, ['5'] = 3, ['6'] = 3, ['7'] = 3, 165 ['8'] = 4, ['9'] = 4, 166 ['A'] = 5, ['a'] = 5, ['C'] = 5, ['c'] = 5, ['D'] = 5, ['d'] = 5, 167 ['B'] = 6, ['b'] = 6, 168 ['E'] = 7, ['e'] = 7, 169 ['F'] = 8, ['f'] = 8, 170 ['L'] = 9, 171 ['l'] = 10, 172 ['P'] = 11, ['p'] = 11, 173 ['U'] = 12, ['u'] = 12, 174 ['X'] = 13, ['x'] = 13, 175 ['+'] = 14, ['-'] = 14, 176 ['.'] = 15, 177 }; 178 179 static char 180 inbuf_peek(void) 181 { 182 return *inp.s; 183 } 184 185 void 186 inbuf_skip(void) 187 { 188 inp.s++; 189 if (inp.s >= inp.e) 190 fill_buffer(); 191 } 192 193 char 194 inbuf_next(void) 195 { 196 char ch = inbuf_peek(); 197 inbuf_skip(); 198 return ch; 199 } 200 201 static void 202 check_size_token(size_t desired_size) 203 { 204 if (token.e + desired_size >= token.l) 205 buf_expand(&token, desired_size); 206 } 207 208 static int 209 cmp_keyword_by_name(const void *key, const void *elem) 210 { 211 return strcmp(key, ((const struct keyword *)elem)->name); 212 } 213 214 static int 215 cmp_type_by_name(const void *key, const void *elem) 216 { 217 return strcmp(key, *((const char *const *)elem)); 218 } 219 220 #ifdef debug 221 const char * 222 token_type_name(token_type ttype) 223 { 224 static const char *const name[] = { 225 "end_of_file", "newline", "lparen_or_lbracket", "rparen_or_rbracket", 226 "unary_op", "binary_op", "postfix_op", "question", 227 "case_label", "colon", 228 "semicolon", "lbrace", "rbrace", "ident", "comma", 229 "comment", "switch_expr", "preprocessing", "form_feed", "decl", 230 "keyword_for_if_while", "keyword_do_else", 231 "if_expr", "while_expr", "for_exprs", 232 "stmt", "stmt_list", "keyword_else", "keyword_do", "do_stmt", 233 "if_expr_stmt", "if_expr_stmt_else", "period", "string_prefix", 234 "storage_class", "funcname", "type_def", "keyword_struct_union_enum" 235 }; 236 237 assert(0 <= ttype && ttype < nitems(name)); 238 239 return name[ttype]; 240 } 241 242 static void 243 debug_print_buf(const char *name, const struct buffer *buf) 244 { 245 if (buf->s < buf->e) { 246 debug_printf(" %s ", name); 247 debug_vis_range("\"", buf->s, buf->e, "\""); 248 } 249 } 250 251 static token_type 252 lexi_end(token_type ttype) 253 { 254 debug_printf("in line %d, lexi returns '%s'", 255 line_no, token_type_name(ttype)); 256 debug_print_buf("token", &token); 257 debug_print_buf("label", &lab); 258 debug_print_buf("code", &code); 259 debug_print_buf("comment", &com); 260 debug_printf("\n"); 261 262 return ttype; 263 } 264 #else 265 #define lexi_end(tk) (tk) 266 #endif 267 268 static void 269 lex_number(void) 270 { 271 for (uint8_t s = 'A'; s != 'f' && s != 'i' && s != 'u';) { 272 uint8_t ch = (uint8_t)*inp.s; 273 if (ch >= nitems(num_lex_row) || num_lex_row[ch] == 0) 274 break; 275 276 uint8_t row = num_lex_row[ch]; 277 if (num_lex_state[row][s - 'A'] == ' ') { 278 /*- 279 * num_lex_state[0][s - 'A'] now indicates the type: 280 * f = floating, i = integer, u = unknown 281 */ 282 break; 283 } 284 285 s = num_lex_state[row][s - 'A']; 286 check_size_token(1); 287 *token.e++ = inbuf_next(); 288 } 289 } 290 291 static void 292 lex_word(void) 293 { 294 while (isalnum((unsigned char)*inp.s) || 295 *inp.s == '\\' || 296 *inp.s == '_' || *inp.s == '$') { 297 298 /* fill_buffer() terminates buffer with newline */ 299 if (*inp.s == '\\') { 300 if (inp.s[1] == '\n') { 301 inp.s += 2; 302 if (inp.s >= inp.e) 303 fill_buffer(); 304 } else 305 break; 306 } 307 308 check_size_token(1); 309 *token.e++ = inbuf_next(); 310 } 311 } 312 313 static void 314 lex_char_or_string(void) 315 { 316 for (char delim = *token.s;;) { 317 if (*inp.s == '\n') { 318 diag(1, "Unterminated literal"); 319 return; 320 } 321 322 check_size_token(2); 323 *token.e++ = inbuf_next(); 324 if (token.e[-1] == delim) 325 return; 326 327 if (token.e[-1] == '\\') { 328 if (*inp.s == '\n') 329 ++line_no; 330 *token.e++ = inbuf_next(); 331 } 332 } 333 } 334 335 /* 336 * This hack attempts to guess whether the current token is in fact a 337 * declaration keyword -- one that has been defined by typedef. 338 */ 339 static bool 340 probably_typedef(const struct parser_state *state) 341 { 342 if (state->p_l_follow != 0) 343 return false; 344 if (state->block_init || state->in_stmt) 345 return false; 346 if (inp.s[0] == '*' && inp.s[1] != '=') 347 goto maybe; 348 if (isalpha((unsigned char)*inp.s)) 349 goto maybe; 350 return false; 351 maybe: 352 return state->last_token == semicolon || 353 state->last_token == lbrace || 354 state->last_token == rbrace; 355 } 356 357 static bool 358 is_typename(void) 359 { 360 if (opt.auto_typedefs) { 361 const char *u; 362 if ((u = strrchr(token.s, '_')) != NULL && strcmp(u, "_t") == 0) 363 return true; 364 } 365 366 if (typenames.len == 0) 367 return false; 368 return bsearch(token.s, typenames.items, (size_t)typenames.len, 369 sizeof(typenames.items[0]), cmp_type_by_name) != NULL; 370 } 371 372 /* Reads the next token, placing it in the global variable "token". */ 373 token_type 374 lexi(struct parser_state *state) 375 { 376 bool unary_delim; /* whether the current token forces a 377 * following operator to be unary */ 378 token_type ttype; 379 380 token.e = token.s; /* point to start of place to save token */ 381 unary_delim = false; 382 state->col_1 = state->last_nl; /* tell world that this token started 383 * in column 1 iff the last thing 384 * scanned was a newline */ 385 state->last_nl = false; 386 387 while (is_hspace(*inp.s)) { 388 state->col_1 = false; 389 inbuf_skip(); 390 } 391 392 /* Scan an alphanumeric token */ 393 if (isalnum((unsigned char)*inp.s) || 394 *inp.s == '_' || *inp.s == '$' || 395 (inp.s[0] == '.' && isdigit((unsigned char)inp.s[1]))) { 396 struct keyword *kw; 397 398 if (isdigit((unsigned char)*inp.s) || 399 (inp.s[0] == '.' && isdigit((unsigned char)inp.s[1]))) { 400 lex_number(); 401 } else { 402 lex_word(); 403 } 404 *token.e = '\0'; 405 406 if (token.s[0] == 'L' && token.s[1] == '\0' && 407 (*inp.s == '"' || *inp.s == '\'')) 408 return lexi_end(string_prefix); 409 410 while (is_hspace(inbuf_peek())) 411 inbuf_skip(); 412 state->keyword = kw_0; 413 414 if (state->last_token == keyword_struct_union_enum && 415 state->p_l_follow == 0) { 416 state->last_u_d = true; 417 return lexi_end(decl); 418 } 419 /* 420 * Operator after identifier is binary unless last token was 'struct' 421 */ 422 state->last_u_d = (state->last_token == keyword_struct_union_enum); 423 424 kw = bsearch(token.s, keywords, nitems(keywords), 425 sizeof(keywords[0]), cmp_keyword_by_name); 426 if (kw == NULL) { 427 if (is_typename()) { 428 state->keyword = kw_type; 429 state->last_u_d = true; 430 goto found_typename; 431 } 432 433 } else { /* we have a keyword */ 434 state->keyword = kw->kind; 435 state->last_u_d = true; 436 437 switch (kw->kind) { 438 case kw_switch: 439 return lexi_end(switch_expr); 440 441 case kw_case_or_default: 442 return lexi_end(case_label); 443 444 case kw_struct_or_union_or_enum: 445 case kw_type: 446 found_typename: 447 if (state->p_l_follow != 0) { 448 /* inside parens: cast, param list, offsetof or sizeof */ 449 state->cast_mask |= (1 << state->p_l_follow) & ~state->not_cast_mask; 450 } 451 if (state->last_token == period || state->last_token == unary_op) { 452 state->keyword = kw_0; 453 break; 454 } 455 if (kw != NULL && kw->kind == kw_struct_or_union_or_enum) 456 return lexi_end(keyword_struct_union_enum); 457 if (state->p_l_follow != 0) 458 break; 459 return lexi_end(decl); 460 461 case kw_for_or_if_or_while: 462 return lexi_end(keyword_for_if_while); 463 464 case kw_do_or_else: 465 return lexi_end(keyword_do_else); 466 467 case kw_storage_class: 468 return lexi_end(storage_class); 469 470 case kw_typedef: 471 return lexi_end(type_def); 472 473 default: /* all others are treated like any other 474 * identifier */ 475 return lexi_end(ident); 476 } /* end of switch */ 477 } /* end of if (found_it) */ 478 479 if (*inp.s == '(' && state->tos <= 1 && state->ind_level == 0 && 480 !state->in_parameter_declaration && !state->block_init) { 481 482 for (char *tp = inp.s; tp < inp.e;) 483 if (*tp++ == ')' && (*tp == ';' || *tp == ',')) 484 goto not_proc; 485 486 strncpy(state->procname, token.s, sizeof state->procname - 1); 487 if (state->in_decl) 488 state->in_parameter_declaration = true; 489 return lexi_end(funcname); 490 not_proc:; 491 492 } else if (probably_typedef(state)) { 493 state->keyword = kw_type; 494 state->last_u_d = true; 495 return lexi_end(decl); 496 } 497 498 if (state->last_token == decl) /* if this is a declared variable, 499 * then following sign is unary */ 500 state->last_u_d = true; /* will make "int a -1" work */ 501 502 return lexi_end(ident); /* the ident is not in the list */ 503 } /* end of processing for alphanum character */ 504 505 /* Scan a non-alphanumeric token */ 506 507 check_size_token(3); /* things like "<<=" */ 508 *token.e++ = inbuf_next(); /* if it is only a one-character token, it is 509 * moved here */ 510 *token.e = '\0'; 511 512 switch (*token.s) { 513 case '\n': 514 unary_delim = state->last_u_d; 515 state->last_nl = true; /* remember that we just had a newline */ 516 /* if data has been exhausted, the newline is a dummy. */ 517 ttype = had_eof ? end_of_file : newline; 518 break; 519 520 case '\'': 521 case '"': 522 lex_char_or_string(); 523 ttype = ident; 524 break; 525 526 case '(': 527 case '[': 528 unary_delim = true; 529 ttype = lparen_or_lbracket; 530 break; 531 532 case ')': 533 case ']': 534 ttype = rparen_or_rbracket; 535 break; 536 537 case '#': 538 unary_delim = state->last_u_d; 539 ttype = preprocessing; 540 break; 541 542 case '?': 543 unary_delim = true; 544 ttype = question; 545 break; 546 547 case ':': 548 ttype = colon; 549 unary_delim = true; 550 break; 551 552 case ';': 553 unary_delim = true; 554 ttype = semicolon; 555 break; 556 557 case '{': 558 unary_delim = true; 559 ttype = lbrace; 560 break; 561 562 case '}': 563 unary_delim = true; 564 ttype = rbrace; 565 break; 566 567 case '\f': 568 unary_delim = state->last_u_d; 569 state->last_nl = true; /* remember this, so we can set 'state->col_1' 570 * right */ 571 ttype = form_feed; 572 break; 573 574 case ',': 575 unary_delim = true; 576 ttype = comma; 577 break; 578 579 case '.': 580 unary_delim = false; 581 ttype = period; 582 break; 583 584 case '-': 585 case '+': /* check for -, +, --, ++ */ 586 ttype = state->last_u_d ? unary_op : binary_op; 587 unary_delim = true; 588 589 if (*inp.s == token.s[0]) { 590 /* check for doubled character */ 591 *token.e++ = *inp.s++; 592 /* buffer overflow will be checked at end of loop */ 593 if (state->last_token == ident || 594 state->last_token == rparen_or_rbracket) { 595 ttype = state->last_u_d ? unary_op : postfix_op; 596 /* check for following ++ or -- */ 597 unary_delim = false; 598 } 599 600 } else if (*inp.s == '=') { 601 /* check for operator += */ 602 *token.e++ = *inp.s++; 603 604 } else if (*inp.s == '>') { 605 /* check for operator -> */ 606 *token.e++ = *inp.s++; 607 unary_delim = false; 608 ttype = unary_op; 609 state->want_blank = false; 610 } 611 break; /* buffer overflow will be checked at end of 612 * switch */ 613 614 case '=': 615 if (state->in_or_st) 616 state->block_init = true; 617 if (*inp.s == '=') { /* == */ 618 *token.e++ = *inp.s++; 619 *token.e = '\0'; 620 } 621 ttype = binary_op; 622 unary_delim = true; 623 break; 624 625 case '>': 626 case '<': 627 case '!': /* ops like <, <<, <=, !=, etc */ 628 if (*inp.s == '>' || *inp.s == '<' || *inp.s == '=') 629 *token.e++ = inbuf_next(); 630 if (*inp.s == '=') 631 *token.e++ = *inp.s++; 632 ttype = state->last_u_d ? unary_op : binary_op; 633 unary_delim = true; 634 break; 635 636 case '*': 637 unary_delim = true; 638 if (!state->last_u_d) { 639 if (*inp.s == '=') 640 *token.e++ = *inp.s++; 641 ttype = binary_op; 642 break; 643 } 644 645 while (*inp.s == '*' || isspace((unsigned char)*inp.s)) { 646 if (*inp.s == '*') { 647 check_size_token(1); 648 *token.e++ = *inp.s; 649 } 650 inbuf_skip(); 651 } 652 653 if (ps.in_decl) { 654 char *tp = inp.s; 655 656 while (isalpha((unsigned char)*tp) || 657 isspace((unsigned char)*tp)) { 658 if (++tp >= inp.e) 659 fill_buffer(); 660 } 661 if (*tp == '(') 662 ps.procname[0] = ' '; 663 } 664 665 ttype = unary_op; 666 break; 667 668 default: 669 if (token.s[0] == '/' && (*inp.s == '*' || *inp.s == '/')) { 670 /* it is start of comment */ 671 *token.e++ = inbuf_next(); 672 673 ttype = comment; 674 unary_delim = state->last_u_d; 675 break; 676 } 677 678 while (token.e[-1] == *inp.s || *inp.s == '=') { 679 /* 680 * handle ||, &&, etc, and also things as in int *****i 681 */ 682 check_size_token(1); 683 *token.e++ = inbuf_next(); 684 } 685 686 ttype = state->last_u_d ? unary_op : binary_op; 687 unary_delim = true; 688 } 689 690 if (inp.s >= inp.e) /* check for input buffer empty */ 691 fill_buffer(); 692 693 state->last_u_d = unary_delim; 694 695 check_size_token(1); 696 *token.e = '\0'; 697 698 return lexi_end(ttype); 699 } 700 701 static int 702 insert_pos(const char *key, const char **arr, unsigned int len) 703 { 704 int lo = 0; 705 int hi = (int)len - 1; 706 707 while (lo <= hi) { 708 int mid = (int)((unsigned)(lo + hi) >> 1); 709 int cmp = strcmp(arr[mid], key); 710 if (cmp < 0) 711 lo = mid + 1; 712 else if (cmp > 0) 713 hi = mid - 1; 714 else 715 return mid; 716 } 717 return -(lo + 1); 718 } 719 720 void 721 add_typename(const char *name) 722 { 723 if (typenames.len >= typenames.cap) { 724 typenames.cap = 16 + 2 * typenames.cap; 725 typenames.items = xrealloc(typenames.items, 726 sizeof(typenames.items[0]) * typenames.cap); 727 } 728 729 int pos = insert_pos(name, typenames.items, typenames.len); 730 if (pos >= 0) 731 return; /* already in the list */ 732 733 pos = -(pos + 1); 734 memmove(typenames.items + pos + 1, typenames.items + pos, 735 sizeof(typenames.items[0]) * (typenames.len++ - (unsigned)pos)); 736 typenames.items[pos] = xstrdup(name); 737 } 738
Indexes created Thu Sep 25 16:09:42 GMT 2025