lexi.c revision 1.173
1 /* $NetBSD: lexi.c,v 1.173 2023/05/11 09:28:53 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 1.173 2023/05/11 09:28:53 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 <stdlib.h> 52 #include <string.h> 53 54 #include "indent.h" 55 56 /* 57 * While inside lexi_alnum, this constant just marks a type, independently of 58 * the parentheses level. 59 */ 60 #define lsym_type lsym_type_outside_parentheses 61 62 /* must be sorted alphabetically, is used in binary search */ 63 static const struct keyword { 64 const char *name; 65 lexer_symbol lsym; 66 } keywords[] = { 67 {"_Bool", lsym_type}, 68 {"_Complex", lsym_type}, 69 {"_Imaginary", lsym_type}, 70 {"auto", lsym_storage_class}, 71 {"bool", lsym_type}, 72 {"break", lsym_word}, 73 {"case", lsym_case_label}, 74 {"char", lsym_type}, 75 {"complex", lsym_type}, 76 {"const", lsym_type}, 77 {"continue", lsym_word}, 78 {"default", lsym_case_label}, 79 {"do", lsym_do}, 80 {"double", lsym_type}, 81 {"else", lsym_else}, 82 {"enum", lsym_tag}, 83 {"extern", lsym_storage_class}, 84 {"float", lsym_type}, 85 {"for", lsym_for}, 86 {"goto", lsym_word}, 87 {"if", lsym_if}, 88 {"imaginary", lsym_type}, 89 {"inline", lsym_word}, 90 {"int", lsym_type}, 91 {"long", lsym_type}, 92 {"offsetof", lsym_offsetof}, 93 {"register", lsym_storage_class}, 94 {"restrict", lsym_word}, 95 {"return", lsym_return}, 96 {"short", lsym_type}, 97 {"signed", lsym_type}, 98 {"sizeof", lsym_sizeof}, 99 {"static", lsym_storage_class}, 100 {"struct", lsym_tag}, 101 {"switch", lsym_switch}, 102 {"typedef", lsym_typedef}, 103 {"union", lsym_tag}, 104 {"unsigned", lsym_type}, 105 {"void", lsym_type}, 106 {"volatile", lsym_type}, 107 {"while", lsym_while} 108 }; 109 110 static struct { 111 const char **items; 112 unsigned int len; 113 unsigned int cap; 114 } typenames; 115 116 /* 117 * The transition table below was rewritten by hand from lx's output, given 118 * the following definitions. lx is Katherine Flavel's lexer generator. 119 * 120 * O = /[0-7]/; D = /[0-9]/; NZ = /[1-9]/; 121 * H = /[a-f0-9]/i; B = /[0-1]/; HP = /0x/i; 122 * BP = /0b/i; E = /e[+\-]?/i D+; P = /p[+\-]?/i D+; 123 * FS = /[fl]/i; IS = /u/i /(l|L|ll|LL)/? | /(l|L|ll|LL)/ /u/i?; 124 * 125 * D+ E FS? -> $float; 126 * D* "." D+ E? FS? -> $float; 127 * D+ "." E? FS? -> $float; HP H+ IS? -> $int; 128 * HP H+ P FS? -> $float; NZ D* IS? -> $int; 129 * HP H* "." H+ P FS? -> $float; "0" O* IS? -> $int; 130 * HP H+ "." P FS -> $float; BP B+ IS? -> $int; 131 */ 132 /* INDENT OFF */ 133 static const unsigned char lex_number_state[][26] = { 134 /* examples: 135 00 136 s 0xx 137 t 00xaa 138 a 11 101100xxa.. 139 r 11ee0001101lbuuxx.a.pp 140 t.01.e+008bLuxll0Ll.aa.p+0 141 states: ABCDEFGHIJKLMNOPQRSTUVWXYZ */ 142 [0] = "uuiifuufiuuiiuiiiiiuiuuuuu", /* (other) */ 143 [1] = "CEIDEHHHIJQ U Q VUVVZZZ", /* 0 */ 144 [2] = "DEIDEHHHIJQ U Q VUVVZZZ", /* 1 */ 145 [3] = "DEIDEHHHIJ U VUVVZZZ", /* 2 3 4 5 6 7 */ 146 [4] = "DEJDEHHHJJ U VUVVZZZ", /* 8 9 */ 147 [5] = " U VUVV ", /* A a C c D d */ 148 [6] = " K U VUVV ", /* B b */ 149 [7] = " FFF FF U VUVV ", /* E e */ 150 [8] = " f f U VUVV f", /* F f */ 151 [9] = " LLf fL PR Li L f", /* L */ 152 [10] = " OOf fO S P O i O f", /* l */ 153 [11] = " FFX ", /* P p */ 154 [12] = " MM M i iiM M ", /* U u */ 155 [13] = " N ", /* X x */ 156 [14] = " G Y ", /* + - */ 157 [15] = "B EE EE T W ", /* . */ 158 /* ABCDEFGHIJKLMNOPQRSTUVWXYZ */ 159 }; 160 /* INDENT ON */ 161 162 static const unsigned char lex_number_row[] = { 163 ['0'] = 1, 164 ['1'] = 2, 165 ['2'] = 3, ['3'] = 3, ['4'] = 3, ['5'] = 3, ['6'] = 3, ['7'] = 3, 166 ['8'] = 4, ['9'] = 4, 167 ['A'] = 5, ['a'] = 5, ['C'] = 5, ['c'] = 5, ['D'] = 5, ['d'] = 5, 168 ['B'] = 6, ['b'] = 6, 169 ['E'] = 7, ['e'] = 7, 170 ['F'] = 8, ['f'] = 8, 171 ['L'] = 9, 172 ['l'] = 10, 173 ['P'] = 11, ['p'] = 11, 174 ['U'] = 12, ['u'] = 12, 175 ['X'] = 13, ['x'] = 13, 176 ['+'] = 14, ['-'] = 14, 177 ['.'] = 15, 178 }; 179 180 static void 181 check_size_token(size_t desired_size) 182 { 183 if (token.e + desired_size >= token.l) 184 buf_expand(&token, desired_size); 185 } 186 187 static void 188 token_add_char(char ch) 189 { 190 check_size_token(1); 191 *token.e++ = ch; 192 } 193 194 #ifdef debug 195 static const char * 196 lsym_name(lexer_symbol sym) 197 { 198 static const char *const name[] = { 199 "eof", 200 "preprocessing", 201 "newline", 202 "form_feed", 203 "comment", 204 "lparen_or_lbracket", 205 "rparen_or_rbracket", 206 "lbrace", 207 "rbrace", 208 "period", 209 "unary_op", 210 "binary_op", 211 "postfix_op", 212 "question", 213 "colon", 214 "comma", 215 "semicolon", 216 "typedef", 217 "storage_class", 218 "type_outside_parentheses", 219 "type_in_parentheses", 220 "tag", 221 "case_label", 222 "sizeof", 223 "offsetof", 224 "word", 225 "funcname", 226 "do", 227 "else", 228 "for", 229 "if", 230 "switch", 231 "while", 232 "return", 233 }; 234 235 return name[sym]; 236 } 237 238 static void 239 debug_print_buf(const char *name, const struct buffer *buf) 240 { 241 if (buf->s < buf->e) { 242 debug_printf("%s ", name); 243 debug_vis_range("\"", buf->s, buf->e, "\"\n"); 244 } 245 } 246 247 static bool 248 debug_full_parser_state(void) 249 { 250 return true; 251 } 252 253 #define debug_ps_bool(name) \ 254 if (ps.name != prev_ps.name) \ 255 debug_println("[%c] -> [%c] ps." #name, \ 256 prev_ps.name ? 'x' : ' ', ps.name ? 'x' : ' '); \ 257 else if (debug_full_parser_state()) \ 258 debug_println(" [%c] ps." #name, ps.name ? 'x' : ' ') 259 #define debug_ps_int(name) \ 260 if (ps.name != prev_ps.name) \ 261 debug_println("%3d -> %3d ps." #name, prev_ps.name, ps.name); \ 262 else if (debug_full_parser_state()) \ 263 debug_println(" %3d ps." #name, ps.name) 264 265 static bool 266 ps_paren_has_changed(const struct parser_state *prev_ps) 267 { 268 const paren_level_props *prev = prev_ps->paren, *curr = ps.paren; 269 270 if (prev_ps->nparen != ps.nparen) 271 return true; 272 273 for (int i = 0; i < ps.nparen; i++) { 274 if (curr[i].indent != prev[i].indent || 275 curr[i].maybe_cast != prev[i].maybe_cast || 276 curr[i].no_cast != prev[i].no_cast) 277 return true; 278 } 279 return false; 280 } 281 282 static void 283 debug_ps_paren(const struct parser_state *prev_ps) 284 { 285 if (!debug_full_parser_state() && !ps_paren_has_changed(prev_ps)) 286 return; 287 288 debug_printf(" ps.paren:"); 289 for (int i = 0; i < ps.nparen; i++) { 290 const paren_level_props *props = ps.paren + i; 291 const char *cast = props->no_cast ? "(no cast)" 292 : props->maybe_cast ? "(cast)" 293 : ""; 294 debug_printf(" %s%d", cast, props->indent); 295 } 296 if (ps.nparen == 0) 297 debug_printf(" none"); 298 debug_println(""); 299 } 300 301 static void 302 debug_lexi(lexer_symbol lsym) 303 { 304 /* 305 * Watch out for 'rolled back parser state' in the debug output; the 306 * differences around these are unreliable. 307 */ 308 static struct parser_state prev_ps; 309 310 debug_println(""); 311 debug_printf("line %d: %s", line_no, lsym_name(lsym)); 312 debug_vis_range(" \"", token.s, token.e, "\"\n"); 313 314 debug_print_buf("label", &lab); 315 debug_print_buf("code", &code); 316 debug_print_buf("comment", &com); 317 318 debug_println(" ps.prev_token = %s", lsym_name(ps.prev_token)); 319 debug_ps_bool(next_col_1); 320 debug_ps_bool(curr_col_1); 321 debug_ps_bool(next_unary); 322 debug_ps_bool(is_function_definition); 323 debug_ps_bool(want_blank); 324 debug_ps_int(line_start_nparen); 325 debug_ps_int(nparen); 326 debug_ps_paren(&prev_ps); 327 328 debug_ps_int(comment_delta); 329 debug_ps_int(n_comment_delta); 330 debug_ps_int(com_ind); 331 332 debug_ps_bool(block_init); 333 debug_ps_int(block_init_level); 334 debug_ps_bool(init_or_struct); 335 336 debug_ps_int(ind_level); 337 debug_ps_int(ind_level_follow); 338 339 debug_ps_int(decl_level); 340 debug_ps_bool(decl_on_line); 341 debug_ps_bool(in_decl); 342 debug_ps_int(just_saw_decl); 343 debug_ps_bool(in_func_def_params); 344 debug_ps_bool(decl_indent_done); 345 346 debug_ps_bool(in_stmt_or_decl); 347 debug_ps_bool(in_stmt_cont); 348 debug_ps_bool(is_case_label); 349 350 prev_ps = ps; 351 } 352 #endif 353 354 static lexer_symbol 355 lexi_end(lexer_symbol lsym) 356 { 357 #ifdef debug 358 debug_lexi(lsym); 359 #endif 360 return lsym; 361 } 362 363 static void 364 lex_number(void) 365 { 366 for (unsigned char s = 'A'; s != 'f' && s != 'i' && s != 'u';) { 367 unsigned char ch = (unsigned char)inp_peek(); 368 if (ch >= array_length(lex_number_row) || lex_number_row[ch] == 0) 369 break; 370 371 unsigned char row = lex_number_row[ch]; 372 if (lex_number_state[row][s - 'A'] == ' ') { 373 /*- 374 * lex_number_state[0][s - 'A'] now indicates the type: 375 * f = floating, i = integer, u = unknown 376 */ 377 return; 378 } 379 380 s = lex_number_state[row][s - 'A']; 381 token_add_char(inp_next()); 382 } 383 } 384 385 static bool 386 is_identifier_start(char ch) 387 { 388 return ch_isalpha(ch) || ch == '_' || ch == '$'; 389 } 390 391 static bool 392 is_identifier_part(char ch) 393 { 394 return ch_isalnum(ch) || ch == '_' || ch == '$'; 395 } 396 397 static void 398 lex_word(void) 399 { 400 for (;;) { 401 if (is_identifier_part(inp_peek())) 402 token_add_char(inp_next()); 403 else if (inp_peek() == '\\' && inp_lookahead(1) == '\n') { 404 inp_skip(); 405 inp_skip(); 406 } else 407 return; 408 } 409 } 410 411 static void 412 lex_char_or_string(void) 413 { 414 for (char delim = token.e[-1];;) { 415 if (inp_peek() == '\n') { 416 diag(1, "Unterminated literal"); 417 return; 418 } 419 420 token_add_char(inp_next()); 421 if (token.e[-1] == delim) 422 return; 423 424 if (token.e[-1] == '\\') { 425 if (inp_peek() == '\n') 426 ++line_no; 427 token_add_char(inp_next()); 428 } 429 } 430 } 431 432 /* Guess whether the current token is a declared type. */ 433 static bool 434 probably_typename(void) 435 { 436 if (ps.prev_token == lsym_storage_class) 437 return true; 438 if (ps.block_init) 439 return false; 440 if (ps.in_stmt_or_decl) /* XXX: this condition looks incorrect */ 441 return false; 442 if (inp_peek() == '*' && inp_lookahead(1) != '=') 443 goto maybe; 444 /* XXX: is_identifier_start */ 445 if (ch_isalpha(inp_peek())) 446 goto maybe; 447 return false; 448 maybe: 449 return ps.prev_token == lsym_semicolon || 450 ps.prev_token == lsym_lbrace || 451 ps.prev_token == lsym_rbrace; 452 } 453 454 static int 455 bsearch_typenames(const char *key) 456 { 457 const char **arr = typenames.items; 458 int lo = 0; 459 int hi = (int)typenames.len - 1; 460 461 while (lo <= hi) { 462 int mid = (int)((unsigned)(lo + hi) >> 1); 463 int cmp = strcmp(arr[mid], key); 464 if (cmp < 0) 465 lo = mid + 1; 466 else if (cmp > 0) 467 hi = mid - 1; 468 else 469 return mid; 470 } 471 return -(lo + 1); 472 } 473 474 static bool 475 is_typename(void) 476 { 477 if (opt.auto_typedefs && 478 token.e - token.s >= 2 && memcmp(token.e - 2, "_t", 2) == 0) 479 return true; 480 481 return bsearch_typenames(token.s) >= 0; 482 } 483 484 static int 485 cmp_keyword_by_name(const void *key, const void *elem) 486 { 487 return strcmp(key, ((const struct keyword *)elem)->name); 488 } 489 490 /* 491 * Looking at something like 'function_name(...)' in a line, guess whether 492 * this starts a function definition or a declaration. 493 */ 494 static bool 495 probably_looking_at_definition(void) 496 { 497 int paren_level = 0; 498 for (const char *p = inp_p(), *e = inp_line_end(); p < e; p++) { 499 if (*p == '(') 500 paren_level++; 501 if (*p == ')' && --paren_level == 0) { 502 p++; 503 504 while (p < e && (ch_isspace(*p) || is_identifier_part(*p))) 505 p++; /* '__dead' or '__unused' */ 506 507 if (p == e) /* func(...) */ 508 break; 509 if (*p == ';') /* func(...); */ 510 return false; 511 if (*p == ',') /* double abs(), pi; */ 512 return false; 513 if (*p == '(') /* func(...) __attribute__((...)) */ 514 paren_level++; /* func(...) __printflike(...) */ 515 else 516 break; /* func(...) { ... */ 517 } 518 } 519 520 /* 521 * To further reduce the cases where indent wrongly treats an incomplete 522 * function declaration as a function definition, thus adding a newline 523 * before the function name, it may be worth looking for parameter names, 524 * as these are often omitted in function declarations and only included 525 * in function definitions. Or just increase the lookahead to more than 526 * just the current line of input, until the next '{'. 527 */ 528 return true; 529 } 530 531 /* Read an alphanumeric token into 'token', or return lsym_eof. */ 532 static lexer_symbol 533 lexi_alnum(void) 534 { 535 if (ch_isdigit(inp_peek()) || 536 (inp_peek() == '.' && ch_isdigit(inp_lookahead(1)))) { 537 lex_number(); 538 } else if (is_identifier_start(inp_peek())) { 539 lex_word(); 540 541 if (token.s[0] == 'L' && token.e - token.s == 1 && 542 (inp_peek() == '"' || inp_peek() == '\'')) { 543 token_add_char(inp_next()); 544 lex_char_or_string(); 545 ps.next_unary = false; 546 547 check_size_token(1); 548 *token.e = '\0'; 549 550 return lsym_word; 551 } 552 } else 553 return lsym_eof; /* just as a placeholder */ 554 555 *token.e = '\0'; 556 557 while (ch_isblank(inp_peek())) 558 inp_skip(); 559 560 ps.next_unary = ps.prev_token == lsym_tag; /* for 'struct s *' */ 561 562 if (ps.prev_token == lsym_tag && ps.nparen == 0) 563 return lsym_type_outside_parentheses; 564 565 const struct keyword *kw = bsearch(token.s, keywords, 566 array_length(keywords), sizeof(keywords[0]), cmp_keyword_by_name); 567 bool is_type = false; 568 if (kw == NULL) { 569 if (is_typename()) { 570 is_type = true; 571 ps.next_unary = true; 572 if (ps.in_enum == in_enum_enum) 573 ps.in_enum = in_enum_type; 574 goto found_typename; 575 } 576 577 } else { /* we have a keyword */ 578 is_type = kw->lsym == lsym_type; 579 ps.next_unary = true; 580 if (kw->lsym != lsym_tag && kw->lsym != lsym_type) 581 return kw->lsym; 582 583 found_typename: 584 if (ps.nparen > 0) { 585 /* inside parentheses: cast, param list, offsetof or sizeof */ 586 if (!ps.paren[ps.nparen - 1].no_cast) 587 ps.paren[ps.nparen - 1].maybe_cast = true; 588 } 589 if (ps.prev_token != lsym_period && ps.prev_token != lsym_unary_op) { 590 if (kw != NULL && kw->lsym == lsym_tag) { 591 if (token.s[0] == 'e' /* enum */) 592 ps.in_enum = in_enum_enum; 593 return lsym_tag; 594 } 595 if (ps.nparen == 0) 596 return lsym_type_outside_parentheses; 597 } 598 } 599 600 if (inp_peek() == '(' && ps.tos <= 1 && ps.ind_level == 0 && 601 !ps.in_func_def_params && !ps.block_init) { 602 603 if (ps.nparen == 0 && probably_looking_at_definition()) { 604 ps.is_function_definition = true; 605 if (ps.in_decl) 606 ps.in_func_def_params = true; 607 return lsym_funcname; 608 } 609 610 } else if (ps.nparen == 0 && probably_typename()) { 611 ps.next_unary = true; 612 return lsym_type_outside_parentheses; 613 } 614 615 return is_type ? lsym_type_in_parentheses : lsym_word; 616 } 617 618 static bool 619 is_asterisk_unary(void) 620 { 621 if (ps.next_unary || ps.in_func_def_params) 622 return true; 623 if (ps.prev_token == lsym_word || 624 ps.prev_token == lsym_rparen_or_rbracket) 625 return false; 626 return ps.in_decl && ps.nparen > 0; 627 } 628 629 static void 630 lex_asterisk_unary(void) 631 { 632 while (inp_peek() == '*' || ch_isspace(inp_peek())) { 633 if (inp_peek() == '*') 634 token_add_char('*'); 635 inp_skip(); 636 } 637 638 if (ps.in_decl) { 639 const char *tp = inp_p(), *e = inp_line_end(); 640 641 while (tp < e) { 642 if (ch_isspace(*tp)) 643 tp++; 644 else if (is_identifier_start(*tp)) { 645 tp++; 646 while (tp < e && is_identifier_part(*tp)) 647 tp++; 648 } else 649 break; 650 } 651 652 if (tp < e && *tp == '(') 653 ps.is_function_definition = true; 654 } 655 } 656 657 /* Reads the next token, placing it in the global variable "token". */ 658 lexer_symbol 659 lexi(void) 660 { 661 token.e = token.s; 662 ps.curr_col_1 = ps.next_col_1; 663 ps.next_col_1 = false; 664 665 while (ch_isblank(inp_peek())) { 666 ps.curr_col_1 = false; 667 inp_skip(); 668 } 669 670 lexer_symbol alnum_lsym = lexi_alnum(); 671 if (alnum_lsym != lsym_eof) 672 return lexi_end(alnum_lsym); 673 674 /* Scan a non-alphanumeric token */ 675 676 check_size_token(3); /* for things like "<<=" */ 677 *token.e++ = inp_next(); 678 *token.e = '\0'; 679 680 lexer_symbol lsym; 681 bool next_unary; 682 683 switch (token.e[-1]) { 684 685 /* INDENT OFF */ 686 case '(': 687 case '[': lsym = lsym_lparen_or_lbracket; next_unary = true; break; 688 case ')': 689 case ']': lsym = lsym_rparen_or_rbracket; next_unary = false; break; 690 case '?': lsym = lsym_question; next_unary = true; break; 691 case ':': lsym = lsym_colon; next_unary = true; break; 692 case ';': lsym = lsym_semicolon; next_unary = true; break; 693 case '{': lsym = lsym_lbrace; next_unary = true; break; 694 case '}': lsym = lsym_rbrace; next_unary = true; break; 695 case ',': lsym = lsym_comma; next_unary = true; break; 696 case '.': lsym = lsym_period; next_unary = false; break; 697 /* INDENT ON */ 698 699 case '\n': 700 /* if data has been exhausted, the '\n' is a dummy. */ 701 lsym = had_eof ? lsym_eof : lsym_newline; 702 next_unary = ps.next_unary; 703 ps.next_col_1 = true; 704 break; 705 706 case '\f': 707 lsym = lsym_form_feed; 708 next_unary = ps.next_unary; 709 ps.next_col_1 = true; 710 break; 711 712 case '#': 713 lsym = lsym_preprocessing; 714 next_unary = ps.next_unary; 715 break; 716 717 case '\'': 718 case '"': 719 lex_char_or_string(); 720 lsym = lsym_word; 721 next_unary = false; 722 break; 723 724 case '-': 725 case '+': 726 lsym = ps.next_unary ? lsym_unary_op : lsym_binary_op; 727 next_unary = true; 728 729 if (inp_peek() == token.e[-1]) { /* '++' or '--' */ 730 *token.e++ = inp_next(); 731 if (ps.prev_token == lsym_word || 732 ps.prev_token == lsym_rparen_or_rbracket) { 733 lsym = ps.next_unary ? lsym_unary_op : lsym_postfix_op; 734 next_unary = false; 735 } 736 737 } else if (inp_peek() == '=') { /* '+=' or '-=' */ 738 *token.e++ = inp_next(); 739 740 } else if (inp_peek() == '>') { /* '->' */ 741 *token.e++ = inp_next(); 742 lsym = lsym_unary_op; 743 next_unary = false; 744 ps.want_blank = false; 745 } 746 break; 747 748 case '=': 749 if (ps.init_or_struct) 750 ps.block_init = true; 751 if (inp_peek() == '=') { /* == */ 752 *token.e++ = inp_next(); 753 *token.e = '\0'; 754 } 755 lsym = lsym_binary_op; 756 next_unary = true; 757 break; 758 759 case '>': 760 case '<': 761 case '!': /* ops like <, <<, <=, !=, etc */ 762 if (inp_peek() == '>' || inp_peek() == '<' || inp_peek() == '=') 763 *token.e++ = inp_next(); 764 if (inp_peek() == '=') 765 *token.e++ = inp_next(); 766 lsym = ps.next_unary ? lsym_unary_op : lsym_binary_op; 767 next_unary = true; 768 break; 769 770 case '*': 771 if (is_asterisk_unary()) { 772 lex_asterisk_unary(); 773 lsym = lsym_unary_op; 774 next_unary = true; 775 } else { 776 if (inp_peek() == '=') 777 *token.e++ = inp_next(); 778 lsym = lsym_binary_op; 779 next_unary = true; 780 } 781 break; 782 783 default: 784 if (token.e[-1] == '/' && (inp_peek() == '*' || inp_peek() == '/')) { 785 *token.e++ = inp_next(); 786 lsym = lsym_comment; 787 next_unary = ps.next_unary; 788 break; 789 } 790 791 /* handle '||', '&&', etc., and also things as in 'int *****i' */ 792 while (token.e[-1] == inp_peek() || inp_peek() == '=') 793 token_add_char(inp_next()); 794 795 lsym = ps.next_unary ? lsym_unary_op : lsym_binary_op; 796 next_unary = true; 797 } 798 799 if (ps.in_enum == in_enum_enum || ps.in_enum == in_enum_type) 800 ps.in_enum = lsym == lsym_lbrace ? in_enum_brace : in_enum_no; 801 if (lsym == lsym_rbrace) 802 ps.in_enum = in_enum_no; 803 804 ps.next_unary = next_unary; 805 806 check_size_token(1); 807 *token.e = '\0'; 808 809 return lexi_end(lsym); 810 } 811 812 void 813 register_typename(const char *name) 814 { 815 if (typenames.len >= typenames.cap) { 816 typenames.cap = 16 + 2 * typenames.cap; 817 typenames.items = xrealloc(typenames.items, 818 sizeof(typenames.items[0]) * typenames.cap); 819 } 820 821 int pos = bsearch_typenames(name); 822 if (pos >= 0) 823 return; /* already in the list */ 824 825 pos = -(pos + 1); 826 memmove(typenames.items + pos + 1, typenames.items + pos, 827 sizeof(typenames.items[0]) * (typenames.len++ - (unsigned)pos)); 828 typenames.items[pos] = xstrdup(name); 829 } 830
Indexes created Thu Oct 23 18:09:57 GMT 2025