lexi.c revision 1.177
1 /* $NetBSD: lexi.c,v 1.177 2023/05/13 09:27:49 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 1.177 2023/05/13 09:27:49 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 } 193 194 195 static lexer_symbol 196 lexi_end(lexer_symbol lsym) 197 { 198 #ifdef debug 199 debug_parser_state(lsym); 200 #endif 201 return lsym; 202 } 203 204 static void 205 lex_number(void) 206 { 207 for (unsigned char s = 'A'; s != 'f' && s != 'i' && s != 'u';) { 208 unsigned char ch = (unsigned char)inp_peek(); 209 if (ch >= array_length(lex_number_row) || lex_number_row[ch] == 0) 210 break; 211 212 unsigned char row = lex_number_row[ch]; 213 if (lex_number_state[row][s - 'A'] == ' ') { 214 /*- 215 * lex_number_state[0][s - 'A'] now indicates the type: 216 * f = floating, i = integer, u = unknown 217 */ 218 return; 219 } 220 221 s = lex_number_state[row][s - 'A']; 222 token_add_char(inp_next()); 223 } 224 } 225 226 static bool 227 is_identifier_start(char ch) 228 { 229 return ch_isalpha(ch) || ch == '_' || ch == '$'; 230 } 231 232 static bool 233 is_identifier_part(char ch) 234 { 235 return ch_isalnum(ch) || ch == '_' || ch == '$'; 236 } 237 238 static void 239 lex_word(void) 240 { 241 for (;;) { 242 if (is_identifier_part(inp_peek())) 243 token_add_char(inp_next()); 244 else if (inp_peek() == '\\' && inp_lookahead(1) == '\n') { 245 inp_skip(); 246 inp_skip(); 247 } else 248 return; 249 } 250 } 251 252 static void 253 lex_char_or_string(void) 254 { 255 for (char delim = token.e[-1];;) { 256 if (inp_peek() == '\n') { 257 diag(1, "Unterminated literal"); 258 return; 259 } 260 261 token_add_char(inp_next()); 262 if (token.e[-1] == delim) 263 return; 264 265 if (token.e[-1] == '\\') { 266 if (inp_peek() == '\n') 267 ++line_no; 268 token_add_char(inp_next()); 269 } 270 } 271 } 272 273 /* Guess whether the current token is a declared type. */ 274 static bool 275 probably_typename(void) 276 { 277 if (ps.prev_token == lsym_storage_class) 278 return true; 279 if (ps.block_init) 280 return false; 281 if (ps.in_stmt_or_decl) /* XXX: this condition looks incorrect */ 282 return false; 283 if (inp_peek() == '*' && inp_lookahead(1) != '=') 284 goto maybe; 285 /* XXX: is_identifier_start */ 286 if (ch_isalpha(inp_peek())) 287 goto maybe; 288 return false; 289 maybe: 290 return ps.prev_token == lsym_semicolon || 291 ps.prev_token == lsym_lbrace || 292 ps.prev_token == lsym_rbrace; 293 } 294 295 static int 296 bsearch_typenames(const char *key) 297 { 298 const char **arr = typenames.items; 299 int lo = 0; 300 int hi = (int)typenames.len - 1; 301 302 while (lo <= hi) { 303 int mid = (int)((unsigned)(lo + hi) >> 1); 304 int cmp = strcmp(arr[mid], key); 305 if (cmp < 0) 306 lo = mid + 1; 307 else if (cmp > 0) 308 hi = mid - 1; 309 else 310 return mid; 311 } 312 return -(lo + 1); 313 } 314 315 static bool 316 is_typename(void) 317 { 318 if (opt.auto_typedefs && 319 token.e - token.s >= 2 && memcmp(token.e - 2, "_t", 2) == 0) 320 return true; 321 322 return bsearch_typenames(token.s) >= 0; 323 } 324 325 static int 326 cmp_keyword_by_name(const void *key, const void *elem) 327 { 328 return strcmp(key, ((const struct keyword *)elem)->name); 329 } 330 331 /* 332 * Looking at something like 'function_name(...)' in a line, guess whether 333 * this starts a function definition or a declaration. 334 */ 335 static bool 336 probably_looking_at_definition(void) 337 { 338 int paren_level = 0; 339 for (const char *p = inp_p(), *e = inp_line_end(); p < e; p++) { 340 if (*p == '(') 341 paren_level++; 342 if (*p == ')' && --paren_level == 0) { 343 p++; 344 345 while (p < e && (ch_isspace(*p) || is_identifier_part(*p))) 346 p++; /* '__dead' or '__unused' */ 347 348 if (p == e) /* func(...) */ 349 break; 350 if (*p == ';') /* func(...); */ 351 return false; 352 if (*p == ',') /* double abs(), pi; */ 353 return false; 354 if (*p == '(') /* func(...) __attribute__((...)) */ 355 paren_level++; /* func(...) __printflike(...) */ 356 else 357 break; /* func(...) { ... */ 358 } 359 } 360 361 /* 362 * To further reduce the cases where indent wrongly treats an incomplete 363 * function declaration as a function definition, thus adding a newline 364 * before the function name, it may be worth looking for parameter names, 365 * as these are often omitted in function declarations and only included 366 * in function definitions. Or just increase the lookahead to more than 367 * just the current line of input, until the next '{'. 368 */ 369 return true; 370 } 371 372 /* Read an alphanumeric token into 'token', or return lsym_eof. */ 373 static lexer_symbol 374 lexi_alnum(void) 375 { 376 if (ch_isdigit(inp_peek()) || 377 (inp_peek() == '.' && ch_isdigit(inp_lookahead(1)))) { 378 lex_number(); 379 } else if (is_identifier_start(inp_peek())) { 380 lex_word(); 381 382 if (token.s[0] == 'L' && token.e - token.s == 1 && 383 (inp_peek() == '"' || inp_peek() == '\'')) { 384 token_add_char(inp_next()); 385 lex_char_or_string(); 386 ps.next_unary = false; 387 388 check_size_token(1); 389 *token.e = '\0'; 390 391 return lsym_word; 392 } 393 } else 394 return lsym_eof; /* just as a placeholder */ 395 396 *token.e = '\0'; 397 398 while (ch_isblank(inp_peek())) 399 inp_skip(); 400 401 ps.next_unary = ps.prev_token == lsym_tag; /* for 'struct s *' */ 402 403 if (ps.prev_token == lsym_tag && ps.nparen == 0) 404 return lsym_type_outside_parentheses; 405 406 const struct keyword *kw = bsearch(token.s, keywords, 407 array_length(keywords), sizeof(keywords[0]), cmp_keyword_by_name); 408 bool is_type = false; 409 if (kw == NULL) { 410 if (is_typename()) { 411 is_type = true; 412 ps.next_unary = true; 413 if (ps.in_enum == in_enum_enum) 414 ps.in_enum = in_enum_type; 415 goto found_typename; 416 } 417 418 } else { /* we have a keyword */ 419 is_type = kw->lsym == lsym_type; 420 ps.next_unary = true; 421 if (kw->lsym != lsym_tag && kw->lsym != lsym_type) 422 return kw->lsym; 423 424 found_typename: 425 if (ps.nparen > 0) { 426 /* inside parentheses: cast, param list, offsetof or sizeof */ 427 if (!ps.paren[ps.nparen - 1].no_cast) 428 ps.paren[ps.nparen - 1].maybe_cast = true; 429 } 430 if (ps.prev_token != lsym_period && ps.prev_token != lsym_unary_op) { 431 if (kw != NULL && kw->lsym == lsym_tag) { 432 if (token.s[0] == 'e' /* enum */) 433 ps.in_enum = in_enum_enum; 434 return lsym_tag; 435 } 436 if (ps.nparen == 0) 437 return lsym_type_outside_parentheses; 438 } 439 } 440 441 if (inp_peek() == '(' && ps.tos <= 1 && ps.ind_level == 0 && 442 !ps.in_func_def_params && !ps.block_init) { 443 444 if (ps.nparen == 0 && probably_looking_at_definition()) { 445 ps.is_function_definition = true; 446 if (ps.in_decl) 447 ps.in_func_def_params = true; 448 return lsym_funcname; 449 } 450 451 } else if (ps.nparen == 0 && probably_typename()) { 452 ps.next_unary = true; 453 return lsym_type_outside_parentheses; 454 } 455 456 return is_type ? lsym_type_in_parentheses : lsym_word; 457 } 458 459 static bool 460 is_asterisk_unary(void) 461 { 462 if (ps.next_unary || ps.in_func_def_params) 463 return true; 464 if (ps.prev_token == lsym_word || 465 ps.prev_token == lsym_rparen_or_rbracket) 466 return false; 467 return ps.in_decl && ps.nparen > 0; 468 } 469 470 static void 471 lex_asterisk_unary(void) 472 { 473 while (inp_peek() == '*' || ch_isspace(inp_peek())) { 474 if (inp_peek() == '*') 475 token_add_char('*'); 476 inp_skip(); 477 } 478 479 if (ps.in_decl) { 480 const char *tp = inp_p(), *e = inp_line_end(); 481 482 while (tp < e) { 483 if (ch_isspace(*tp)) 484 tp++; 485 else if (is_identifier_start(*tp)) { 486 tp++; 487 while (tp < e && is_identifier_part(*tp)) 488 tp++; 489 } else 490 break; 491 } 492 493 if (tp < e && *tp == '(') 494 ps.is_function_definition = true; 495 } 496 } 497 498 /* Reads the next token, placing it in the global variable "token". */ 499 lexer_symbol 500 lexi(void) 501 { 502 token.e = token.s; 503 ps.curr_col_1 = ps.next_col_1; 504 ps.next_col_1 = false; 505 506 while (ch_isblank(inp_peek())) { 507 ps.curr_col_1 = false; 508 inp_skip(); 509 } 510 511 lexer_symbol alnum_lsym = lexi_alnum(); 512 if (alnum_lsym != lsym_eof) 513 return lexi_end(alnum_lsym); 514 515 /* Scan a non-alphanumeric token */ 516 517 check_size_token(3); /* for things like "<<=" */ 518 *token.e++ = inp_next(); 519 *token.e = '\0'; 520 521 lexer_symbol lsym; 522 bool next_unary; 523 524 switch (token.e[-1]) { 525 526 /* INDENT OFF */ 527 case '(': 528 case '[': lsym = lsym_lparen_or_lbracket; next_unary = true; break; 529 case ')': 530 case ']': lsym = lsym_rparen_or_rbracket; next_unary = false; break; 531 case '?': lsym = lsym_question; next_unary = true; break; 532 case ':': lsym = lsym_colon; next_unary = true; break; 533 case ';': lsym = lsym_semicolon; next_unary = true; break; 534 case '{': lsym = lsym_lbrace; next_unary = true; break; 535 case '}': lsym = lsym_rbrace; next_unary = true; break; 536 case ',': lsym = lsym_comma; next_unary = true; break; 537 case '.': lsym = lsym_period; next_unary = false; break; 538 /* INDENT ON */ 539 540 case '\n': 541 /* if data has been exhausted, the '\n' is a dummy. */ 542 lsym = had_eof ? lsym_eof : lsym_newline; 543 next_unary = ps.next_unary; 544 ps.next_col_1 = true; 545 break; 546 547 case '\f': 548 lsym = lsym_form_feed; 549 next_unary = ps.next_unary; 550 ps.next_col_1 = true; 551 break; 552 553 case '#': 554 lsym = lsym_preprocessing; 555 next_unary = ps.next_unary; 556 break; 557 558 case '\'': 559 case '"': 560 lex_char_or_string(); 561 lsym = lsym_word; 562 next_unary = false; 563 break; 564 565 case '-': 566 case '+': 567 lsym = ps.next_unary ? lsym_unary_op : lsym_binary_op; 568 next_unary = true; 569 570 if (inp_peek() == token.e[-1]) { /* '++' or '--' */ 571 *token.e++ = inp_next(); 572 if (ps.prev_token == lsym_word || 573 ps.prev_token == lsym_rparen_or_rbracket) { 574 lsym = ps.next_unary ? lsym_unary_op : lsym_postfix_op; 575 next_unary = false; 576 } 577 578 } else if (inp_peek() == '=') { /* '+=' or '-=' */ 579 *token.e++ = inp_next(); 580 581 } else if (inp_peek() == '>') { /* '->' */ 582 *token.e++ = inp_next(); 583 lsym = lsym_unary_op; 584 next_unary = false; 585 ps.want_blank = false; 586 } 587 break; 588 589 case '=': 590 if (ps.init_or_struct) 591 ps.block_init = true; 592 if (inp_peek() == '=') { /* == */ 593 *token.e++ = inp_next(); 594 *token.e = '\0'; 595 } 596 lsym = lsym_binary_op; 597 next_unary = true; 598 break; 599 600 case '>': 601 case '<': 602 case '!': /* ops like <, <<, <=, !=, etc */ 603 if (inp_peek() == '>' || inp_peek() == '<' || inp_peek() == '=') 604 *token.e++ = inp_next(); 605 if (inp_peek() == '=') 606 *token.e++ = inp_next(); 607 lsym = ps.next_unary ? lsym_unary_op : lsym_binary_op; 608 next_unary = true; 609 break; 610 611 case '*': 612 if (is_asterisk_unary()) { 613 lex_asterisk_unary(); 614 lsym = lsym_unary_op; 615 next_unary = true; 616 } else { 617 if (inp_peek() == '=') 618 *token.e++ = inp_next(); 619 lsym = lsym_binary_op; 620 next_unary = true; 621 } 622 break; 623 624 default: 625 if (token.e[-1] == '/' && (inp_peek() == '*' || inp_peek() == '/')) { 626 *token.e++ = inp_next(); 627 lsym = lsym_comment; 628 next_unary = ps.next_unary; 629 break; 630 } 631 632 /* handle '||', '&&', etc., and also things as in 'int *****i' */ 633 while (token.e[-1] == inp_peek() || inp_peek() == '=') 634 token_add_char(inp_next()); 635 636 lsym = ps.next_unary ? lsym_unary_op : lsym_binary_op; 637 next_unary = true; 638 } 639 640 if (ps.in_enum == in_enum_enum || ps.in_enum == in_enum_type) 641 ps.in_enum = lsym == lsym_lbrace ? in_enum_brace : in_enum_no; 642 if (lsym == lsym_rbrace) 643 ps.in_enum = in_enum_no; 644 645 ps.next_unary = next_unary; 646 647 check_size_token(1); 648 *token.e = '\0'; 649 650 return lexi_end(lsym); 651 } 652 653 void 654 register_typename(const char *name) 655 { 656 if (typenames.len >= typenames.cap) { 657 typenames.cap = 16 + 2 * typenames.cap; 658 typenames.items = xrealloc(typenames.items, 659 sizeof(typenames.items[0]) * typenames.cap); 660 } 661 662 int pos = bsearch_typenames(name); 663 if (pos >= 0) 664 return; /* already in the list */ 665 666 pos = -(pos + 1); 667 memmove(typenames.items + pos + 1, typenames.items + pos, 668 sizeof(typenames.items[0]) * (typenames.len++ - (unsigned)pos)); 669 typenames.items[pos] = xstrdup(name); 670 } 671
Indexes created Fri Sep 26 20:09:58 GMT 2025