1 /*- 2 * SPDX-License-Identifier: BSD-4-Clause 3 * 4 * Copyright (c) 1985 Sun Microsystems, Inc. 5 * Copyright (c) 1976 Board of Trustees of the University of Illinois. 6 * Copyright (c) 1980, 1993 7 * The Regents of the University of California. All rights reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 3. All advertising materials mentioning features or use of this software 18 * must display the following acknowledgement: 19 * This product includes software developed by the University of 20 * California, Berkeley and its contributors. 21 * 4. Neither the name of the University nor the names of its contributors 22 * may be used to endorse or promote products derived from this software 23 * without specific prior written permission. 24 * 25 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 27 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 28 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 29 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 30 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 31 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 32 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 33 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 34 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 35 * SUCH DAMAGE. 36 */ 37 38 #include <sys/param.h> 39 #include <sys/capsicum.h> 40 #include <capsicum_helpers.h> 41 #include <err.h> 42 #include <errno.h> 43 #include <fcntl.h> 44 #include <unistd.h> 45 #include <stdio.h> 46 #include <stdlib.h> 47 #include <string.h> 48 #include <ctype.h> 49 #include "indent_globs.h" 50 #include "indent_codes.h" 51 #include "indent.h" 52 53 /* Globals */ 54 FILE *input, *output; 55 char *labbuf, *s_lab, *e_lab, *l_lab; 56 char *codebuf, *s_code, *e_code, *l_code; 57 char *combuf, *s_com, *e_com, *l_com; 58 char *tokenbuf, *s_token, *e_token, *l_token; 59 char *in_buffer, *in_buffer_limit; 60 char *buf_ptr, *buf_end; 61 62 char sc_buf[sc_size]; 63 64 char *save_com, *sc_end; 65 char *bp_save; 66 char *be_save; 67 68 struct options opt; 69 int line_no; 70 71 struct parser_state ps; 72 int ifdef_level; 73 struct parser_state state_stack[5]; 74 struct parser_state match_state[5]; 75 76 77 static void bakcopy(void); 78 static void indent_declaration(int, int); 79 80 const char *in_name = "Standard Input"; /* will always point to name of input 81 * file */ 82 const char *out_name = "Standard Output"; /* will always point to name 83 * of output file */ 84 const char *simple_backup_suffix = ".BAK"; /* Suffix to use for backup 85 * files */ 86 char bakfile[MAXPATHLEN] = ""; 87 88 int 89 main(int argc, char **argv) 90 { 91 cap_rights_t rights; 92 93 int dec_ind; /* current indentation for declarations */ 94 int di_stack[20]; /* a stack of structure indentation levels */ 95 int force_nl; /* when true, code must be broken */ 96 int hd_type = 0; /* used to store type of stmt for if (...), 97 * for (...), etc */ 98 int i; /* local loop counter */ 99 int scase; /* set to true when we see a case, so we will 100 * know what to do with the following colon */ 101 int sp_sw; /* when true, we are in the expression of 102 * if(...), while(...), etc. */ 103 int squest; /* when this is positive, we have seen a ? 104 * without the matching : in a <c>?<s>:<s> 105 * construct */ 106 const char *t_ptr; /* used for copying tokens */ 107 int tabs_to_var; /* true if using tabs to indent to var name */ 108 int type_code; /* the type of token, returned by lexi */ 109 110 int last_else = 0; /* true iff last keyword was an else */ 111 const char *profile_name = NULL; 112 const char *envval = NULL; 113 struct parser_state transient_state; /* a copy for lookup */ 114 115 /*-----------------------------------------------*\ 116 | INITIALIZATION | 117 \*-----------------------------------------------*/ 118 119 found_err = 0; 120 121 ps.p_stack[0] = stmt; /* this is the parser's stack */ 122 ps.last_nl = true; /* this is true if the last thing scanned was 123 * a newline */ 124 ps.last_token = semicolon; 125 combuf = (char *) malloc(bufsize); 126 if (combuf == NULL) 127 err(1, NULL); 128 labbuf = (char *) malloc(bufsize); 129 if (labbuf == NULL) 130 err(1, NULL); 131 codebuf = (char *) malloc(bufsize); 132 if (codebuf == NULL) 133 err(1, NULL); 134 tokenbuf = (char *) malloc(bufsize); 135 if (tokenbuf == NULL) 136 err(1, NULL); 137 alloc_typenames(); 138 init_constant_tt(); 139 l_com = combuf + bufsize - 5; 140 l_lab = labbuf + bufsize - 5; 141 l_code = codebuf + bufsize - 5; 142 l_token = tokenbuf + bufsize - 5; 143 combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and 144 * comment buffers */ 145 combuf[1] = codebuf[1] = labbuf[1] = '\0'; 146 opt.else_if = 1; /* Default else-if special processing to on */ 147 s_lab = e_lab = labbuf + 1; 148 s_code = e_code = codebuf + 1; 149 s_com = e_com = combuf + 1; 150 s_token = e_token = tokenbuf + 1; 151 152 in_buffer = (char *) malloc(10); 153 if (in_buffer == NULL) 154 err(1, NULL); 155 in_buffer_limit = in_buffer + 8; 156 buf_ptr = buf_end = in_buffer; 157 line_no = 1; 158 had_eof = ps.in_decl = ps.decl_on_line = break_comma = false; 159 sp_sw = force_nl = false; 160 ps.in_or_st = false; 161 ps.bl_line = true; 162 dec_ind = 0; 163 di_stack[ps.dec_nest = 0] = 0; 164 ps.want_blank = ps.in_stmt = ps.ind_stmt = false; 165 166 scase = ps.pcase = false; 167 squest = 0; 168 sc_end = NULL; 169 bp_save = NULL; 170 be_save = NULL; 171 172 output = NULL; 173 tabs_to_var = 0; 174 175 envval = getenv("SIMPLE_BACKUP_SUFFIX"); 176 if (envval) 177 simple_backup_suffix = envval; 178 179 /*--------------------------------------------------*\ 180 | COMMAND LINE SCAN | 181 \*--------------------------------------------------*/ 182 183 #ifdef undef 184 max_col = 78; /* -l78 */ 185 lineup_to_parens = 1; /* -lp */ 186 lineup_to_parens_always = 0; /* -nlpl */ 187 ps.ljust_decl = 0; /* -ndj */ 188 ps.com_ind = 33; /* -c33 */ 189 star_comment_cont = 1; /* -sc */ 190 ps.ind_size = 8; /* -i8 */ 191 verbose = 0; 192 ps.decl_indent = 16; /* -di16 */ 193 ps.local_decl_indent = -1; /* if this is not set to some nonnegative value 194 * by an arg, we will set this equal to 195 * ps.decl_ind */ 196 ps.indent_parameters = 1; /* -ip */ 197 ps.decl_com_ind = 0; /* if this is not set to some positive value 198 * by an arg, we will set this equal to 199 * ps.com_ind */ 200 btype_2 = 1; /* -br */ 201 cuddle_else = 1; /* -ce */ 202 ps.unindent_displace = 0; /* -d0 */ 203 ps.case_indent = 0; /* -cli0 */ 204 format_block_comments = 1; /* -fcb */ 205 format_col1_comments = 1; /* -fc1 */ 206 procnames_start_line = 1; /* -psl */ 207 proc_calls_space = 0; /* -npcs */ 208 comment_delimiter_on_blankline = 1; /* -cdb */ 209 ps.leave_comma = 1; /* -nbc */ 210 #endif 211 212 for (i = 1; i < argc; ++i) 213 if (strcmp(argv[i], "-npro") == 0) 214 break; 215 else if (argv[i][0] == '-' && argv[i][1] == 'P' && argv[i][2] != '\0') 216 profile_name = argv[i]; /* non-empty -P (set profile) */ 217 set_defaults(); 218 if (i >= argc) 219 set_profile(profile_name); 220 221 for (i = 1; i < argc; ++i) { 222 223 /* 224 * look thru args (if any) for changes to defaults 225 */ 226 if (argv[i][0] != '-') {/* no flag on parameter */ 227 if (input == NULL) { /* we must have the input file */ 228 in_name = argv[i]; /* remember name of input file */ 229 input = fopen(in_name, "r"); 230 if (input == NULL) /* check for open error */ 231 err(1, "%s", in_name); 232 continue; 233 } 234 else if (output == NULL) { /* we have the output file */ 235 out_name = argv[i]; /* remember name of output file */ 236 if (strcmp(in_name, out_name) == 0) { /* attempt to overwrite 237 * the file */ 238 errx(1, "input and output files must be different"); 239 } 240 output = fopen(out_name, "w"); 241 if (output == NULL) /* check for create error */ 242 err(1, "%s", out_name); 243 continue; 244 } 245 errx(1, "unknown parameter: %s", argv[i]); 246 } 247 else 248 set_option(argv[i]); 249 } /* end of for */ 250 if (input == NULL) 251 input = stdin; 252 if (output == NULL) { 253 if (input == stdin) 254 output = stdout; 255 else { 256 out_name = in_name; 257 bakcopy(); 258 } 259 } 260 261 /* Restrict input/output descriptors and enter Capsicum sandbox. */ 262 cap_rights_init(&rights, CAP_FSTAT, CAP_WRITE); 263 if (caph_rights_limit(fileno(output), &rights) < 0) 264 err(EXIT_FAILURE, "unable to limit rights for %s", out_name); 265 cap_rights_init(&rights, CAP_FSTAT, CAP_READ); 266 if (caph_rights_limit(fileno(input), &rights) < 0) 267 err(EXIT_FAILURE, "unable to limit rights for %s", in_name); 268 if (caph_enter() < 0) 269 err(EXIT_FAILURE, "unable to enter capability mode"); 270 271 if (opt.com_ind <= 1) 272 opt.com_ind = 2; /* don't put normal comments before column 2 */ 273 if (opt.block_comment_max_col <= 0) 274 opt.block_comment_max_col = opt.max_col; 275 if (opt.local_decl_indent < 0) /* if not specified by user, set this */ 276 opt.local_decl_indent = opt.decl_indent; 277 if (opt.decl_com_ind <= 0) /* if not specified by user, set this */ 278 opt.decl_com_ind = opt.ljust_decl ? (opt.com_ind <= 10 ? 2 : opt.com_ind - 8) : opt.com_ind; 279 if (opt.continuation_indent == 0) 280 opt.continuation_indent = opt.ind_size; 281 fill_buffer(); /* get first batch of stuff into input buffer */ 282 283 parse(semicolon); 284 { 285 char *p = buf_ptr; 286 int col = 1; 287 288 while (1) { 289 if (*p == ' ') 290 col++; 291 else if (*p == '\t') 292 col = opt.tabsize * (1 + (col - 1) / opt.tabsize) + 1; 293 else 294 break; 295 p++; 296 } 297 if (col > opt.ind_size) 298 ps.ind_level = ps.i_l_follow = col / opt.ind_size; 299 } 300 301 /* 302 * START OF MAIN LOOP 303 */ 304 305 while (1) { /* this is the main loop. it will go until we 306 * reach eof */ 307 int comment_buffered = false; 308 309 type_code = lexi(&ps); /* lexi reads one token. The actual 310 * characters read are stored in "token". lexi 311 * returns a code indicating the type of token */ 312 313 /* 314 * The following code moves newlines and comments following an if (), 315 * while (), else, etc. up to the start of the following stmt to 316 * a buffer. This allows proper handling of both kinds of brace 317 * placement (-br, -bl) and cuddling "else" (-ce). 318 */ 319 320 while (ps.search_brace) { 321 switch (type_code) { 322 case newline: 323 if (sc_end == NULL) { 324 save_com = sc_buf; 325 save_com[0] = save_com[1] = ' '; 326 sc_end = &save_com[2]; 327 } 328 *sc_end++ = '\n'; 329 /* 330 * We may have inherited a force_nl == true from the previous 331 * token (like a semicolon). But once we know that a newline 332 * has been scanned in this loop, force_nl should be false. 333 * 334 * However, the force_nl == true must be preserved if newline 335 * is never scanned in this loop, so this assignment cannot be 336 * done earlier. 337 */ 338 force_nl = false; 339 case form_feed: 340 break; 341 case comment: 342 if (sc_end == NULL) { 343 /* 344 * Copy everything from the start of the line, because 345 * pr_comment() will use that to calculate original 346 * indentation of a boxed comment. 347 */ 348 memcpy(sc_buf, in_buffer, buf_ptr - in_buffer - 4); 349 save_com = sc_buf + (buf_ptr - in_buffer - 4); 350 save_com[0] = save_com[1] = ' '; 351 sc_end = &save_com[2]; 352 } 353 comment_buffered = true; 354 *sc_end++ = '/'; /* copy in start of comment */ 355 *sc_end++ = '*'; 356 for (;;) { /* loop until we get to the end of the comment */ 357 *sc_end = *buf_ptr++; 358 if (buf_ptr >= buf_end) 359 fill_buffer(); 360 if (*sc_end++ == '*' && *buf_ptr == '/') 361 break; /* we are at end of comment */ 362 if (sc_end >= &save_com[sc_size]) { /* check for temp buffer 363 * overflow */ 364 diag2(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever"); 365 fflush(output); 366 exit(1); 367 } 368 } 369 *sc_end++ = '/'; /* add ending slash */ 370 if (++buf_ptr >= buf_end) /* get past / in buffer */ 371 fill_buffer(); 372 break; 373 case lbrace: 374 /* 375 * Put KNF-style lbraces before the buffered up tokens and 376 * jump out of this loop in order to avoid copying the token 377 * again under the default case of the switch below. 378 */ 379 if (sc_end != NULL && opt.btype_2) { 380 save_com[0] = '{'; 381 /* 382 * Originally the lbrace may have been alone on its own 383 * line, but it will be moved into "the else's line", so 384 * if there was a newline resulting from the "{" before, 385 * it must be scanned now and ignored. 386 */ 387 while (isspace((unsigned char)*buf_ptr)) { 388 if (++buf_ptr >= buf_end) 389 fill_buffer(); 390 if (*buf_ptr == '\n') 391 break; 392 } 393 goto sw_buffer; 394 } 395 /* FALLTHROUGH */ 396 default: /* it is the start of a normal statement */ 397 { 398 int remove_newlines; 399 400 remove_newlines = 401 /* "} else" */ 402 (type_code == sp_nparen && *token == 'e' && 403 e_code != s_code && e_code[-1] == '}') 404 /* "else if" */ 405 || (type_code == sp_paren && *token == 'i' && 406 last_else && opt.else_if); 407 if (remove_newlines) 408 force_nl = false; 409 if (sc_end == NULL) { /* ignore buffering if 410 * comment wasn't saved up */ 411 ps.search_brace = false; 412 goto check_type; 413 } 414 while (sc_end > save_com && isblank((unsigned char)sc_end[-1])) { 415 sc_end--; 416 } 417 if (opt.swallow_optional_blanklines || 418 (!comment_buffered && remove_newlines)) { 419 force_nl = !remove_newlines; 420 while (sc_end > save_com && sc_end[-1] == '\n') { 421 sc_end--; 422 } 423 } 424 if (force_nl) { /* if we should insert a nl here, put 425 * it into the buffer */ 426 force_nl = false; 427 --line_no; /* this will be re-increased when the 428 * newline is read from the buffer */ 429 *sc_end++ = '\n'; 430 *sc_end++ = ' '; 431 if (opt.verbose) /* print error msg if the line was 432 * not already broken */ 433 diag2(0, "Line broken"); 434 } 435 for (t_ptr = token; *t_ptr; ++t_ptr) 436 *sc_end++ = *t_ptr; 437 438 sw_buffer: 439 ps.search_brace = false; /* stop looking for start of 440 * stmt */ 441 bp_save = buf_ptr; /* save current input buffer */ 442 be_save = buf_end; 443 buf_ptr = save_com; /* fix so that subsequent calls to 444 * lexi will take tokens out of 445 * save_com */ 446 *sc_end++ = ' ';/* add trailing blank, just in case */ 447 buf_end = sc_end; 448 sc_end = NULL; 449 break; 450 } 451 } /* end of switch */ 452 /* 453 * We must make this check, just in case there was an unexpected 454 * EOF. 455 */ 456 if (type_code != 0) { 457 /* 458 * The only intended purpose of calling lexi() below is to 459 * categorize the next token in order to decide whether to 460 * continue buffering forthcoming tokens. Once the buffering 461 * is over, lexi() will be called again elsewhere on all of 462 * the tokens - this time for normal processing. 463 * 464 * Calling it for this purpose is a bug, because lexi() also 465 * changes the parser state and discards leading whitespace, 466 * which is needed mostly for comment-related considerations. 467 * 468 * Work around the former problem by giving lexi() a copy of 469 * the current parser state and discard it if the call turned 470 * out to be just a look ahead. 471 * 472 * Work around the latter problem by copying all whitespace 473 * characters into the buffer so that the later lexi() call 474 * will read them. 475 */ 476 if (sc_end != NULL) { 477 while (*buf_ptr == ' ' || *buf_ptr == '\t') { 478 *sc_end++ = *buf_ptr++; 479 if (sc_end >= &save_com[sc_size]) { 480 errx(1, "input too long"); 481 } 482 } 483 if (buf_ptr >= buf_end) { 484 fill_buffer(); 485 } 486 } 487 transient_state = ps; 488 type_code = lexi(&transient_state); /* read another token */ 489 if (type_code != newline && type_code != form_feed && 490 type_code != comment && !transient_state.search_brace) { 491 ps = transient_state; 492 } 493 } 494 } /* end of while (search_brace) */ 495 last_else = 0; 496 check_type: 497 if (type_code == 0) { /* we got eof */ 498 if (s_lab != e_lab || s_code != e_code 499 || s_com != e_com) /* must dump end of line */ 500 dump_line(); 501 if (ps.tos > 1) /* check for balanced braces */ 502 diag2(1, "Stuff missing from end of file"); 503 504 if (opt.verbose) { 505 printf("There were %d output lines and %d comments\n", 506 ps.out_lines, ps.out_coms); 507 printf("(Lines with comments)/(Lines with code): %6.3f\n", 508 (1.0 * ps.com_lines) / code_lines); 509 } 510 fflush(output); 511 exit(found_err); 512 } 513 if ( 514 (type_code != comment) && 515 (type_code != newline) && 516 (type_code != preesc) && 517 (type_code != form_feed)) { 518 if (force_nl && 519 (type_code != semicolon) && 520 (type_code != lbrace || !opt.btype_2)) { 521 /* we should force a broken line here */ 522 if (opt.verbose) 523 diag2(0, "Line broken"); 524 dump_line(); 525 ps.want_blank = false; /* dont insert blank at line start */ 526 force_nl = false; 527 } 528 ps.in_stmt = true; /* turn on flag which causes an extra level of 529 * indentation. this is turned off by a ; or 530 * '}' */ 531 if (s_com != e_com) { /* the turkey has embedded a comment 532 * in a line. fix it */ 533 int len = e_com - s_com; 534 535 CHECK_SIZE_CODE(len + 3); 536 *e_code++ = ' '; 537 memcpy(e_code, s_com, len); 538 e_code += len; 539 *e_code++ = ' '; 540 *e_code = '\0'; /* null terminate code sect */ 541 ps.want_blank = false; 542 e_com = s_com; 543 } 544 } 545 else if (type_code != comment) /* preserve force_nl thru a comment */ 546 force_nl = false; /* cancel forced newline after newline, form 547 * feed, etc */ 548 549 550 551 /*-----------------------------------------------------*\ 552 | do switch on type of token scanned | 553 \*-----------------------------------------------------*/ 554 CHECK_SIZE_CODE(3); /* maximum number of increments of e_code 555 * before the next CHECK_SIZE_CODE or 556 * dump_line() is 2. After that there's the 557 * final increment for the null character. */ 558 switch (type_code) { /* now, decide what to do with the token */ 559 560 case form_feed: /* found a form feed in line */ 561 ps.use_ff = true; /* a form feed is treated much like a newline */ 562 dump_line(); 563 ps.want_blank = false; 564 break; 565 566 case newline: 567 if (ps.last_token != comma || ps.p_l_follow > 0 568 || !opt.leave_comma || ps.block_init || !break_comma || s_com != e_com) { 569 dump_line(); 570 ps.want_blank = false; 571 } 572 ++line_no; /* keep track of input line number */ 573 break; 574 575 case lparen: /* got a '(' or '[' */ 576 /* count parens to make Healy happy */ 577 if (++ps.p_l_follow == nitems(ps.paren_indents)) { 578 diag3(0, "Reached internal limit of %d unclosed parens", 579 nitems(ps.paren_indents)); 580 ps.p_l_follow--; 581 } 582 if (*token == '[') 583 /* not a function pointer declaration or a function call */; 584 else if (ps.in_decl && !ps.block_init && !ps.dumped_decl_indent && 585 ps.procname[0] == '\0' && ps.paren_level == 0) { 586 /* function pointer declarations */ 587 indent_declaration(dec_ind, tabs_to_var); 588 ps.dumped_decl_indent = true; 589 } 590 else if (ps.want_blank && 591 ((ps.last_token != ident && ps.last_token != funcname) || 592 opt.proc_calls_space || 593 /* offsetof (1) is never allowed a space; sizeof (2) gets 594 * one iff -bs; all other keywords (>2) always get a space 595 * before lparen */ 596 ps.keyword + opt.Bill_Shannon > 2)) 597 *e_code++ = ' '; 598 ps.want_blank = false; 599 *e_code++ = token[0]; 600 ps.paren_indents[ps.p_l_follow - 1] = count_spaces_until(1, s_code, e_code) - 1; 601 if (sp_sw && ps.p_l_follow == 1 && opt.extra_expression_indent 602 && ps.paren_indents[0] < 2 * opt.ind_size) 603 ps.paren_indents[0] = 2 * opt.ind_size; 604 if (ps.in_or_st && *token == '(' && ps.tos <= 2) { 605 /* 606 * this is a kluge to make sure that declarations will be 607 * aligned right if proc decl has an explicit type on it, i.e. 608 * "int a(x) {..." 609 */ 610 parse(semicolon); /* I said this was a kluge... */ 611 ps.in_or_st = false; /* turn off flag for structure decl or 612 * initialization */ 613 } 614 /* parenthesized type following sizeof or offsetof is not a cast */ 615 if (ps.keyword == 1 || ps.keyword == 2) 616 ps.not_cast_mask |= 1 << ps.p_l_follow; 617 break; 618 619 case rparen: /* got a ')' or ']' */ 620 if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.not_cast_mask) { 621 ps.last_u_d = true; 622 ps.cast_mask &= (1 << ps.p_l_follow) - 1; 623 ps.want_blank = opt.space_after_cast; 624 } else 625 ps.want_blank = true; 626 ps.not_cast_mask &= (1 << ps.p_l_follow) - 1; 627 if (--ps.p_l_follow < 0) { 628 ps.p_l_follow = 0; 629 diag3(0, "Extra %c", *token); 630 } 631 if (e_code == s_code) /* if the paren starts the line */ 632 ps.paren_level = ps.p_l_follow; /* then indent it */ 633 634 *e_code++ = token[0]; 635 636 if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if 637 * (...), or some such */ 638 sp_sw = false; 639 force_nl = true;/* must force newline after if */ 640 ps.last_u_d = true; /* inform lexi that a following 641 * operator is unary */ 642 ps.in_stmt = false; /* dont use stmt continuation 643 * indentation */ 644 645 parse(hd_type); /* let parser worry about if, or whatever */ 646 } 647 ps.search_brace = opt.btype_2; /* this should ensure that 648 * constructs such as main(){...} 649 * and int[]{...} have their braces 650 * put in the right place */ 651 break; 652 653 case unary_op: /* this could be any unary operation */ 654 if (!ps.dumped_decl_indent && ps.in_decl && !ps.block_init && 655 ps.procname[0] == '\0' && ps.paren_level == 0) { 656 /* pointer declarations */ 657 658 /* 659 * if this is a unary op in a declaration, we should indent 660 * this token 661 */ 662 for (i = 0; token[i]; ++i) 663 /* find length of token */; 664 indent_declaration(dec_ind - i, tabs_to_var); 665 ps.dumped_decl_indent = true; 666 } 667 else if (ps.want_blank) 668 *e_code++ = ' '; 669 670 { 671 int len = e_token - s_token; 672 673 CHECK_SIZE_CODE(len); 674 memcpy(e_code, token, len); 675 e_code += len; 676 } 677 ps.want_blank = false; 678 break; 679 680 case binary_op: /* any binary operation */ 681 { 682 int len = e_token - s_token; 683 684 CHECK_SIZE_CODE(len + 1); 685 if (ps.want_blank) 686 *e_code++ = ' '; 687 memcpy(e_code, token, len); 688 e_code += len; 689 } 690 ps.want_blank = true; 691 break; 692 693 case postop: /* got a trailing ++ or -- */ 694 *e_code++ = token[0]; 695 *e_code++ = token[1]; 696 ps.want_blank = true; 697 break; 698 699 case question: /* got a ? */ 700 squest++; /* this will be used when a later colon 701 * appears so we can distinguish the 702 * <c>?<n>:<n> construct */ 703 if (ps.want_blank) 704 *e_code++ = ' '; 705 *e_code++ = '?'; 706 ps.want_blank = true; 707 break; 708 709 case casestmt: /* got word 'case' or 'default' */ 710 scase = true; /* so we can process the later colon properly */ 711 goto copy_id; 712 713 case colon: /* got a ':' */ 714 if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */ 715 --squest; 716 if (ps.want_blank) 717 *e_code++ = ' '; 718 *e_code++ = ':'; 719 ps.want_blank = true; 720 break; 721 } 722 if (ps.in_or_st) { 723 *e_code++ = ':'; 724 ps.want_blank = false; 725 break; 726 } 727 ps.in_stmt = false; /* seeing a label does not imply we are in a 728 * stmt */ 729 /* 730 * turn everything so far into a label 731 */ 732 { 733 int len = e_code - s_code; 734 735 CHECK_SIZE_LAB(len + 3); 736 memcpy(e_lab, s_code, len); 737 e_lab += len; 738 *e_lab++ = ':'; 739 *e_lab = '\0'; 740 e_code = s_code; 741 } 742 force_nl = ps.pcase = scase; /* ps.pcase will be used by 743 * dump_line to decide how to 744 * indent the label. force_nl 745 * will force a case n: to be 746 * on a line by itself */ 747 scase = false; 748 ps.want_blank = false; 749 break; 750 751 case semicolon: /* got a ';' */ 752 if (ps.dec_nest == 0) 753 ps.in_or_st = false;/* we are not in an initialization or 754 * structure declaration */ 755 scase = false; /* these will only need resetting in an error */ 756 squest = 0; 757 if (ps.last_token == rparen) 758 ps.in_parameter_declaration = 0; 759 ps.cast_mask = 0; 760 ps.not_cast_mask = 0; 761 ps.block_init = 0; 762 ps.block_init_level = 0; 763 ps.just_saw_decl--; 764 765 if (ps.in_decl && s_code == e_code && !ps.block_init && 766 !ps.dumped_decl_indent && ps.paren_level == 0) { 767 /* indent stray semicolons in declarations */ 768 indent_declaration(dec_ind - 1, tabs_to_var); 769 ps.dumped_decl_indent = true; 770 } 771 772 ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level 773 * structure declaration, we 774 * arent any more */ 775 776 if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) { 777 778 /* 779 * This should be true iff there were unbalanced parens in the 780 * stmt. It is a bit complicated, because the semicolon might 781 * be in a for stmt 782 */ 783 diag2(1, "Unbalanced parens"); 784 ps.p_l_follow = 0; 785 if (sp_sw) { /* this is a check for an if, while, etc. with 786 * unbalanced parens */ 787 sp_sw = false; 788 parse(hd_type); /* dont lose the if, or whatever */ 789 } 790 } 791 *e_code++ = ';'; 792 ps.want_blank = true; 793 ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the 794 * middle of a stmt */ 795 796 if (!sp_sw) { /* if not if for (;;) */ 797 parse(semicolon); /* let parser know about end of stmt */ 798 force_nl = true;/* force newline after an end of stmt */ 799 } 800 break; 801 802 case lbrace: /* got a '{' */ 803 ps.in_stmt = false; /* dont indent the {} */ 804 if (!ps.block_init) 805 force_nl = true;/* force other stuff on same line as '{' onto 806 * new line */ 807 else if (ps.block_init_level <= 0) 808 ps.block_init_level = 1; 809 else 810 ps.block_init_level++; 811 812 if (s_code != e_code && !ps.block_init) { 813 if (!opt.btype_2) { 814 dump_line(); 815 ps.want_blank = false; 816 } 817 else if (ps.in_parameter_declaration && !ps.in_or_st) { 818 ps.i_l_follow = 0; 819 if (opt.function_brace_split) { /* dump the line prior 820 * to the brace ... */ 821 dump_line(); 822 ps.want_blank = false; 823 } else /* add a space between the decl and brace */ 824 ps.want_blank = true; 825 } 826 } 827 if (ps.in_parameter_declaration) 828 prefix_blankline_requested = 0; 829 830 if (ps.p_l_follow > 0) { /* check for preceding unbalanced 831 * parens */ 832 diag2(1, "Unbalanced parens"); 833 ps.p_l_follow = 0; 834 if (sp_sw) { /* check for unclosed if, for, etc. */ 835 sp_sw = false; 836 parse(hd_type); 837 ps.ind_level = ps.i_l_follow; 838 } 839 } 840 if (s_code == e_code) 841 ps.ind_stmt = false; /* dont put extra indentation on line 842 * with '{' */ 843 if (ps.in_decl && ps.in_or_st) { /* this is either a structure 844 * declaration or an init */ 845 di_stack[ps.dec_nest] = dec_ind; 846 if (++ps.dec_nest == nitems(di_stack)) { 847 diag3(0, "Reached internal limit of %d struct levels", 848 nitems(di_stack)); 849 ps.dec_nest--; 850 } 851 /* ? dec_ind = 0; */ 852 } 853 else { 854 ps.decl_on_line = false; /* we can't be in the middle of 855 * a declaration, so don't do 856 * special indentation of 857 * comments */ 858 if (opt.blanklines_after_declarations_at_proctop 859 && ps.in_parameter_declaration) 860 postfix_blankline_requested = 1; 861 ps.in_parameter_declaration = 0; 862 ps.in_decl = false; 863 } 864 dec_ind = 0; 865 parse(lbrace); /* let parser know about this */ 866 if (ps.want_blank) /* put a blank before '{' if '{' is not at 867 * start of line */ 868 *e_code++ = ' '; 869 ps.want_blank = false; 870 *e_code++ = '{'; 871 ps.just_saw_decl = 0; 872 break; 873 874 case rbrace: /* got a '}' */ 875 if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be 876 * omitted in 877 * declarations */ 878 parse(semicolon); 879 if (ps.p_l_follow) {/* check for unclosed if, for, else. */ 880 diag2(1, "Unbalanced parens"); 881 ps.p_l_follow = 0; 882 sp_sw = false; 883 } 884 ps.just_saw_decl = 0; 885 ps.block_init_level--; 886 if (s_code != e_code && !ps.block_init) { /* '}' must be first on 887 * line */ 888 if (opt.verbose) 889 diag2(0, "Line broken"); 890 dump_line(); 891 } 892 *e_code++ = '}'; 893 ps.want_blank = true; 894 ps.in_stmt = ps.ind_stmt = false; 895 if (ps.dec_nest > 0) { /* we are in multi-level structure 896 * declaration */ 897 dec_ind = di_stack[--ps.dec_nest]; 898 if (ps.dec_nest == 0 && !ps.in_parameter_declaration) 899 ps.just_saw_decl = 2; 900 ps.in_decl = true; 901 } 902 prefix_blankline_requested = 0; 903 parse(rbrace); /* let parser know about this */ 904 ps.search_brace = opt.cuddle_else && ps.p_stack[ps.tos] == ifhead 905 && ps.il[ps.tos] >= ps.ind_level; 906 if (ps.tos <= 1 && opt.blanklines_after_procs && ps.dec_nest <= 0) 907 postfix_blankline_requested = 1; 908 break; 909 910 case swstmt: /* got keyword "switch" */ 911 sp_sw = true; 912 hd_type = swstmt; /* keep this for when we have seen the 913 * expression */ 914 goto copy_id; /* go move the token into buffer */ 915 916 case sp_paren: /* token is if, while, for */ 917 sp_sw = true; /* the interesting stuff is done after the 918 * expression is scanned */ 919 hd_type = (*token == 'i' ? ifstmt : 920 (*token == 'w' ? whilestmt : forstmt)); 921 922 /* 923 * remember the type of header for later use by parser 924 */ 925 goto copy_id; /* copy the token into line */ 926 927 case sp_nparen: /* got else, do */ 928 ps.in_stmt = false; 929 if (*token == 'e') { 930 if (e_code != s_code && (!opt.cuddle_else || e_code[-1] != '}')) { 931 if (opt.verbose) 932 diag2(0, "Line broken"); 933 dump_line();/* make sure this starts a line */ 934 ps.want_blank = false; 935 } 936 force_nl = true;/* also, following stuff must go onto new line */ 937 last_else = 1; 938 parse(elselit); 939 } 940 else { 941 if (e_code != s_code) { /* make sure this starts a line */ 942 if (opt.verbose) 943 diag2(0, "Line broken"); 944 dump_line(); 945 ps.want_blank = false; 946 } 947 force_nl = true;/* also, following stuff must go onto new line */ 948 last_else = 0; 949 parse(dolit); 950 } 951 goto copy_id; /* move the token into line */ 952 953 case type_def: 954 case storage: 955 prefix_blankline_requested = 0; 956 goto copy_id; 957 958 case structure: 959 if (ps.p_l_follow > 0) 960 goto copy_id; 961 /* FALLTHROUGH */ 962 case decl: /* we have a declaration type (int, etc.) */ 963 parse(decl); /* let parser worry about indentation */ 964 if (ps.last_token == rparen && ps.tos <= 1) { 965 if (s_code != e_code) { 966 dump_line(); 967 ps.want_blank = 0; 968 } 969 } 970 if (ps.in_parameter_declaration && opt.indent_parameters && ps.dec_nest == 0) { 971 ps.ind_level = ps.i_l_follow = 1; 972 ps.ind_stmt = 0; 973 } 974 ps.in_or_st = true; /* this might be a structure or initialization 975 * declaration */ 976 ps.in_decl = ps.decl_on_line = ps.last_token != type_def; 977 if ( /* !ps.in_or_st && */ ps.dec_nest <= 0) 978 ps.just_saw_decl = 2; 979 prefix_blankline_requested = 0; 980 for (i = 0; token[i++];); /* get length of token */ 981 982 if (ps.ind_level == 0 || ps.dec_nest > 0) { 983 /* global variable or struct member in local variable */ 984 dec_ind = opt.decl_indent > 0 ? opt.decl_indent : i; 985 tabs_to_var = (opt.use_tabs ? opt.decl_indent > 0 : 0); 986 } else { 987 /* local variable */ 988 dec_ind = opt.local_decl_indent > 0 ? opt.local_decl_indent : i; 989 tabs_to_var = (opt.use_tabs ? opt.local_decl_indent > 0 : 0); 990 } 991 goto copy_id; 992 993 case funcname: 994 case ident: /* got an identifier or constant */ 995 if (ps.in_decl) { 996 if (type_code == funcname) { 997 ps.in_decl = false; 998 if (opt.procnames_start_line && s_code != e_code) { 999 *e_code = '\0'; 1000 dump_line(); 1001 } 1002 else if (ps.want_blank) { 1003 *e_code++ = ' '; 1004 } 1005 ps.want_blank = false; 1006 } 1007 else if (!ps.block_init && !ps.dumped_decl_indent && 1008 ps.paren_level == 0) { /* if we are in a declaration, we 1009 * must indent identifier */ 1010 indent_declaration(dec_ind, tabs_to_var); 1011 ps.dumped_decl_indent = true; 1012 ps.want_blank = false; 1013 } 1014 } 1015 else if (sp_sw && ps.p_l_follow == 0) { 1016 sp_sw = false; 1017 force_nl = true; 1018 ps.last_u_d = true; 1019 ps.in_stmt = false; 1020 parse(hd_type); 1021 } 1022 copy_id: 1023 { 1024 int len = e_token - s_token; 1025 1026 CHECK_SIZE_CODE(len + 1); 1027 if (ps.want_blank) 1028 *e_code++ = ' '; 1029 memcpy(e_code, s_token, len); 1030 e_code += len; 1031 } 1032 if (type_code != funcname) 1033 ps.want_blank = true; 1034 break; 1035 1036 case strpfx: 1037 { 1038 int len = e_token - s_token; 1039 1040 CHECK_SIZE_CODE(len + 1); 1041 if (ps.want_blank) 1042 *e_code++ = ' '; 1043 memcpy(e_code, token, len); 1044 e_code += len; 1045 } 1046 ps.want_blank = false; 1047 break; 1048 1049 case period: /* treat a period kind of like a binary 1050 * operation */ 1051 *e_code++ = '.'; /* move the period into line */ 1052 ps.want_blank = false; /* dont put a blank after a period */ 1053 break; 1054 1055 case comma: 1056 ps.want_blank = (s_code != e_code); /* only put blank after comma 1057 * if comma does not start the 1058 * line */ 1059 if (ps.in_decl && ps.procname[0] == '\0' && !ps.block_init && 1060 !ps.dumped_decl_indent && ps.paren_level == 0) { 1061 /* indent leading commas and not the actual identifiers */ 1062 indent_declaration(dec_ind - 1, tabs_to_var); 1063 ps.dumped_decl_indent = true; 1064 } 1065 *e_code++ = ','; 1066 if (ps.p_l_follow == 0) { 1067 if (ps.block_init_level <= 0) 1068 ps.block_init = 0; 1069 if (break_comma && (!opt.leave_comma || 1070 count_spaces_until(compute_code_target(), s_code, e_code) > 1071 opt.max_col - opt.tabsize)) 1072 force_nl = true; 1073 } 1074 break; 1075 1076 case preesc: /* got the character '#' */ 1077 if ((s_com != e_com) || 1078 (s_lab != e_lab) || 1079 (s_code != e_code)) 1080 dump_line(); 1081 CHECK_SIZE_LAB(1); 1082 *e_lab++ = '#'; /* move whole line to 'label' buffer */ 1083 { 1084 int in_comment = 0; 1085 int com_start = 0; 1086 char quote = 0; 1087 int com_end = 0; 1088 1089 while (*buf_ptr == ' ' || *buf_ptr == '\t') { 1090 buf_ptr++; 1091 if (buf_ptr >= buf_end) 1092 fill_buffer(); 1093 } 1094 while (*buf_ptr != '\n' || (in_comment && !had_eof)) { 1095 CHECK_SIZE_LAB(2); 1096 *e_lab = *buf_ptr++; 1097 if (buf_ptr >= buf_end) 1098 fill_buffer(); 1099 switch (*e_lab++) { 1100 case BACKSLASH: 1101 if (!in_comment) { 1102 *e_lab++ = *buf_ptr++; 1103 if (buf_ptr >= buf_end) 1104 fill_buffer(); 1105 } 1106 break; 1107 case '/': 1108 if (*buf_ptr == '*' && !in_comment && !quote) { 1109 in_comment = 1; 1110 *e_lab++ = *buf_ptr++; 1111 com_start = e_lab - s_lab - 2; 1112 } 1113 break; 1114 case '"': 1115 if (quote == '"') 1116 quote = 0; 1117 break; 1118 case '\'': 1119 if (quote == '\'') 1120 quote = 0; 1121 break; 1122 case '*': 1123 if (*buf_ptr == '/' && in_comment) { 1124 in_comment = 0; 1125 *e_lab++ = *buf_ptr++; 1126 com_end = e_lab - s_lab; 1127 } 1128 break; 1129 } 1130 } 1131 1132 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1133 e_lab--; 1134 if (e_lab - s_lab == com_end && bp_save == NULL) { 1135 /* comment on preprocessor line */ 1136 if (sc_end == NULL) { /* if this is the first comment, 1137 * we must set up the buffer */ 1138 save_com = sc_buf; 1139 sc_end = &save_com[0]; 1140 } 1141 else { 1142 *sc_end++ = '\n'; /* add newline between 1143 * comments */ 1144 *sc_end++ = ' '; 1145 --line_no; 1146 } 1147 if (sc_end - save_com + com_end - com_start > sc_size) 1148 errx(1, "input too long"); 1149 memmove(sc_end, s_lab + com_start, com_end - com_start); 1150 sc_end += com_end - com_start; 1151 e_lab = s_lab + com_start; 1152 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1153 e_lab--; 1154 bp_save = buf_ptr; /* save current input buffer */ 1155 be_save = buf_end; 1156 buf_ptr = save_com; /* fix so that subsequent calls to 1157 * lexi will take tokens out of 1158 * save_com */ 1159 *sc_end++ = ' '; /* add trailing blank, just in case */ 1160 buf_end = sc_end; 1161 sc_end = NULL; 1162 } 1163 CHECK_SIZE_LAB(1); 1164 *e_lab = '\0'; /* null terminate line */ 1165 ps.pcase = false; 1166 } 1167 1168 if (strncmp(s_lab, "#if", 3) == 0) { /* also ifdef, ifndef */ 1169 if ((size_t)ifdef_level < nitems(state_stack)) { 1170 match_state[ifdef_level].tos = -1; 1171 state_stack[ifdef_level++] = ps; 1172 } 1173 else 1174 diag2(1, "#if stack overflow"); 1175 } 1176 else if (strncmp(s_lab, "#el", 3) == 0) { /* else, elif */ 1177 if (ifdef_level <= 0) 1178 diag2(1, s_lab[3] == 'i' ? "Unmatched #elif" : "Unmatched #else"); 1179 else { 1180 match_state[ifdef_level - 1] = ps; 1181 ps = state_stack[ifdef_level - 1]; 1182 } 1183 } 1184 else if (strncmp(s_lab, "#endif", 6) == 0) { 1185 if (ifdef_level <= 0) 1186 diag2(1, "Unmatched #endif"); 1187 else 1188 ifdef_level--; 1189 } else { 1190 struct directives { 1191 int size; 1192 const char *string; 1193 } 1194 recognized[] = { 1195 {7, "include"}, 1196 {6, "define"}, 1197 {5, "undef"}, 1198 {4, "line"}, 1199 {5, "error"}, 1200 {6, "pragma"} 1201 }; 1202 int d = nitems(recognized); 1203 while (--d >= 0) 1204 if (strncmp(s_lab + 1, recognized[d].string, recognized[d].size) == 0) 1205 break; 1206 if (d < 0) { 1207 diag2(1, "Unrecognized cpp directive"); 1208 break; 1209 } 1210 } 1211 if (opt.blanklines_around_conditional_compilation) { 1212 postfix_blankline_requested++; 1213 n_real_blanklines = 0; 1214 } 1215 else { 1216 postfix_blankline_requested = 0; 1217 prefix_blankline_requested = 0; 1218 } 1219 break; /* subsequent processing of the newline 1220 * character will cause the line to be printed */ 1221 1222 case comment: /* we have gotten a / followed by * this is a biggie */ 1223 pr_comment(); 1224 break; 1225 } /* end of big switch stmt */ 1226 1227 *e_code = '\0'; /* make sure code section is null terminated */ 1228 if (type_code != comment && type_code != newline && type_code != preesc) 1229 ps.last_token = type_code; 1230 } /* end of main while (1) loop */ 1231 } 1232 1233 /* 1234 * copy input file to backup file if in_name is /blah/blah/blah/file, then 1235 * backup file will be ".Bfile" then make the backup file the input and 1236 * original input file the output 1237 */ 1238 static void 1239 bakcopy(void) 1240 { 1241 int n, 1242 bakchn; 1243 char buff[8 * 1024]; 1244 const char *p; 1245 1246 /* construct file name .Bfile */ 1247 for (p = in_name; *p; p++); /* skip to end of string */ 1248 while (p > in_name && *p != '/') /* find last '/' */ 1249 p--; 1250 if (*p == '/') 1251 p++; 1252 sprintf(bakfile, "%s%s", p, simple_backup_suffix); 1253 1254 /* copy in_name to backup file */ 1255 bakchn = creat(bakfile, 0600); 1256 if (bakchn < 0) 1257 err(1, "%s", bakfile); 1258 while ((n = read(fileno(input), buff, sizeof(buff))) > 0) 1259 if (write(bakchn, buff, n) != n) 1260 err(1, "%s", bakfile); 1261 if (n < 0) 1262 err(1, "%s", in_name); 1263 close(bakchn); 1264 fclose(input); 1265 1266 /* re-open backup file as the input file */ 1267 input = fopen(bakfile, "r"); 1268 if (input == NULL) 1269 err(1, "%s", bakfile); 1270 /* now the original input file will be the output */ 1271 output = fopen(in_name, "w"); 1272 if (output == NULL) { 1273 unlink(bakfile); 1274 err(1, "%s", in_name); 1275 } 1276 } 1277 1278 static void 1279 indent_declaration(int cur_dec_ind, int tabs_to_var) 1280 { 1281 int pos = e_code - s_code; 1282 char *startpos = e_code; 1283 1284 /* 1285 * get the tab math right for indentations that are not multiples of tabsize 1286 */ 1287 if ((ps.ind_level * opt.ind_size) % opt.tabsize != 0) { 1288 pos += (ps.ind_level * opt.ind_size) % opt.tabsize; 1289 cur_dec_ind += (ps.ind_level * opt.ind_size) % opt.tabsize; 1290 } 1291 if (tabs_to_var) { 1292 int tpos; 1293 1294 CHECK_SIZE_CODE(cur_dec_ind / opt.tabsize); 1295 while ((tpos = opt.tabsize * (1 + pos / opt.tabsize)) <= cur_dec_ind) { 1296 *e_code++ = '\t'; 1297 pos = tpos; 1298 } 1299 } 1300 CHECK_SIZE_CODE(cur_dec_ind - pos + 1); 1301 while (pos < cur_dec_ind) { 1302 *e_code++ = ' '; 1303 pos++; 1304 } 1305 if (e_code == startpos && ps.want_blank) { 1306 *e_code++ = ' '; 1307 ps.want_blank = false; 1308 } 1309 } 1310