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 #if 0 39 #ifndef lint 40 static char sccsid[] = "@(#)indent.c 5.17 (Berkeley) 6/7/93"; 41 #endif /* not lint */ 42 #endif 43 44 #include <sys/cdefs.h> 45 __FBSDID("$FreeBSD$"); 46 47 #include <sys/param.h> 48 #include <sys/capsicum.h> 49 #include <capsicum_helpers.h> 50 #include <err.h> 51 #include <errno.h> 52 #include <fcntl.h> 53 #include <unistd.h> 54 #include <stdio.h> 55 #include <stdlib.h> 56 #include <string.h> 57 #include <ctype.h> 58 #include "indent_globs.h" 59 #include "indent_codes.h" 60 #include "indent.h" 61 62 static void bakcopy(void); 63 static void indent_declaration(int, int); 64 65 const char *in_name = "Standard Input"; /* will always point to name of input 66 * file */ 67 const char *out_name = "Standard Output"; /* will always point to name 68 * of output file */ 69 const char *simple_backup_suffix = ".BAK"; /* Suffix to use for backup 70 * files */ 71 char bakfile[MAXPATHLEN] = ""; 72 73 int 74 main(int argc, char **argv) 75 { 76 cap_rights_t rights; 77 78 int dec_ind; /* current indentation for declarations */ 79 int di_stack[20]; /* a stack of structure indentation levels */ 80 int force_nl; /* when true, code must be broken */ 81 int hd_type = 0; /* used to store type of stmt for if (...), 82 * for (...), etc */ 83 int i; /* local loop counter */ 84 int scase; /* set to true when we see a case, so we will 85 * know what to do with the following colon */ 86 int sp_sw; /* when true, we are in the expression of 87 * if(...), while(...), etc. */ 88 int squest; /* when this is positive, we have seen a ? 89 * without the matching : in a <c>?<s>:<s> 90 * construct */ 91 const char *t_ptr; /* used for copying tokens */ 92 int tabs_to_var; /* true if using tabs to indent to var name */ 93 int type_code; /* the type of token, returned by lexi */ 94 95 int last_else = 0; /* true iff last keyword was an else */ 96 const char *profile_name = NULL; 97 const char *envval = NULL; 98 struct parser_state transient_state; /* a copy for lookup */ 99 100 /*-----------------------------------------------*\ 101 | INITIALIZATION | 102 \*-----------------------------------------------*/ 103 104 found_err = 0; 105 106 ps.p_stack[0] = stmt; /* this is the parser's stack */ 107 ps.last_nl = true; /* this is true if the last thing scanned was 108 * a newline */ 109 ps.last_token = semicolon; 110 combuf = (char *) malloc(bufsize); 111 if (combuf == NULL) 112 err(1, NULL); 113 labbuf = (char *) malloc(bufsize); 114 if (labbuf == NULL) 115 err(1, NULL); 116 codebuf = (char *) malloc(bufsize); 117 if (codebuf == NULL) 118 err(1, NULL); 119 tokenbuf = (char *) malloc(bufsize); 120 if (tokenbuf == NULL) 121 err(1, NULL); 122 alloc_typenames(); 123 l_com = combuf + bufsize - 5; 124 l_lab = labbuf + bufsize - 5; 125 l_code = codebuf + bufsize - 5; 126 l_token = tokenbuf + bufsize - 5; 127 combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and 128 * comment buffers */ 129 combuf[1] = codebuf[1] = labbuf[1] = '\0'; 130 opt.else_if = 1; /* Default else-if special processing to on */ 131 s_lab = e_lab = labbuf + 1; 132 s_code = e_code = codebuf + 1; 133 s_com = e_com = combuf + 1; 134 s_token = e_token = tokenbuf + 1; 135 136 in_buffer = (char *) malloc(10); 137 if (in_buffer == NULL) 138 err(1, NULL); 139 in_buffer_limit = in_buffer + 8; 140 buf_ptr = buf_end = in_buffer; 141 line_no = 1; 142 had_eof = ps.in_decl = ps.decl_on_line = break_comma = false; 143 sp_sw = force_nl = false; 144 ps.in_or_st = false; 145 ps.bl_line = true; 146 dec_ind = 0; 147 di_stack[ps.dec_nest = 0] = 0; 148 ps.want_blank = ps.in_stmt = ps.ind_stmt = false; 149 150 scase = ps.pcase = false; 151 squest = 0; 152 sc_end = NULL; 153 bp_save = NULL; 154 be_save = NULL; 155 156 output = NULL; 157 tabs_to_var = 0; 158 159 envval = getenv("SIMPLE_BACKUP_SUFFIX"); 160 if (envval) 161 simple_backup_suffix = envval; 162 163 /*--------------------------------------------------*\ 164 | COMMAND LINE SCAN | 165 \*--------------------------------------------------*/ 166 167 #ifdef undef 168 max_col = 78; /* -l78 */ 169 lineup_to_parens = 1; /* -lp */ 170 lineup_to_parens_always = 0; /* -nlpl */ 171 ps.ljust_decl = 0; /* -ndj */ 172 ps.com_ind = 33; /* -c33 */ 173 star_comment_cont = 1; /* -sc */ 174 ps.ind_size = 8; /* -i8 */ 175 verbose = 0; 176 ps.decl_indent = 16; /* -di16 */ 177 ps.local_decl_indent = -1; /* if this is not set to some nonnegative value 178 * by an arg, we will set this equal to 179 * ps.decl_ind */ 180 ps.indent_parameters = 1; /* -ip */ 181 ps.decl_com_ind = 0; /* if this is not set to some positive value 182 * by an arg, we will set this equal to 183 * ps.com_ind */ 184 btype_2 = 1; /* -br */ 185 cuddle_else = 1; /* -ce */ 186 ps.unindent_displace = 0; /* -d0 */ 187 ps.case_indent = 0; /* -cli0 */ 188 format_block_comments = 1; /* -fcb */ 189 format_col1_comments = 1; /* -fc1 */ 190 procnames_start_line = 1; /* -psl */ 191 proc_calls_space = 0; /* -npcs */ 192 comment_delimiter_on_blankline = 1; /* -cdb */ 193 ps.leave_comma = 1; /* -nbc */ 194 #endif 195 196 for (i = 1; i < argc; ++i) 197 if (strcmp(argv[i], "-npro") == 0) 198 break; 199 else if (argv[i][0] == '-' && argv[i][1] == 'P' && argv[i][2] != '\0') 200 profile_name = argv[i]; /* non-empty -P (set profile) */ 201 set_defaults(); 202 if (i >= argc) 203 set_profile(profile_name); 204 205 for (i = 1; i < argc; ++i) { 206 207 /* 208 * look thru args (if any) for changes to defaults 209 */ 210 if (argv[i][0] != '-') {/* no flag on parameter */ 211 if (input == NULL) { /* we must have the input file */ 212 in_name = argv[i]; /* remember name of input file */ 213 input = fopen(in_name, "r"); 214 if (input == NULL) /* check for open error */ 215 err(1, "%s", in_name); 216 continue; 217 } 218 else if (output == NULL) { /* we have the output file */ 219 out_name = argv[i]; /* remember name of output file */ 220 if (strcmp(in_name, out_name) == 0) { /* attempt to overwrite 221 * the file */ 222 errx(1, "input and output files must be different"); 223 } 224 output = fopen(out_name, "w"); 225 if (output == NULL) /* check for create error */ 226 err(1, "%s", out_name); 227 continue; 228 } 229 errx(1, "unknown parameter: %s", argv[i]); 230 } 231 else 232 set_option(argv[i]); 233 } /* end of for */ 234 if (input == NULL) 235 input = stdin; 236 if (output == NULL) { 237 if (input == stdin) 238 output = stdout; 239 else { 240 out_name = in_name; 241 bakcopy(); 242 } 243 } 244 245 /* Restrict input/output descriptors and enter Capsicum sandbox. */ 246 cap_rights_init(&rights, CAP_FSTAT, CAP_WRITE); 247 if (cap_rights_limit(fileno(output), &rights) < 0 && errno != ENOSYS) 248 err(EXIT_FAILURE, "unable to limit rights for %s", out_name); 249 cap_rights_init(&rights, CAP_FSTAT, CAP_READ); 250 if (cap_rights_limit(fileno(input), &rights) < 0 && errno != ENOSYS) 251 err(EXIT_FAILURE, "unable to limit rights for %s", in_name); 252 if (caph_enter() < 0) 253 err(EXIT_FAILURE, "unable to enter capability mode"); 254 255 if (opt.com_ind <= 1) 256 opt.com_ind = 2; /* don't put normal comments before column 2 */ 257 if (opt.block_comment_max_col <= 0) 258 opt.block_comment_max_col = opt.max_col; 259 if (opt.local_decl_indent < 0) /* if not specified by user, set this */ 260 opt.local_decl_indent = opt.decl_indent; 261 if (opt.decl_com_ind <= 0) /* if not specified by user, set this */ 262 opt.decl_com_ind = opt.ljust_decl ? (opt.com_ind <= 10 ? 2 : opt.com_ind - 8) : opt.com_ind; 263 if (opt.continuation_indent == 0) 264 opt.continuation_indent = opt.ind_size; 265 fill_buffer(); /* get first batch of stuff into input buffer */ 266 267 parse(semicolon); 268 { 269 char *p = buf_ptr; 270 int col = 1; 271 272 while (1) { 273 if (*p == ' ') 274 col++; 275 else if (*p == '\t') 276 col = opt.tabsize * (1 + (col - 1) / opt.tabsize) + 1; 277 else 278 break; 279 p++; 280 } 281 if (col > opt.ind_size) 282 ps.ind_level = ps.i_l_follow = col / opt.ind_size; 283 } 284 285 /* 286 * START OF MAIN LOOP 287 */ 288 289 while (1) { /* this is the main loop. it will go until we 290 * reach eof */ 291 int comment_buffered = false; 292 293 type_code = lexi(&ps); /* lexi reads one token. The actual 294 * characters read are stored in "token". lexi 295 * returns a code indicating the type of token */ 296 297 /* 298 * The following code moves newlines and comments following an if (), 299 * while (), else, etc. up to the start of the following stmt to 300 * a buffer. This allows proper handling of both kinds of brace 301 * placement (-br, -bl) and cuddling "else" (-ce). 302 */ 303 304 while (ps.search_brace) { 305 switch (type_code) { 306 case newline: 307 if (sc_end == NULL) { 308 save_com = sc_buf; 309 save_com[0] = save_com[1] = ' '; 310 sc_end = &save_com[2]; 311 } 312 *sc_end++ = '\n'; 313 /* 314 * We may have inherited a force_nl == true from the previous 315 * token (like a semicolon). But once we know that a newline 316 * has been scanned in this loop, force_nl should be false. 317 * 318 * However, the force_nl == true must be preserved if newline 319 * is never scanned in this loop, so this assignment cannot be 320 * done earlier. 321 */ 322 force_nl = false; 323 case form_feed: 324 break; 325 case comment: 326 if (sc_end == NULL) { 327 /* 328 * Copy everything from the start of the line, because 329 * pr_comment() will use that to calculate original 330 * indentation of a boxed comment. 331 */ 332 memcpy(sc_buf, in_buffer, buf_ptr - in_buffer - 4); 333 save_com = sc_buf + (buf_ptr - in_buffer - 4); 334 save_com[0] = save_com[1] = ' '; 335 sc_end = &save_com[2]; 336 } 337 comment_buffered = true; 338 *sc_end++ = '/'; /* copy in start of comment */ 339 *sc_end++ = '*'; 340 for (;;) { /* loop until we get to the end of the comment */ 341 *sc_end = *buf_ptr++; 342 if (buf_ptr >= buf_end) 343 fill_buffer(); 344 if (*sc_end++ == '*' && *buf_ptr == '/') 345 break; /* we are at end of comment */ 346 if (sc_end >= &save_com[sc_size]) { /* check for temp buffer 347 * overflow */ 348 diag2(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever"); 349 fflush(output); 350 exit(1); 351 } 352 } 353 *sc_end++ = '/'; /* add ending slash */ 354 if (++buf_ptr >= buf_end) /* get past / in buffer */ 355 fill_buffer(); 356 break; 357 case lbrace: 358 /* 359 * Put KNF-style lbraces before the buffered up tokens and 360 * jump out of this loop in order to avoid copying the token 361 * again under the default case of the switch below. 362 */ 363 if (sc_end != NULL && opt.btype_2) { 364 save_com[0] = '{'; 365 /* 366 * Originally the lbrace may have been alone on its own 367 * line, but it will be moved into "the else's line", so 368 * if there was a newline resulting from the "{" before, 369 * it must be scanned now and ignored. 370 */ 371 while (isspace((unsigned char)*buf_ptr)) { 372 if (++buf_ptr >= buf_end) 373 fill_buffer(); 374 if (*buf_ptr == '\n') 375 break; 376 } 377 goto sw_buffer; 378 } 379 /* FALLTHROUGH */ 380 default: /* it is the start of a normal statement */ 381 { 382 int remove_newlines; 383 384 remove_newlines = 385 /* "} else" */ 386 (type_code == sp_nparen && *token == 'e' && 387 e_code != s_code && e_code[-1] == '}') 388 /* "else if" */ 389 || (type_code == sp_paren && *token == 'i' && 390 last_else && opt.else_if); 391 if (remove_newlines) 392 force_nl = false; 393 if (sc_end == NULL) { /* ignore buffering if 394 * comment wasn't saved up */ 395 ps.search_brace = false; 396 goto check_type; 397 } 398 while (sc_end > save_com && isblank((unsigned char)sc_end[-1])) { 399 sc_end--; 400 } 401 if (opt.swallow_optional_blanklines || 402 (!comment_buffered && remove_newlines)) { 403 force_nl = !remove_newlines; 404 while (sc_end > save_com && sc_end[-1] == '\n') { 405 sc_end--; 406 } 407 } 408 if (force_nl) { /* if we should insert a nl here, put 409 * it into the buffer */ 410 force_nl = false; 411 --line_no; /* this will be re-increased when the 412 * newline is read from the buffer */ 413 *sc_end++ = '\n'; 414 *sc_end++ = ' '; 415 if (opt.verbose) /* print error msg if the line was 416 * not already broken */ 417 diag2(0, "Line broken"); 418 } 419 for (t_ptr = token; *t_ptr; ++t_ptr) 420 *sc_end++ = *t_ptr; 421 422 sw_buffer: 423 ps.search_brace = false; /* stop looking for start of 424 * stmt */ 425 bp_save = buf_ptr; /* save current input buffer */ 426 be_save = buf_end; 427 buf_ptr = save_com; /* fix so that subsequent calls to 428 * lexi will take tokens out of 429 * save_com */ 430 *sc_end++ = ' ';/* add trailing blank, just in case */ 431 buf_end = sc_end; 432 sc_end = NULL; 433 break; 434 } 435 } /* end of switch */ 436 /* 437 * We must make this check, just in case there was an unexpected 438 * EOF. 439 */ 440 if (type_code != 0) { 441 /* 442 * The only intended purpose of calling lexi() below is to 443 * categorize the next token in order to decide whether to 444 * continue buffering forthcoming tokens. Once the buffering 445 * is over, lexi() will be called again elsewhere on all of 446 * the tokens - this time for normal processing. 447 * 448 * Calling it for this purpose is a bug, because lexi() also 449 * changes the parser state and discards leading whitespace, 450 * which is needed mostly for comment-related considerations. 451 * 452 * Work around the former problem by giving lexi() a copy of 453 * the current parser state and discard it if the call turned 454 * out to be just a look ahead. 455 * 456 * Work around the latter problem by copying all whitespace 457 * characters into the buffer so that the later lexi() call 458 * will read them. 459 */ 460 if (sc_end != NULL) { 461 while (*buf_ptr == ' ' || *buf_ptr == '\t') { 462 *sc_end++ = *buf_ptr++; 463 if (sc_end >= &save_com[sc_size]) { 464 errx(1, "input too long"); 465 } 466 } 467 if (buf_ptr >= buf_end) { 468 fill_buffer(); 469 } 470 } 471 transient_state = ps; 472 type_code = lexi(&transient_state); /* read another token */ 473 if (type_code != newline && type_code != form_feed && 474 type_code != comment && !transient_state.search_brace) { 475 ps = transient_state; 476 } 477 } 478 } /* end of while (search_brace) */ 479 last_else = 0; 480 check_type: 481 if (type_code == 0) { /* we got eof */ 482 if (s_lab != e_lab || s_code != e_code 483 || s_com != e_com) /* must dump end of line */ 484 dump_line(); 485 if (ps.tos > 1) /* check for balanced braces */ 486 diag2(1, "Stuff missing from end of file"); 487 488 if (opt.verbose) { 489 printf("There were %d output lines and %d comments\n", 490 ps.out_lines, ps.out_coms); 491 printf("(Lines with comments)/(Lines with code): %6.3f\n", 492 (1.0 * ps.com_lines) / code_lines); 493 } 494 fflush(output); 495 exit(found_err); 496 } 497 if ( 498 (type_code != comment) && 499 (type_code != newline) && 500 (type_code != preesc) && 501 (type_code != form_feed)) { 502 if (force_nl && 503 (type_code != semicolon) && 504 (type_code != lbrace || !opt.btype_2)) { 505 /* we should force a broken line here */ 506 if (opt.verbose) 507 diag2(0, "Line broken"); 508 dump_line(); 509 ps.want_blank = false; /* dont insert blank at line start */ 510 force_nl = false; 511 } 512 ps.in_stmt = true; /* turn on flag which causes an extra level of 513 * indentation. this is turned off by a ; or 514 * '}' */ 515 if (s_com != e_com) { /* the turkey has embedded a comment 516 * in a line. fix it */ 517 int len = e_com - s_com; 518 519 CHECK_SIZE_CODE(len + 3); 520 *e_code++ = ' '; 521 memcpy(e_code, s_com, len); 522 e_code += len; 523 *e_code++ = ' '; 524 *e_code = '\0'; /* null terminate code sect */ 525 ps.want_blank = false; 526 e_com = s_com; 527 } 528 } 529 else if (type_code != comment) /* preserve force_nl thru a comment */ 530 force_nl = false; /* cancel forced newline after newline, form 531 * feed, etc */ 532 533 534 535 /*-----------------------------------------------------*\ 536 | do switch on type of token scanned | 537 \*-----------------------------------------------------*/ 538 CHECK_SIZE_CODE(3); /* maximum number of increments of e_code 539 * before the next CHECK_SIZE_CODE or 540 * dump_line() is 2. After that there's the 541 * final increment for the null character. */ 542 switch (type_code) { /* now, decide what to do with the token */ 543 544 case form_feed: /* found a form feed in line */ 545 ps.use_ff = true; /* a form feed is treated much like a newline */ 546 dump_line(); 547 ps.want_blank = false; 548 break; 549 550 case newline: 551 if (ps.last_token != comma || ps.p_l_follow > 0 552 || !opt.leave_comma || ps.block_init || !break_comma || s_com != e_com) { 553 dump_line(); 554 ps.want_blank = false; 555 } 556 ++line_no; /* keep track of input line number */ 557 break; 558 559 case lparen: /* got a '(' or '[' */ 560 /* count parens to make Healy happy */ 561 if (++ps.p_l_follow == nitems(ps.paren_indents)) { 562 diag3(0, "Reached internal limit of %d unclosed parens", 563 nitems(ps.paren_indents)); 564 ps.p_l_follow--; 565 } 566 if (*token == '[') 567 /* not a function pointer declaration or a function call */; 568 else if (ps.in_decl && !ps.block_init && !ps.dumped_decl_indent && 569 ps.procname[0] == '\0' && ps.paren_level == 0) { 570 /* function pointer declarations */ 571 indent_declaration(dec_ind, tabs_to_var); 572 ps.dumped_decl_indent = true; 573 } 574 else if (ps.want_blank && 575 ((ps.last_token != ident && ps.last_token != funcname) || 576 opt.proc_calls_space || 577 /* offsetof (1) is never allowed a space; sizeof (2) gets 578 * one iff -bs; all other keywords (>2) always get a space 579 * before lparen */ 580 ps.keyword + opt.Bill_Shannon > 2)) 581 *e_code++ = ' '; 582 ps.want_blank = false; 583 *e_code++ = token[0]; 584 ps.paren_indents[ps.p_l_follow - 1] = count_spaces_until(1, s_code, e_code) - 1; 585 if (sp_sw && ps.p_l_follow == 1 && opt.extra_expression_indent 586 && ps.paren_indents[0] < 2 * opt.ind_size) 587 ps.paren_indents[0] = 2 * opt.ind_size; 588 if (ps.in_or_st && *token == '(' && ps.tos <= 2) { 589 /* 590 * this is a kluge to make sure that declarations will be 591 * aligned right if proc decl has an explicit type on it, i.e. 592 * "int a(x) {..." 593 */ 594 parse(semicolon); /* I said this was a kluge... */ 595 ps.in_or_st = false; /* turn off flag for structure decl or 596 * initialization */ 597 } 598 /* parenthesized type following sizeof or offsetof is not a cast */ 599 if (ps.keyword == 1 || ps.keyword == 2) 600 ps.not_cast_mask |= 1 << ps.p_l_follow; 601 break; 602 603 case rparen: /* got a ')' or ']' */ 604 if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.not_cast_mask) { 605 ps.last_u_d = true; 606 ps.cast_mask &= (1 << ps.p_l_follow) - 1; 607 ps.want_blank = opt.space_after_cast; 608 } else 609 ps.want_blank = true; 610 ps.not_cast_mask &= (1 << ps.p_l_follow) - 1; 611 if (--ps.p_l_follow < 0) { 612 ps.p_l_follow = 0; 613 diag3(0, "Extra %c", *token); 614 } 615 if (e_code == s_code) /* if the paren starts the line */ 616 ps.paren_level = ps.p_l_follow; /* then indent it */ 617 618 *e_code++ = token[0]; 619 620 if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if 621 * (...), or some such */ 622 sp_sw = false; 623 force_nl = true;/* must force newline after if */ 624 ps.last_u_d = true; /* inform lexi that a following 625 * operator is unary */ 626 ps.in_stmt = false; /* dont use stmt continuation 627 * indentation */ 628 629 parse(hd_type); /* let parser worry about if, or whatever */ 630 } 631 ps.search_brace = opt.btype_2; /* this should ensure that 632 * constructs such as main(){...} 633 * and int[]{...} have their braces 634 * put in the right place */ 635 break; 636 637 case unary_op: /* this could be any unary operation */ 638 if (!ps.dumped_decl_indent && ps.in_decl && !ps.block_init && 639 ps.procname[0] == '\0' && ps.paren_level == 0) { 640 /* pointer declarations */ 641 642 /* 643 * if this is a unary op in a declaration, we should indent 644 * this token 645 */ 646 for (i = 0; token[i]; ++i) 647 /* find length of token */; 648 indent_declaration(dec_ind - i, tabs_to_var); 649 ps.dumped_decl_indent = true; 650 } 651 else if (ps.want_blank) 652 *e_code++ = ' '; 653 654 { 655 int len = e_token - s_token; 656 657 CHECK_SIZE_CODE(len); 658 memcpy(e_code, token, len); 659 e_code += len; 660 } 661 ps.want_blank = false; 662 break; 663 664 case binary_op: /* any binary operation */ 665 { 666 int len = e_token - s_token; 667 668 CHECK_SIZE_CODE(len + 1); 669 if (ps.want_blank) 670 *e_code++ = ' '; 671 memcpy(e_code, token, len); 672 e_code += len; 673 } 674 ps.want_blank = true; 675 break; 676 677 case postop: /* got a trailing ++ or -- */ 678 *e_code++ = token[0]; 679 *e_code++ = token[1]; 680 ps.want_blank = true; 681 break; 682 683 case question: /* got a ? */ 684 squest++; /* this will be used when a later colon 685 * appears so we can distinguish the 686 * <c>?<n>:<n> construct */ 687 if (ps.want_blank) 688 *e_code++ = ' '; 689 *e_code++ = '?'; 690 ps.want_blank = true; 691 break; 692 693 case casestmt: /* got word 'case' or 'default' */ 694 scase = true; /* so we can process the later colon properly */ 695 goto copy_id; 696 697 case colon: /* got a ':' */ 698 if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */ 699 --squest; 700 if (ps.want_blank) 701 *e_code++ = ' '; 702 *e_code++ = ':'; 703 ps.want_blank = true; 704 break; 705 } 706 if (ps.in_or_st) { 707 *e_code++ = ':'; 708 ps.want_blank = false; 709 break; 710 } 711 ps.in_stmt = false; /* seeing a label does not imply we are in a 712 * stmt */ 713 /* 714 * turn everything so far into a label 715 */ 716 { 717 int len = e_code - s_code; 718 719 CHECK_SIZE_LAB(len + 3); 720 memcpy(e_lab, s_code, len); 721 e_lab += len; 722 *e_lab++ = ':'; 723 *e_lab = '\0'; 724 e_code = s_code; 725 } 726 force_nl = ps.pcase = scase; /* ps.pcase will be used by 727 * dump_line to decide how to 728 * indent the label. force_nl 729 * will force a case n: to be 730 * on a line by itself */ 731 scase = false; 732 ps.want_blank = false; 733 break; 734 735 case semicolon: /* got a ';' */ 736 if (ps.dec_nest == 0) 737 ps.in_or_st = false;/* we are not in an initialization or 738 * structure declaration */ 739 scase = false; /* these will only need resetting in an error */ 740 squest = 0; 741 if (ps.last_token == rparen) 742 ps.in_parameter_declaration = 0; 743 ps.cast_mask = 0; 744 ps.not_cast_mask = 0; 745 ps.block_init = 0; 746 ps.block_init_level = 0; 747 ps.just_saw_decl--; 748 749 if (ps.in_decl && s_code == e_code && !ps.block_init && 750 !ps.dumped_decl_indent && ps.paren_level == 0) { 751 /* indent stray semicolons in declarations */ 752 indent_declaration(dec_ind - 1, tabs_to_var); 753 ps.dumped_decl_indent = true; 754 } 755 756 ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level 757 * structure declaration, we 758 * arent any more */ 759 760 if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) { 761 762 /* 763 * This should be true iff there were unbalanced parens in the 764 * stmt. It is a bit complicated, because the semicolon might 765 * be in a for stmt 766 */ 767 diag2(1, "Unbalanced parens"); 768 ps.p_l_follow = 0; 769 if (sp_sw) { /* this is a check for an if, while, etc. with 770 * unbalanced parens */ 771 sp_sw = false; 772 parse(hd_type); /* dont lose the if, or whatever */ 773 } 774 } 775 *e_code++ = ';'; 776 ps.want_blank = true; 777 ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the 778 * middle of a stmt */ 779 780 if (!sp_sw) { /* if not if for (;;) */ 781 parse(semicolon); /* let parser know about end of stmt */ 782 force_nl = true;/* force newline after an end of stmt */ 783 } 784 break; 785 786 case lbrace: /* got a '{' */ 787 ps.in_stmt = false; /* dont indent the {} */ 788 if (!ps.block_init) 789 force_nl = true;/* force other stuff on same line as '{' onto 790 * new line */ 791 else if (ps.block_init_level <= 0) 792 ps.block_init_level = 1; 793 else 794 ps.block_init_level++; 795 796 if (s_code != e_code && !ps.block_init) { 797 if (!opt.btype_2) { 798 dump_line(); 799 ps.want_blank = false; 800 } 801 else if (ps.in_parameter_declaration && !ps.in_or_st) { 802 ps.i_l_follow = 0; 803 if (opt.function_brace_split) { /* dump the line prior 804 * to the brace ... */ 805 dump_line(); 806 ps.want_blank = false; 807 } else /* add a space between the decl and brace */ 808 ps.want_blank = true; 809 } 810 } 811 if (ps.in_parameter_declaration) 812 prefix_blankline_requested = 0; 813 814 if (ps.p_l_follow > 0) { /* check for preceding unbalanced 815 * parens */ 816 diag2(1, "Unbalanced parens"); 817 ps.p_l_follow = 0; 818 if (sp_sw) { /* check for unclosed if, for, etc. */ 819 sp_sw = false; 820 parse(hd_type); 821 ps.ind_level = ps.i_l_follow; 822 } 823 } 824 if (s_code == e_code) 825 ps.ind_stmt = false; /* dont put extra indentation on line 826 * with '{' */ 827 if (ps.in_decl && ps.in_or_st) { /* this is either a structure 828 * declaration or an init */ 829 di_stack[ps.dec_nest] = dec_ind; 830 if (++ps.dec_nest == nitems(di_stack)) { 831 diag3(0, "Reached internal limit of %d struct levels", 832 nitems(di_stack)); 833 ps.dec_nest--; 834 } 835 /* ? dec_ind = 0; */ 836 } 837 else { 838 ps.decl_on_line = false; /* we can't be in the middle of 839 * a declaration, so don't do 840 * special indentation of 841 * comments */ 842 if (opt.blanklines_after_declarations_at_proctop 843 && ps.in_parameter_declaration) 844 postfix_blankline_requested = 1; 845 ps.in_parameter_declaration = 0; 846 ps.in_decl = false; 847 } 848 dec_ind = 0; 849 parse(lbrace); /* let parser know about this */ 850 if (ps.want_blank) /* put a blank before '{' if '{' is not at 851 * start of line */ 852 *e_code++ = ' '; 853 ps.want_blank = false; 854 *e_code++ = '{'; 855 ps.just_saw_decl = 0; 856 break; 857 858 case rbrace: /* got a '}' */ 859 if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be 860 * omitted in 861 * declarations */ 862 parse(semicolon); 863 if (ps.p_l_follow) {/* check for unclosed if, for, else. */ 864 diag2(1, "Unbalanced parens"); 865 ps.p_l_follow = 0; 866 sp_sw = false; 867 } 868 ps.just_saw_decl = 0; 869 ps.block_init_level--; 870 if (s_code != e_code && !ps.block_init) { /* '}' must be first on 871 * line */ 872 if (opt.verbose) 873 diag2(0, "Line broken"); 874 dump_line(); 875 } 876 *e_code++ = '}'; 877 ps.want_blank = true; 878 ps.in_stmt = ps.ind_stmt = false; 879 if (ps.dec_nest > 0) { /* we are in multi-level structure 880 * declaration */ 881 dec_ind = di_stack[--ps.dec_nest]; 882 if (ps.dec_nest == 0 && !ps.in_parameter_declaration) 883 ps.just_saw_decl = 2; 884 ps.in_decl = true; 885 } 886 prefix_blankline_requested = 0; 887 parse(rbrace); /* let parser know about this */ 888 ps.search_brace = opt.cuddle_else && ps.p_stack[ps.tos] == ifhead 889 && ps.il[ps.tos] >= ps.ind_level; 890 if (ps.tos <= 1 && opt.blanklines_after_procs && ps.dec_nest <= 0) 891 postfix_blankline_requested = 1; 892 break; 893 894 case swstmt: /* got keyword "switch" */ 895 sp_sw = true; 896 hd_type = swstmt; /* keep this for when we have seen the 897 * expression */ 898 goto copy_id; /* go move the token into buffer */ 899 900 case sp_paren: /* token is if, while, for */ 901 sp_sw = true; /* the interesting stuff is done after the 902 * expression is scanned */ 903 hd_type = (*token == 'i' ? ifstmt : 904 (*token == 'w' ? whilestmt : forstmt)); 905 906 /* 907 * remember the type of header for later use by parser 908 */ 909 goto copy_id; /* copy the token into line */ 910 911 case sp_nparen: /* got else, do */ 912 ps.in_stmt = false; 913 if (*token == 'e') { 914 if (e_code != s_code && (!opt.cuddle_else || e_code[-1] != '}')) { 915 if (opt.verbose) 916 diag2(0, "Line broken"); 917 dump_line();/* make sure this starts a line */ 918 ps.want_blank = false; 919 } 920 force_nl = true;/* also, following stuff must go onto new line */ 921 last_else = 1; 922 parse(elselit); 923 } 924 else { 925 if (e_code != s_code) { /* make sure this starts a line */ 926 if (opt.verbose) 927 diag2(0, "Line broken"); 928 dump_line(); 929 ps.want_blank = false; 930 } 931 force_nl = true;/* also, following stuff must go onto new line */ 932 last_else = 0; 933 parse(dolit); 934 } 935 goto copy_id; /* move the token into line */ 936 937 case type_def: 938 case storage: 939 prefix_blankline_requested = 0; 940 goto copy_id; 941 942 case structure: 943 if (ps.p_l_follow > 0) 944 goto copy_id; 945 case decl: /* we have a declaration type (int, etc.) */ 946 parse(decl); /* let parser worry about indentation */ 947 if (ps.last_token == rparen && ps.tos <= 1) { 948 if (s_code != e_code) { 949 dump_line(); 950 ps.want_blank = 0; 951 } 952 } 953 if (ps.in_parameter_declaration && opt.indent_parameters && ps.dec_nest == 0) { 954 ps.ind_level = ps.i_l_follow = 1; 955 ps.ind_stmt = 0; 956 } 957 ps.in_or_st = true; /* this might be a structure or initialization 958 * declaration */ 959 ps.in_decl = ps.decl_on_line = ps.last_token != type_def; 960 if ( /* !ps.in_or_st && */ ps.dec_nest <= 0) 961 ps.just_saw_decl = 2; 962 prefix_blankline_requested = 0; 963 for (i = 0; token[i++];); /* get length of token */ 964 965 if (ps.ind_level == 0 || ps.dec_nest > 0) { 966 /* global variable or struct member in local variable */ 967 dec_ind = opt.decl_indent > 0 ? opt.decl_indent : i; 968 tabs_to_var = (opt.use_tabs ? opt.decl_indent > 0 : 0); 969 } else { 970 /* local variable */ 971 dec_ind = opt.local_decl_indent > 0 ? opt.local_decl_indent : i; 972 tabs_to_var = (opt.use_tabs ? opt.local_decl_indent > 0 : 0); 973 } 974 goto copy_id; 975 976 case funcname: 977 case ident: /* got an identifier or constant */ 978 if (ps.in_decl) { 979 if (type_code == funcname) { 980 ps.in_decl = false; 981 if (opt.procnames_start_line && s_code != e_code) { 982 *e_code = '\0'; 983 dump_line(); 984 } 985 else if (ps.want_blank) { 986 *e_code++ = ' '; 987 } 988 ps.want_blank = false; 989 } 990 else if (!ps.block_init && !ps.dumped_decl_indent && 991 ps.paren_level == 0) { /* if we are in a declaration, we 992 * must indent identifier */ 993 indent_declaration(dec_ind, tabs_to_var); 994 ps.dumped_decl_indent = true; 995 ps.want_blank = false; 996 } 997 } 998 else if (sp_sw && ps.p_l_follow == 0) { 999 sp_sw = false; 1000 force_nl = true; 1001 ps.last_u_d = true; 1002 ps.in_stmt = false; 1003 parse(hd_type); 1004 } 1005 copy_id: 1006 { 1007 int len = e_token - s_token; 1008 1009 CHECK_SIZE_CODE(len + 1); 1010 if (ps.want_blank) 1011 *e_code++ = ' '; 1012 memcpy(e_code, s_token, len); 1013 e_code += len; 1014 } 1015 if (type_code != funcname) 1016 ps.want_blank = true; 1017 break; 1018 1019 case strpfx: 1020 { 1021 int len = e_token - s_token; 1022 1023 CHECK_SIZE_CODE(len + 1); 1024 if (ps.want_blank) 1025 *e_code++ = ' '; 1026 memcpy(e_code, token, len); 1027 e_code += len; 1028 } 1029 ps.want_blank = false; 1030 break; 1031 1032 case period: /* treat a period kind of like a binary 1033 * operation */ 1034 *e_code++ = '.'; /* move the period into line */ 1035 ps.want_blank = false; /* dont put a blank after a period */ 1036 break; 1037 1038 case comma: 1039 ps.want_blank = (s_code != e_code); /* only put blank after comma 1040 * if comma does not start the 1041 * line */ 1042 if (ps.in_decl && ps.procname[0] == '\0' && !ps.block_init && 1043 !ps.dumped_decl_indent && ps.paren_level == 0) { 1044 /* indent leading commas and not the actual identifiers */ 1045 indent_declaration(dec_ind - 1, tabs_to_var); 1046 ps.dumped_decl_indent = true; 1047 } 1048 *e_code++ = ','; 1049 if (ps.p_l_follow == 0) { 1050 if (ps.block_init_level <= 0) 1051 ps.block_init = 0; 1052 if (break_comma && (!opt.leave_comma || 1053 count_spaces_until(compute_code_target(), s_code, e_code) > 1054 opt.max_col - opt.tabsize)) 1055 force_nl = true; 1056 } 1057 break; 1058 1059 case preesc: /* got the character '#' */ 1060 if ((s_com != e_com) || 1061 (s_lab != e_lab) || 1062 (s_code != e_code)) 1063 dump_line(); 1064 CHECK_SIZE_LAB(1); 1065 *e_lab++ = '#'; /* move whole line to 'label' buffer */ 1066 { 1067 int in_comment = 0; 1068 int com_start = 0; 1069 char quote = 0; 1070 int com_end = 0; 1071 1072 while (*buf_ptr == ' ' || *buf_ptr == '\t') { 1073 buf_ptr++; 1074 if (buf_ptr >= buf_end) 1075 fill_buffer(); 1076 } 1077 while (*buf_ptr != '\n' || (in_comment && !had_eof)) { 1078 CHECK_SIZE_LAB(2); 1079 *e_lab = *buf_ptr++; 1080 if (buf_ptr >= buf_end) 1081 fill_buffer(); 1082 switch (*e_lab++) { 1083 case BACKSLASH: 1084 if (!in_comment) { 1085 *e_lab++ = *buf_ptr++; 1086 if (buf_ptr >= buf_end) 1087 fill_buffer(); 1088 } 1089 break; 1090 case '/': 1091 if (*buf_ptr == '*' && !in_comment && !quote) { 1092 in_comment = 1; 1093 *e_lab++ = *buf_ptr++; 1094 com_start = e_lab - s_lab - 2; 1095 } 1096 break; 1097 case '"': 1098 if (quote == '"') 1099 quote = 0; 1100 break; 1101 case '\'': 1102 if (quote == '\'') 1103 quote = 0; 1104 break; 1105 case '*': 1106 if (*buf_ptr == '/' && in_comment) { 1107 in_comment = 0; 1108 *e_lab++ = *buf_ptr++; 1109 com_end = e_lab - s_lab; 1110 } 1111 break; 1112 } 1113 } 1114 1115 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1116 e_lab--; 1117 if (e_lab - s_lab == com_end && bp_save == NULL) { 1118 /* comment on preprocessor line */ 1119 if (sc_end == NULL) { /* if this is the first comment, 1120 * we must set up the buffer */ 1121 save_com = sc_buf; 1122 sc_end = &save_com[0]; 1123 } 1124 else { 1125 *sc_end++ = '\n'; /* add newline between 1126 * comments */ 1127 *sc_end++ = ' '; 1128 --line_no; 1129 } 1130 if (sc_end - save_com + com_end - com_start > sc_size) 1131 errx(1, "input too long"); 1132 memmove(sc_end, s_lab + com_start, com_end - com_start); 1133 sc_end += com_end - com_start; 1134 e_lab = s_lab + com_start; 1135 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1136 e_lab--; 1137 bp_save = buf_ptr; /* save current input buffer */ 1138 be_save = buf_end; 1139 buf_ptr = save_com; /* fix so that subsequent calls to 1140 * lexi will take tokens out of 1141 * save_com */ 1142 *sc_end++ = ' '; /* add trailing blank, just in case */ 1143 buf_end = sc_end; 1144 sc_end = NULL; 1145 } 1146 CHECK_SIZE_LAB(1); 1147 *e_lab = '\0'; /* null terminate line */ 1148 ps.pcase = false; 1149 } 1150 1151 if (strncmp(s_lab, "#if", 3) == 0) { /* also ifdef, ifndef */ 1152 if ((size_t)ifdef_level < nitems(state_stack)) { 1153 match_state[ifdef_level].tos = -1; 1154 state_stack[ifdef_level++] = ps; 1155 } 1156 else 1157 diag2(1, "#if stack overflow"); 1158 } 1159 else if (strncmp(s_lab, "#el", 3) == 0) { /* else, elif */ 1160 if (ifdef_level <= 0) 1161 diag2(1, s_lab[3] == 'i' ? "Unmatched #elif" : "Unmatched #else"); 1162 else { 1163 match_state[ifdef_level - 1] = ps; 1164 ps = state_stack[ifdef_level - 1]; 1165 } 1166 } 1167 else if (strncmp(s_lab, "#endif", 6) == 0) { 1168 if (ifdef_level <= 0) 1169 diag2(1, "Unmatched #endif"); 1170 else 1171 ifdef_level--; 1172 } else { 1173 struct directives { 1174 int size; 1175 const char *string; 1176 } 1177 recognized[] = { 1178 {7, "include"}, 1179 {6, "define"}, 1180 {5, "undef"}, 1181 {4, "line"}, 1182 {5, "error"}, 1183 {6, "pragma"} 1184 }; 1185 int d = nitems(recognized); 1186 while (--d >= 0) 1187 if (strncmp(s_lab + 1, recognized[d].string, recognized[d].size) == 0) 1188 break; 1189 if (d < 0) { 1190 diag2(1, "Unrecognized cpp directive"); 1191 break; 1192 } 1193 } 1194 if (opt.blanklines_around_conditional_compilation) { 1195 postfix_blankline_requested++; 1196 n_real_blanklines = 0; 1197 } 1198 else { 1199 postfix_blankline_requested = 0; 1200 prefix_blankline_requested = 0; 1201 } 1202 break; /* subsequent processing of the newline 1203 * character will cause the line to be printed */ 1204 1205 case comment: /* we have gotten a / followed by * this is a biggie */ 1206 pr_comment(); 1207 break; 1208 } /* end of big switch stmt */ 1209 1210 *e_code = '\0'; /* make sure code section is null terminated */ 1211 if (type_code != comment && type_code != newline && type_code != preesc) 1212 ps.last_token = type_code; 1213 } /* end of main while (1) loop */ 1214 } 1215 1216 /* 1217 * copy input file to backup file if in_name is /blah/blah/blah/file, then 1218 * backup file will be ".Bfile" then make the backup file the input and 1219 * original input file the output 1220 */ 1221 static void 1222 bakcopy(void) 1223 { 1224 int n, 1225 bakchn; 1226 char buff[8 * 1024]; 1227 const char *p; 1228 1229 /* construct file name .Bfile */ 1230 for (p = in_name; *p; p++); /* skip to end of string */ 1231 while (p > in_name && *p != '/') /* find last '/' */ 1232 p--; 1233 if (*p == '/') 1234 p++; 1235 sprintf(bakfile, "%s%s", p, simple_backup_suffix); 1236 1237 /* copy in_name to backup file */ 1238 bakchn = creat(bakfile, 0600); 1239 if (bakchn < 0) 1240 err(1, "%s", bakfile); 1241 while ((n = read(fileno(input), buff, sizeof(buff))) > 0) 1242 if (write(bakchn, buff, n) != n) 1243 err(1, "%s", bakfile); 1244 if (n < 0) 1245 err(1, "%s", in_name); 1246 close(bakchn); 1247 fclose(input); 1248 1249 /* re-open backup file as the input file */ 1250 input = fopen(bakfile, "r"); 1251 if (input == NULL) 1252 err(1, "%s", bakfile); 1253 /* now the original input file will be the output */ 1254 output = fopen(in_name, "w"); 1255 if (output == NULL) { 1256 unlink(bakfile); 1257 err(1, "%s", in_name); 1258 } 1259 } 1260 1261 static void 1262 indent_declaration(int cur_dec_ind, int tabs_to_var) 1263 { 1264 int pos = e_code - s_code; 1265 char *startpos = e_code; 1266 1267 /* 1268 * get the tab math right for indentations that are not multiples of tabsize 1269 */ 1270 if ((ps.ind_level * opt.ind_size) % opt.tabsize != 0) { 1271 pos += (ps.ind_level * opt.ind_size) % opt.tabsize; 1272 cur_dec_ind += (ps.ind_level * opt.ind_size) % opt.tabsize; 1273 } 1274 if (tabs_to_var) { 1275 int tpos; 1276 1277 CHECK_SIZE_CODE(cur_dec_ind / opt.tabsize); 1278 while ((tpos = opt.tabsize * (1 + pos / opt.tabsize)) <= cur_dec_ind) { 1279 *e_code++ = '\t'; 1280 pos = tpos; 1281 } 1282 } 1283 CHECK_SIZE_CODE(cur_dec_ind - pos + 1); 1284 while (pos < cur_dec_ind) { 1285 *e_code++ = ' '; 1286 pos++; 1287 } 1288 if (e_code == startpos && ps.want_blank) { 1289 *e_code++ = ' '; 1290 ps.want_blank = false; 1291 } 1292 } 1293