1/*- 2 * Copyright (c) 1985 Sun Microsystems, Inc. 3 * Copyright (c) 1976 Board of Trustees of the University of Illinois. 4 * Copyright (c) 1980, 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. All advertising materials mentioning features or use of this software 16 * must display the following acknowledgement: 17 * This product includes software developed by the University of 18 * California, Berkeley and its contributors. 19 * 4. Neither the name of the University nor the names of its contributors 20 * may be used to endorse or promote products derived from this software 21 * without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 */ 35 36#ifndef lint 37static const char copyright[] = 38"@(#) Copyright (c) 1985 Sun Microsystems, Inc.\n\ 39@(#) Copyright (c) 1976 Board of Trustees of the University of Illinois.\n\ 40@(#) Copyright (c) 1980, 1993\n\ 41 The Regents of the University of California. All rights reserved.\n"; 42#endif /* not lint */ 43 44#if 0 45#ifndef lint 46static char sccsid[] = "@(#)indent.c 5.17 (Berkeley) 6/7/93"; 47#endif /* not lint */ 48#endif 49 50#include <sys/cdefs.h> 51__FBSDID("$FreeBSD: stable/11/usr.bin/indent/indent.c 331722 2018-03-29 02:50:57Z eadler $"); 52 53#include <sys/capsicum.h> 54#include <sys/param.h> 55#include <err.h> 56#include <errno.h> 57#include <fcntl.h> 58#include <unistd.h> 59#include <stdio.h> 60#include <stdlib.h> 61#include <string.h> 62#include <ctype.h> 63#include "indent_globs.h" 64#include "indent_codes.h" 65#include "indent.h" 66 67static void bakcopy(void); 68static void indent_declaration(int, int); 69 70const char *in_name = "Standard Input"; /* will always point to name of input 71 * file */ 72const char *out_name = "Standard Output"; /* will always point to name 73 * of output file */ 74const char *simple_backup_suffix = ".BAK"; /* Suffix to use for backup 75 * files */ 76char bakfile[MAXPATHLEN] = ""; 77 78int 79main(int argc, char **argv) 80{ 81 cap_rights_t rights; 82 83 int dec_ind; /* current indentation for declarations */ 84 int di_stack[20]; /* a stack of structure indentation levels */ 85 int flushed_nl; /* used when buffering up comments to remember 86 * that a newline was passed over */ 87 int force_nl; /* when true, code must be broken */ 88 int hd_type = 0; /* used to store type of stmt for if (...), 89 * for (...), etc */ 90 int i; /* local loop counter */ 91 int scase; /* set to true when we see a case, so we will 92 * know what to do with the following colon */ 93 int sp_sw; /* when true, we are in the expression of 94 * if(...), while(...), etc. */ 95 int squest; /* when this is positive, we have seen a ? 96 * without the matching : in a <c>?<s>:<s> 97 * construct */ 98 const char *t_ptr; /* used for copying tokens */ 99 int tabs_to_var; /* true if using tabs to indent to var name */ 100 int type_code; /* the type of token, returned by lexi */ 101 102 int last_else = 0; /* true iff last keyword was an else */ 103 const char *envval = NULL; 104 105 /*-----------------------------------------------*\ 106 | INITIALIZATION | 107 \*-----------------------------------------------*/ 108 109 found_err = 0; 110 111 ps.p_stack[0] = stmt; /* this is the parser's stack */ 112 ps.last_nl = true; /* this is true if the last thing scanned was 113 * a newline */ 114 ps.last_token = semicolon; 115 combuf = (char *) malloc(bufsize); 116 if (combuf == NULL) 117 err(1, NULL); 118 labbuf = (char *) malloc(bufsize); 119 if (labbuf == NULL) 120 err(1, NULL); 121 codebuf = (char *) malloc(bufsize); 122 if (codebuf == NULL) 123 err(1, NULL); 124 tokenbuf = (char *) malloc(bufsize); 125 if (tokenbuf == NULL) 126 err(1, NULL); 127 l_com = combuf + bufsize - 5; 128 l_lab = labbuf + bufsize - 5; 129 l_code = codebuf + bufsize - 5; 130 l_token = tokenbuf + bufsize - 5; 131 combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and 132 * comment buffers */ 133 combuf[1] = codebuf[1] = labbuf[1] = '\0'; 134 ps.else_if = 1; /* Default else-if special processing to on */ 135 s_lab = e_lab = labbuf + 1; 136 s_code = e_code = codebuf + 1; 137 s_com = e_com = combuf + 1; 138 s_token = e_token = tokenbuf + 1; 139 140 in_buffer = (char *) malloc(10); 141 if (in_buffer == NULL) 142 err(1, NULL); 143 in_buffer_limit = in_buffer + 8; 144 buf_ptr = buf_end = in_buffer; 145 line_no = 1; 146 had_eof = ps.in_decl = ps.decl_on_line = break_comma = false; 147 sp_sw = force_nl = false; 148 ps.in_or_st = false; 149 ps.bl_line = true; 150 dec_ind = 0; 151 di_stack[ps.dec_nest = 0] = 0; 152 ps.want_blank = ps.in_stmt = ps.ind_stmt = false; 153 154 scase = ps.pcase = false; 155 squest = 0; 156 sc_end = NULL; 157 bp_save = NULL; 158 be_save = NULL; 159 160 output = NULL; 161 tabs_to_var = 0; 162 163 envval = getenv("SIMPLE_BACKUP_SUFFIX"); 164 if (envval) 165 simple_backup_suffix = envval; 166 167 /*--------------------------------------------------*\ 168 | COMMAND LINE SCAN | 169 \*--------------------------------------------------*/ 170 171#ifdef undef 172 max_col = 78; /* -l78 */ 173 lineup_to_parens = 1; /* -lp */ 174 ps.ljust_decl = 0; /* -ndj */ 175 ps.com_ind = 33; /* -c33 */ 176 star_comment_cont = 1; /* -sc */ 177 ps.ind_size = 8; /* -i8 */ 178 verbose = 0; 179 ps.decl_indent = 16; /* -di16 */ 180 ps.local_decl_indent = -1; /* if this is not set to some nonnegative value 181 * by an arg, we will set this equal to 182 * ps.decl_ind */ 183 ps.indent_parameters = 1; /* -ip */ 184 ps.decl_com_ind = 0; /* if this is not set to some positive value 185 * by an arg, we will set this equal to 186 * ps.com_ind */ 187 btype_2 = 1; /* -br */ 188 cuddle_else = 1; /* -ce */ 189 ps.unindent_displace = 0; /* -d0 */ 190 ps.case_indent = 0; /* -cli0 */ 191 format_block_comments = 1; /* -fcb */ 192 format_col1_comments = 1; /* -fc1 */ 193 procnames_start_line = 1; /* -psl */ 194 proc_calls_space = 0; /* -npcs */ 195 comment_delimiter_on_blankline = 1; /* -cdb */ 196 ps.leave_comma = 1; /* -nbc */ 197#endif 198 199 for (i = 1; i < argc; ++i) 200 if (strcmp(argv[i], "-npro") == 0) 201 break; 202 set_defaults(); 203 if (i >= argc) 204 set_profile(); 205 206 for (i = 1; i < argc; ++i) { 207 208 /* 209 * look thru args (if any) for changes to defaults 210 */ 211 if (argv[i][0] != '-') {/* no flag on parameter */ 212 if (input == NULL) { /* we must have the input file */ 213 in_name = argv[i]; /* remember name of input file */ 214 input = fopen(in_name, "r"); 215 if (input == NULL) /* check for open error */ 216 err(1, "%s", in_name); 217 continue; 218 } 219 else if (output == NULL) { /* we have the output file */ 220 out_name = argv[i]; /* remember name of output file */ 221 if (strcmp(in_name, out_name) == 0) { /* attempt to overwrite 222 * the file */ 223 errx(1, "input and output files must be different"); 224 } 225 output = fopen(out_name, "w"); 226 if (output == NULL) /* check for create error */ 227 err(1, "%s", out_name); 228 continue; 229 } 230 errx(1, "unknown parameter: %s", argv[i]); 231 } 232 else 233 set_option(argv[i]); 234 } /* end of for */ 235 if (input == NULL) 236 input = stdin; 237 if (output == NULL) { 238 if (troff || input == stdin) 239 output = stdout; 240 else { 241 out_name = in_name; 242 bakcopy(); 243 } 244 } 245 246 /* Restrict input/output descriptors and enter Capsicum sandbox. */ 247 cap_rights_init(&rights, CAP_FSTAT, CAP_WRITE); 248 if (cap_rights_limit(fileno(output), &rights) < 0 && errno != ENOSYS) 249 err(EXIT_FAILURE, "unable to limit rights for %s", out_name); 250 cap_rights_init(&rights, CAP_FSTAT, CAP_READ); 251 if (cap_rights_limit(fileno(input), &rights) < 0 && errno != ENOSYS) 252 err(EXIT_FAILURE, "unable to limit rights for %s", in_name); 253 if (cap_enter() < 0 && errno != ENOSYS) 254 err(EXIT_FAILURE, "unable to enter capability mode"); 255 256 if (ps.com_ind <= 1) 257 ps.com_ind = 2; /* dont put normal comments before column 2 */ 258 if (troff) { 259 if (bodyf.font[0] == 0) 260 parsefont(&bodyf, "R"); 261 if (scomf.font[0] == 0) 262 parsefont(&scomf, "I"); 263 if (blkcomf.font[0] == 0) 264 blkcomf = scomf, blkcomf.size += 2; 265 if (boxcomf.font[0] == 0) 266 boxcomf = blkcomf; 267 if (stringf.font[0] == 0) 268 parsefont(&stringf, "L"); 269 if (keywordf.font[0] == 0) 270 parsefont(&keywordf, "B"); 271 writefdef(&bodyf, 'B'); 272 writefdef(&scomf, 'C'); 273 writefdef(&blkcomf, 'L'); 274 writefdef(&boxcomf, 'X'); 275 writefdef(&stringf, 'S'); 276 writefdef(&keywordf, 'K'); 277 } 278 if (block_comment_max_col <= 0) 279 block_comment_max_col = max_col; 280 if (ps.local_decl_indent < 0) /* if not specified by user, set this */ 281 ps.local_decl_indent = ps.decl_indent; 282 if (ps.decl_com_ind <= 0) /* if not specified by user, set this */ 283 ps.decl_com_ind = ps.ljust_decl ? (ps.com_ind <= 10 ? 2 : ps.com_ind - 8) : ps.com_ind; 284 if (continuation_indent == 0) 285 continuation_indent = ps.ind_size; 286 fill_buffer(); /* get first batch of stuff into input buffer */ 287 288 parse(semicolon); 289 { 290 char *p = buf_ptr; 291 int col = 1; 292 293 while (1) { 294 if (*p == ' ') 295 col++; 296 else if (*p == '\t') 297 col = ((col - 1) & ~7) + 9; 298 else 299 break; 300 p++; 301 } 302 if (col > ps.ind_size) 303 ps.ind_level = ps.i_l_follow = col / ps.ind_size; 304 } 305 if (troff) { 306 const char *p = in_name, 307 *beg = in_name; 308 309 while (*p) 310 if (*p++ == '/') 311 beg = p; 312 fprintf(output, ".Fn \"%s\"\n", beg); 313 } 314 /* 315 * START OF MAIN LOOP 316 */ 317 318 while (1) { /* this is the main loop. it will go until we 319 * reach eof */ 320 int is_procname; 321 322 type_code = lexi(); /* lexi reads one token. The actual 323 * characters read are stored in "token". lexi 324 * returns a code indicating the type of token */ 325 is_procname = ps.procname[0]; 326 327 /* 328 * The following code moves everything following an if (), while (), 329 * else, etc. up to the start of the following stmt to a buffer. This 330 * allows proper handling of both kinds of brace placement. 331 */ 332 333 flushed_nl = false; 334 while (ps.search_brace) { /* if we scanned an if(), while(), 335 * etc., we might need to copy stuff 336 * into a buffer we must loop, copying 337 * stuff into save_com, until we find 338 * the start of the stmt which follows 339 * the if, or whatever */ 340 switch (type_code) { 341 case newline: 342 ++line_no; 343 if (sc_end != NULL) { /* dump comment, if any */ 344 *sc_end++ = '\n'; /* newlines are needed in this case */ 345 goto sw_buffer; 346 } 347 flushed_nl = true; 348 case form_feed: 349 break; /* form feeds and newlines found here will be 350 * ignored */ 351 352 case lbrace: /* this is a brace that starts the compound 353 * stmt */ 354 if (sc_end == NULL) { /* ignore buffering if a comment wasn't 355 * stored up */ 356 ps.search_brace = false; 357 goto check_type; 358 } 359 if (btype_2) { 360 save_com[0] = '{'; /* we either want to put the brace 361 * right after the if */ 362 goto sw_buffer; /* go to common code to get out of 363 * this loop */ 364 } 365 case comment: /* we have a comment, so we must copy it into 366 * the buffer */ 367 if (!flushed_nl || sc_end != NULL) { 368 if (sc_end == NULL) { /* if this is the first comment, we 369 * must set up the buffer */ 370 save_com[0] = save_com[1] = ' '; 371 sc_end = &(save_com[2]); 372 } 373 else { 374 *sc_end++ = '\n'; /* add newline between 375 * comments */ 376 *sc_end++ = ' '; 377 --line_no; 378 } 379 *sc_end++ = '/'; /* copy in start of comment */ 380 *sc_end++ = '*'; 381 382 for (;;) { /* loop until we get to the end of the comment */ 383 *sc_end = *buf_ptr++; 384 if (buf_ptr >= buf_end) 385 fill_buffer(); 386 387 if (*sc_end++ == '*' && *buf_ptr == '/') 388 break; /* we are at end of comment */ 389 390 if (sc_end >= &(save_com[sc_size])) { /* check for temp buffer 391 * overflow */ 392 diag2(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever"); 393 fflush(output); 394 exit(1); 395 } 396 } 397 *sc_end++ = '/'; /* add ending slash */ 398 if (++buf_ptr >= buf_end) /* get past / in buffer */ 399 fill_buffer(); 400 break; 401 } 402 default: /* it is the start of a normal statement */ 403 if (flushed_nl) /* if we flushed a newline, make sure it is 404 * put back */ 405 force_nl = true; 406 if ((type_code == sp_paren && *token == 'i' 407 && last_else && ps.else_if) 408 || (type_code == sp_nparen && *token == 'e' 409 && e_code != s_code && e_code[-1] == '}')) 410 force_nl = false; 411 412 if (sc_end == NULL) { /* ignore buffering if comment wasn't 413 * saved up */ 414 ps.search_brace = false; 415 goto check_type; 416 } 417 if (force_nl) { /* if we should insert a nl here, put it into 418 * the buffer */ 419 force_nl = false; 420 --line_no; /* this will be re-increased when the nl is 421 * read from the buffer */ 422 *sc_end++ = '\n'; 423 *sc_end++ = ' '; 424 if (verbose && !flushed_nl) /* print error msg if the line 425 * was not already broken */ 426 diag2(0, "Line broken"); 427 flushed_nl = false; 428 } 429 for (t_ptr = token; *t_ptr; ++t_ptr) 430 *sc_end++ = *t_ptr; /* copy token into temp buffer */ 431 ps.procname[0] = 0; 432 433 sw_buffer: 434 ps.search_brace = false; /* stop looking for start of 435 * stmt */ 436 bp_save = buf_ptr; /* save current input buffer */ 437 be_save = buf_end; 438 buf_ptr = save_com; /* fix so that subsequent calls to 439 * lexi will take tokens out of 440 * save_com */ 441 *sc_end++ = ' ';/* add trailing blank, just in case */ 442 buf_end = sc_end; 443 sc_end = NULL; 444 break; 445 } /* end of switch */ 446 if (type_code != 0) /* we must make this check, just in case there 447 * was an unexpected EOF */ 448 type_code = lexi(); /* read another token */ 449 /* if (ps.search_brace) ps.procname[0] = 0; */ 450 if ((is_procname = ps.procname[0]) && flushed_nl 451 && !procnames_start_line && ps.in_decl 452 && type_code == ident) 453 flushed_nl = 0; 454 } /* end of while (search_brace) */ 455 last_else = 0; 456check_type: 457 if (type_code == 0) { /* we got eof */ 458 if (s_lab != e_lab || s_code != e_code 459 || s_com != e_com) /* must dump end of line */ 460 dump_line(); 461 if (ps.tos > 1) /* check for balanced braces */ 462 diag2(1, "Stuff missing from end of file"); 463 464 if (verbose) { 465 printf("There were %d output lines and %d comments\n", 466 ps.out_lines, ps.out_coms); 467 printf("(Lines with comments)/(Lines with code): %6.3f\n", 468 (1.0 * ps.com_lines) / code_lines); 469 } 470 fflush(output); 471 exit(found_err); 472 } 473 if ( 474 (type_code != comment) && 475 (type_code != newline) && 476 (type_code != preesc) && 477 (type_code != form_feed)) { 478 if (force_nl && 479 (type_code != semicolon) && 480 (type_code != lbrace || !btype_2)) { 481 /* we should force a broken line here */ 482 if (verbose && !flushed_nl) 483 diag2(0, "Line broken"); 484 flushed_nl = false; 485 dump_line(); 486 ps.want_blank = false; /* dont insert blank at line start */ 487 force_nl = false; 488 } 489 ps.in_stmt = true; /* turn on flag which causes an extra level of 490 * indentation. this is turned off by a ; or 491 * '}' */ 492 if (s_com != e_com) { /* the turkey has embedded a comment 493 * in a line. fix it */ 494 *e_code++ = ' '; 495 for (t_ptr = s_com; *t_ptr; ++t_ptr) { 496 CHECK_SIZE_CODE; 497 *e_code++ = *t_ptr; 498 } 499 *e_code++ = ' '; 500 *e_code = '\0'; /* null terminate code sect */ 501 ps.want_blank = false; 502 e_com = s_com; 503 } 504 } 505 else if (type_code != comment) /* preserve force_nl thru a comment */ 506 force_nl = false; /* cancel forced newline after newline, form 507 * feed, etc */ 508 509 510 511 /*-----------------------------------------------------*\ 512 | do switch on type of token scanned | 513 \*-----------------------------------------------------*/ 514 CHECK_SIZE_CODE; 515 switch (type_code) { /* now, decide what to do with the token */ 516 517 case form_feed: /* found a form feed in line */ 518 ps.use_ff = true; /* a form feed is treated much like a newline */ 519 dump_line(); 520 ps.want_blank = false; 521 break; 522 523 case newline: 524 if (ps.last_token != comma || ps.p_l_follow > 0 525 || !ps.leave_comma || ps.block_init || !break_comma || s_com != e_com) { 526 dump_line(); 527 ps.want_blank = false; 528 } 529 ++line_no; /* keep track of input line number */ 530 break; 531 532 case lparen: /* got a '(' or '[' */ 533 /* count parens to make Healy happy */ 534 if (++ps.p_l_follow == nitems(ps.paren_indents)) { 535 diag3(0, "Reached internal limit of %d unclosed parens", 536 nitems(ps.paren_indents)); 537 ps.p_l_follow--; 538 } 539 if (ps.want_blank && *token != '[' && 540 (ps.last_token != ident || proc_calls_space || 541 /* offsetof (1) is never allowed a space; sizeof (2) gets 542 * one iff -bs; all other keywords (>2) always get a space 543 * before lparen */ 544 (ps.keyword + Bill_Shannon > 2))) 545 *e_code++ = ' '; 546 ps.want_blank = false; 547 if (ps.in_decl && !ps.block_init && !ps.dumped_decl_indent && 548 !is_procname) { 549 /* function pointer declarations */ 550 if (troff) { 551 sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token); 552 e_code += strlen(e_code); 553 } 554 else { 555 indent_declaration(dec_ind, tabs_to_var); 556 } 557 ps.dumped_decl_indent = true; 558 } 559 if (!troff) 560 *e_code++ = token[0]; 561 ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code; 562 if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent 563 && ps.paren_indents[0] < 2 * ps.ind_size) 564 ps.paren_indents[0] = 2 * ps.ind_size; 565 if (ps.in_or_st && *token == '(' && ps.tos <= 2) { 566 /* 567 * this is a kluge to make sure that declarations will be 568 * aligned right if proc decl has an explicit type on it, i.e. 569 * "int a(x) {..." 570 */ 571 parse(semicolon); /* I said this was a kluge... */ 572 ps.in_or_st = false; /* turn off flag for structure decl or 573 * initialization */ 574 } 575 /* parenthesized type following sizeof or offsetof is not a cast */ 576 if (ps.keyword == 1 || ps.keyword == 2) 577 ps.not_cast_mask |= 1 << ps.p_l_follow; 578 break; 579 580 case rparen: /* got a ')' or ']' */ 581 rparen_count--; 582 if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.not_cast_mask) { 583 ps.last_u_d = true; 584 ps.cast_mask &= (1 << ps.p_l_follow) - 1; 585 ps.want_blank = false; 586 } else 587 ps.want_blank = true; 588 ps.not_cast_mask &= (1 << ps.p_l_follow) - 1; 589 if (--ps.p_l_follow < 0) { 590 ps.p_l_follow = 0; 591 diag3(0, "Extra %c", *token); 592 } 593 if (e_code == s_code) /* if the paren starts the line */ 594 ps.paren_level = ps.p_l_follow; /* then indent it */ 595 596 *e_code++ = token[0]; 597 598 if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if 599 * (...), or some such */ 600 sp_sw = false; 601 force_nl = true;/* must force newline after if */ 602 ps.last_u_d = true; /* inform lexi that a following 603 * operator is unary */ 604 ps.in_stmt = false; /* dont use stmt continuation 605 * indentation */ 606 607 parse(hd_type); /* let parser worry about if, or whatever */ 608 } 609 ps.search_brace = btype_2; /* this should insure that constructs 610 * such as main(){...} and int[]{...} 611 * have their braces put in the right 612 * place */ 613 break; 614 615 case unary_op: /* this could be any unary operation */ 616 if (!ps.dumped_decl_indent && ps.in_decl && !is_procname && 617 !ps.block_init) { 618 /* pointer declarations */ 619 if (troff) { 620 if (ps.want_blank) 621 *e_code++ = ' '; 622 sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, 623 token); 624 e_code += strlen(e_code); 625 } 626 else { 627 /* if this is a unary op in a declaration, we should 628 * indent this token */ 629 for (i = 0; token[i]; ++i) 630 /* find length of token */; 631 indent_declaration(dec_ind - i, tabs_to_var); 632 } 633 ps.dumped_decl_indent = true; 634 } 635 else if (ps.want_blank) 636 *e_code++ = ' '; 637 { 638 const char *res = token; 639 640 if (troff && token[0] == '-' && token[1] == '>') 641 res = "\\(->"; 642 for (t_ptr = res; *t_ptr; ++t_ptr) { 643 CHECK_SIZE_CODE; 644 *e_code++ = *t_ptr; 645 } 646 } 647 ps.want_blank = false; 648 break; 649 650 case binary_op: /* any binary operation */ 651 if (ps.want_blank) 652 *e_code++ = ' '; 653 { 654 const char *res = token; 655 656 if (troff) 657 switch (token[0]) { 658 case '<': 659 if (token[1] == '=') 660 res = "\\(<="; 661 break; 662 case '>': 663 if (token[1] == '=') 664 res = "\\(>="; 665 break; 666 case '!': 667 if (token[1] == '=') 668 res = "\\(!="; 669 break; 670 case '|': 671 if (token[1] == '|') 672 res = "\\(br\\(br"; 673 else if (token[1] == 0) 674 res = "\\(br"; 675 break; 676 } 677 for (t_ptr = res; *t_ptr; ++t_ptr) { 678 CHECK_SIZE_CODE; 679 *e_code++ = *t_ptr; /* move the operator */ 680 } 681 } 682 ps.want_blank = true; 683 break; 684 685 case postop: /* got a trailing ++ or -- */ 686 *e_code++ = token[0]; 687 *e_code++ = token[1]; 688 ps.want_blank = true; 689 break; 690 691 case question: /* got a ? */ 692 squest++; /* this will be used when a later colon 693 * appears so we can distinguish the 694 * <c>?<n>:<n> construct */ 695 if (ps.want_blank) 696 *e_code++ = ' '; 697 *e_code++ = '?'; 698 ps.want_blank = true; 699 break; 700 701 case casestmt: /* got word 'case' or 'default' */ 702 scase = true; /* so we can process the later colon properly */ 703 goto copy_id; 704 705 case colon: /* got a ':' */ 706 if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */ 707 --squest; 708 if (ps.want_blank) 709 *e_code++ = ' '; 710 *e_code++ = ':'; 711 ps.want_blank = true; 712 break; 713 } 714 if (ps.in_or_st) { 715 *e_code++ = ':'; 716 ps.want_blank = false; 717 break; 718 } 719 ps.in_stmt = false; /* seeing a label does not imply we are in a 720 * stmt */ 721 for (t_ptr = s_code; *t_ptr; ++t_ptr) 722 *e_lab++ = *t_ptr; /* turn everything so far into a label */ 723 e_code = s_code; 724 *e_lab++ = ':'; 725 *e_lab++ = ' '; 726 *e_lab = '\0'; 727 728 force_nl = ps.pcase = scase; /* ps.pcase will be used by 729 * dump_line to decide how to 730 * indent the label. force_nl 731 * will force a case n: to be 732 * on a line by itself */ 733 scase = false; 734 ps.want_blank = false; 735 break; 736 737 case semicolon: /* got a ';' */ 738 if (ps.dec_nest == 0) 739 ps.in_or_st = false;/* we are not in an initialization or 740 * structure declaration */ 741 scase = false; /* these will only need resetting in an error */ 742 squest = 0; 743 if (ps.last_token == rparen && rparen_count == 0) 744 ps.in_parameter_declaration = 0; 745 ps.cast_mask = 0; 746 ps.not_cast_mask = 0; 747 ps.block_init = 0; 748 ps.block_init_level = 0; 749 ps.just_saw_decl--; 750 751 if (ps.in_decl && s_code == e_code && !ps.block_init && 752 !ps.dumped_decl_indent) { 753 /* indent stray semicolons in declarations */ 754 indent_declaration(dec_ind - 1, tabs_to_var); 755 ps.dumped_decl_indent = true; 756 } 757 758 ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level 759 * structure declaration, we 760 * arent any more */ 761 762 if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) { 763 764 /* 765 * This should be true iff there were unbalanced parens in the 766 * stmt. It is a bit complicated, because the semicolon might 767 * be in a for stmt 768 */ 769 diag2(1, "Unbalanced parens"); 770 ps.p_l_follow = 0; 771 if (sp_sw) { /* this is a check for an if, while, etc. with 772 * unbalanced parens */ 773 sp_sw = false; 774 parse(hd_type); /* dont lose the if, or whatever */ 775 } 776 } 777 *e_code++ = ';'; 778 ps.want_blank = true; 779 ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the 780 * middle of a stmt */ 781 782 if (!sp_sw) { /* if not if for (;;) */ 783 parse(semicolon); /* let parser know about end of stmt */ 784 force_nl = true;/* force newline after an end of stmt */ 785 } 786 break; 787 788 case lbrace: /* got a '{' */ 789 ps.in_stmt = false; /* dont indent the {} */ 790 if (!ps.block_init) 791 force_nl = true;/* force other stuff on same line as '{' onto 792 * new line */ 793 else if (ps.block_init_level <= 0) 794 ps.block_init_level = 1; 795 else 796 ps.block_init_level++; 797 798 if (s_code != e_code && !ps.block_init) { 799 if (!btype_2) { 800 dump_line(); 801 ps.want_blank = false; 802 } 803 else if (ps.in_parameter_declaration && !ps.in_or_st) { 804 ps.i_l_follow = 0; 805 if (function_brace_split) { /* dump the line prior to the 806 * brace ... */ 807 dump_line(); 808 ps.want_blank = false; 809 } else /* add a space between the decl and brace */ 810 ps.want_blank = true; 811 } 812 } 813 if (ps.in_parameter_declaration) 814 prefix_blankline_requested = 0; 815 816 if (ps.p_l_follow > 0) { /* check for preceding unbalanced 817 * parens */ 818 diag2(1, "Unbalanced parens"); 819 ps.p_l_follow = 0; 820 if (sp_sw) { /* check for unclosed if, for, etc. */ 821 sp_sw = false; 822 parse(hd_type); 823 ps.ind_level = ps.i_l_follow; 824 } 825 } 826 if (s_code == e_code) 827 ps.ind_stmt = false; /* dont put extra indentation on line 828 * with '{' */ 829 if (ps.in_decl && ps.in_or_st) { /* this is either a structure 830 * declaration or an init */ 831 di_stack[ps.dec_nest++] = dec_ind; 832 /* ? dec_ind = 0; */ 833 } 834 else { 835 ps.decl_on_line = false; /* we can't be in the middle of 836 * a declaration, so don't do 837 * special indentation of 838 * comments */ 839 if (blanklines_after_declarations_at_proctop 840 && ps.in_parameter_declaration) 841 postfix_blankline_requested = 1; 842 ps.in_parameter_declaration = 0; 843 } 844 dec_ind = 0; 845 parse(lbrace); /* let parser know about this */ 846 if (ps.want_blank) /* put a blank before '{' if '{' is not at 847 * start of line */ 848 *e_code++ = ' '; 849 ps.want_blank = false; 850 *e_code++ = '{'; 851 ps.just_saw_decl = 0; 852 break; 853 854 case rbrace: /* got a '}' */ 855 if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be 856 * omitted in 857 * declarations */ 858 parse(semicolon); 859 if (ps.p_l_follow) {/* check for unclosed if, for, else. */ 860 diag2(1, "Unbalanced parens"); 861 ps.p_l_follow = 0; 862 sp_sw = false; 863 } 864 ps.just_saw_decl = 0; 865 ps.block_init_level--; 866 if (s_code != e_code && !ps.block_init) { /* '}' must be first on 867 * line */ 868 if (verbose) 869 diag2(0, "Line broken"); 870 dump_line(); 871 } 872 *e_code++ = '}'; 873 ps.want_blank = true; 874 ps.in_stmt = ps.ind_stmt = false; 875 if (ps.dec_nest > 0) { /* we are in multi-level structure 876 * declaration */ 877 dec_ind = di_stack[--ps.dec_nest]; 878 if (ps.dec_nest == 0 && !ps.in_parameter_declaration) 879 ps.just_saw_decl = 2; 880 ps.in_decl = true; 881 } 882 prefix_blankline_requested = 0; 883 parse(rbrace); /* let parser know about this */ 884 ps.search_brace = cuddle_else && ps.p_stack[ps.tos] == ifhead 885 && ps.il[ps.tos] >= ps.ind_level; 886 if (ps.tos <= 1 && blanklines_after_procs && ps.dec_nest <= 0) 887 postfix_blankline_requested = 1; 888 break; 889 890 case swstmt: /* got keyword "switch" */ 891 sp_sw = true; 892 hd_type = swstmt; /* keep this for when we have seen the 893 * expression */ 894 goto copy_id; /* go move the token into buffer */ 895 896 case sp_paren: /* token is if, while, for */ 897 sp_sw = true; /* the interesting stuff is done after the 898 * expression is scanned */ 899 hd_type = (*token == 'i' ? ifstmt : 900 (*token == 'w' ? whilestmt : forstmt)); 901 902 /* 903 * remember the type of header for later use by parser 904 */ 905 goto copy_id; /* copy the token into line */ 906 907 case sp_nparen: /* got else, do */ 908 ps.in_stmt = false; 909 if (*token == 'e') { 910 if (e_code != s_code && (!cuddle_else || e_code[-1] != '}')) { 911 if (verbose) 912 diag2(0, "Line broken"); 913 dump_line();/* make sure this starts a line */ 914 ps.want_blank = false; 915 } 916 force_nl = true;/* also, following stuff must go onto new line */ 917 last_else = 1; 918 parse(elselit); 919 } 920 else { 921 if (e_code != s_code) { /* make sure this starts a line */ 922 if (verbose) 923 diag2(0, "Line broken"); 924 dump_line(); 925 ps.want_blank = false; 926 } 927 force_nl = true;/* also, following stuff must go onto new line */ 928 last_else = 0; 929 parse(dolit); 930 } 931 goto copy_id; /* move the token into line */ 932 933 case decl: /* we have a declaration type (int, register, 934 * etc.) */ 935 parse(decl); /* let parser worry about indentation */ 936 if (ps.last_token == rparen && ps.tos <= 1) { 937 ps.in_parameter_declaration = 1; 938 if (s_code != e_code) { 939 dump_line(); 940 ps.want_blank = 0; 941 } 942 } 943 if (ps.in_parameter_declaration && ps.indent_parameters && ps.dec_nest == 0) { 944 ps.ind_level = ps.i_l_follow = 1; 945 ps.ind_stmt = 0; 946 } 947 ps.in_or_st = true; /* this might be a structure or initialization 948 * declaration */ 949 ps.in_decl = ps.decl_on_line = true; 950 if ( /* !ps.in_or_st && */ ps.dec_nest <= 0) 951 ps.just_saw_decl = 2; 952 prefix_blankline_requested = 0; 953 for (i = 0; token[i++];); /* get length of token */ 954 955 if (ps.ind_level == 0 || ps.dec_nest > 0) { 956 /* global variable or struct member in local variable */ 957 dec_ind = ps.decl_indent > 0 ? ps.decl_indent : i; 958 tabs_to_var = (use_tabs ? ps.decl_indent > 0 : 0); 959 } else { 960 /* local variable */ 961 dec_ind = ps.local_decl_indent > 0 ? ps.local_decl_indent : i; 962 tabs_to_var = (use_tabs ? ps.local_decl_indent > 0 : 0); 963 } 964 goto copy_id; 965 966 case ident: /* got an identifier or constant */ 967 if (ps.in_decl) { /* if we are in a declaration, we must indent 968 * identifier */ 969 if (is_procname == 0 || !procnames_start_line) { 970 if (!ps.block_init && !ps.dumped_decl_indent) { 971 if (troff) { 972 if (ps.want_blank) 973 *e_code++ = ' '; 974 sprintf(e_code, "\n.De %dp+\200p\n", dec_ind * 7); 975 e_code += strlen(e_code); 976 } else 977 indent_declaration(dec_ind, tabs_to_var); 978 ps.dumped_decl_indent = true; 979 ps.want_blank = false; 980 } 981 } else { 982 if (ps.want_blank) 983 *e_code++ = ' '; 984 ps.want_blank = false; 985 if (dec_ind && s_code != e_code) { 986 *e_code = '\0'; 987 dump_line(); 988 } 989 dec_ind = 0; 990 } 991 } 992 else if (sp_sw && ps.p_l_follow == 0) { 993 sp_sw = false; 994 force_nl = true; 995 ps.last_u_d = true; 996 ps.in_stmt = false; 997 parse(hd_type); 998 } 999 copy_id: 1000 if (ps.want_blank) 1001 *e_code++ = ' '; 1002 if (troff && ps.keyword) { 1003 e_code = chfont(&bodyf, &keywordf, e_code); 1004 for (t_ptr = token; *t_ptr; ++t_ptr) { 1005 CHECK_SIZE_CODE; 1006 *e_code++ = keywordf.allcaps && islower(*t_ptr) 1007 ? toupper(*t_ptr) : *t_ptr; 1008 } 1009 e_code = chfont(&keywordf, &bodyf, e_code); 1010 } 1011 else 1012 for (t_ptr = token; *t_ptr; ++t_ptr) { 1013 CHECK_SIZE_CODE; 1014 *e_code++ = *t_ptr; 1015 } 1016 ps.want_blank = true; 1017 break; 1018 1019 case strpfx: 1020 if (ps.want_blank) 1021 *e_code++ = ' '; 1022 for (t_ptr = token; *t_ptr; ++t_ptr) { 1023 CHECK_SIZE_CODE; 1024 *e_code++ = *t_ptr; 1025 } 1026 ps.want_blank = false; 1027 break; 1028 1029 case period: /* treat a period kind of like a binary 1030 * operation */ 1031 *e_code++ = '.'; /* move the period into line */ 1032 ps.want_blank = false; /* dont put a blank after a period */ 1033 break; 1034 1035 case comma: 1036 ps.want_blank = (s_code != e_code); /* only put blank after comma 1037 * if comma does not start the 1038 * line */ 1039 if (ps.in_decl && is_procname == 0 && !ps.block_init && 1040 !ps.dumped_decl_indent) { 1041 /* indent leading commas and not the actual identifiers */ 1042 indent_declaration(dec_ind - 1, tabs_to_var); 1043 ps.dumped_decl_indent = true; 1044 } 1045 *e_code++ = ','; 1046 if (ps.p_l_follow == 0) { 1047 if (ps.block_init_level <= 0) 1048 ps.block_init = 0; 1049 if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8)) 1050 force_nl = true; 1051 } 1052 break; 1053 1054 case preesc: /* got the character '#' */ 1055 if ((s_com != e_com) || 1056 (s_lab != e_lab) || 1057 (s_code != e_code)) 1058 dump_line(); 1059 *e_lab++ = '#'; /* move whole line to 'label' buffer */ 1060 { 1061 int in_comment = 0; 1062 int com_start = 0; 1063 char quote = 0; 1064 int com_end = 0; 1065 1066 while (*buf_ptr == ' ' || *buf_ptr == '\t') { 1067 buf_ptr++; 1068 if (buf_ptr >= buf_end) 1069 fill_buffer(); 1070 } 1071 while (*buf_ptr != '\n' || (in_comment && !had_eof)) { 1072 CHECK_SIZE_LAB; 1073 *e_lab = *buf_ptr++; 1074 if (buf_ptr >= buf_end) 1075 fill_buffer(); 1076 switch (*e_lab++) { 1077 case BACKSLASH: 1078 if (troff) 1079 *e_lab++ = BACKSLASH; 1080 if (!in_comment) { 1081 *e_lab++ = *buf_ptr++; 1082 if (buf_ptr >= buf_end) 1083 fill_buffer(); 1084 } 1085 break; 1086 case '/': 1087 if (*buf_ptr == '*' && !in_comment && !quote) { 1088 in_comment = 1; 1089 *e_lab++ = *buf_ptr++; 1090 com_start = e_lab - s_lab - 2; 1091 } 1092 break; 1093 case '"': 1094 if (quote == '"') 1095 quote = 0; 1096 break; 1097 case '\'': 1098 if (quote == '\'') 1099 quote = 0; 1100 break; 1101 case '*': 1102 if (*buf_ptr == '/' && in_comment) { 1103 in_comment = 0; 1104 *e_lab++ = *buf_ptr++; 1105 com_end = e_lab - s_lab; 1106 } 1107 break; 1108 } 1109 } 1110 1111 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1112 e_lab--; 1113 if (e_lab - s_lab == com_end && bp_save == NULL) { 1114 /* comment on preprocessor line */ 1115 if (sc_end == NULL) /* if this is the first comment, we 1116 * must set up the buffer */ 1117 sc_end = &(save_com[0]); 1118 else { 1119 *sc_end++ = '\n'; /* add newline between 1120 * comments */ 1121 *sc_end++ = ' '; 1122 --line_no; 1123 } 1124 bcopy(s_lab + com_start, sc_end, com_end - com_start); 1125 sc_end += com_end - com_start; 1126 if (sc_end >= &save_com[sc_size]) 1127 abort(); 1128 e_lab = s_lab + com_start; 1129 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1130 e_lab--; 1131 bp_save = buf_ptr; /* save current input buffer */ 1132 be_save = buf_end; 1133 buf_ptr = save_com; /* fix so that subsequent calls to 1134 * lexi will take tokens out of 1135 * save_com */ 1136 *sc_end++ = ' '; /* add trailing blank, just in case */ 1137 buf_end = sc_end; 1138 sc_end = NULL; 1139 } 1140 *e_lab = '\0'; /* null terminate line */ 1141 ps.pcase = false; 1142 } 1143 1144 if (strncmp(s_lab, "#if", 3) == 0) { /* also ifdef, ifndef */ 1145 if ((size_t)ifdef_level < nitems(state_stack)) { 1146 match_state[ifdef_level].tos = -1; 1147 state_stack[ifdef_level++] = ps; 1148 } 1149 else 1150 diag2(1, "#if stack overflow"); 1151 } 1152 else if (strncmp(s_lab, "#el", 3) == 0) { /* else, elif */ 1153 if (ifdef_level <= 0) 1154 diag2(1, s_lab[3] == 'i' ? "Unmatched #elif" : "Unmatched #else"); 1155 else { 1156 match_state[ifdef_level - 1] = ps; 1157 ps = state_stack[ifdef_level - 1]; 1158 } 1159 } 1160 else if (strncmp(s_lab, "#endif", 6) == 0) { 1161 if (ifdef_level <= 0) 1162 diag2(1, "Unmatched #endif"); 1163 else 1164 ifdef_level--; 1165 } else { 1166 struct directives { 1167 int size; 1168 const char *string; 1169 } 1170 recognized[] = { 1171 {7, "include"}, 1172 {6, "define"}, 1173 {5, "undef"}, 1174 {4, "line"}, 1175 {5, "error"}, 1176 {6, "pragma"} 1177 }; 1178 int d = nitems(recognized); 1179 while (--d >= 0) 1180 if (strncmp(s_lab + 1, recognized[d].string, recognized[d].size) == 0) 1181 break; 1182 if (d < 0) { 1183 diag2(1, "Unrecognized cpp directive"); 1184 break; 1185 } 1186 } 1187 if (blanklines_around_conditional_compilation) { 1188 postfix_blankline_requested++; 1189 n_real_blanklines = 0; 1190 } 1191 else { 1192 postfix_blankline_requested = 0; 1193 prefix_blankline_requested = 0; 1194 } 1195 break; /* subsequent processing of the newline 1196 * character will cause the line to be printed */ 1197 1198 case comment: /* we have gotten a / followed by * this is a biggie */ 1199 if (flushed_nl) { /* we should force a broken line here */ 1200 dump_line(); 1201 ps.want_blank = false; /* dont insert blank at line start */ 1202 force_nl = false; 1203 } 1204 pr_comment(); 1205 break; 1206 } /* end of big switch stmt */ 1207 1208 *e_code = '\0'; /* make sure code section is null terminated */ 1209 if (type_code != comment && type_code != newline && type_code != preesc) 1210 ps.last_token = type_code; 1211 } /* end of main while (1) loop */ 1212} 1213 1214/* 1215 * copy input file to backup file if in_name is /blah/blah/blah/file, then 1216 * backup file will be ".Bfile" then make the backup file the input and 1217 * original input file the output 1218 */ 1219static void 1220bakcopy(void) 1221{ 1222 int n, 1223 bakchn; 1224 char buff[8 * 1024]; 1225 const char *p; 1226 1227 /* construct file name .Bfile */ 1228 for (p = in_name; *p; p++); /* skip to end of string */ 1229 while (p > in_name && *p != '/') /* find last '/' */ 1230 p--; 1231 if (*p == '/') 1232 p++; 1233 sprintf(bakfile, "%s%s", p, simple_backup_suffix); 1234 1235 /* copy in_name to backup file */ 1236 bakchn = creat(bakfile, 0600); 1237 if (bakchn < 0) 1238 err(1, "%s", bakfile); 1239 while ((n = read(fileno(input), buff, sizeof(buff))) > 0) 1240 if (write(bakchn, buff, n) != n) 1241 err(1, "%s", bakfile); 1242 if (n < 0) 1243 err(1, "%s", in_name); 1244 close(bakchn); 1245 fclose(input); 1246 1247 /* re-open backup file as the input file */ 1248 input = fopen(bakfile, "r"); 1249 if (input == NULL) 1250 err(1, "%s", bakfile); 1251 /* now the original input file will be the output */ 1252 output = fopen(in_name, "w"); 1253 if (output == NULL) { 1254 unlink(bakfile); 1255 err(1, "%s", in_name); 1256 } 1257} 1258 1259static void 1260indent_declaration(int cur_dec_ind, int tabs_to_var) 1261{ 1262 int pos = e_code - s_code; 1263 char *startpos = e_code; 1264 1265 /* 1266 * get the tab math right for indentations that are not multiples of 8 1267 */ 1268 if ((ps.ind_level * ps.ind_size) % 8 != 0) { 1269 pos += (ps.ind_level * ps.ind_size) % 8; 1270 cur_dec_ind += (ps.ind_level * ps.ind_size) % 8; 1271 } 1272 if (tabs_to_var) 1273 while ((pos & ~7) + 8 <= cur_dec_ind) { 1274 CHECK_SIZE_CODE; 1275 *e_code++ = '\t'; 1276 pos = (pos & ~7) + 8; 1277 } 1278 while (pos < cur_dec_ind) { 1279 CHECK_SIZE_CODE; 1280 *e_code++ = ' '; 1281 pos++; 1282 } 1283 if (e_code == startpos && ps.want_blank) { 1284 *e_code++ = ' '; 1285 ps.want_blank = false; 1286 } 1287} 1288