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. Neither the name of the University nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 * 31 * @(#) Copyright (c) 1985 Sun Microsystems, Inc. 32 * @(#) Copyright (c) 1976 Board of Trustees of the University of Illinois. 33 * @(#) Copyright (c) 1980, 1993 The Regents of the University of California. All rights reserved. 34 * @(#)indent.c 5.17 (Berkeley) 6/7/93 35 * $FreeBSD: src/usr.bin/indent/indent.c,v 1.26 2010/03/31 16:55:47 avg Exp $ 36 */ 37 38 #include <sys/param.h> 39 #include <err.h> 40 #include <fcntl.h> 41 #include <unistd.h> 42 #include <stdio.h> 43 #include <stdlib.h> 44 #include <string.h> 45 #include <ctype.h> 46 #include "indent_globs.h" 47 #include "indent_codes.h" 48 #include "indent.h" 49 50 static void bakcopy(void); 51 52 const char *in_name = "Standard Input"; /* will always point to name of input 53 * file */ 54 const char *out_name = "Standard Output"; /* will always point to name 55 * of output file */ 56 char bakfile[MAXPATHLEN] = ""; 57 58 int 59 main(int argc, char **argv) 60 { 61 62 int dec_ind; /* current indentation for declarations */ 63 int di_stack[20]; /* a stack of structure indentation levels */ 64 int flushed_nl; /* used when buffering up comments to remember 65 * that a newline was passed over */ 66 int force_nl; /* when true, code must be broken */ 67 int hd_type = 0; /* used to store type of stmt for if (...), 68 * for (...), etc */ 69 int i; /* local loop counter */ 70 int scase; /* set to true when we see a case, so we will 71 * know what to do with the following colon */ 72 int sp_sw; /* when true, we are in the expression of 73 * if(...), while(...), etc. */ 74 int squest; /* when this is positive, we have seen a ? 75 * without the matching : in a <c>?<s>:<s> 76 * construct */ 77 const char *t_ptr; /* used for copying tokens */ 78 int tabs_to_var; /* true if using tabs to indent to var name */ 79 int type_code; /* the type of token, returned by lexi */ 80 81 int last_else = 0; /* true iff last keyword was an else */ 82 83 84 /*-----------------------------------------------*\ 85 | INITIALIZATION | 86 \*-----------------------------------------------*/ 87 88 found_err = 0; 89 90 ps.p_stack[0] = stmt; /* this is the parser's stack */ 91 ps.last_nl = true; /* this is true if the last thing scanned was 92 * a newline */ 93 ps.last_token = semicolon; 94 combuf = (char *) malloc(bufsize); 95 if (combuf == NULL) 96 err(1, NULL); 97 labbuf = (char *) malloc(bufsize); 98 if (labbuf == NULL) 99 err(1, NULL); 100 codebuf = (char *) malloc(bufsize); 101 if (codebuf == NULL) 102 err(1, NULL); 103 tokenbuf = (char *) malloc(bufsize); 104 if (tokenbuf == NULL) 105 err(1, NULL); 106 l_com = combuf + bufsize - 5; 107 l_lab = labbuf + bufsize - 5; 108 l_code = codebuf + bufsize - 5; 109 l_token = tokenbuf + bufsize - 5; 110 combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and 111 * comment buffers */ 112 combuf[1] = codebuf[1] = labbuf[1] = '\0'; 113 ps.else_if = 1; /* Default else-if special processing to on */ 114 s_lab = e_lab = labbuf + 1; 115 s_code = e_code = codebuf + 1; 116 s_com = e_com = combuf + 1; 117 s_token = e_token = tokenbuf + 1; 118 119 in_buffer = (char *) malloc(10); 120 if (in_buffer == NULL) 121 err(1, NULL); 122 in_buffer_limit = in_buffer + 8; 123 buf_ptr = buf_end = in_buffer; 124 line_no = 1; 125 had_eof = ps.in_decl = ps.decl_on_line = break_comma = false; 126 sp_sw = force_nl = false; 127 ps.in_or_st = false; 128 ps.bl_line = true; 129 dec_ind = 0; 130 di_stack[ps.dec_nest = 0] = 0; 131 ps.want_blank = ps.in_stmt = ps.ind_stmt = false; 132 133 scase = ps.pcase = false; 134 squest = 0; 135 sc_end = NULL; 136 bp_save = NULL; 137 be_save = NULL; 138 139 output = NULL; 140 tabs_to_var = 0; 141 142 /*--------------------------------------------------*\ 143 | COMMAND LINE SCAN | 144 \*--------------------------------------------------*/ 145 146 #ifdef undef 147 max_col = 78; /* -l78 */ 148 lineup_to_parens = 1; /* -lp */ 149 ps.ljust_decl = 0; /* -ndj */ 150 ps.com_ind = 33; /* -c33 */ 151 star_comment_cont = 1; /* -sc */ 152 ps.ind_size = 8; /* -i8 */ 153 verbose = 0; 154 ps.decl_indent = 16; /* -di16 */ 155 ps.local_decl_indent = -1; /* if this is not set to some nonnegative value 156 * by an arg, we will set this equal to 157 * ps.decl_ind */ 158 ps.indent_parameters = 1; /* -ip */ 159 ps.decl_com_ind = 0; /* if this is not set to some positive value 160 * by an arg, we will set this equal to 161 * ps.com_ind */ 162 btype_2 = 1; /* -br */ 163 cuddle_else = 1; /* -ce */ 164 ps.unindent_displace = 0; /* -d0 */ 165 ps.case_indent = 0; /* -cli0 */ 166 format_block_comments = 1; /* -fcb */ 167 format_col1_comments = 1; /* -fc1 */ 168 procnames_start_line = 1; /* -psl */ 169 proc_calls_space = 0; /* -npcs */ 170 comment_delimiter_on_blankline = 1; /* -cdb */ 171 ps.leave_comma = 1; /* -nbc */ 172 #endif 173 174 for (i = 1; i < argc; ++i) 175 if (strcmp(argv[i], "-npro") == 0) 176 break; 177 set_defaults(); 178 if (i >= argc) 179 set_profile(); 180 181 for (i = 1; i < argc; ++i) { 182 183 /* 184 * look thru args (if any) for changes to defaults 185 */ 186 if (argv[i][0] != '-') { /* no flag on parameter */ 187 if (input == NULL) { /* we must have the input file */ 188 in_name = argv[i]; /* remember name of input file */ 189 input = fopen(in_name, "r"); 190 if (input == NULL) /* check for open error */ 191 err(1, "%s", in_name); 192 continue; 193 } 194 else if (output == NULL) { /* we have the output file */ 195 out_name = argv[i]; /* remember name of output file */ 196 if (strcmp(in_name, out_name) == 0) { /* attempt to overwrite 197 * the file */ 198 errx(1, "input and output files must be different"); 199 } 200 output = fopen(out_name, "w"); 201 if (output == NULL) /* check for create error */ 202 err(1, "%s", out_name); 203 continue; 204 } 205 errx(1, "unknown parameter: %s", argv[i]); 206 } 207 else 208 set_option(argv[i]); 209 } /* end of for */ 210 if (input == NULL) 211 input = stdin; 212 if (output == NULL) { 213 if (troff || input == stdin) 214 output = stdout; 215 else { 216 out_name = in_name; 217 bakcopy(); 218 } 219 } 220 if (ps.com_ind <= 1) 221 ps.com_ind = 2; /* dont put normal comments before column 2 */ 222 if (troff) { 223 if (bodyf.font[0] == 0) 224 parsefont(&bodyf, "R"); 225 if (scomf.font[0] == 0) 226 parsefont(&scomf, "I"); 227 if (blkcomf.font[0] == 0) 228 blkcomf = scomf, blkcomf.size += 2; 229 if (boxcomf.font[0] == 0) 230 boxcomf = blkcomf; 231 if (stringf.font[0] == 0) 232 parsefont(&stringf, "L"); 233 if (keywordf.font[0] == 0) 234 parsefont(&keywordf, "B"); 235 writefdef(&bodyf, 'B'); 236 writefdef(&scomf, 'C'); 237 writefdef(&blkcomf, 'L'); 238 writefdef(&boxcomf, 'X'); 239 writefdef(&stringf, 'S'); 240 writefdef(&keywordf, 'K'); 241 } 242 if (block_comment_max_col <= 0) 243 block_comment_max_col = max_col; 244 if (ps.local_decl_indent < 0) /* if not specified by user, set this */ 245 ps.local_decl_indent = ps.decl_indent; 246 if (ps.decl_com_ind <= 0) /* if not specified by user, set this */ 247 ps.decl_com_ind = ps.ljust_decl ? (ps.com_ind <= 10 ? 2 : ps.com_ind - 8) : ps.com_ind; 248 if (continuation_indent == 0) 249 continuation_indent = ps.ind_size; 250 fill_buffer(); /* get first batch of stuff into input buffer */ 251 252 parse(semicolon); 253 { 254 char *p = buf_ptr; 255 int col = 1; 256 257 while (1) { 258 if (*p == ' ') 259 col++; 260 else if (*p == '\t') 261 col = ((col - 1) & ~7) + 9; 262 else 263 break; 264 p++; 265 } 266 if (col > ps.ind_size) 267 ps.ind_level = ps.i_l_follow = col / ps.ind_size; 268 } 269 if (troff) { 270 const char *p = in_name, 271 *beg = in_name; 272 273 while (*p) 274 if (*p++ == '/') 275 beg = p; 276 fprintf(output, ".Fn \"%s\"\n", beg); 277 } 278 /* 279 * START OF MAIN LOOP 280 */ 281 282 while (1) { /* this is the main loop. it will go until we 283 * reach eof */ 284 int is_procname; 285 286 type_code = lexi(); /* lexi reads one token. The actual 287 * characters read are stored in "token". lexi 288 * returns a code indicating the type of token */ 289 is_procname = ps.procname[0]; 290 291 /* 292 * The following code moves everything following an if (), while (), 293 * else, etc. up to the start of the following stmt to a buffer. This 294 * allows proper handling of both kinds of brace placement. 295 */ 296 297 flushed_nl = false; 298 while (ps.search_brace) { /* if we scanned an if(), while(), 299 * etc., we might need to copy stuff 300 * into a buffer we must loop, copying 301 * stuff into save_com, until we find 302 * the start of the stmt which follows 303 * the if, or whatever */ 304 switch (type_code) { 305 case newline: 306 ++line_no; 307 flushed_nl = true; 308 case form_feed: 309 break; /* form feeds and newlines found here will be 310 * ignored */ 311 312 case lbrace: /* this is a brace that starts the compound 313 * stmt */ 314 if (sc_end == NULL) { /* ignore buffering if a comment wasn't 315 * stored up */ 316 ps.search_brace = false; 317 goto check_type; 318 } 319 if (btype_2) { 320 save_com[0] = '{'; /* we either want to put the brace 321 * right after the if */ 322 goto sw_buffer; /* go to common code to get out of 323 * this loop */ 324 } 325 /* FALLTHROUGH */ 326 case comment: /* we have a comment, so we must copy it into 327 * the buffer */ 328 if (!flushed_nl || sc_end != NULL) { 329 if (sc_end == NULL) { /* if this is the first comment, we 330 * must set up the buffer */ 331 save_com[0] = save_com[1] = ' '; 332 sc_end = &(save_com[2]); 333 } 334 else { 335 *sc_end++ = '\n'; /* add newline between 336 * comments */ 337 *sc_end++ = ' '; 338 --line_no; 339 } 340 *sc_end++ = '/'; /* copy in start of comment */ 341 *sc_end++ = '*'; 342 343 for (;;) { /* loop until we get to the end of the comment */ 344 *sc_end = *buf_ptr++; 345 if (buf_ptr >= buf_end) 346 fill_buffer(); 347 348 if (*sc_end++ == '*' && *buf_ptr == '/') 349 break; /* we are at end of comment */ 350 351 if (sc_end >= &(save_com[sc_size])) { /* check for temp buffer 352 * overflow */ 353 diag2(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever"); 354 fflush(output); 355 exit(1); 356 } 357 } 358 *sc_end++ = '/'; /* add ending slash */ 359 if (++buf_ptr >= buf_end) /* get past / in buffer */ 360 fill_buffer(); 361 break; 362 } 363 /* FALLTHROUGH */ 364 default: /* it is the start of a normal statement */ 365 if (flushed_nl) /* if we flushed a newline, make sure it is 366 * put back */ 367 force_nl = true; 368 if ((type_code == sp_paren && *token == 'i' 369 && last_else && ps.else_if) 370 || (type_code == sp_nparen && *token == 'e' 371 && e_code != s_code && e_code[-1] == '}')) 372 force_nl = false; 373 374 if (sc_end == NULL) { /* ignore buffering if comment wasn't 375 * saved up */ 376 ps.search_brace = false; 377 goto check_type; 378 } 379 if (force_nl) { /* if we should insert a nl here, put it into 380 * the buffer */ 381 force_nl = false; 382 --line_no; /* this will be re-increased when the nl is 383 * read from the buffer */ 384 *sc_end++ = '\n'; 385 *sc_end++ = ' '; 386 if (verbose && !flushed_nl) /* print error msg if the line 387 * was not already broken */ 388 diag2(0, "Line broken"); 389 flushed_nl = false; 390 } 391 for (t_ptr = token; *t_ptr; ++t_ptr) 392 *sc_end++ = *t_ptr; /* copy token into temp buffer */ 393 ps.procname[0] = 0; 394 395 sw_buffer: 396 ps.search_brace = false; /* stop looking for start of 397 * stmt */ 398 bp_save = buf_ptr; /* save current input buffer */ 399 be_save = buf_end; 400 buf_ptr = save_com; /* fix so that subsequent calls to 401 * lexi will take tokens out of 402 * save_com */ 403 *sc_end++ = ' ';/* add trailing blank, just in case */ 404 buf_end = sc_end; 405 sc_end = NULL; 406 break; 407 } /* end of switch */ 408 if (type_code != 0) /* we must make this check, just in case there 409 * was an unexpected EOF */ 410 type_code = lexi(); /* read another token */ 411 /* if (ps.search_brace) ps.procname[0] = 0; */ 412 if ((is_procname = ps.procname[0]) && flushed_nl 413 && !procnames_start_line && ps.in_decl 414 && type_code == ident) 415 flushed_nl = 0; 416 } /* end of while (search_brace) */ 417 last_else = 0; 418 check_type: 419 if (type_code == 0) { /* we got eof */ 420 if (s_lab != e_lab || s_code != e_code 421 || s_com != e_com) /* must dump end of line */ 422 dump_line(); 423 if (ps.tos > 1) /* check for balanced braces */ 424 diag2(1, "Stuff missing from end of file"); 425 426 if (verbose) { 427 printf("There were %d output lines and %d comments\n", 428 ps.out_lines, ps.out_coms); 429 printf("(Lines with comments)/(Lines with code): %6.3f\n", 430 (1.0 * ps.com_lines) / code_lines); 431 } 432 fflush(output); 433 exit(found_err); 434 } 435 if ( 436 (type_code != comment) && 437 (type_code != newline) && 438 (type_code != preesc) && 439 (type_code != form_feed)) { 440 if (force_nl && 441 (type_code != semicolon) && 442 (type_code != lbrace || !btype_2)) { 443 /* we should force a broken line here */ 444 if (verbose && !flushed_nl) 445 diag2(0, "Line broken"); 446 flushed_nl = false; 447 dump_line(); 448 ps.want_blank = false; /* dont insert blank at line start */ 449 force_nl = false; 450 } 451 ps.in_stmt = true; /* turn on flag which causes an extra level of 452 * indentation. this is turned off by a ; or 453 * '}' */ 454 if (s_com != e_com) { /* the turkey has embedded a comment 455 * in a line. fix it */ 456 *e_code++ = ' '; 457 for (t_ptr = s_com; *t_ptr; ++t_ptr) { 458 CHECK_SIZE_CODE; 459 *e_code++ = *t_ptr; 460 } 461 *e_code++ = ' '; 462 *e_code = '\0'; /* null terminate code sect */ 463 ps.want_blank = false; 464 e_com = s_com; 465 } 466 } 467 else if (type_code != comment) /* preserve force_nl thru a comment */ 468 force_nl = false; /* cancel forced newline after newline, form 469 * feed, etc */ 470 471 472 473 /*-----------------------------------------------------*\ 474 | do switch on type of token scanned | 475 \*-----------------------------------------------------*/ 476 CHECK_SIZE_CODE; 477 switch (type_code) { /* now, decide what to do with the token */ 478 479 case form_feed: /* found a form feed in line */ 480 ps.use_ff = true; /* a form feed is treated much like a newline */ 481 dump_line(); 482 ps.want_blank = false; 483 break; 484 485 case newline: 486 if (ps.last_token != comma || ps.p_l_follow > 0 487 || !ps.leave_comma || ps.block_init || !break_comma || s_com != e_com) { 488 dump_line(); 489 ps.want_blank = false; 490 } 491 ++line_no; /* keep track of input line number */ 492 break; 493 494 case lparen: /* got a '(' or '[' */ 495 ++ps.p_l_follow; /* count parens to make Healy happy */ 496 if (ps.want_blank && *token != '[' && 497 (ps.last_token != ident || proc_calls_space 498 || (ps.its_a_keyword && (!ps.sizeof_keyword || Bill_Shannon)))) 499 *e_code++ = ' '; 500 if (ps.in_decl && !ps.block_init) 501 if (troff && !ps.dumped_decl_indent && !is_procname && ps.last_token == decl) { 502 ps.dumped_decl_indent = 1; 503 sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token); 504 e_code += strlen(e_code); 505 } 506 else { 507 while ((e_code - s_code) < dec_ind) { 508 CHECK_SIZE_CODE; 509 *e_code++ = ' '; 510 } 511 *e_code++ = token[0]; 512 } 513 else 514 *e_code++ = token[0]; 515 ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code; 516 if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent 517 && ps.paren_indents[0] < 2 * ps.ind_size) 518 ps.paren_indents[0] = 2 * ps.ind_size; 519 ps.want_blank = false; 520 if (ps.in_or_st && *token == '(' && ps.tos <= 2) { 521 /* 522 * this is a kluge to make sure that declarations will be 523 * aligned right if proc decl has an explicit type on it, i.e. 524 * "int a(x) {..." 525 */ 526 parse(semicolon); /* I said this was a kluge... */ 527 ps.in_or_st = false; /* turn off flag for structure decl or 528 * initialization */ 529 } 530 if (ps.sizeof_keyword) 531 ps.sizeof_mask |= 1 << ps.p_l_follow; 532 break; 533 534 case rparen: /* got a ')' or ']' */ 535 rparen_count--; 536 if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.sizeof_mask) { 537 ps.last_u_d = true; 538 ps.cast_mask &= (1 << ps.p_l_follow) - 1; 539 ps.want_blank = false; 540 } else 541 ps.want_blank = true; 542 ps.sizeof_mask &= (1 << ps.p_l_follow) - 1; 543 if (--ps.p_l_follow < 0) { 544 ps.p_l_follow = 0; 545 diag3(0, "Extra %c", *token); 546 } 547 if (e_code == s_code) /* if the paren starts the line */ 548 ps.paren_level = ps.p_l_follow; /* then indent it */ 549 550 *e_code++ = token[0]; 551 552 if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if 553 * (...), or some such */ 554 sp_sw = false; 555 force_nl = true;/* must force newline after if */ 556 ps.last_u_d = true; /* inform lexi that a following 557 * operator is unary */ 558 ps.in_stmt = false; /* dont use stmt continuation 559 * indentation */ 560 561 parse(hd_type); /* let parser worry about if, or whatever */ 562 } 563 ps.search_brace = btype_2; /* this should insure that constructs 564 * such as main(){...} and int[]{...} 565 * have their braces put in the right 566 * place */ 567 break; 568 569 case unary_op: /* this could be any unary operation */ 570 if (ps.want_blank) 571 *e_code++ = ' '; 572 573 if (troff && !ps.dumped_decl_indent && ps.in_decl && !is_procname) { 574 sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token); 575 ps.dumped_decl_indent = 1; 576 e_code += strlen(e_code); 577 } 578 else { 579 const char *res = token; 580 581 if (ps.in_decl && !ps.block_init) { /* if this is a unary op 582 * in a declaration, we 583 * should indent this 584 * token */ 585 for (i = 0; token[i]; ++i); /* find length of token */ 586 while ((e_code - s_code) < (dec_ind - i)) { 587 CHECK_SIZE_CODE; 588 *e_code++ = ' '; /* pad it */ 589 } 590 } 591 if (troff && token[0] == '-' && token[1] == '>') 592 res = "\\(->"; 593 for (t_ptr = res; *t_ptr; ++t_ptr) { 594 CHECK_SIZE_CODE; 595 *e_code++ = *t_ptr; 596 } 597 } 598 ps.want_blank = false; 599 break; 600 601 case binary_op: /* any binary operation */ 602 if (ps.want_blank) 603 *e_code++ = ' '; 604 { 605 const char *res = token; 606 607 if (troff) 608 switch (token[0]) { 609 case '<': 610 if (token[1] == '=') 611 res = "\\(<="; 612 break; 613 case '>': 614 if (token[1] == '=') 615 res = "\\(>="; 616 break; 617 case '!': 618 if (token[1] == '=') 619 res = "\\(!="; 620 break; 621 case '|': 622 if (token[1] == '|') 623 res = "\\(br\\(br"; 624 else if (token[1] == 0) 625 res = "\\(br"; 626 break; 627 } 628 for (t_ptr = res; *t_ptr; ++t_ptr) { 629 CHECK_SIZE_CODE; 630 *e_code++ = *t_ptr; /* move the operator */ 631 } 632 } 633 ps.want_blank = true; 634 break; 635 636 case postop: /* got a trailing ++ or -- */ 637 *e_code++ = token[0]; 638 *e_code++ = token[1]; 639 ps.want_blank = true; 640 break; 641 642 case question: /* got a ? */ 643 squest++; /* this will be used when a later colon 644 * appears so we can distinguish the 645 * <c>?<n>:<n> construct */ 646 if (ps.want_blank) 647 *e_code++ = ' '; 648 *e_code++ = '?'; 649 ps.want_blank = true; 650 break; 651 652 case casestmt: /* got word 'case' or 'default' */ 653 scase = true; /* so we can process the later colon properly */ 654 goto copy_id; 655 656 case colon: /* got a ':' */ 657 if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */ 658 --squest; 659 if (ps.want_blank) 660 *e_code++ = ' '; 661 *e_code++ = ':'; 662 ps.want_blank = true; 663 break; 664 } 665 if (ps.in_or_st) { 666 *e_code++ = ':'; 667 ps.want_blank = false; 668 break; 669 } 670 ps.in_stmt = false; /* seeing a label does not imply we are in a 671 * stmt */ 672 for (t_ptr = s_code; *t_ptr; ++t_ptr) 673 *e_lab++ = *t_ptr; /* turn everything so far into a label */ 674 e_code = s_code; 675 *e_lab++ = ':'; 676 *e_lab++ = ' '; 677 *e_lab = '\0'; 678 679 force_nl = ps.pcase = scase; /* ps.pcase will be used by 680 * dump_line to decide how to 681 * indent the label. force_nl 682 * will force a case n: to be 683 * on a line by itself */ 684 scase = false; 685 ps.want_blank = false; 686 break; 687 688 case semicolon: /* got a ';' */ 689 ps.in_or_st = false;/* we are not in an initialization or 690 * structure declaration */ 691 scase = false; /* these will only need resetting in an error */ 692 squest = 0; 693 if (ps.last_token == rparen && rparen_count == 0) 694 ps.in_parameter_declaration = 0; 695 ps.cast_mask = 0; 696 ps.sizeof_mask = 0; 697 ps.block_init = 0; 698 ps.block_init_level = 0; 699 ps.just_saw_decl--; 700 701 if (ps.in_decl && s_code == e_code && !ps.block_init) 702 while ((e_code - s_code) < (dec_ind - 1)) { 703 CHECK_SIZE_CODE; 704 *e_code++ = ' '; 705 } 706 707 ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level 708 * structure declaration, we 709 * arent any more */ 710 711 if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) { 712 713 /* 714 * This should be true iff there were unbalanced parens in the 715 * stmt. It is a bit complicated, because the semicolon might 716 * be in a for stmt 717 */ 718 diag2(1, "Unbalanced parens"); 719 ps.p_l_follow = 0; 720 if (sp_sw) { /* this is a check for an if, while, etc. with 721 * unbalanced parens */ 722 sp_sw = false; 723 parse(hd_type); /* dont lose the if, or whatever */ 724 } 725 } 726 *e_code++ = ';'; 727 ps.want_blank = true; 728 ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the 729 * middle of a stmt */ 730 731 if (!sp_sw) { /* if not if for (;;) */ 732 parse(semicolon); /* let parser know about end of stmt */ 733 force_nl = true;/* force newline after an end of stmt */ 734 } 735 break; 736 737 case lbrace: /* got a '{' */ 738 ps.in_stmt = false; /* dont indent the {} */ 739 if (!ps.block_init) 740 force_nl = true;/* force other stuff on same line as '{' onto 741 * new line */ 742 else if (ps.block_init_level <= 0) 743 ps.block_init_level = 1; 744 else 745 ps.block_init_level++; 746 747 if (s_code != e_code && !ps.block_init) { 748 if (!btype_2) { 749 dump_line(); 750 ps.want_blank = false; 751 } 752 else if (ps.in_parameter_declaration && !ps.in_or_st) { 753 ps.i_l_follow = 0; 754 if (function_brace_split) { /* dump the line prior to the 755 * brace ... */ 756 dump_line(); 757 ps.want_blank = false; 758 } else /* add a space between the decl and brace */ 759 ps.want_blank = true; 760 } 761 } 762 if (ps.in_parameter_declaration) 763 prefix_blankline_requested = 0; 764 765 if (ps.p_l_follow > 0) { /* check for preceding unbalanced 766 * parens */ 767 diag2(1, "Unbalanced parens"); 768 ps.p_l_follow = 0; 769 if (sp_sw) { /* check for unclosed if, for, etc. */ 770 sp_sw = false; 771 parse(hd_type); 772 ps.ind_level = ps.i_l_follow; 773 } 774 } 775 if (s_code == e_code) 776 ps.ind_stmt = false; /* dont put extra indentation on line 777 * with '{' */ 778 if (ps.in_decl && ps.in_or_st) { /* this is either a structure 779 * declaration or an init */ 780 di_stack[ps.dec_nest++] = dec_ind; 781 /* ? dec_ind = 0; */ 782 } 783 else { 784 ps.decl_on_line = false; /* we can't be in the middle of 785 * a declaration, so don't do 786 * special indentation of 787 * comments */ 788 if (blanklines_after_declarations_at_proctop 789 && ps.in_parameter_declaration) 790 postfix_blankline_requested = 1; 791 ps.in_parameter_declaration = 0; 792 } 793 dec_ind = 0; 794 parse(lbrace); /* let parser know about this */ 795 if (ps.want_blank) /* put a blank before '{' if '{' is not at 796 * start of line */ 797 *e_code++ = ' '; 798 ps.want_blank = false; 799 *e_code++ = '{'; 800 ps.just_saw_decl = 0; 801 break; 802 803 case rbrace: /* got a '}' */ 804 if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be 805 * omitted in 806 * declarations */ 807 parse(semicolon); 808 if (ps.p_l_follow) {/* check for unclosed if, for, else. */ 809 diag2(1, "Unbalanced parens"); 810 ps.p_l_follow = 0; 811 sp_sw = false; 812 } 813 ps.just_saw_decl = 0; 814 ps.block_init_level--; 815 if (s_code != e_code && !ps.block_init) { /* '}' must be first on 816 * line */ 817 if (verbose) 818 diag2(0, "Line broken"); 819 dump_line(); 820 } 821 *e_code++ = '}'; 822 ps.want_blank = true; 823 ps.in_stmt = ps.ind_stmt = false; 824 if (ps.dec_nest > 0) { /* we are in multi-level structure 825 * declaration */ 826 dec_ind = di_stack[--ps.dec_nest]; 827 if (ps.dec_nest == 0 && !ps.in_parameter_declaration) 828 ps.just_saw_decl = 2; 829 ps.in_decl = true; 830 } 831 prefix_blankline_requested = 0; 832 parse(rbrace); /* let parser know about this */ 833 ps.search_brace = cuddle_else && ps.p_stack[ps.tos] == ifhead 834 && ps.il[ps.tos] >= ps.ind_level; 835 if (ps.tos <= 1 && blanklines_after_procs && ps.dec_nest <= 0) 836 postfix_blankline_requested = 1; 837 break; 838 839 case swstmt: /* got keyword "switch" */ 840 sp_sw = true; 841 hd_type = swstmt; /* keep this for when we have seen the 842 * expression */ 843 goto copy_id; /* go move the token into buffer */ 844 845 case sp_paren: /* token is if, while, for */ 846 sp_sw = true; /* the interesting stuff is done after the 847 * expression is scanned */ 848 hd_type = (*token == 'i' ? ifstmt : 849 (*token == 'w' ? whilestmt : forstmt)); 850 851 /* 852 * remember the type of header for later use by parser 853 */ 854 goto copy_id; /* copy the token into line */ 855 856 case sp_nparen: /* got else, do */ 857 ps.in_stmt = false; 858 if (*token == 'e') { 859 if (e_code != s_code && (!cuddle_else || e_code[-1] != '}')) { 860 if (verbose) 861 diag2(0, "Line broken"); 862 dump_line();/* make sure this starts a line */ 863 ps.want_blank = false; 864 } 865 force_nl = true;/* also, following stuff must go onto new line */ 866 last_else = 1; 867 parse(elselit); 868 } 869 else { 870 if (e_code != s_code) { /* make sure this starts a line */ 871 if (verbose) 872 diag2(0, "Line broken"); 873 dump_line(); 874 ps.want_blank = false; 875 } 876 force_nl = true;/* also, following stuff must go onto new line */ 877 last_else = 0; 878 parse(dolit); 879 } 880 goto copy_id; /* move the token into line */ 881 882 case decl: /* we have a declaration type (int, register, 883 * etc.) */ 884 parse(decl); /* let parser worry about indentation */ 885 if (ps.last_token == rparen && ps.tos <= 1) { 886 ps.in_parameter_declaration = 1; 887 if (s_code != e_code) { 888 dump_line(); 889 ps.want_blank = 0; 890 } 891 } 892 if (ps.in_parameter_declaration && ps.indent_parameters && ps.dec_nest == 0) { 893 ps.ind_level = ps.i_l_follow = 1; 894 ps.ind_stmt = 0; 895 } 896 ps.in_or_st = true; /* this might be a structure or initialization 897 * declaration */ 898 ps.in_decl = ps.decl_on_line = true; 899 if ( /* !ps.in_or_st && */ ps.dec_nest <= 0) 900 ps.just_saw_decl = 2; 901 prefix_blankline_requested = 0; 902 for (i = 0; token[i++];); /* get length of token */ 903 904 if (ps.ind_level == 0 || ps.dec_nest > 0) { 905 /* global variable or struct member in local variable */ 906 dec_ind = ps.decl_indent > 0 ? ps.decl_indent : i; 907 tabs_to_var = (use_tabs ? ps.decl_indent > 0 : 0); 908 } else { 909 /* local variable */ 910 dec_ind = ps.local_decl_indent > 0 ? ps.local_decl_indent : i; 911 tabs_to_var = (use_tabs ? ps.local_decl_indent > 0 : 0); 912 } 913 goto copy_id; 914 915 case ident: /* got an identifier or constant */ 916 if (ps.in_decl) { /* if we are in a declaration, we must indent 917 * identifier */ 918 if (is_procname == 0 || !procnames_start_line) { 919 if (!ps.block_init) { 920 if (troff && !ps.dumped_decl_indent) { 921 if (ps.want_blank) 922 *e_code++ = ' '; 923 ps.want_blank = false; 924 sprintf(e_code, "\n.De %dp+\200p\n", dec_ind * 7); 925 ps.dumped_decl_indent = 1; 926 e_code += strlen(e_code); 927 } else { 928 int cur_dec_ind; 929 int pos, startpos; 930 931 /* 932 * in order to get the tab math right for 933 * indentations that are not multiples of 8 we 934 * need to modify both startpos and dec_ind 935 * (cur_dec_ind) here by eight minus the 936 * remainder of the current starting column 937 * divided by eight. This seems to be a 938 * properly working fix 939 */ 940 startpos = e_code - s_code; 941 cur_dec_ind = dec_ind; 942 pos = startpos; 943 if ((ps.ind_level * ps.ind_size) % 8 != 0) { 944 pos += (ps.ind_level * ps.ind_size) % 8; 945 cur_dec_ind += (ps.ind_level * ps.ind_size) % 8; 946 } 947 948 if (tabs_to_var) { 949 while ((pos & ~7) + 8 <= cur_dec_ind) { 950 CHECK_SIZE_CODE; 951 *e_code++ = '\t'; 952 pos = (pos & ~7) + 8; 953 } 954 } 955 while (pos < cur_dec_ind) { 956 CHECK_SIZE_CODE; 957 *e_code++ = ' '; 958 pos++; 959 } 960 if (ps.want_blank && e_code - s_code == startpos) 961 *e_code++ = ' '; 962 ps.want_blank = false; 963 } 964 } 965 } else { 966 if (ps.want_blank) 967 *e_code++ = ' '; 968 ps.want_blank = false; 969 if (dec_ind && s_code != e_code) 970 dump_line(); 971 dec_ind = 0; 972 } 973 } 974 else if (sp_sw && ps.p_l_follow == 0) { 975 sp_sw = false; 976 force_nl = true; 977 ps.last_u_d = true; 978 ps.in_stmt = false; 979 parse(hd_type); 980 } 981 copy_id: 982 if (ps.want_blank) 983 *e_code++ = ' '; 984 if (troff && ps.its_a_keyword) { 985 e_code = chfont(&bodyf, &keywordf, e_code); 986 for (t_ptr = token; *t_ptr; ++t_ptr) { 987 CHECK_SIZE_CODE; 988 *e_code++ = keywordf.allcaps && islower(*t_ptr) 989 ? toupper(*t_ptr) : *t_ptr; 990 } 991 e_code = chfont(&keywordf, &bodyf, e_code); 992 } 993 else 994 for (t_ptr = token; *t_ptr; ++t_ptr) { 995 CHECK_SIZE_CODE; 996 *e_code++ = *t_ptr; 997 } 998 ps.want_blank = true; 999 break; 1000 1001 case period: /* treat a period kind of like a binary 1002 * operation */ 1003 *e_code++ = '.'; /* move the period into line */ 1004 ps.want_blank = false; /* dont put a blank after a period */ 1005 break; 1006 1007 case comma: 1008 ps.want_blank = (s_code != e_code); /* only put blank after comma 1009 * if comma does not start the 1010 * line */ 1011 if (ps.in_decl && is_procname == 0 && !ps.block_init) 1012 while ((e_code - s_code) < (dec_ind - 1)) { 1013 CHECK_SIZE_CODE; 1014 *e_code++ = ' '; 1015 } 1016 1017 *e_code++ = ','; 1018 if (ps.p_l_follow == 0) { 1019 if (ps.block_init_level <= 0) 1020 ps.block_init = 0; 1021 if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8)) 1022 force_nl = true; 1023 } 1024 break; 1025 1026 case preesc: /* got the character '#' */ 1027 if ((s_com != e_com) || 1028 (s_lab != e_lab) || 1029 (s_code != e_code)) 1030 dump_line(); 1031 *e_lab++ = '#'; /* move whole line to 'label' buffer */ 1032 { 1033 int in_comment = 0; 1034 int com_start = 0; 1035 char quote = 0; 1036 int com_end = 0; 1037 1038 while (*buf_ptr == ' ' || *buf_ptr == '\t') { 1039 buf_ptr++; 1040 if (buf_ptr >= buf_end) 1041 fill_buffer(); 1042 } 1043 while (*buf_ptr != '\n' || (in_comment && !had_eof)) { 1044 CHECK_SIZE_LAB; 1045 *e_lab = *buf_ptr++; 1046 if (buf_ptr >= buf_end) 1047 fill_buffer(); 1048 switch (*e_lab++) { 1049 case BACKSLASH: 1050 if (troff) 1051 *e_lab++ = BACKSLASH; 1052 if (!in_comment) { 1053 *e_lab++ = *buf_ptr++; 1054 if (buf_ptr >= buf_end) 1055 fill_buffer(); 1056 } 1057 break; 1058 case '/': 1059 if (*buf_ptr == '*' && !in_comment && !quote) { 1060 in_comment = 1; 1061 *e_lab++ = *buf_ptr++; 1062 com_start = e_lab - s_lab - 2; 1063 } 1064 break; 1065 case '"': 1066 if (quote == '"') 1067 quote = 0; 1068 break; 1069 case '\'': 1070 if (quote == '\'') 1071 quote = 0; 1072 break; 1073 case '*': 1074 if (*buf_ptr == '/' && in_comment) { 1075 in_comment = 0; 1076 *e_lab++ = *buf_ptr++; 1077 com_end = e_lab - s_lab; 1078 } 1079 break; 1080 } 1081 } 1082 1083 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1084 e_lab--; 1085 if (e_lab - s_lab == com_end && bp_save == NULL) { 1086 /* comment on preprocessor line */ 1087 if (sc_end == NULL) /* if this is the first comment, we 1088 * must set up the buffer */ 1089 sc_end = &(save_com[0]); 1090 else { 1091 *sc_end++ = '\n'; /* add newline between 1092 * comments */ 1093 *sc_end++ = ' '; 1094 --line_no; 1095 } 1096 bcopy(s_lab + com_start, sc_end, com_end - com_start); 1097 sc_end += com_end - com_start; 1098 if (sc_end >= &save_com[sc_size]) 1099 abort(); 1100 e_lab = s_lab + com_start; 1101 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1102 e_lab--; 1103 bp_save = buf_ptr; /* save current input buffer */ 1104 be_save = buf_end; 1105 buf_ptr = save_com; /* fix so that subsequent calls to 1106 * lexi will take tokens out of 1107 * save_com */ 1108 *sc_end++ = ' '; /* add trailing blank, just in case */ 1109 buf_end = sc_end; 1110 sc_end = NULL; 1111 } 1112 *e_lab = '\0'; /* null terminate line */ 1113 ps.pcase = false; 1114 } 1115 1116 if (strncmp(s_lab, "#if", 3) == 0) { 1117 if (blanklines_around_conditional_compilation) { 1118 int c; 1119 prefix_blankline_requested++; 1120 while ((c = getc(input)) == '\n'); 1121 ungetc(c, input); 1122 } 1123 if ((size_t)ifdef_level < sizeof(state_stack)/sizeof(state_stack[0])) { 1124 match_state[ifdef_level].tos = -1; 1125 state_stack[ifdef_level++] = ps; 1126 } 1127 else 1128 diag2(1, "#if stack overflow"); 1129 } 1130 else if (strncmp(s_lab, "#else", 5) == 0) 1131 if (ifdef_level <= 0) 1132 diag2(1, "Unmatched #else"); 1133 else { 1134 match_state[ifdef_level - 1] = ps; 1135 ps = state_stack[ifdef_level - 1]; 1136 } 1137 else if (strncmp(s_lab, "#endif", 6) == 0) { 1138 if (ifdef_level <= 0) 1139 diag2(1, "Unmatched #endif"); 1140 else { 1141 ifdef_level--; 1142 1143 #ifdef undef 1144 /* 1145 * This match needs to be more intelligent before the 1146 * message is useful 1147 */ 1148 if (match_state[ifdef_level].tos >= 0 1149 && bcmp(&ps, &match_state[ifdef_level], sizeof ps)) 1150 diag2(0, "Syntactically inconsistent #ifdef alternatives"); 1151 #endif 1152 } 1153 if (blanklines_around_conditional_compilation) { 1154 postfix_blankline_requested++; 1155 n_real_blanklines = 0; 1156 } 1157 } 1158 break; /* subsequent processing of the newline 1159 * character will cause the line to be printed */ 1160 1161 case comment: /* we have gotten a / followed by * this is a biggie */ 1162 if (flushed_nl) { /* we should force a broken line here */ 1163 flushed_nl = false; 1164 dump_line(); 1165 ps.want_blank = false; /* dont insert blank at line start */ 1166 force_nl = false; 1167 } 1168 pr_comment(); 1169 break; 1170 } /* end of big switch stmt */ 1171 1172 *e_code = '\0'; /* make sure code section is null terminated */ 1173 if (type_code != comment && type_code != newline && type_code != preesc) 1174 ps.last_token = type_code; 1175 } /* end of main while (1) loop */ 1176 } 1177 1178 /* 1179 * copy input file to backup file if in_name is /blah/blah/blah/file, then 1180 * backup file will be ".Bfile" then make the backup file the input and 1181 * original input file the output 1182 */ 1183 static void 1184 bakcopy(void) 1185 { 1186 int n, 1187 bakchn; 1188 char buff[8 * 1024]; 1189 const char *p; 1190 1191 /* construct file name .Bfile */ 1192 for (p = in_name; *p; p++); /* skip to end of string */ 1193 while (p > in_name && *p != '/') /* find last '/' */ 1194 p--; 1195 if (*p == '/') 1196 p++; 1197 sprintf(bakfile, "%s.BAK", p); 1198 1199 /* copy in_name to backup file */ 1200 bakchn = creat(bakfile, 0600); 1201 if (bakchn < 0) 1202 err(1, "%s", bakfile); 1203 while ((n = read(fileno(input), buff, sizeof buff)) != 0) 1204 if (write(bakchn, buff, n) != n) 1205 err(1, "%s", bakfile); 1206 if (n < 0) 1207 err(1, "%s", in_name); 1208 close(bakchn); 1209 fclose(input); 1210 1211 /* re-open backup file as the input file */ 1212 input = fopen(bakfile, "r"); 1213 if (input == NULL) 1214 err(1, "%s", bakfile); 1215 /* now the original input file will be the output */ 1216 output = fopen(in_name, "w"); 1217 if (output == NULL) { 1218 unlink(bakfile); 1219 err(1, "%s", in_name); 1220 } 1221 } 1222