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 case comment: /* we have a comment, so we must copy it into 326 * the buffer */ 327 if (!flushed_nl || sc_end != NULL) { 328 if (sc_end == NULL) { /* if this is the first comment, we 329 * must set up the buffer */ 330 save_com[0] = save_com[1] = ' '; 331 sc_end = &(save_com[2]); 332 } 333 else { 334 *sc_end++ = '\n'; /* add newline between 335 * comments */ 336 *sc_end++ = ' '; 337 --line_no; 338 } 339 *sc_end++ = '/'; /* copy in start of comment */ 340 *sc_end++ = '*'; 341 342 for (;;) { /* loop until we get to the end of the comment */ 343 *sc_end = *buf_ptr++; 344 if (buf_ptr >= buf_end) 345 fill_buffer(); 346 347 if (*sc_end++ == '*' && *buf_ptr == '/') 348 break; /* we are at end of comment */ 349 350 if (sc_end >= &(save_com[sc_size])) { /* check for temp buffer 351 * overflow */ 352 diag2(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever"); 353 fflush(output); 354 exit(1); 355 } 356 } 357 *sc_end++ = '/'; /* add ending slash */ 358 if (++buf_ptr >= buf_end) /* get past / in buffer */ 359 fill_buffer(); 360 break; 361 } 362 default: /* it is the start of a normal statement */ 363 if (flushed_nl) /* if we flushed a newline, make sure it is 364 * put back */ 365 force_nl = true; 366 if ((type_code == sp_paren && *token == 'i' 367 && last_else && ps.else_if) 368 || (type_code == sp_nparen && *token == 'e' 369 && e_code != s_code && e_code[-1] == '}')) 370 force_nl = false; 371 372 if (sc_end == NULL) { /* ignore buffering if comment wasn't 373 * saved up */ 374 ps.search_brace = false; 375 goto check_type; 376 } 377 if (force_nl) { /* if we should insert a nl here, put it into 378 * the buffer */ 379 force_nl = false; 380 --line_no; /* this will be re-increased when the nl is 381 * read from the buffer */ 382 *sc_end++ = '\n'; 383 *sc_end++ = ' '; 384 if (verbose && !flushed_nl) /* print error msg if the line 385 * was not already broken */ 386 diag2(0, "Line broken"); 387 flushed_nl = false; 388 } 389 for (t_ptr = token; *t_ptr; ++t_ptr) 390 *sc_end++ = *t_ptr; /* copy token into temp buffer */ 391 ps.procname[0] = 0; 392 393 sw_buffer: 394 ps.search_brace = false; /* stop looking for start of 395 * stmt */ 396 bp_save = buf_ptr; /* save current input buffer */ 397 be_save = buf_end; 398 buf_ptr = save_com; /* fix so that subsequent calls to 399 * lexi will take tokens out of 400 * save_com */ 401 *sc_end++ = ' ';/* add trailing blank, just in case */ 402 buf_end = sc_end; 403 sc_end = NULL; 404 break; 405 } /* end of switch */ 406 if (type_code != 0) /* we must make this check, just in case there 407 * was an unexpected EOF */ 408 type_code = lexi(); /* read another token */ 409 /* if (ps.search_brace) ps.procname[0] = 0; */ 410 if ((is_procname = ps.procname[0]) && flushed_nl 411 && !procnames_start_line && ps.in_decl 412 && type_code == ident) 413 flushed_nl = 0; 414 } /* end of while (search_brace) */ 415 last_else = 0; 416 check_type: 417 if (type_code == 0) { /* we got eof */ 418 if (s_lab != e_lab || s_code != e_code 419 || s_com != e_com) /* must dump end of line */ 420 dump_line(); 421 if (ps.tos > 1) /* check for balanced braces */ 422 diag2(1, "Stuff missing from end of file"); 423 424 if (verbose) { 425 printf("There were %d output lines and %d comments\n", 426 ps.out_lines, ps.out_coms); 427 printf("(Lines with comments)/(Lines with code): %6.3f\n", 428 (1.0 * ps.com_lines) / code_lines); 429 } 430 fflush(output); 431 exit(found_err); 432 } 433 if ( 434 (type_code != comment) && 435 (type_code != newline) && 436 (type_code != preesc) && 437 (type_code != form_feed)) { 438 if (force_nl && 439 (type_code != semicolon) && 440 (type_code != lbrace || !btype_2)) { 441 /* we should force a broken line here */ 442 if (verbose && !flushed_nl) 443 diag2(0, "Line broken"); 444 flushed_nl = false; 445 dump_line(); 446 ps.want_blank = false; /* dont insert blank at line start */ 447 force_nl = false; 448 } 449 ps.in_stmt = true; /* turn on flag which causes an extra level of 450 * indentation. this is turned off by a ; or 451 * '}' */ 452 if (s_com != e_com) { /* the turkey has embedded a comment 453 * in a line. fix it */ 454 *e_code++ = ' '; 455 for (t_ptr = s_com; *t_ptr; ++t_ptr) { 456 CHECK_SIZE_CODE; 457 *e_code++ = *t_ptr; 458 } 459 *e_code++ = ' '; 460 *e_code = '\0'; /* null terminate code sect */ 461 ps.want_blank = false; 462 e_com = s_com; 463 } 464 } 465 else if (type_code != comment) /* preserve force_nl thru a comment */ 466 force_nl = false; /* cancel forced newline after newline, form 467 * feed, etc */ 468 469 470 471 /*-----------------------------------------------------*\ 472 | do switch on type of token scanned | 473 \*-----------------------------------------------------*/ 474 CHECK_SIZE_CODE; 475 switch (type_code) { /* now, decide what to do with the token */ 476 477 case form_feed: /* found a form feed in line */ 478 ps.use_ff = true; /* a form feed is treated much like a newline */ 479 dump_line(); 480 ps.want_blank = false; 481 break; 482 483 case newline: 484 if (ps.last_token != comma || ps.p_l_follow > 0 485 || !ps.leave_comma || ps.block_init || !break_comma || s_com != e_com) { 486 dump_line(); 487 ps.want_blank = false; 488 } 489 ++line_no; /* keep track of input line number */ 490 break; 491 492 case lparen: /* got a '(' or '[' */ 493 ++ps.p_l_follow; /* count parens to make Healy happy */ 494 if (ps.want_blank && *token != '[' && 495 (ps.last_token != ident || proc_calls_space 496 || (ps.its_a_keyword && (!ps.sizeof_keyword || Bill_Shannon)))) 497 *e_code++ = ' '; 498 if (ps.in_decl && !ps.block_init) 499 if (troff && !ps.dumped_decl_indent && !is_procname && ps.last_token == decl) { 500 ps.dumped_decl_indent = 1; 501 sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token); 502 e_code += strlen(e_code); 503 } 504 else { 505 while ((e_code - s_code) < dec_ind) { 506 CHECK_SIZE_CODE; 507 *e_code++ = ' '; 508 } 509 *e_code++ = token[0]; 510 } 511 else 512 *e_code++ = token[0]; 513 ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code; 514 if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent 515 && ps.paren_indents[0] < 2 * ps.ind_size) 516 ps.paren_indents[0] = 2 * ps.ind_size; 517 ps.want_blank = false; 518 if (ps.in_or_st && *token == '(' && ps.tos <= 2) { 519 /* 520 * this is a kluge to make sure that declarations will be 521 * aligned right if proc decl has an explicit type on it, i.e. 522 * "int a(x) {..." 523 */ 524 parse(semicolon); /* I said this was a kluge... */ 525 ps.in_or_st = false; /* turn off flag for structure decl or 526 * initialization */ 527 } 528 if (ps.sizeof_keyword) 529 ps.sizeof_mask |= 1 << ps.p_l_follow; 530 break; 531 532 case rparen: /* got a ')' or ']' */ 533 rparen_count--; 534 if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.sizeof_mask) { 535 ps.last_u_d = true; 536 ps.cast_mask &= (1 << ps.p_l_follow) - 1; 537 ps.want_blank = false; 538 } else 539 ps.want_blank = true; 540 ps.sizeof_mask &= (1 << ps.p_l_follow) - 1; 541 if (--ps.p_l_follow < 0) { 542 ps.p_l_follow = 0; 543 diag3(0, "Extra %c", *token); 544 } 545 if (e_code == s_code) /* if the paren starts the line */ 546 ps.paren_level = ps.p_l_follow; /* then indent it */ 547 548 *e_code++ = token[0]; 549 550 if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if 551 * (...), or some such */ 552 sp_sw = false; 553 force_nl = true;/* must force newline after if */ 554 ps.last_u_d = true; /* inform lexi that a following 555 * operator is unary */ 556 ps.in_stmt = false; /* dont use stmt continuation 557 * indentation */ 558 559 parse(hd_type); /* let parser worry about if, or whatever */ 560 } 561 ps.search_brace = btype_2; /* this should insure that constructs 562 * such as main(){...} and int[]{...} 563 * have their braces put in the right 564 * place */ 565 break; 566 567 case unary_op: /* this could be any unary operation */ 568 if (ps.want_blank) 569 *e_code++ = ' '; 570 571 if (troff && !ps.dumped_decl_indent && ps.in_decl && !is_procname) { 572 sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token); 573 ps.dumped_decl_indent = 1; 574 e_code += strlen(e_code); 575 } 576 else { 577 const char *res = token; 578 579 if (ps.in_decl && !ps.block_init) { /* if this is a unary op 580 * in a declaration, we 581 * should indent this 582 * token */ 583 for (i = 0; token[i]; ++i); /* find length of token */ 584 while ((e_code - s_code) < (dec_ind - i)) { 585 CHECK_SIZE_CODE; 586 *e_code++ = ' '; /* pad it */ 587 } 588 } 589 if (troff && token[0] == '-' && token[1] == '>') 590 res = "\\(->"; 591 for (t_ptr = res; *t_ptr; ++t_ptr) { 592 CHECK_SIZE_CODE; 593 *e_code++ = *t_ptr; 594 } 595 } 596 ps.want_blank = false; 597 break; 598 599 case binary_op: /* any binary operation */ 600 if (ps.want_blank) 601 *e_code++ = ' '; 602 { 603 const char *res = token; 604 605 if (troff) 606 switch (token[0]) { 607 case '<': 608 if (token[1] == '=') 609 res = "\\(<="; 610 break; 611 case '>': 612 if (token[1] == '=') 613 res = "\\(>="; 614 break; 615 case '!': 616 if (token[1] == '=') 617 res = "\\(!="; 618 break; 619 case '|': 620 if (token[1] == '|') 621 res = "\\(br\\(br"; 622 else if (token[1] == 0) 623 res = "\\(br"; 624 break; 625 } 626 for (t_ptr = res; *t_ptr; ++t_ptr) { 627 CHECK_SIZE_CODE; 628 *e_code++ = *t_ptr; /* move the operator */ 629 } 630 } 631 ps.want_blank = true; 632 break; 633 634 case postop: /* got a trailing ++ or -- */ 635 *e_code++ = token[0]; 636 *e_code++ = token[1]; 637 ps.want_blank = true; 638 break; 639 640 case question: /* got a ? */ 641 squest++; /* this will be used when a later colon 642 * appears so we can distinguish the 643 * <c>?<n>:<n> construct */ 644 if (ps.want_blank) 645 *e_code++ = ' '; 646 *e_code++ = '?'; 647 ps.want_blank = true; 648 break; 649 650 case casestmt: /* got word 'case' or 'default' */ 651 scase = true; /* so we can process the later colon properly */ 652 goto copy_id; 653 654 case colon: /* got a ':' */ 655 if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */ 656 --squest; 657 if (ps.want_blank) 658 *e_code++ = ' '; 659 *e_code++ = ':'; 660 ps.want_blank = true; 661 break; 662 } 663 if (ps.in_or_st) { 664 *e_code++ = ':'; 665 ps.want_blank = false; 666 break; 667 } 668 ps.in_stmt = false; /* seeing a label does not imply we are in a 669 * stmt */ 670 for (t_ptr = s_code; *t_ptr; ++t_ptr) 671 *e_lab++ = *t_ptr; /* turn everything so far into a label */ 672 e_code = s_code; 673 *e_lab++ = ':'; 674 *e_lab++ = ' '; 675 *e_lab = '\0'; 676 677 force_nl = ps.pcase = scase; /* ps.pcase will be used by 678 * dump_line to decide how to 679 * indent the label. force_nl 680 * will force a case n: to be 681 * on a line by itself */ 682 scase = false; 683 ps.want_blank = false; 684 break; 685 686 case semicolon: /* got a ';' */ 687 ps.in_or_st = false;/* we are not in an initialization or 688 * structure declaration */ 689 scase = false; /* these will only need resetting in an error */ 690 squest = 0; 691 if (ps.last_token == rparen && rparen_count == 0) 692 ps.in_parameter_declaration = 0; 693 ps.cast_mask = 0; 694 ps.sizeof_mask = 0; 695 ps.block_init = 0; 696 ps.block_init_level = 0; 697 ps.just_saw_decl--; 698 699 if (ps.in_decl && s_code == e_code && !ps.block_init) 700 while ((e_code - s_code) < (dec_ind - 1)) { 701 CHECK_SIZE_CODE; 702 *e_code++ = ' '; 703 } 704 705 ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level 706 * structure declaration, we 707 * arent any more */ 708 709 if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) { 710 711 /* 712 * This should be true iff there were unbalanced parens in the 713 * stmt. It is a bit complicated, because the semicolon might 714 * be in a for stmt 715 */ 716 diag2(1, "Unbalanced parens"); 717 ps.p_l_follow = 0; 718 if (sp_sw) { /* this is a check for an if, while, etc. with 719 * unbalanced parens */ 720 sp_sw = false; 721 parse(hd_type); /* dont lose the if, or whatever */ 722 } 723 } 724 *e_code++ = ';'; 725 ps.want_blank = true; 726 ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the 727 * middle of a stmt */ 728 729 if (!sp_sw) { /* if not if for (;;) */ 730 parse(semicolon); /* let parser know about end of stmt */ 731 force_nl = true;/* force newline after an end of stmt */ 732 } 733 break; 734 735 case lbrace: /* got a '{' */ 736 ps.in_stmt = false; /* dont indent the {} */ 737 if (!ps.block_init) 738 force_nl = true;/* force other stuff on same line as '{' onto 739 * new line */ 740 else if (ps.block_init_level <= 0) 741 ps.block_init_level = 1; 742 else 743 ps.block_init_level++; 744 745 if (s_code != e_code && !ps.block_init) { 746 if (!btype_2) { 747 dump_line(); 748 ps.want_blank = false; 749 } 750 else if (ps.in_parameter_declaration && !ps.in_or_st) { 751 ps.i_l_follow = 0; 752 if (function_brace_split) { /* dump the line prior to the 753 * brace ... */ 754 dump_line(); 755 ps.want_blank = false; 756 } else /* add a space between the decl and brace */ 757 ps.want_blank = true; 758 } 759 } 760 if (ps.in_parameter_declaration) 761 prefix_blankline_requested = 0; 762 763 if (ps.p_l_follow > 0) { /* check for preceding unbalanced 764 * parens */ 765 diag2(1, "Unbalanced parens"); 766 ps.p_l_follow = 0; 767 if (sp_sw) { /* check for unclosed if, for, etc. */ 768 sp_sw = false; 769 parse(hd_type); 770 ps.ind_level = ps.i_l_follow; 771 } 772 } 773 if (s_code == e_code) 774 ps.ind_stmt = false; /* dont put extra indentation on line 775 * with '{' */ 776 if (ps.in_decl && ps.in_or_st) { /* this is either a structure 777 * declaration or an init */ 778 di_stack[ps.dec_nest++] = dec_ind; 779 /* ? dec_ind = 0; */ 780 } 781 else { 782 ps.decl_on_line = false; /* we can't be in the middle of 783 * a declaration, so don't do 784 * special indentation of 785 * comments */ 786 if (blanklines_after_declarations_at_proctop 787 && ps.in_parameter_declaration) 788 postfix_blankline_requested = 1; 789 ps.in_parameter_declaration = 0; 790 } 791 dec_ind = 0; 792 parse(lbrace); /* let parser know about this */ 793 if (ps.want_blank) /* put a blank before '{' if '{' is not at 794 * start of line */ 795 *e_code++ = ' '; 796 ps.want_blank = false; 797 *e_code++ = '{'; 798 ps.just_saw_decl = 0; 799 break; 800 801 case rbrace: /* got a '}' */ 802 if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be 803 * omitted in 804 * declarations */ 805 parse(semicolon); 806 if (ps.p_l_follow) {/* check for unclosed if, for, else. */ 807 diag2(1, "Unbalanced parens"); 808 ps.p_l_follow = 0; 809 sp_sw = false; 810 } 811 ps.just_saw_decl = 0; 812 ps.block_init_level--; 813 if (s_code != e_code && !ps.block_init) { /* '}' must be first on 814 * line */ 815 if (verbose) 816 diag2(0, "Line broken"); 817 dump_line(); 818 } 819 *e_code++ = '}'; 820 ps.want_blank = true; 821 ps.in_stmt = ps.ind_stmt = false; 822 if (ps.dec_nest > 0) { /* we are in multi-level structure 823 * declaration */ 824 dec_ind = di_stack[--ps.dec_nest]; 825 if (ps.dec_nest == 0 && !ps.in_parameter_declaration) 826 ps.just_saw_decl = 2; 827 ps.in_decl = true; 828 } 829 prefix_blankline_requested = 0; 830 parse(rbrace); /* let parser know about this */ 831 ps.search_brace = cuddle_else && ps.p_stack[ps.tos] == ifhead 832 && ps.il[ps.tos] >= ps.ind_level; 833 if (ps.tos <= 1 && blanklines_after_procs && ps.dec_nest <= 0) 834 postfix_blankline_requested = 1; 835 break; 836 837 case swstmt: /* got keyword "switch" */ 838 sp_sw = true; 839 hd_type = swstmt; /* keep this for when we have seen the 840 * expression */ 841 goto copy_id; /* go move the token into buffer */ 842 843 case sp_paren: /* token is if, while, for */ 844 sp_sw = true; /* the interesting stuff is done after the 845 * expression is scanned */ 846 hd_type = (*token == 'i' ? ifstmt : 847 (*token == 'w' ? whilestmt : forstmt)); 848 849 /* 850 * remember the type of header for later use by parser 851 */ 852 goto copy_id; /* copy the token into line */ 853 854 case sp_nparen: /* got else, do */ 855 ps.in_stmt = false; 856 if (*token == 'e') { 857 if (e_code != s_code && (!cuddle_else || e_code[-1] != '}')) { 858 if (verbose) 859 diag2(0, "Line broken"); 860 dump_line();/* make sure this starts a line */ 861 ps.want_blank = false; 862 } 863 force_nl = true;/* also, following stuff must go onto new line */ 864 last_else = 1; 865 parse(elselit); 866 } 867 else { 868 if (e_code != s_code) { /* make sure this starts a line */ 869 if (verbose) 870 diag2(0, "Line broken"); 871 dump_line(); 872 ps.want_blank = false; 873 } 874 force_nl = true;/* also, following stuff must go onto new line */ 875 last_else = 0; 876 parse(dolit); 877 } 878 goto copy_id; /* move the token into line */ 879 880 case decl: /* we have a declaration type (int, register, 881 * etc.) */ 882 parse(decl); /* let parser worry about indentation */ 883 if (ps.last_token == rparen && ps.tos <= 1) { 884 ps.in_parameter_declaration = 1; 885 if (s_code != e_code) { 886 dump_line(); 887 ps.want_blank = 0; 888 } 889 } 890 if (ps.in_parameter_declaration && ps.indent_parameters && ps.dec_nest == 0) { 891 ps.ind_level = ps.i_l_follow = 1; 892 ps.ind_stmt = 0; 893 } 894 ps.in_or_st = true; /* this might be a structure or initialization 895 * declaration */ 896 ps.in_decl = ps.decl_on_line = true; 897 if ( /* !ps.in_or_st && */ ps.dec_nest <= 0) 898 ps.just_saw_decl = 2; 899 prefix_blankline_requested = 0; 900 for (i = 0; token[i++];); /* get length of token */ 901 902 if (ps.ind_level == 0 || ps.dec_nest > 0) { 903 /* global variable or struct member in local variable */ 904 dec_ind = ps.decl_indent > 0 ? ps.decl_indent : i; 905 tabs_to_var = (use_tabs ? ps.decl_indent > 0 : 0); 906 } else { 907 /* local variable */ 908 dec_ind = ps.local_decl_indent > 0 ? ps.local_decl_indent : i; 909 tabs_to_var = (use_tabs ? ps.local_decl_indent > 0 : 0); 910 } 911 goto copy_id; 912 913 case ident: /* got an identifier or constant */ 914 if (ps.in_decl) { /* if we are in a declaration, we must indent 915 * identifier */ 916 if (is_procname == 0 || !procnames_start_line) { 917 if (!ps.block_init) { 918 if (troff && !ps.dumped_decl_indent) { 919 if (ps.want_blank) 920 *e_code++ = ' '; 921 ps.want_blank = false; 922 sprintf(e_code, "\n.De %dp+\200p\n", dec_ind * 7); 923 ps.dumped_decl_indent = 1; 924 e_code += strlen(e_code); 925 } else { 926 int cur_dec_ind; 927 int pos, startpos; 928 929 /* 930 * in order to get the tab math right for 931 * indentations that are not multiples of 8 we 932 * need to modify both startpos and dec_ind 933 * (cur_dec_ind) here by eight minus the 934 * remainder of the current starting column 935 * divided by eight. This seems to be a 936 * properly working fix 937 */ 938 startpos = e_code - s_code; 939 cur_dec_ind = dec_ind; 940 pos = startpos; 941 if ((ps.ind_level * ps.ind_size) % 8 != 0) { 942 pos += (ps.ind_level * ps.ind_size) % 8; 943 cur_dec_ind += (ps.ind_level * ps.ind_size) % 8; 944 } 945 946 if (tabs_to_var) { 947 while ((pos & ~7) + 8 <= cur_dec_ind) { 948 CHECK_SIZE_CODE; 949 *e_code++ = '\t'; 950 pos = (pos & ~7) + 8; 951 } 952 } 953 while (pos < cur_dec_ind) { 954 CHECK_SIZE_CODE; 955 *e_code++ = ' '; 956 pos++; 957 } 958 if (ps.want_blank && e_code - s_code == startpos) 959 *e_code++ = ' '; 960 ps.want_blank = false; 961 } 962 } 963 } else { 964 if (ps.want_blank) 965 *e_code++ = ' '; 966 ps.want_blank = false; 967 if (dec_ind && s_code != e_code) 968 dump_line(); 969 dec_ind = 0; 970 } 971 } 972 else if (sp_sw && ps.p_l_follow == 0) { 973 sp_sw = false; 974 force_nl = true; 975 ps.last_u_d = true; 976 ps.in_stmt = false; 977 parse(hd_type); 978 } 979 copy_id: 980 if (ps.want_blank) 981 *e_code++ = ' '; 982 if (troff && ps.its_a_keyword) { 983 e_code = chfont(&bodyf, &keywordf, e_code); 984 for (t_ptr = token; *t_ptr; ++t_ptr) { 985 CHECK_SIZE_CODE; 986 *e_code++ = keywordf.allcaps && islower(*t_ptr) 987 ? toupper(*t_ptr) : *t_ptr; 988 } 989 e_code = chfont(&keywordf, &bodyf, e_code); 990 } 991 else 992 for (t_ptr = token; *t_ptr; ++t_ptr) { 993 CHECK_SIZE_CODE; 994 *e_code++ = *t_ptr; 995 } 996 ps.want_blank = true; 997 break; 998 999 case period: /* treat a period kind of like a binary 1000 * operation */ 1001 *e_code++ = '.'; /* move the period into line */ 1002 ps.want_blank = false; /* dont put a blank after a period */ 1003 break; 1004 1005 case comma: 1006 ps.want_blank = (s_code != e_code); /* only put blank after comma 1007 * if comma does not start the 1008 * line */ 1009 if (ps.in_decl && is_procname == 0 && !ps.block_init) 1010 while ((e_code - s_code) < (dec_ind - 1)) { 1011 CHECK_SIZE_CODE; 1012 *e_code++ = ' '; 1013 } 1014 1015 *e_code++ = ','; 1016 if (ps.p_l_follow == 0) { 1017 if (ps.block_init_level <= 0) 1018 ps.block_init = 0; 1019 if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8)) 1020 force_nl = true; 1021 } 1022 break; 1023 1024 case preesc: /* got the character '#' */ 1025 if ((s_com != e_com) || 1026 (s_lab != e_lab) || 1027 (s_code != e_code)) 1028 dump_line(); 1029 *e_lab++ = '#'; /* move whole line to 'label' buffer */ 1030 { 1031 int in_comment = 0; 1032 int com_start = 0; 1033 char quote = 0; 1034 int com_end = 0; 1035 1036 while (*buf_ptr == ' ' || *buf_ptr == '\t') { 1037 buf_ptr++; 1038 if (buf_ptr >= buf_end) 1039 fill_buffer(); 1040 } 1041 while (*buf_ptr != '\n' || (in_comment && !had_eof)) { 1042 CHECK_SIZE_LAB; 1043 *e_lab = *buf_ptr++; 1044 if (buf_ptr >= buf_end) 1045 fill_buffer(); 1046 switch (*e_lab++) { 1047 case BACKSLASH: 1048 if (troff) 1049 *e_lab++ = BACKSLASH; 1050 if (!in_comment) { 1051 *e_lab++ = *buf_ptr++; 1052 if (buf_ptr >= buf_end) 1053 fill_buffer(); 1054 } 1055 break; 1056 case '/': 1057 if (*buf_ptr == '*' && !in_comment && !quote) { 1058 in_comment = 1; 1059 *e_lab++ = *buf_ptr++; 1060 com_start = e_lab - s_lab - 2; 1061 } 1062 break; 1063 case '"': 1064 if (quote == '"') 1065 quote = 0; 1066 break; 1067 case '\'': 1068 if (quote == '\'') 1069 quote = 0; 1070 break; 1071 case '*': 1072 if (*buf_ptr == '/' && in_comment) { 1073 in_comment = 0; 1074 *e_lab++ = *buf_ptr++; 1075 com_end = e_lab - s_lab; 1076 } 1077 break; 1078 } 1079 } 1080 1081 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1082 e_lab--; 1083 if (e_lab - s_lab == com_end && bp_save == NULL) { 1084 /* comment on preprocessor line */ 1085 if (sc_end == NULL) /* if this is the first comment, we 1086 * must set up the buffer */ 1087 sc_end = &(save_com[0]); 1088 else { 1089 *sc_end++ = '\n'; /* add newline between 1090 * comments */ 1091 *sc_end++ = ' '; 1092 --line_no; 1093 } 1094 bcopy(s_lab + com_start, sc_end, com_end - com_start); 1095 sc_end += com_end - com_start; 1096 if (sc_end >= &save_com[sc_size]) 1097 abort(); 1098 e_lab = s_lab + com_start; 1099 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1100 e_lab--; 1101 bp_save = buf_ptr; /* save current input buffer */ 1102 be_save = buf_end; 1103 buf_ptr = save_com; /* fix so that subsequent calls to 1104 * lexi will take tokens out of 1105 * save_com */ 1106 *sc_end++ = ' '; /* add trailing blank, just in case */ 1107 buf_end = sc_end; 1108 sc_end = NULL; 1109 } 1110 *e_lab = '\0'; /* null terminate line */ 1111 ps.pcase = false; 1112 } 1113 1114 if (strncmp(s_lab, "#if", 3) == 0) { 1115 if (blanklines_around_conditional_compilation) { 1116 int c; 1117 prefix_blankline_requested++; 1118 while ((c = getc(input)) == '\n'); 1119 ungetc(c, input); 1120 } 1121 if ((size_t)ifdef_level < sizeof(state_stack)/sizeof(state_stack[0])) { 1122 match_state[ifdef_level].tos = -1; 1123 state_stack[ifdef_level++] = ps; 1124 } 1125 else 1126 diag2(1, "#if stack overflow"); 1127 } 1128 else if (strncmp(s_lab, "#else", 5) == 0) 1129 if (ifdef_level <= 0) 1130 diag2(1, "Unmatched #else"); 1131 else { 1132 match_state[ifdef_level - 1] = ps; 1133 ps = state_stack[ifdef_level - 1]; 1134 } 1135 else if (strncmp(s_lab, "#endif", 6) == 0) { 1136 if (ifdef_level <= 0) 1137 diag2(1, "Unmatched #endif"); 1138 else { 1139 ifdef_level--; 1140 1141 #ifdef undef 1142 /* 1143 * This match needs to be more intelligent before the 1144 * message is useful 1145 */ 1146 if (match_state[ifdef_level].tos >= 0 1147 && bcmp(&ps, &match_state[ifdef_level], sizeof ps)) 1148 diag2(0, "Syntactically inconsistent #ifdef alternatives"); 1149 #endif 1150 } 1151 if (blanklines_around_conditional_compilation) { 1152 postfix_blankline_requested++; 1153 n_real_blanklines = 0; 1154 } 1155 } 1156 break; /* subsequent processing of the newline 1157 * character will cause the line to be printed */ 1158 1159 case comment: /* we have gotten a / followed by * this is a biggie */ 1160 if (flushed_nl) { /* we should force a broken line here */ 1161 flushed_nl = false; 1162 dump_line(); 1163 ps.want_blank = false; /* dont insert blank at line start */ 1164 force_nl = false; 1165 } 1166 pr_comment(); 1167 break; 1168 } /* end of big switch stmt */ 1169 1170 *e_code = '\0'; /* make sure code section is null terminated */ 1171 if (type_code != comment && type_code != newline && type_code != preesc) 1172 ps.last_token = type_code; 1173 } /* end of main while (1) loop */ 1174 } 1175 1176 /* 1177 * copy input file to backup file if in_name is /blah/blah/blah/file, then 1178 * backup file will be ".Bfile" then make the backup file the input and 1179 * original input file the output 1180 */ 1181 static void 1182 bakcopy(void) 1183 { 1184 int n, 1185 bakchn; 1186 char buff[8 * 1024]; 1187 const char *p; 1188 1189 /* construct file name .Bfile */ 1190 for (p = in_name; *p; p++); /* skip to end of string */ 1191 while (p > in_name && *p != '/') /* find last '/' */ 1192 p--; 1193 if (*p == '/') 1194 p++; 1195 sprintf(bakfile, "%s.BAK", p); 1196 1197 /* copy in_name to backup file */ 1198 bakchn = creat(bakfile, 0600); 1199 if (bakchn < 0) 1200 err(1, "%s", bakfile); 1201 while ((n = read(fileno(input), buff, sizeof buff)) != 0) 1202 if (write(bakchn, buff, n) != n) 1203 err(1, "%s", bakfile); 1204 if (n < 0) 1205 err(1, "%s", in_name); 1206 close(bakchn); 1207 fclose(input); 1208 1209 /* re-open backup file as the input file */ 1210 input = fopen(bakfile, "r"); 1211 if (input == NULL) 1212 err(1, "%s", bakfile); 1213 /* now the original input file will be the output */ 1214 output = fopen(in_name, "w"); 1215 if (output == NULL) { 1216 unlink(bakfile); 1217 err(1, "%s", in_name); 1218 } 1219 } 1220