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 * @(#) Copyright (c) 1985 Sun Microsystems, Inc. 36 * @(#) Copyright (c) 1976 Board of Trustees of the University of Illinois. 37 * @(#) Copyright (c) 1980, 1993 The Regents of the University of California. All rights reserved. 38 * @(#)indent.c 5.17 (Berkeley) 6/7/93 39 * $FreeBSD: src/usr.bin/indent/indent.c,v 1.26 2010/03/31 16:55:47 avg Exp $ 40 */ 41 42 #include <sys/param.h> 43 #include <err.h> 44 #include <fcntl.h> 45 #include <unistd.h> 46 #include <stdio.h> 47 #include <stdlib.h> 48 #include <string.h> 49 #include <ctype.h> 50 #include "indent_globs.h" 51 #include "indent_codes.h" 52 #include "indent.h" 53 54 static void bakcopy(void); 55 56 const char *in_name = "Standard Input"; /* will always point to name of input 57 * file */ 58 const char *out_name = "Standard Output"; /* will always point to name 59 * of output file */ 60 char bakfile[MAXPATHLEN] = ""; 61 62 int 63 main(int argc, char **argv) 64 { 65 66 int dec_ind; /* current indentation for declarations */ 67 int di_stack[20]; /* a stack of structure indentation levels */ 68 int flushed_nl; /* used when buffering up comments to remember 69 * that a newline was passed over */ 70 int force_nl; /* when true, code must be broken */ 71 int hd_type = 0; /* used to store type of stmt for if (...), 72 * for (...), etc */ 73 int i; /* local loop counter */ 74 int scase; /* set to true when we see a case, so we will 75 * know what to do with the following colon */ 76 int sp_sw; /* when true, we are in the expression of 77 * if(...), while(...), etc. */ 78 int squest; /* when this is positive, we have seen a ? 79 * without the matching : in a <c>?<s>:<s> 80 * construct */ 81 const char *t_ptr; /* used for copying tokens */ 82 int tabs_to_var; /* true if using tabs to indent to var name */ 83 int type_code; /* the type of token, returned by lexi */ 84 85 int last_else = 0; /* true iff last keyword was an else */ 86 87 88 /*-----------------------------------------------*\ 89 | INITIALIZATION | 90 \*-----------------------------------------------*/ 91 92 found_err = 0; 93 94 ps.p_stack[0] = stmt; /* this is the parser's stack */ 95 ps.last_nl = true; /* this is true if the last thing scanned was 96 * a newline */ 97 ps.last_token = semicolon; 98 combuf = (char *) malloc(bufsize); 99 if (combuf == NULL) 100 err(1, NULL); 101 labbuf = (char *) malloc(bufsize); 102 if (labbuf == NULL) 103 err(1, NULL); 104 codebuf = (char *) malloc(bufsize); 105 if (codebuf == NULL) 106 err(1, NULL); 107 tokenbuf = (char *) malloc(bufsize); 108 if (tokenbuf == NULL) 109 err(1, NULL); 110 l_com = combuf + bufsize - 5; 111 l_lab = labbuf + bufsize - 5; 112 l_code = codebuf + bufsize - 5; 113 l_token = tokenbuf + bufsize - 5; 114 combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and 115 * comment buffers */ 116 combuf[1] = codebuf[1] = labbuf[1] = '\0'; 117 ps.else_if = 1; /* Default else-if special processing to on */ 118 s_lab = e_lab = labbuf + 1; 119 s_code = e_code = codebuf + 1; 120 s_com = e_com = combuf + 1; 121 s_token = e_token = tokenbuf + 1; 122 123 in_buffer = (char *) malloc(10); 124 if (in_buffer == NULL) 125 err(1, NULL); 126 in_buffer_limit = in_buffer + 8; 127 buf_ptr = buf_end = in_buffer; 128 line_no = 1; 129 had_eof = ps.in_decl = ps.decl_on_line = break_comma = false; 130 sp_sw = force_nl = false; 131 ps.in_or_st = false; 132 ps.bl_line = true; 133 dec_ind = 0; 134 di_stack[ps.dec_nest = 0] = 0; 135 ps.want_blank = ps.in_stmt = ps.ind_stmt = false; 136 137 scase = ps.pcase = false; 138 squest = 0; 139 sc_end = 0; 140 bp_save = 0; 141 be_save = 0; 142 143 output = 0; 144 tabs_to_var = 0; 145 146 /*--------------------------------------------------*\ 147 | COMMAND LINE SCAN | 148 \*--------------------------------------------------*/ 149 150 #ifdef undef 151 max_col = 78; /* -l78 */ 152 lineup_to_parens = 1; /* -lp */ 153 ps.ljust_decl = 0; /* -ndj */ 154 ps.com_ind = 33; /* -c33 */ 155 star_comment_cont = 1; /* -sc */ 156 ps.ind_size = 8; /* -i8 */ 157 verbose = 0; 158 ps.decl_indent = 16; /* -di16 */ 159 ps.local_decl_indent = -1; /* if this is not set to some nonnegative value 160 * by an arg, we will set this equal to 161 * ps.decl_ind */ 162 ps.indent_parameters = 1; /* -ip */ 163 ps.decl_com_ind = 0; /* if this is not set to some positive value 164 * by an arg, we will set this equal to 165 * ps.com_ind */ 166 btype_2 = 1; /* -br */ 167 cuddle_else = 1; /* -ce */ 168 ps.unindent_displace = 0; /* -d0 */ 169 ps.case_indent = 0; /* -cli0 */ 170 format_block_comments = 1; /* -fcb */ 171 format_col1_comments = 1; /* -fc1 */ 172 procnames_start_line = 1; /* -psl */ 173 proc_calls_space = 0; /* -npcs */ 174 comment_delimiter_on_blankline = 1; /* -cdb */ 175 ps.leave_comma = 1; /* -nbc */ 176 #endif 177 178 for (i = 1; i < argc; ++i) 179 if (strcmp(argv[i], "-npro") == 0) 180 break; 181 set_defaults(); 182 if (i >= argc) 183 set_profile(); 184 185 for (i = 1; i < argc; ++i) { 186 187 /* 188 * look thru args (if any) for changes to defaults 189 */ 190 if (argv[i][0] != '-') {/* no flag on parameter */ 191 if (input == 0) { /* we must have the input file */ 192 in_name = argv[i]; /* remember name of input file */ 193 input = fopen(in_name, "r"); 194 if (input == 0) /* check for open error */ 195 err(1, "%s", in_name); 196 continue; 197 } 198 else if (output == 0) { /* we have the output file */ 199 out_name = argv[i]; /* remember name of output file */ 200 if (strcmp(in_name, out_name) == 0) { /* attempt to overwrite 201 * the file */ 202 errx(1, "input and output files must be different"); 203 } 204 output = fopen(out_name, "w"); 205 if (output == 0) /* check for create error */ 206 err(1, "%s", out_name); 207 continue; 208 } 209 errx(1, "unknown parameter: %s", argv[i]); 210 } 211 else 212 set_option(argv[i]); 213 } /* end of for */ 214 if (input == 0) 215 input = stdin; 216 if (output == 0) { 217 if (troff || input == stdin) 218 output = stdout; 219 else { 220 out_name = in_name; 221 bakcopy(); 222 } 223 } 224 if (ps.com_ind <= 1) 225 ps.com_ind = 2; /* dont put normal comments before column 2 */ 226 if (troff) { 227 if (bodyf.font[0] == 0) 228 parsefont(&bodyf, "R"); 229 if (scomf.font[0] == 0) 230 parsefont(&scomf, "I"); 231 if (blkcomf.font[0] == 0) 232 blkcomf = scomf, blkcomf.size += 2; 233 if (boxcomf.font[0] == 0) 234 boxcomf = blkcomf; 235 if (stringf.font[0] == 0) 236 parsefont(&stringf, "L"); 237 if (keywordf.font[0] == 0) 238 parsefont(&keywordf, "B"); 239 writefdef(&bodyf, 'B'); 240 writefdef(&scomf, 'C'); 241 writefdef(&blkcomf, 'L'); 242 writefdef(&boxcomf, 'X'); 243 writefdef(&stringf, 'S'); 244 writefdef(&keywordf, 'K'); 245 } 246 if (block_comment_max_col <= 0) 247 block_comment_max_col = max_col; 248 if (ps.local_decl_indent < 0) /* if not specified by user, set this */ 249 ps.local_decl_indent = ps.decl_indent; 250 if (ps.decl_com_ind <= 0) /* if not specified by user, set this */ 251 ps.decl_com_ind = ps.ljust_decl ? (ps.com_ind <= 10 ? 2 : ps.com_ind - 8) : ps.com_ind; 252 if (continuation_indent == 0) 253 continuation_indent = ps.ind_size; 254 fill_buffer(); /* get first batch of stuff into input buffer */ 255 256 parse(semicolon); 257 { 258 char *p = buf_ptr; 259 int col = 1; 260 261 while (1) { 262 if (*p == ' ') 263 col++; 264 else if (*p == '\t') 265 col = ((col - 1) & ~7) + 9; 266 else 267 break; 268 p++; 269 } 270 if (col > ps.ind_size) 271 ps.ind_level = ps.i_l_follow = col / ps.ind_size; 272 } 273 if (troff) { 274 const char *p = in_name, 275 *beg = in_name; 276 277 while (*p) 278 if (*p++ == '/') 279 beg = p; 280 fprintf(output, ".Fn \"%s\"\n", beg); 281 } 282 /* 283 * START OF MAIN LOOP 284 */ 285 286 while (1) { /* this is the main loop. it will go until we 287 * reach eof */ 288 int is_procname; 289 290 type_code = lexi(); /* lexi reads one token. The actual 291 * characters read are stored in "token". lexi 292 * returns a code indicating the type of token */ 293 is_procname = ps.procname[0]; 294 295 /* 296 * The following code moves everything following an if (), while (), 297 * else, etc. up to the start of the following stmt to a buffer. This 298 * allows proper handling of both kinds of brace placement. 299 */ 300 301 flushed_nl = false; 302 while (ps.search_brace) { /* if we scanned an if(), while(), 303 * etc., we might need to copy stuff 304 * into a buffer we must loop, copying 305 * stuff into save_com, until we find 306 * the start of the stmt which follows 307 * the if, or whatever */ 308 switch (type_code) { 309 case newline: 310 ++line_no; 311 flushed_nl = true; 312 case form_feed: 313 break; /* form feeds and newlines found here will be 314 * ignored */ 315 316 case lbrace: /* this is a brace that starts the compound 317 * stmt */ 318 if (sc_end == 0) { /* ignore buffering if a comment wasnt 319 * stored up */ 320 ps.search_brace = false; 321 goto check_type; 322 } 323 if (btype_2) { 324 save_com[0] = '{'; /* we either want to put the brace 325 * right after the if */ 326 goto sw_buffer; /* go to common code to get out of 327 * this loop */ 328 } 329 case comment: /* we have a comment, so we must copy it into 330 * the buffer */ 331 if (!flushed_nl || sc_end != 0) { 332 if (sc_end == 0) { /* if this is the first comment, we 333 * must set up the buffer */ 334 save_com[0] = save_com[1] = ' '; 335 sc_end = &(save_com[2]); 336 } 337 else { 338 *sc_end++ = '\n'; /* add newline between 339 * comments */ 340 *sc_end++ = ' '; 341 --line_no; 342 } 343 *sc_end++ = '/'; /* copy in start of comment */ 344 *sc_end++ = '*'; 345 346 for (;;) { /* loop until we get to the end of the comment */ 347 *sc_end = *buf_ptr++; 348 if (buf_ptr >= buf_end) 349 fill_buffer(); 350 351 if (*sc_end++ == '*' && *buf_ptr == '/') 352 break; /* we are at end of comment */ 353 354 if (sc_end >= &(save_com[sc_size])) { /* check for temp buffer 355 * overflow */ 356 diag2(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever"); 357 fflush(output); 358 exit(1); 359 } 360 } 361 *sc_end++ = '/'; /* add ending slash */ 362 if (++buf_ptr >= buf_end) /* get past / in buffer */ 363 fill_buffer(); 364 break; 365 } 366 default: /* it is the start of a normal statement */ 367 if (flushed_nl) /* if we flushed a newline, make sure it is 368 * put back */ 369 force_nl = true; 370 if ((type_code == sp_paren && *token == 'i' 371 && last_else && ps.else_if) 372 || (type_code == sp_nparen && *token == 'e' 373 && e_code != s_code && e_code[-1] == '}')) 374 force_nl = false; 375 376 if (sc_end == 0) { /* ignore buffering if comment wasnt 377 * saved up */ 378 ps.search_brace = false; 379 goto check_type; 380 } 381 if (force_nl) { /* if we should insert a nl here, put it into 382 * the buffer */ 383 force_nl = false; 384 --line_no; /* this will be re-increased when the nl is 385 * read from the buffer */ 386 *sc_end++ = '\n'; 387 *sc_end++ = ' '; 388 if (verbose && !flushed_nl) /* print error msg if the line 389 * was not already broken */ 390 diag2(0, "Line broken"); 391 flushed_nl = false; 392 } 393 for (t_ptr = token; *t_ptr; ++t_ptr) 394 *sc_end++ = *t_ptr; /* copy token into temp buffer */ 395 ps.procname[0] = 0; 396 397 sw_buffer: 398 ps.search_brace = false; /* stop looking for start of 399 * stmt */ 400 bp_save = buf_ptr; /* save current input buffer */ 401 be_save = buf_end; 402 buf_ptr = save_com; /* fix so that subsequent calls to 403 * lexi will take tokens out of 404 * save_com */ 405 *sc_end++ = ' ';/* add trailing blank, just in case */ 406 buf_end = sc_end; 407 sc_end = 0; 408 break; 409 } /* end of switch */ 410 if (type_code != 0) /* we must make this check, just in case there 411 * was an unexpected EOF */ 412 type_code = lexi(); /* read another token */ 413 /* if (ps.search_brace) ps.procname[0] = 0; */ 414 if ((is_procname = ps.procname[0]) && flushed_nl 415 && !procnames_start_line && ps.in_decl 416 && type_code == ident) 417 flushed_nl = 0; 418 } /* end of while (search_brace) */ 419 last_else = 0; 420 check_type: 421 if (type_code == 0) { /* we got eof */ 422 if (s_lab != e_lab || s_code != e_code 423 || s_com != e_com) /* must dump end of line */ 424 dump_line(); 425 if (ps.tos > 1) /* check for balanced braces */ 426 diag2(1, "Stuff missing from end of file"); 427 428 if (verbose) { 429 printf("There were %d output lines and %d comments\n", 430 ps.out_lines, ps.out_coms); 431 printf("(Lines with comments)/(Lines with code): %6.3f\n", 432 (1.0 * ps.com_lines) / code_lines); 433 } 434 fflush(output); 435 exit(found_err); 436 } 437 if ( 438 (type_code != comment) && 439 (type_code != newline) && 440 (type_code != preesc) && 441 (type_code != form_feed)) { 442 if (force_nl && 443 (type_code != semicolon) && 444 (type_code != lbrace || !btype_2)) { 445 /* we should force a broken line here */ 446 if (verbose && !flushed_nl) 447 diag2(0, "Line broken"); 448 flushed_nl = false; 449 dump_line(); 450 ps.want_blank = false; /* dont insert blank at line start */ 451 force_nl = false; 452 } 453 ps.in_stmt = true; /* turn on flag which causes an extra level of 454 * indentation. this is turned off by a ; or 455 * '}' */ 456 if (s_com != e_com) { /* the turkey has embedded a comment 457 * in a line. fix it */ 458 *e_code++ = ' '; 459 for (t_ptr = s_com; *t_ptr; ++t_ptr) { 460 CHECK_SIZE_CODE; 461 *e_code++ = *t_ptr; 462 } 463 *e_code++ = ' '; 464 *e_code = '\0'; /* null terminate code sect */ 465 ps.want_blank = false; 466 e_com = s_com; 467 } 468 } 469 else if (type_code != comment) /* preserve force_nl thru a comment */ 470 force_nl = false; /* cancel forced newline after newline, form 471 * feed, etc */ 472 473 474 475 /*-----------------------------------------------------*\ 476 | do switch on type of token scanned | 477 \*-----------------------------------------------------*/ 478 CHECK_SIZE_CODE; 479 switch (type_code) { /* now, decide what to do with the token */ 480 481 case form_feed: /* found a form feed in line */ 482 ps.use_ff = true; /* a form feed is treated much like a newline */ 483 dump_line(); 484 ps.want_blank = false; 485 break; 486 487 case newline: 488 if (ps.last_token != comma || ps.p_l_follow > 0 489 || !ps.leave_comma || ps.block_init || !break_comma || s_com != e_com) { 490 dump_line(); 491 ps.want_blank = false; 492 } 493 ++line_no; /* keep track of input line number */ 494 break; 495 496 case lparen: /* got a '(' or '[' */ 497 ++ps.p_l_follow; /* count parens to make Healy happy */ 498 if (ps.want_blank && *token != '[' && 499 (ps.last_token != ident || proc_calls_space 500 || (ps.its_a_keyword && (!ps.sizeof_keyword || Bill_Shannon)))) 501 *e_code++ = ' '; 502 if (ps.in_decl && !ps.block_init) 503 if (troff && !ps.dumped_decl_indent && !is_procname && ps.last_token == decl) { 504 ps.dumped_decl_indent = 1; 505 sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token); 506 e_code += strlen(e_code); 507 } 508 else { 509 while ((e_code - s_code) < dec_ind) { 510 CHECK_SIZE_CODE; 511 *e_code++ = ' '; 512 } 513 *e_code++ = token[0]; 514 } 515 else 516 *e_code++ = token[0]; 517 ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code; 518 if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent 519 && ps.paren_indents[0] < 2 * ps.ind_size) 520 ps.paren_indents[0] = 2 * ps.ind_size; 521 ps.want_blank = false; 522 if (ps.in_or_st && *token == '(' && ps.tos <= 2) { 523 /* 524 * this is a kluge to make sure that declarations will be 525 * aligned right if proc decl has an explicit type on it, i.e. 526 * "int a(x) {..." 527 */ 528 parse(semicolon); /* I said this was a kluge... */ 529 ps.in_or_st = false; /* turn off flag for structure decl or 530 * initialization */ 531 } 532 if (ps.sizeof_keyword) 533 ps.sizeof_mask |= 1 << ps.p_l_follow; 534 break; 535 536 case rparen: /* got a ')' or ']' */ 537 rparen_count--; 538 if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.sizeof_mask) { 539 ps.last_u_d = true; 540 ps.cast_mask &= (1 << ps.p_l_follow) - 1; 541 ps.want_blank = false; 542 } else 543 ps.want_blank = true; 544 ps.sizeof_mask &= (1 << ps.p_l_follow) - 1; 545 if (--ps.p_l_follow < 0) { 546 ps.p_l_follow = 0; 547 diag3(0, "Extra %c", *token); 548 } 549 if (e_code == s_code) /* if the paren starts the line */ 550 ps.paren_level = ps.p_l_follow; /* then indent it */ 551 552 *e_code++ = token[0]; 553 554 if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if 555 * (...), or some such */ 556 sp_sw = false; 557 force_nl = true;/* must force newline after if */ 558 ps.last_u_d = true; /* inform lexi that a following 559 * operator is unary */ 560 ps.in_stmt = false; /* dont use stmt continuation 561 * indentation */ 562 563 parse(hd_type); /* let parser worry about if, or whatever */ 564 } 565 ps.search_brace = btype_2; /* this should insure that constructs 566 * such as main(){...} and int[]{...} 567 * have their braces put in the right 568 * place */ 569 break; 570 571 case unary_op: /* this could be any unary operation */ 572 if (ps.want_blank) 573 *e_code++ = ' '; 574 575 if (troff && !ps.dumped_decl_indent && ps.in_decl && !is_procname) { 576 sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token); 577 ps.dumped_decl_indent = 1; 578 e_code += strlen(e_code); 579 } 580 else { 581 const char *res = token; 582 583 if (ps.in_decl && !ps.block_init) { /* if this is a unary op 584 * in a declaration, we 585 * should indent this 586 * token */ 587 for (i = 0; token[i]; ++i); /* find length of token */ 588 while ((e_code - s_code) < (dec_ind - i)) { 589 CHECK_SIZE_CODE; 590 *e_code++ = ' '; /* pad it */ 591 } 592 } 593 if (troff && token[0] == '-' && token[1] == '>') 594 res = "\\(->"; 595 for (t_ptr = res; *t_ptr; ++t_ptr) { 596 CHECK_SIZE_CODE; 597 *e_code++ = *t_ptr; 598 } 599 } 600 ps.want_blank = false; 601 break; 602 603 case binary_op: /* any binary operation */ 604 if (ps.want_blank) 605 *e_code++ = ' '; 606 { 607 const char *res = token; 608 609 if (troff) 610 switch (token[0]) { 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 = "\\(!="; 622 break; 623 case '|': 624 if (token[1] == '|') 625 res = "\\(br\\(br"; 626 else if (token[1] == 0) 627 res = "\\(br"; 628 break; 629 } 630 for (t_ptr = res; *t_ptr; ++t_ptr) { 631 CHECK_SIZE_CODE; 632 *e_code++ = *t_ptr; /* move the operator */ 633 } 634 } 635 ps.want_blank = true; 636 break; 637 638 case postop: /* got a trailing ++ or -- */ 639 *e_code++ = token[0]; 640 *e_code++ = token[1]; 641 ps.want_blank = true; 642 break; 643 644 case question: /* got a ? */ 645 squest++; /* this will be used when a later colon 646 * appears so we can distinguish the 647 * <c>?<n>:<n> construct */ 648 if (ps.want_blank) 649 *e_code++ = ' '; 650 *e_code++ = '?'; 651 ps.want_blank = true; 652 break; 653 654 case casestmt: /* got word 'case' or 'default' */ 655 scase = true; /* so we can process the later colon properly */ 656 goto copy_id; 657 658 case colon: /* got a ':' */ 659 if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */ 660 --squest; 661 if (ps.want_blank) 662 *e_code++ = ' '; 663 *e_code++ = ':'; 664 ps.want_blank = true; 665 break; 666 } 667 if (ps.in_or_st) { 668 *e_code++ = ':'; 669 ps.want_blank = false; 670 break; 671 } 672 ps.in_stmt = false; /* seeing a label does not imply we are in a 673 * stmt */ 674 for (t_ptr = s_code; *t_ptr; ++t_ptr) 675 *e_lab++ = *t_ptr; /* turn everything so far into a label */ 676 e_code = s_code; 677 *e_lab++ = ':'; 678 *e_lab++ = ' '; 679 *e_lab = '\0'; 680 681 force_nl = ps.pcase = scase; /* ps.pcase will be used by 682 * dump_line to decide how to 683 * indent the label. force_nl 684 * will force a case n: to be 685 * on a line by itself */ 686 scase = false; 687 ps.want_blank = false; 688 break; 689 690 case semicolon: /* got a ';' */ 691 ps.in_or_st = false;/* we are not in an initialization or 692 * structure declaration */ 693 scase = false; /* these will only need resetting in an error */ 694 squest = 0; 695 if (ps.last_token == rparen && rparen_count == 0) 696 ps.in_parameter_declaration = 0; 697 ps.cast_mask = 0; 698 ps.sizeof_mask = 0; 699 ps.block_init = 0; 700 ps.block_init_level = 0; 701 ps.just_saw_decl--; 702 703 if (ps.in_decl && s_code == e_code && !ps.block_init) 704 while ((e_code - s_code) < (dec_ind - 1)) { 705 CHECK_SIZE_CODE; 706 *e_code++ = ' '; 707 } 708 709 ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level 710 * structure declaration, we 711 * arent any more */ 712 713 if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) { 714 715 /* 716 * This should be true iff there were unbalanced parens in the 717 * stmt. It is a bit complicated, because the semicolon might 718 * be in a for stmt 719 */ 720 diag2(1, "Unbalanced parens"); 721 ps.p_l_follow = 0; 722 if (sp_sw) { /* this is a check for an if, while, etc. with 723 * unbalanced parens */ 724 sp_sw = false; 725 parse(hd_type); /* dont lose the if, or whatever */ 726 } 727 } 728 *e_code++ = ';'; 729 ps.want_blank = true; 730 ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the 731 * middle of a stmt */ 732 733 if (!sp_sw) { /* if not if for (;;) */ 734 parse(semicolon); /* let parser know about end of stmt */ 735 force_nl = true;/* force newline after an end of stmt */ 736 } 737 break; 738 739 case lbrace: /* got a '{' */ 740 ps.in_stmt = false; /* dont indent the {} */ 741 if (!ps.block_init) 742 force_nl = true;/* force other stuff on same line as '{' onto 743 * new line */ 744 else if (ps.block_init_level <= 0) 745 ps.block_init_level = 1; 746 else 747 ps.block_init_level++; 748 749 if (s_code != e_code && !ps.block_init) { 750 if (!btype_2) { 751 dump_line(); 752 ps.want_blank = false; 753 } 754 else if (ps.in_parameter_declaration && !ps.in_or_st) { 755 ps.i_l_follow = 0; 756 if (function_brace_split) { /* dump the line prior to the 757 * brace ... */ 758 dump_line(); 759 ps.want_blank = false; 760 } else /* add a space between the decl and brace */ 761 ps.want_blank = true; 762 } 763 } 764 if (ps.in_parameter_declaration) 765 prefix_blankline_requested = 0; 766 767 if (ps.p_l_follow > 0) { /* check for preceding unbalanced 768 * parens */ 769 diag2(1, "Unbalanced parens"); 770 ps.p_l_follow = 0; 771 if (sp_sw) { /* check for unclosed if, for, etc. */ 772 sp_sw = false; 773 parse(hd_type); 774 ps.ind_level = ps.i_l_follow; 775 } 776 } 777 if (s_code == e_code) 778 ps.ind_stmt = false; /* dont put extra indentation on line 779 * with '{' */ 780 if (ps.in_decl && ps.in_or_st) { /* this is either a structure 781 * declaration or an init */ 782 di_stack[ps.dec_nest++] = dec_ind; 783 /* ? dec_ind = 0; */ 784 } 785 else { 786 ps.decl_on_line = false; /* we cant be in the middle of 787 * a declaration, so dont do 788 * special indentation of 789 * comments */ 790 if (blanklines_after_declarations_at_proctop 791 && ps.in_parameter_declaration) 792 postfix_blankline_requested = 1; 793 ps.in_parameter_declaration = 0; 794 } 795 dec_ind = 0; 796 parse(lbrace); /* let parser know about this */ 797 if (ps.want_blank) /* put a blank before '{' if '{' is not at 798 * start of line */ 799 *e_code++ = ' '; 800 ps.want_blank = false; 801 *e_code++ = '{'; 802 ps.just_saw_decl = 0; 803 break; 804 805 case rbrace: /* got a '}' */ 806 if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be 807 * omitted in 808 * declarations */ 809 parse(semicolon); 810 if (ps.p_l_follow) {/* check for unclosed if, for, else. */ 811 diag2(1, "Unbalanced parens"); 812 ps.p_l_follow = 0; 813 sp_sw = false; 814 } 815 ps.just_saw_decl = 0; 816 ps.block_init_level--; 817 if (s_code != e_code && !ps.block_init) { /* '}' must be first on 818 * line */ 819 if (verbose) 820 diag2(0, "Line broken"); 821 dump_line(); 822 } 823 *e_code++ = '}'; 824 ps.want_blank = true; 825 ps.in_stmt = ps.ind_stmt = false; 826 if (ps.dec_nest > 0) { /* we are in multi-level structure 827 * declaration */ 828 dec_ind = di_stack[--ps.dec_nest]; 829 if (ps.dec_nest == 0 && !ps.in_parameter_declaration) 830 ps.just_saw_decl = 2; 831 ps.in_decl = true; 832 } 833 prefix_blankline_requested = 0; 834 parse(rbrace); /* let parser know about this */ 835 ps.search_brace = cuddle_else && ps.p_stack[ps.tos] == ifhead 836 && ps.il[ps.tos] >= ps.ind_level; 837 if (ps.tos <= 1 && blanklines_after_procs && ps.dec_nest <= 0) 838 postfix_blankline_requested = 1; 839 break; 840 841 case swstmt: /* got keyword "switch" */ 842 sp_sw = true; 843 hd_type = swstmt; /* keep this for when we have seen the 844 * expression */ 845 goto copy_id; /* go move the token into buffer */ 846 847 case sp_paren: /* token is if, while, for */ 848 sp_sw = true; /* the interesting stuff is done after the 849 * expression is scanned */ 850 hd_type = (*token == 'i' ? ifstmt : 851 (*token == 'w' ? whilestmt : forstmt)); 852 853 /* 854 * remember the type of header for later use by parser 855 */ 856 goto copy_id; /* copy the token into line */ 857 858 case sp_nparen: /* got else, do */ 859 ps.in_stmt = false; 860 if (*token == 'e') { 861 if (e_code != s_code && (!cuddle_else || e_code[-1] != '}')) { 862 if (verbose) 863 diag2(0, "Line broken"); 864 dump_line();/* make sure this starts a line */ 865 ps.want_blank = false; 866 } 867 force_nl = true;/* also, following stuff must go onto new line */ 868 last_else = 1; 869 parse(elselit); 870 } 871 else { 872 if (e_code != s_code) { /* make sure this starts a line */ 873 if (verbose) 874 diag2(0, "Line broken"); 875 dump_line(); 876 ps.want_blank = false; 877 } 878 force_nl = true;/* also, following stuff must go onto new line */ 879 last_else = 0; 880 parse(dolit); 881 } 882 goto copy_id; /* move the token into line */ 883 884 case decl: /* we have a declaration type (int, register, 885 * etc.) */ 886 parse(decl); /* let parser worry about indentation */ 887 if (ps.last_token == rparen && ps.tos <= 1) { 888 ps.in_parameter_declaration = 1; 889 if (s_code != e_code) { 890 dump_line(); 891 ps.want_blank = 0; 892 } 893 } 894 if (ps.in_parameter_declaration && ps.indent_parameters && ps.dec_nest == 0) { 895 ps.ind_level = ps.i_l_follow = 1; 896 ps.ind_stmt = 0; 897 } 898 ps.in_or_st = true; /* this might be a structure or initialization 899 * declaration */ 900 ps.in_decl = ps.decl_on_line = true; 901 if ( /* !ps.in_or_st && */ ps.dec_nest <= 0) 902 ps.just_saw_decl = 2; 903 prefix_blankline_requested = 0; 904 for (i = 0; token[i++];); /* get length of token */ 905 906 if (ps.ind_level == 0 || ps.dec_nest > 0) { 907 /* global variable or struct member in local variable */ 908 dec_ind = ps.decl_indent > 0 ? ps.decl_indent : i; 909 tabs_to_var = (use_tabs ? ps.decl_indent > 0 : 0); 910 } else { 911 /* local variable */ 912 dec_ind = ps.local_decl_indent > 0 ? ps.local_decl_indent : i; 913 tabs_to_var = (use_tabs ? ps.local_decl_indent > 0 : 0); 914 } 915 goto copy_id; 916 917 case ident: /* got an identifier or constant */ 918 if (ps.in_decl) { /* if we are in a declaration, we must indent 919 * identifier */ 920 if (is_procname == 0 || !procnames_start_line) { 921 if (!ps.block_init) { 922 if (troff && !ps.dumped_decl_indent) { 923 if (ps.want_blank) 924 *e_code++ = ' '; 925 ps.want_blank = false; 926 sprintf(e_code, "\n.De %dp+\200p\n", dec_ind * 7); 927 ps.dumped_decl_indent = 1; 928 e_code += strlen(e_code); 929 } else { 930 int cur_dec_ind; 931 int pos, startpos; 932 933 /* 934 * in order to get the tab math right for 935 * indentations that are not multiples of 8 we 936 * need to modify both startpos and dec_ind 937 * (cur_dec_ind) here by eight minus the 938 * remainder of the current starting column 939 * divided by eight. This seems to be a 940 * properly working fix 941 */ 942 startpos = e_code - s_code; 943 cur_dec_ind = dec_ind; 944 pos = startpos; 945 if ((ps.ind_level * ps.ind_size) % 8 != 0) { 946 pos += (ps.ind_level * ps.ind_size) % 8; 947 cur_dec_ind += (ps.ind_level * ps.ind_size) % 8; 948 } 949 950 if (tabs_to_var) { 951 while ((pos & ~7) + 8 <= cur_dec_ind) { 952 CHECK_SIZE_CODE; 953 *e_code++ = '\t'; 954 pos = (pos & ~7) + 8; 955 } 956 } 957 while (pos < cur_dec_ind) { 958 CHECK_SIZE_CODE; 959 *e_code++ = ' '; 960 pos++; 961 } 962 if (ps.want_blank && e_code - s_code == startpos) 963 *e_code++ = ' '; 964 ps.want_blank = false; 965 } 966 } 967 } else { 968 if (ps.want_blank) 969 *e_code++ = ' '; 970 ps.want_blank = false; 971 if (dec_ind && s_code != e_code) 972 dump_line(); 973 dec_ind = 0; 974 } 975 } 976 else if (sp_sw && ps.p_l_follow == 0) { 977 sp_sw = false; 978 force_nl = true; 979 ps.last_u_d = true; 980 ps.in_stmt = false; 981 parse(hd_type); 982 } 983 copy_id: 984 if (ps.want_blank) 985 *e_code++ = ' '; 986 if (troff && ps.its_a_keyword) { 987 e_code = chfont(&bodyf, &keywordf, e_code); 988 for (t_ptr = token; *t_ptr; ++t_ptr) { 989 CHECK_SIZE_CODE; 990 *e_code++ = keywordf.allcaps && islower(*t_ptr) 991 ? toupper(*t_ptr) : *t_ptr; 992 } 993 e_code = chfont(&keywordf, &bodyf, e_code); 994 } 995 else 996 for (t_ptr = token; *t_ptr; ++t_ptr) { 997 CHECK_SIZE_CODE; 998 *e_code++ = *t_ptr; 999 } 1000 ps.want_blank = true; 1001 break; 1002 1003 case period: /* treat a period kind of like a binary 1004 * operation */ 1005 *e_code++ = '.'; /* move the period into line */ 1006 ps.want_blank = false; /* dont put a blank after a period */ 1007 break; 1008 1009 case comma: 1010 ps.want_blank = (s_code != e_code); /* only put blank after comma 1011 * if comma does not start the 1012 * line */ 1013 if (ps.in_decl && is_procname == 0 && !ps.block_init) 1014 while ((e_code - s_code) < (dec_ind - 1)) { 1015 CHECK_SIZE_CODE; 1016 *e_code++ = ' '; 1017 } 1018 1019 *e_code++ = ','; 1020 if (ps.p_l_follow == 0) { 1021 if (ps.block_init_level <= 0) 1022 ps.block_init = 0; 1023 if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8)) 1024 force_nl = true; 1025 } 1026 break; 1027 1028 case preesc: /* got the character '#' */ 1029 if ((s_com != e_com) || 1030 (s_lab != e_lab) || 1031 (s_code != e_code)) 1032 dump_line(); 1033 *e_lab++ = '#'; /* move whole line to 'label' buffer */ 1034 { 1035 int in_comment = 0; 1036 int com_start = 0; 1037 char quote = 0; 1038 int com_end = 0; 1039 1040 while (*buf_ptr == ' ' || *buf_ptr == '\t') { 1041 buf_ptr++; 1042 if (buf_ptr >= buf_end) 1043 fill_buffer(); 1044 } 1045 while (*buf_ptr != '\n' || (in_comment && !had_eof)) { 1046 CHECK_SIZE_LAB; 1047 *e_lab = *buf_ptr++; 1048 if (buf_ptr >= buf_end) 1049 fill_buffer(); 1050 switch (*e_lab++) { 1051 case BACKSLASH: 1052 if (troff) 1053 *e_lab++ = BACKSLASH; 1054 if (!in_comment) { 1055 *e_lab++ = *buf_ptr++; 1056 if (buf_ptr >= buf_end) 1057 fill_buffer(); 1058 } 1059 break; 1060 case '/': 1061 if (*buf_ptr == '*' && !in_comment && !quote) { 1062 in_comment = 1; 1063 *e_lab++ = *buf_ptr++; 1064 com_start = e_lab - s_lab - 2; 1065 } 1066 break; 1067 case '"': 1068 if (quote == '"') 1069 quote = 0; 1070 break; 1071 case '\'': 1072 if (quote == '\'') 1073 quote = 0; 1074 break; 1075 case '*': 1076 if (*buf_ptr == '/' && in_comment) { 1077 in_comment = 0; 1078 *e_lab++ = *buf_ptr++; 1079 com_end = e_lab - s_lab; 1080 } 1081 break; 1082 } 1083 } 1084 1085 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1086 e_lab--; 1087 if (e_lab - s_lab == com_end && bp_save == 0) { /* comment on 1088 * preprocessor line */ 1089 if (sc_end == 0) /* if this is the first comment, we 1090 * must set up the buffer */ 1091 sc_end = &(save_com[0]); 1092 else { 1093 *sc_end++ = '\n'; /* add newline between 1094 * comments */ 1095 *sc_end++ = ' '; 1096 --line_no; 1097 } 1098 bcopy(s_lab + com_start, sc_end, com_end - com_start); 1099 sc_end += com_end - com_start; 1100 if (sc_end >= &save_com[sc_size]) 1101 abort(); 1102 e_lab = s_lab + com_start; 1103 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t')) 1104 e_lab--; 1105 bp_save = buf_ptr; /* save current input buffer */ 1106 be_save = buf_end; 1107 buf_ptr = save_com; /* fix so that subsequent calls to 1108 * lexi will take tokens out of 1109 * save_com */ 1110 *sc_end++ = ' '; /* add trailing blank, just in case */ 1111 buf_end = sc_end; 1112 sc_end = 0; 1113 } 1114 *e_lab = '\0'; /* null terminate line */ 1115 ps.pcase = false; 1116 } 1117 1118 if (strncmp(s_lab, "#if", 3) == 0) { 1119 if (blanklines_around_conditional_compilation) { 1120 int c; 1121 prefix_blankline_requested++; 1122 while ((c = getc(input)) == '\n'); 1123 ungetc(c, input); 1124 } 1125 if ((size_t)ifdef_level < sizeof(state_stack)/sizeof(state_stack[0])) { 1126 match_state[ifdef_level].tos = -1; 1127 state_stack[ifdef_level++] = ps; 1128 } 1129 else 1130 diag2(1, "#if stack overflow"); 1131 } 1132 else if (strncmp(s_lab, "#else", 5) == 0) 1133 if (ifdef_level <= 0) 1134 diag2(1, "Unmatched #else"); 1135 else { 1136 match_state[ifdef_level - 1] = ps; 1137 ps = state_stack[ifdef_level - 1]; 1138 } 1139 else if (strncmp(s_lab, "#endif", 6) == 0) { 1140 if (ifdef_level <= 0) 1141 diag2(1, "Unmatched #endif"); 1142 else { 1143 ifdef_level--; 1144 1145 #ifdef undef 1146 /* 1147 * This match needs to be more intelligent before the 1148 * message is useful 1149 */ 1150 if (match_state[ifdef_level].tos >= 0 1151 && bcmp(&ps, &match_state[ifdef_level], sizeof ps)) 1152 diag2(0, "Syntactically inconsistent #ifdef alternatives"); 1153 #endif 1154 } 1155 if (blanklines_around_conditional_compilation) { 1156 postfix_blankline_requested++; 1157 n_real_blanklines = 0; 1158 } 1159 } 1160 break; /* subsequent processing of the newline 1161 * character will cause the line to be printed */ 1162 1163 case comment: /* we have gotten a / followed by * this is a biggie */ 1164 if (flushed_nl) { /* we should force a broken line here */ 1165 flushed_nl = false; 1166 dump_line(); 1167 ps.want_blank = false; /* dont insert blank at line start */ 1168 force_nl = false; 1169 } 1170 pr_comment(); 1171 break; 1172 } /* end of big switch stmt */ 1173 1174 *e_code = '\0'; /* make sure code section is null terminated */ 1175 if (type_code != comment && type_code != newline && type_code != preesc) 1176 ps.last_token = type_code; 1177 } /* end of main while (1) loop */ 1178 } 1179 1180 /* 1181 * copy input file to backup file if in_name is /blah/blah/blah/file, then 1182 * backup file will be ".Bfile" then make the backup file the input and 1183 * original input file the output 1184 */ 1185 static void 1186 bakcopy(void) 1187 { 1188 int n, 1189 bakchn; 1190 char buff[8 * 1024]; 1191 const char *p; 1192 1193 /* construct file name .Bfile */ 1194 for (p = in_name; *p; p++); /* skip to end of string */ 1195 while (p > in_name && *p != '/') /* find last '/' */ 1196 p--; 1197 if (*p == '/') 1198 p++; 1199 sprintf(bakfile, "%s.BAK", p); 1200 1201 /* copy in_name to backup file */ 1202 bakchn = creat(bakfile, 0600); 1203 if (bakchn < 0) 1204 err(1, "%s", bakfile); 1205 while ((n = read(fileno(input), buff, sizeof buff)) != 0) 1206 if (write(bakchn, buff, n) != n) 1207 err(1, "%s", bakfile); 1208 if (n < 0) 1209 err(1, "%s", in_name); 1210 close(bakchn); 1211 fclose(input); 1212 1213 /* re-open backup file as the input file */ 1214 input = fopen(bakfile, "r"); 1215 if (input == 0) 1216 err(1, "%s", bakfile); 1217 /* now the original input file will be the output */ 1218 output = fopen(in_name, "w"); 1219 if (output == 0) { 1220 unlink(bakfile); 1221 err(1, "%s", in_name); 1222 } 1223 } 1224