indent.c revision 1.93 1 /* $NetBSD: indent.c,v 1.93 2021/09/30 20:58:26 rillig Exp $ */
2
3 /*-
4 * SPDX-License-Identifier: BSD-4-Clause
5 *
6 * Copyright (c) 1985 Sun Microsystems, Inc.
7 * Copyright (c) 1976 Board of Trustees of the University of Illinois.
8 * Copyright (c) 1980, 1993
9 * The Regents of the University of California. All rights reserved.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. All advertising materials mentioning features or use of this software
20 * must display the following acknowledgement:
21 * This product includes software developed by the University of
22 * California, Berkeley and its contributors.
23 * 4. Neither the name of the University nor the names of its contributors
24 * may be used to endorse or promote products derived from this software
25 * without specific prior written permission.
26 *
27 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
29 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
30 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
31 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
32 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
33 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
34 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
35 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
36 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
37 * SUCH DAMAGE.
38 */
39
40 #if 0
41 static char sccsid[] = "@(#)indent.c 5.17 (Berkeley) 6/7/93";
42 #endif
43
44 #include <sys/cdefs.h>
45 #if defined(__NetBSD__)
46 __RCSID("$NetBSD: indent.c,v 1.93 2021/09/30 20:58:26 rillig Exp $");
47 #elif defined(__FreeBSD__)
48 __FBSDID("$FreeBSD: head/usr.bin/indent/indent.c 340138 2018-11-04 19:24:49Z oshogbo $");
49 #endif
50
51 #include <sys/param.h>
52 #if HAVE_CAPSICUM
53 #include <sys/capsicum.h>
54 #include <capsicum_helpers.h>
55 #endif
56 #include <ctype.h>
57 #include <err.h>
58 #include <errno.h>
59 #include <fcntl.h>
60 #include <stdio.h>
61 #include <stdlib.h>
62 #include <string.h>
63 #include <unistd.h>
64
65 #include "indent.h"
66
67 struct options opt = {
68 .btype_2 = true,
69 .comment_delimiter_on_blankline = true,
70 .cuddle_else = true,
71 .comment_column = 33,
72 .decl_indent = 16,
73 .else_if = true,
74 .function_brace_split = true,
75 .format_col1_comments = true,
76 .format_block_comments = true,
77 .indent_parameters = true,
78 .indent_size = 8,
79 .local_decl_indent = -1,
80 .lineup_to_parens = true,
81 .procnames_start_line = true,
82 .star_comment_cont = true,
83 .tabsize = 8,
84 .max_line_length = 78,
85 .use_tabs = true,
86 };
87
88 struct parser_state ps;
89
90 struct buffer lab;
91 struct buffer code;
92 struct buffer com;
93 struct buffer token;
94
95 char *in_buffer;
96 char *in_buffer_limit;
97 char *buf_ptr;
98 char *buf_end;
99
100 char sc_buf[sc_size];
101 char *save_com;
102 static char *sc_end; /* pointer into save_com buffer */
103
104 char *bp_save;
105 char *be_save;
106
107 bool found_err;
108 int n_real_blanklines;
109 bool prefix_blankline_requested;
110 bool postfix_blankline_requested;
111 bool break_comma;
112 float case_ind;
113 bool had_eof;
114 int line_no;
115 bool inhibit_formatting;
116
117 static int ifdef_level;
118 static struct parser_state state_stack[5];
119
120 FILE *input;
121 FILE *output;
122
123 static void bakcopy(void);
124 static void indent_declaration(int, bool);
125
126 static const char *in_name = "Standard Input";
127 static const char *out_name = "Standard Output";
128 static const char *backup_suffix = ".BAK";
129 static char bakfile[MAXPATHLEN] = "";
130
131 static void
132 check_size_code(size_t desired_size)
133 {
134 if (code.e + desired_size >= code.l)
135 buf_expand(&code, desired_size);
136 }
137
138 static void
139 check_size_label(size_t desired_size)
140 {
141 if (lab.e + desired_size >= lab.l)
142 buf_expand(&lab, desired_size);
143 }
144
145 #if HAVE_CAPSICUM
146 static void
147 init_capsicum(void)
148 {
149 cap_rights_t rights;
150
151 /* Restrict input/output descriptors and enter Capsicum sandbox. */
152 cap_rights_init(&rights, CAP_FSTAT, CAP_WRITE);
153 if (caph_rights_limit(fileno(output), &rights) < 0)
154 err(EXIT_FAILURE, "unable to limit rights for %s", out_name);
155 cap_rights_init(&rights, CAP_FSTAT, CAP_READ);
156 if (caph_rights_limit(fileno(input), &rights) < 0)
157 err(EXIT_FAILURE, "unable to limit rights for %s", in_name);
158 if (caph_enter() < 0)
159 err(EXIT_FAILURE, "unable to enter capability mode");
160 }
161 #endif
162
163 static void
164 search_brace_newline(bool *inout_force_nl)
165 {
166 if (sc_end == NULL) {
167 save_com = sc_buf;
168 save_com[0] = save_com[1] = ' ';
169 sc_end = &save_com[2];
170 }
171 *sc_end++ = '\n';
172
173 /*
174 * We may have inherited a force_nl == true from the previous token (like
175 * a semicolon). But once we know that a newline has been scanned in this
176 * loop, force_nl should be false.
177 *
178 * However, the force_nl == true must be preserved if newline is never
179 * scanned in this loop, so this assignment cannot be done earlier.
180 */
181 *inout_force_nl = false;
182 }
183
184 static void
185 search_brace_comment(bool *inout_comment_buffered)
186 {
187 if (sc_end == NULL) {
188 /*
189 * Copy everything from the start of the line, because
190 * process_comment() will use that to calculate original indentation
191 * of a boxed comment.
192 */
193 memcpy(sc_buf, in_buffer, (size_t)(buf_ptr - in_buffer) - 4);
194 save_com = sc_buf + (buf_ptr - in_buffer - 4);
195 save_com[0] = save_com[1] = ' ';
196 sc_end = &save_com[2];
197 }
198 *inout_comment_buffered = true;
199 *sc_end++ = '/'; /* copy in start of comment */
200 *sc_end++ = '*';
201 for (;;) { /* loop until the end of the comment */
202 *sc_end = *buf_ptr++;
203 if (buf_ptr >= buf_end)
204 fill_buffer();
205 if (*sc_end++ == '*' && *buf_ptr == '/')
206 break; /* we are at end of comment */
207 if (sc_end >= &save_com[sc_size]) { /* check for temp buffer
208 * overflow */
209 diag(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever");
210 fflush(output);
211 exit(1);
212 }
213 }
214 *sc_end++ = '/'; /* add ending slash */
215 if (++buf_ptr >= buf_end) /* get past / in buffer */
216 fill_buffer();
217 }
218
219 static bool
220 search_brace_lbrace(void)
221 {
222 /*
223 * Put KNF-style lbraces before the buffered up tokens and jump out of
224 * this loop in order to avoid copying the token again under the default
225 * case of the switch below.
226 */
227 if (sc_end != NULL && opt.btype_2) {
228 save_com[0] = '{';
229 /*
230 * Originally the lbrace may have been alone on its own line, but it
231 * will be moved into "the else's line", so if there was a newline
232 * resulting from the "{" before, it must be scanned now and ignored.
233 */
234 while (isspace((unsigned char)*buf_ptr)) {
235 if (++buf_ptr >= buf_end)
236 fill_buffer();
237 if (*buf_ptr == '\n')
238 break;
239 }
240 return true;
241 }
242 return false;
243 }
244
245 static bool
246 search_brace_other(token_type ttype, bool *inout_force_nl,
247 bool comment_buffered, bool last_else)
248 {
249 bool remove_newlines;
250
251 remove_newlines =
252 /* "} else" */
253 (ttype == keyword_do_else && *token.s == 'e' &&
254 code.e != code.s && code.e[-1] == '}')
255 /* "else if" */
256 || (ttype == keyword_for_if_while &&
257 *token.s == 'i' && last_else && opt.else_if);
258 if (remove_newlines)
259 *inout_force_nl = false;
260 if (sc_end == NULL) { /* ignore buffering if comment wasn't saved
261 * up */
262 ps.search_brace = false;
263 return false;
264 }
265 while (sc_end > save_com && isblank((unsigned char)sc_end[-1])) {
266 sc_end--;
267 }
268 if (opt.swallow_optional_blanklines ||
269 (!comment_buffered && remove_newlines)) {
270 *inout_force_nl = !remove_newlines;
271 while (sc_end > save_com && sc_end[-1] == '\n') {
272 sc_end--;
273 }
274 }
275 if (*inout_force_nl) { /* if we should insert a nl here, put it into
276 * the buffer */
277 *inout_force_nl = false;
278 --line_no; /* this will be re-increased when the newline
279 * is read from the buffer */
280 *sc_end++ = '\n';
281 *sc_end++ = ' ';
282 if (opt.verbose) /* print error msg if the line was not already
283 * broken */
284 diag(0, "Line broken");
285 }
286 for (const char *t_ptr = token.s; *t_ptr != '\0'; ++t_ptr)
287 *sc_end++ = *t_ptr;
288 return true;
289 }
290
291 static void
292 switch_buffer(void)
293 {
294 ps.search_brace = false; /* stop looking for start of stmt */
295 bp_save = buf_ptr; /* save current input buffer */
296 be_save = buf_end;
297 buf_ptr = save_com; /* fix so that subsequent calls to lexi will
298 * take tokens out of save_com */
299 *sc_end++ = ' '; /* add trailing blank, just in case */
300 buf_end = sc_end;
301 sc_end = NULL;
302 debug_println("switched buf_ptr to save_com");
303 }
304
305 static void
306 search_brace_lookahead(token_type *inout_ttype)
307 {
308 /*
309 * We must make this check, just in case there was an unexpected EOF.
310 */
311 if (*inout_ttype != end_of_file) {
312 /*
313 * The only intended purpose of calling lexi() below is to categorize
314 * the next token in order to decide whether to continue buffering
315 * forthcoming tokens. Once the buffering is over, lexi() will be
316 * called again elsewhere on all of the tokens - this time for normal
317 * processing.
318 *
319 * Calling it for this purpose is a bug, because lexi() also changes
320 * the parser state and discards leading whitespace, which is needed
321 * mostly for comment-related considerations.
322 *
323 * Work around the former problem by giving lexi() a copy of the
324 * current parser state and discard it if the call turned out to be
325 * just a look ahead.
326 *
327 * Work around the latter problem by copying all whitespace characters
328 * into the buffer so that the later lexi() call will read them.
329 */
330 if (sc_end != NULL) {
331 while (*buf_ptr == ' ' || *buf_ptr == '\t') {
332 *sc_end++ = *buf_ptr++;
333 if (sc_end >= &save_com[sc_size]) {
334 errx(1, "input too long");
335 }
336 }
337 if (buf_ptr >= buf_end) {
338 fill_buffer();
339 }
340 }
341
342 struct parser_state transient_state;
343 transient_state = ps;
344 *inout_ttype = lexi(&transient_state); /* read another token */
345 if (*inout_ttype != newline && *inout_ttype != form_feed &&
346 *inout_ttype != comment && !transient_state.search_brace) {
347 ps = transient_state;
348 }
349 }
350 }
351
352 static void
353 search_brace(token_type *inout_ttype, bool *inout_force_nl,
354 bool *inout_comment_buffered, bool *inout_last_else)
355 {
356 while (ps.search_brace) {
357 switch (*inout_ttype) {
358 case newline:
359 search_brace_newline(inout_force_nl);
360 break;
361 case form_feed:
362 break;
363 case comment:
364 search_brace_comment(inout_comment_buffered);
365 break;
366 case lbrace:
367 if (search_brace_lbrace())
368 goto sw_buffer;
369 /* FALLTHROUGH */
370 default: /* it is the start of a normal statement */
371 if (!search_brace_other(*inout_ttype, inout_force_nl,
372 *inout_comment_buffered, *inout_last_else))
373 return;
374 sw_buffer:
375 switch_buffer();
376 }
377 search_brace_lookahead(inout_ttype);
378 }
379
380 *inout_last_else = false;
381 }
382
383 static void
384 buf_init(struct buffer *buf)
385 {
386 buf->buf = xmalloc(bufsize);
387 buf->buf[0] = ' '; /* allow accessing buf->e[-1] */
388 buf->buf[1] = '\0';
389 buf->s = buf->buf + 1;
390 buf->e = buf->s;
391 buf->l = buf->buf + bufsize - 5; /* safety margin, though unreliable */
392 }
393
394 void
395 buf_expand(struct buffer *buf, size_t desired_size)
396 {
397 size_t nsize = buf->l - buf->s + 400 + desired_size;
398 size_t code_len = buf->e - buf->s;
399 buf->buf = xrealloc(buf->buf, nsize);
400 buf->e = buf->buf + code_len + 1;
401 buf->l = buf->buf + nsize - 5;
402 buf->s = buf->buf + 1;
403 }
404
405 static void
406 main_init_globals(void)
407 {
408 found_err = false;
409
410 ps.p_stack[0] = stmt; /* this is the parser's stack */
411 ps.last_nl = true; /* this is true if the last thing scanned was
412 * a newline */
413 ps.last_token = semicolon;
414 buf_init(&com);
415 buf_init(&lab);
416 buf_init(&code);
417 buf_init(&token);
418 opt.else_if = true; /* XXX: redundant? */
419
420 in_buffer = xmalloc(10);
421 in_buffer_limit = in_buffer + 8;
422 buf_ptr = buf_end = in_buffer;
423 line_no = 1;
424 had_eof = ps.in_decl = ps.decl_on_line = (break_comma = false);
425 ps.in_or_st = false;
426 ps.want_blank = ps.in_stmt = ps.ind_stmt = false;
427
428 ps.pcase = false;
429 sc_end = NULL;
430 bp_save = NULL;
431 be_save = NULL;
432
433 output = NULL;
434
435 const char *suffix = getenv("SIMPLE_BACKUP_SUFFIX");
436 if (suffix != NULL)
437 backup_suffix = suffix;
438 }
439
440 static void
441 main_parse_command_line(int argc, char **argv)
442 {
443 int i;
444 const char *profile_name = NULL;
445
446 for (i = 1; i < argc; ++i)
447 if (strcmp(argv[i], "-npro") == 0)
448 break;
449 else if (argv[i][0] == '-' && argv[i][1] == 'P' && argv[i][2] != '\0')
450 profile_name = argv[i]; /* non-empty -P (set profile) */
451 if (i >= argc)
452 set_profile(profile_name);
453
454 for (i = 1; i < argc; ++i) {
455 if (argv[i][0] == '-') {
456 set_option(argv[i]);
457
458 } else if (input == NULL) {
459 in_name = argv[i];
460 input = fopen(in_name, "r");
461 if (input == NULL)
462 err(1, "%s", in_name);
463
464 } else if (output == NULL) {
465 out_name = argv[i];
466 if (strcmp(in_name, out_name) == 0)
467 errx(1, "input and output files must be different");
468 output = fopen(out_name, "w");
469 if (output == NULL)
470 err(1, "%s", out_name);
471
472 } else
473 errx(1, "unknown parameter: %s", argv[i]);
474 }
475
476 if (input == NULL)
477 input = stdin;
478 if (output == NULL) {
479 if (input == stdin)
480 output = stdout;
481 else {
482 out_name = in_name;
483 bakcopy();
484 }
485 }
486
487 if (opt.comment_column <= 1)
488 opt.comment_column = 2; /* don't put normal comments before column 2 */
489 if (opt.block_comment_max_line_length <= 0)
490 opt.block_comment_max_line_length = opt.max_line_length;
491 if (opt.local_decl_indent < 0) /* if not specified by user, set this */
492 opt.local_decl_indent = opt.decl_indent;
493 if (opt.decl_comment_column <= 0) /* if not specified by user, set this */
494 opt.decl_comment_column = opt.ljust_decl
495 ? (opt.comment_column <= 10 ? 2 : opt.comment_column - 8)
496 : opt.comment_column;
497 if (opt.continuation_indent == 0)
498 opt.continuation_indent = opt.indent_size;
499 }
500
501 static void
502 main_prepare_parsing(void)
503 {
504 fill_buffer(); /* get first batch of stuff into input buffer */
505
506 parse(semicolon);
507
508 char *p = buf_ptr;
509 int col = 1;
510
511 for (;;) {
512 if (*p == ' ')
513 col++;
514 else if (*p == '\t')
515 col = opt.tabsize * (1 + (col - 1) / opt.tabsize) + 1;
516 else
517 break;
518 p++;
519 }
520 if (col > opt.indent_size)
521 ps.ind_level = ps.ind_level_follow = col / opt.indent_size;
522 }
523
524 static void __attribute__((__noreturn__))
525 process_end_of_file(void)
526 {
527 if (lab.s != lab.e || code.s != code.e || com.s != com.e)
528 dump_line();
529
530 if (ps.tos > 1) /* check for balanced braces */
531 diag(1, "Stuff missing from end of file");
532
533 if (opt.verbose) {
534 printf("There were %d output lines and %d comments\n",
535 ps.stats.lines, ps.stats.comments);
536 printf("(Lines with comments)/(Lines with code): %6.3f\n",
537 (1.0 * ps.stats.comment_lines) / ps.stats.code_lines);
538 }
539
540 fflush(output);
541 exit(found_err ? EXIT_FAILURE : EXIT_SUCCESS);
542 }
543
544 static void
545 process_comment_in_code(token_type ttype, bool *inout_force_nl)
546 {
547 if (*inout_force_nl &&
548 ttype != semicolon &&
549 (ttype != lbrace || !opt.btype_2)) {
550
551 /* we should force a broken line here */
552 if (opt.verbose)
553 diag(0, "Line broken");
554 dump_line();
555 ps.want_blank = false; /* dont insert blank at line start */
556 *inout_force_nl = false;
557 }
558
559 ps.in_stmt = true; /* turn on flag which causes an extra level of
560 * indentation. this is turned off by a ';' or
561 * '}' */
562 if (com.s != com.e) { /* the turkey has embedded a comment in a
563 * line. fix it */
564 size_t len = com.e - com.s;
565
566 check_size_code(len + 3);
567 *code.e++ = ' ';
568 memcpy(code.e, com.s, len);
569 code.e += len;
570 *code.e++ = ' ';
571 *code.e = '\0';
572 ps.want_blank = false;
573 com.e = com.s;
574 }
575 }
576
577 static void
578 process_form_feed(void)
579 {
580 ps.use_ff = true; /* a form feed is treated much like a newline */
581 dump_line();
582 ps.want_blank = false;
583 }
584
585 static void
586 process_newline(void)
587 {
588 if (ps.last_token != comma || ps.p_l_follow > 0 || opt.break_after_comma
589 || ps.block_init || !break_comma || com.s != com.e) {
590 dump_line();
591 ps.want_blank = false;
592 }
593 ++line_no; /* keep track of input line number */
594 }
595
596 static void
597 process_lparen_or_lbracket(int dec_ind, bool tabs_to_var, bool sp_sw)
598 {
599 /* count parens to make Healy happy */
600 if (++ps.p_l_follow == nitems(ps.paren_indents)) {
601 diag(0, "Reached internal limit of %zu unclosed parens",
602 nitems(ps.paren_indents));
603 ps.p_l_follow--;
604 }
605 if (token.s[0] == '(' && ps.in_decl
606 && !ps.block_init && !ps.dumped_decl_indent &&
607 ps.procname[0] == '\0' && ps.paren_level == 0) {
608 /* function pointer declarations */
609 indent_declaration(dec_ind, tabs_to_var);
610 ps.dumped_decl_indent = true;
611 } else if (ps.want_blank &&
612 ((ps.last_token != ident && ps.last_token != funcname) ||
613 opt.proc_calls_space ||
614 (ps.keyword == kw_sizeof ? opt.blank_after_sizeof :
615 ps.keyword != kw_0 && ps.keyword != kw_offsetof)))
616 *code.e++ = ' ';
617 ps.want_blank = false;
618 *code.e++ = token.s[0];
619
620 ps.paren_indents[ps.p_l_follow - 1] =
621 indentation_after_range(0, code.s, code.e);
622 debug_println("paren_indent[%d] is now %d",
623 ps.p_l_follow - 1, ps.paren_indents[ps.p_l_follow - 1]);
624
625 if (sp_sw && ps.p_l_follow == 1 && opt.extra_expression_indent
626 && ps.paren_indents[0] < 2 * opt.indent_size) {
627 ps.paren_indents[0] = 2 * opt.indent_size;
628 debug_println("paren_indent[0] is now %d", ps.paren_indents[0]);
629 }
630 if (ps.in_or_st && *token.s == '(' && ps.tos <= 2) {
631 /*
632 * this is a kluge to make sure that declarations will be aligned
633 * right if proc decl has an explicit type on it, i.e. "int a(x) {..."
634 */
635 parse(semicolon); /* I said this was a kluge... */
636 ps.in_or_st = false; /* turn off flag for structure decl or
637 * initialization */
638 }
639 /* parenthesized type following sizeof or offsetof is not a cast */
640 if (ps.keyword == kw_offsetof || ps.keyword == kw_sizeof)
641 ps.not_cast_mask |= 1 << ps.p_l_follow;
642 }
643
644 static void
645 process_rparen_or_rbracket(bool *inout_sp_sw, bool *inout_force_nl,
646 token_type hd_type)
647 {
648 if ((ps.cast_mask & (1 << ps.p_l_follow) & ~ps.not_cast_mask) != 0) {
649 ps.last_u_d = true;
650 ps.cast_mask &= (1 << ps.p_l_follow) - 1;
651 ps.want_blank = opt.space_after_cast;
652 } else
653 ps.want_blank = true;
654 ps.not_cast_mask &= (1 << ps.p_l_follow) - 1;
655
656 if (--ps.p_l_follow < 0) {
657 ps.p_l_follow = 0;
658 diag(0, "Extra %c", *token.s);
659 }
660
661 if (code.e == code.s) /* if the paren starts the line */
662 ps.paren_level = ps.p_l_follow; /* then indent it */
663
664 *code.e++ = token.s[0];
665
666 if (*inout_sp_sw && (ps.p_l_follow == 0)) { /* check for end of if (...),
667 * or some such */
668 *inout_sp_sw = false;
669 *inout_force_nl = true; /* must force newline after if */
670 ps.last_u_d = true; /* inform lexi that a following operator is
671 * unary */
672 ps.in_stmt = false; /* dont use stmt continuation indentation */
673
674 parse(hd_type); /* let parser worry about if, or whatever */
675 }
676 ps.search_brace = opt.btype_2; /* this should ensure that constructs
677 * such as main(){...} and int[]{...}
678 * have their braces put in the right
679 * place */
680 }
681
682 static void
683 process_unary_op(int dec_ind, bool tabs_to_var)
684 {
685 if (!ps.dumped_decl_indent && ps.in_decl && !ps.block_init &&
686 ps.procname[0] == '\0' && ps.paren_level == 0) {
687 /* pointer declarations */
688 indent_declaration(dec_ind - (int)strlen(token.s), tabs_to_var);
689 ps.dumped_decl_indent = true;
690 } else if (ps.want_blank)
691 *code.e++ = ' ';
692
693 {
694 size_t len = token.e - token.s;
695
696 check_size_code(len);
697 memcpy(code.e, token.s, len);
698 code.e += len;
699 }
700 ps.want_blank = false;
701 }
702
703 static void
704 process_binary_op(void)
705 {
706 size_t len = token.e - token.s;
707
708 check_size_code(len + 1);
709 if (ps.want_blank)
710 *code.e++ = ' ';
711 memcpy(code.e, token.s, len);
712 code.e += len;
713
714 ps.want_blank = true;
715 }
716
717 static void
718 process_postfix_op(void)
719 {
720 *code.e++ = token.s[0];
721 *code.e++ = token.s[1];
722 ps.want_blank = true;
723 }
724
725 static void
726 process_question(int *inout_squest)
727 {
728 (*inout_squest)++; /* this will be used when a later colon
729 * appears so we can distinguish the
730 * <c>?<n>:<n> construct */
731 if (ps.want_blank)
732 *code.e++ = ' ';
733 *code.e++ = '?';
734 ps.want_blank = true;
735 }
736
737 static void
738 process_colon(int *inout_squest, bool *inout_force_nl, bool *inout_scase)
739 {
740 if (*inout_squest > 0) { /* it is part of the <c>?<n>: <n> construct */
741 --*inout_squest;
742 if (ps.want_blank)
743 *code.e++ = ' ';
744 *code.e++ = ':';
745 ps.want_blank = true;
746 return;
747 }
748 if (ps.in_or_st) {
749 *code.e++ = ':';
750 ps.want_blank = false;
751 return;
752 }
753 ps.in_stmt = false; /* seeing a label does not imply we are in a
754 * stmt */
755 /*
756 * turn everything so far into a label
757 */
758 {
759 size_t len = code.e - code.s;
760
761 check_size_label(len + 3);
762 memcpy(lab.e, code.s, len);
763 lab.e += len;
764 *lab.e++ = ':';
765 *lab.e = '\0';
766 code.e = code.s;
767 }
768 *inout_force_nl = ps.pcase = *inout_scase; /* ps.pcase will be used by
769 * dump_line to decide how to
770 * indent the label. force_nl
771 * will force a case n: to be
772 * on a line by itself */
773 *inout_scase = false;
774 ps.want_blank = false;
775 }
776
777 static void
778 process_semicolon(bool *inout_scase, int *inout_squest, int dec_ind,
779 bool tabs_to_var, bool *inout_sp_sw,
780 token_type hd_type,
781 bool *inout_force_nl)
782 {
783 if (ps.decl_nest == 0)
784 ps.in_or_st = false; /* we are not in an initialization or
785 * structure declaration */
786 *inout_scase = false; /* these will only need resetting in an error */
787 *inout_squest = 0;
788 if (ps.last_token == rparen)
789 ps.in_parameter_declaration = false;
790 ps.cast_mask = 0;
791 ps.not_cast_mask = 0;
792 ps.block_init = false;
793 ps.block_init_level = 0;
794 ps.just_saw_decl--;
795
796 if (ps.in_decl && code.s == code.e && !ps.block_init &&
797 !ps.dumped_decl_indent && ps.paren_level == 0) {
798 /* indent stray semicolons in declarations */
799 indent_declaration(dec_ind - 1, tabs_to_var);
800 ps.dumped_decl_indent = true;
801 }
802
803 ps.in_decl = (ps.decl_nest > 0); /* if we were in a first level
804 * structure declaration, we arent any
805 * more */
806
807 if ((!*inout_sp_sw || hd_type != for_exprs) && ps.p_l_follow > 0) {
808
809 /*
810 * This should be true iff there were unbalanced parens in the stmt.
811 * It is a bit complicated, because the semicolon might be in a for
812 * stmt
813 */
814 diag(1, "Unbalanced parens");
815 ps.p_l_follow = 0;
816 if (*inout_sp_sw) { /* this is a check for an if, while, etc. with
817 * unbalanced parens */
818 *inout_sp_sw = false;
819 parse(hd_type); /* dont lose the if, or whatever */
820 }
821 }
822 *code.e++ = ';';
823 ps.want_blank = true;
824 ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the middle of a
825 * stmt */
826
827 if (!*inout_sp_sw) { /* if not if for (;;) */
828 parse(semicolon); /* let parser know about end of stmt */
829 *inout_force_nl = true; /* force newline after an end of stmt */
830 }
831 }
832
833 static void
834 process_lbrace(bool *inout_force_nl, bool *inout_sp_sw, token_type hd_type,
835 int *di_stack, int di_stack_cap, int *inout_dec_ind)
836 {
837 ps.in_stmt = false; /* dont indent the {} */
838 if (!ps.block_init)
839 *inout_force_nl = true; /* force other stuff on same line as '{' onto
840 * new line */
841 else if (ps.block_init_level <= 0)
842 ps.block_init_level = 1;
843 else
844 ps.block_init_level++;
845
846 if (code.s != code.e && !ps.block_init) {
847 if (!opt.btype_2) {
848 dump_line();
849 ps.want_blank = false;
850 } else if (ps.in_parameter_declaration && !ps.in_or_st) {
851 ps.ind_level_follow = 0;
852 if (opt.function_brace_split) { /* dump the line prior to the
853 * brace ... */
854 dump_line();
855 ps.want_blank = false;
856 } else /* add a space between the decl and brace */
857 ps.want_blank = true;
858 }
859 }
860 if (ps.in_parameter_declaration)
861 prefix_blankline_requested = false;
862
863 if (ps.p_l_follow > 0) { /* check for preceding unbalanced parens */
864 diag(1, "Unbalanced parens");
865 ps.p_l_follow = 0;
866 if (*inout_sp_sw) { /* check for unclosed if, for, etc. */
867 *inout_sp_sw = false;
868 parse(hd_type);
869 ps.ind_level = ps.ind_level_follow;
870 }
871 }
872 if (code.s == code.e)
873 ps.ind_stmt = false; /* dont put extra indentation on line
874 * with '{' */
875 if (ps.in_decl && ps.in_or_st) { /* this is either a structure
876 * declaration or an init */
877 di_stack[ps.decl_nest] = *inout_dec_ind;
878 if (++ps.decl_nest == di_stack_cap) {
879 diag(0, "Reached internal limit of %d struct levels",
880 di_stack_cap);
881 ps.decl_nest--;
882 }
883 /* ? dec_ind = 0; */
884 } else {
885 ps.decl_on_line = false; /* we can't be in the middle of a
886 * declaration, so don't do special
887 * indentation of comments */
888 if (opt.blanklines_after_declarations_at_proctop
889 && ps.in_parameter_declaration)
890 postfix_blankline_requested = true;
891 ps.in_parameter_declaration = false;
892 ps.in_decl = false;
893 }
894 *inout_dec_ind = 0;
895 parse(lbrace); /* let parser know about this */
896 if (ps.want_blank) /* put a blank before '{' if '{' is not at
897 * start of line */
898 *code.e++ = ' ';
899 ps.want_blank = false;
900 *code.e++ = '{';
901 ps.just_saw_decl = 0;
902 }
903
904 static void
905 process_rbrace(bool *inout_sp_sw, int *inout_dec_ind, const int *di_stack)
906 {
907 if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be
908 * omitted in
909 * declarations */
910 parse(semicolon);
911 if (ps.p_l_follow != 0) { /* check for unclosed if, for, else. */
912 diag(1, "Unbalanced parens");
913 ps.p_l_follow = 0;
914 *inout_sp_sw = false;
915 }
916 ps.just_saw_decl = 0;
917 ps.block_init_level--;
918 if (code.s != code.e && !ps.block_init) { /* '}' must be first on line */
919 if (opt.verbose)
920 diag(0, "Line broken");
921 dump_line();
922 }
923 *code.e++ = '}';
924 ps.want_blank = true;
925 ps.in_stmt = ps.ind_stmt = false;
926 if (ps.decl_nest > 0) { /* we are in multi-level structure declaration */
927 *inout_dec_ind = di_stack[--ps.decl_nest];
928 if (ps.decl_nest == 0 && !ps.in_parameter_declaration)
929 ps.just_saw_decl = 2;
930 ps.in_decl = true;
931 }
932 prefix_blankline_requested = false;
933 parse(rbrace); /* let parser know about this */
934 ps.search_brace = opt.cuddle_else
935 && ps.p_stack[ps.tos] == if_expr_stmt
936 && ps.il[ps.tos] >= ps.ind_level;
937 if (ps.tos <= 1 && opt.blanklines_after_procs && ps.decl_nest <= 0)
938 postfix_blankline_requested = true;
939 }
940
941 static void
942 process_keyword_do_else(bool *inout_force_nl, bool *inout_last_else)
943 {
944 ps.in_stmt = false;
945 if (*token.s == 'e') {
946 if (code.e != code.s && (!opt.cuddle_else || code.e[-1] != '}')) {
947 if (opt.verbose)
948 diag(0, "Line broken");
949 dump_line(); /* make sure this starts a line */
950 ps.want_blank = false;
951 }
952 *inout_force_nl = true;/* also, following stuff must go onto new line */
953 *inout_last_else = true;
954 parse(keyword_else);
955 } else {
956 if (code.e != code.s) { /* make sure this starts a line */
957 if (opt.verbose)
958 diag(0, "Line broken");
959 dump_line();
960 ps.want_blank = false;
961 }
962 *inout_force_nl = true;/* also, following stuff must go onto new line */
963 *inout_last_else = false;
964 parse(keyword_do);
965 }
966 }
967
968 static void
969 process_decl(int *out_dec_ind, bool *out_tabs_to_var)
970 {
971 parse(decl); /* let parser worry about indentation */
972 if (ps.last_token == rparen && ps.tos <= 1) {
973 if (code.s != code.e) {
974 dump_line();
975 ps.want_blank = false;
976 }
977 }
978 if (ps.in_parameter_declaration && opt.indent_parameters &&
979 ps.decl_nest == 0) {
980 ps.ind_level = ps.ind_level_follow = 1;
981 ps.ind_stmt = false;
982 }
983 ps.in_or_st = true; /* this might be a structure or initialization
984 * declaration */
985 ps.in_decl = ps.decl_on_line = ps.last_token != type_def;
986 if ( /* !ps.in_or_st && */ ps.decl_nest <= 0)
987 ps.just_saw_decl = 2;
988 prefix_blankline_requested = false;
989
990 int len = (int)strlen(token.s) + 1;
991 int ind = ps.ind_level == 0 || ps.decl_nest > 0
992 ? opt.decl_indent /* global variable or local member */
993 : opt.local_decl_indent; /* local variable */
994 *out_dec_ind = ind > 0 ? ind : len;
995 *out_tabs_to_var = opt.use_tabs ? ind > 0 : false;
996 }
997
998 static void
999 process_ident(token_type ttype, int dec_ind, bool tabs_to_var,
1000 bool *inout_sp_sw, bool *inout_force_nl, token_type hd_type)
1001 {
1002 if (ps.in_decl) {
1003 if (ttype == funcname) {
1004 ps.in_decl = false;
1005 if (opt.procnames_start_line && code.s != code.e) {
1006 *code.e = '\0';
1007 dump_line();
1008 } else if (ps.want_blank) {
1009 *code.e++ = ' ';
1010 }
1011 ps.want_blank = false;
1012 } else if (!ps.block_init && !ps.dumped_decl_indent &&
1013 ps.paren_level == 0) { /* if we are in a declaration, we must
1014 * indent identifier */
1015 indent_declaration(dec_ind, tabs_to_var);
1016 ps.dumped_decl_indent = true;
1017 ps.want_blank = false;
1018 }
1019 } else if (*inout_sp_sw && ps.p_l_follow == 0) {
1020 *inout_sp_sw = false;
1021 *inout_force_nl = true;
1022 ps.last_u_d = true;
1023 ps.in_stmt = false;
1024 parse(hd_type);
1025 }
1026 }
1027
1028 static void
1029 copy_id(void)
1030 {
1031 size_t len = token.e - token.s;
1032
1033 check_size_code(len + 1);
1034 if (ps.want_blank)
1035 *code.e++ = ' ';
1036 memcpy(code.e, token.s, len);
1037 code.e += len;
1038 }
1039
1040 static void
1041 process_string_prefix(void)
1042 {
1043 size_t len = token.e - token.s;
1044
1045 check_size_code(len + 1);
1046 if (ps.want_blank)
1047 *code.e++ = ' ';
1048 memcpy(code.e, token.s, len);
1049 code.e += len;
1050
1051 ps.want_blank = false;
1052 }
1053
1054 static void
1055 process_period(void)
1056 {
1057 if (code.e[-1] == ',')
1058 *code.e++ = ' ';
1059 *code.e++ = '.'; /* move the period into line */
1060 ps.want_blank = false; /* dont put a blank after a period */
1061 }
1062
1063 static void
1064 process_comma(int dec_ind, bool tabs_to_var, bool *inout_force_nl)
1065 {
1066 ps.want_blank = (code.s != code.e); /* only put blank after comma if comma
1067 * does not start the line */
1068 if (ps.in_decl && ps.procname[0] == '\0' && !ps.block_init &&
1069 !ps.dumped_decl_indent && ps.paren_level == 0) {
1070 /* indent leading commas and not the actual identifiers */
1071 indent_declaration(dec_ind - 1, tabs_to_var);
1072 ps.dumped_decl_indent = true;
1073 }
1074 *code.e++ = ',';
1075 if (ps.p_l_follow == 0) {
1076 if (ps.block_init_level <= 0)
1077 ps.block_init = false;
1078 if (break_comma && (opt.break_after_comma ||
1079 indentation_after_range(
1080 compute_code_indent(), code.s, code.e)
1081 >= opt.max_line_length - opt.tabsize))
1082 *inout_force_nl = true;
1083 }
1084 }
1085
1086 static void
1087 process_preprocessing(void)
1088 {
1089 if (com.s != com.e || lab.s != lab.e || code.s != code.e)
1090 dump_line();
1091 check_size_label(1);
1092 *lab.e++ = '#'; /* move whole line to 'label' buffer */
1093
1094 {
1095 bool in_comment = false;
1096 int com_start = 0;
1097 char quote = '\0';
1098 int com_end = 0;
1099
1100 while (*buf_ptr == ' ' || *buf_ptr == '\t') {
1101 buf_ptr++;
1102 if (buf_ptr >= buf_end)
1103 fill_buffer();
1104 }
1105 while (*buf_ptr != '\n' || (in_comment && !had_eof)) {
1106 check_size_label(2);
1107 *lab.e = *buf_ptr++;
1108 if (buf_ptr >= buf_end)
1109 fill_buffer();
1110 switch (*lab.e++) {
1111 case '\\':
1112 if (!in_comment) {
1113 *lab.e++ = *buf_ptr++;
1114 if (buf_ptr >= buf_end)
1115 fill_buffer();
1116 }
1117 break;
1118 case '/':
1119 if (*buf_ptr == '*' && !in_comment && quote == '\0') {
1120 in_comment = true;
1121 *lab.e++ = *buf_ptr++;
1122 com_start = (int)(lab.e - lab.s) - 2;
1123 }
1124 break;
1125 case '"':
1126 if (quote == '"')
1127 quote = '\0';
1128 else if (quote == '\0')
1129 quote = '"';
1130 break;
1131 case '\'':
1132 if (quote == '\'')
1133 quote = '\0';
1134 else if (quote == '\0')
1135 quote = '\'';
1136 break;
1137 case '*':
1138 if (*buf_ptr == '/' && in_comment) {
1139 in_comment = false;
1140 *lab.e++ = *buf_ptr++;
1141 com_end = (int)(lab.e - lab.s);
1142 }
1143 break;
1144 }
1145 }
1146
1147 while (lab.e > lab.s && (lab.e[-1] == ' ' || lab.e[-1] == '\t'))
1148 lab.e--;
1149 if (lab.e - lab.s == com_end && bp_save == NULL) {
1150 /* comment on preprocessor line */
1151 if (sc_end == NULL) { /* if this is the first comment, we
1152 * must set up the buffer */
1153 save_com = sc_buf;
1154 sc_end = &save_com[0];
1155 } else {
1156 *sc_end++ = '\n'; /* add newline between comments */
1157 *sc_end++ = ' ';
1158 --line_no;
1159 }
1160 if (sc_end - save_com + com_end - com_start > sc_size)
1161 errx(1, "input too long");
1162 memmove(sc_end, lab.s + com_start, (size_t)(com_end - com_start));
1163 sc_end += com_end - com_start;
1164 lab.e = lab.s + com_start;
1165 while (lab.e > lab.s && (lab.e[-1] == ' ' || lab.e[-1] == '\t'))
1166 lab.e--;
1167 bp_save = buf_ptr; /* save current input buffer */
1168 be_save = buf_end;
1169 buf_ptr = save_com; /* fix so that subsequent calls to lexi will
1170 * take tokens out of save_com */
1171 *sc_end++ = ' '; /* add trailing blank, just in case */
1172 buf_end = sc_end;
1173 sc_end = NULL;
1174 debug_println("switched buf_ptr to save_com");
1175 }
1176 check_size_label(1);
1177 *lab.e = '\0'; /* null terminate line */
1178 ps.pcase = false;
1179 }
1180
1181 if (strncmp(lab.s, "#if", 3) == 0) { /* also ifdef, ifndef */
1182 if ((size_t)ifdef_level < nitems(state_stack))
1183 state_stack[ifdef_level++] = ps;
1184 else
1185 diag(1, "#if stack overflow");
1186 } else if (strncmp(lab.s, "#el", 3) == 0) { /* else, elif */
1187 if (ifdef_level <= 0)
1188 diag(1, lab.s[3] == 'i' ? "Unmatched #elif" : "Unmatched #else");
1189 else
1190 ps = state_stack[ifdef_level - 1];
1191 } else if (strncmp(lab.s, "#endif", 6) == 0) {
1192 if (ifdef_level <= 0)
1193 diag(1, "Unmatched #endif");
1194 else
1195 ifdef_level--;
1196 } else {
1197 if (strncmp(lab.s + 1, "pragma", 6) != 0 &&
1198 strncmp(lab.s + 1, "error", 5) != 0 &&
1199 strncmp(lab.s + 1, "line", 4) != 0 &&
1200 strncmp(lab.s + 1, "undef", 5) != 0 &&
1201 strncmp(lab.s + 1, "define", 6) != 0 &&
1202 strncmp(lab.s + 1, "include", 7) != 0) {
1203 diag(1, "Unrecognized cpp directive");
1204 return;
1205 }
1206 }
1207 if (opt.blanklines_around_conditional_compilation) {
1208 postfix_blankline_requested = true;
1209 n_real_blanklines = 0;
1210 } else {
1211 postfix_blankline_requested = false;
1212 prefix_blankline_requested = false;
1213 }
1214
1215 /*
1216 * subsequent processing of the newline character will cause the line to
1217 * be printed
1218 */
1219 }
1220
1221 static void __attribute__((__noreturn__))
1222 main_loop(void)
1223 {
1224 token_type ttype;
1225 bool force_nl; /* when true, code must be broken */
1226 bool last_else = false; /* true iff last keyword was an else */
1227 int dec_ind; /* current indentation for declarations */
1228 int di_stack[20]; /* a stack of structure indentation levels */
1229 bool tabs_to_var; /* true if using tabs to indent to var name */
1230 bool sp_sw; /* when true, we are in the expression of
1231 * if(...), while(...), etc. */
1232 token_type hd_type = end_of_file; /* used to store type of stmt for if
1233 * (...), for (...), etc */
1234 int squest; /* when this is positive, we have seen a '?'
1235 * without the matching ':' in a <c>?<s>:<s>
1236 * construct */
1237 bool scase; /* set to true when we see a 'case', so we
1238 * know what to do with the following colon */
1239
1240 sp_sw = force_nl = false;
1241 dec_ind = 0;
1242 di_stack[ps.decl_nest = 0] = 0;
1243 scase = false;
1244 squest = 0;
1245 tabs_to_var = false;
1246
1247 for (;;) { /* this is the main loop. it will go until we
1248 * reach eof */
1249 bool comment_buffered = false;
1250
1251 ttype = lexi(&ps); /* Read the next token. The actual characters
1252 * read are stored in "token". */
1253
1254 /*
1255 * The following code moves newlines and comments following an if (),
1256 * while (), else, etc. up to the start of the following stmt to a
1257 * buffer. This allows proper handling of both kinds of brace
1258 * placement (-br, -bl) and cuddling "else" (-ce).
1259 */
1260 search_brace(&ttype, &force_nl, &comment_buffered, &last_else);
1261
1262 if (ttype == end_of_file) {
1263 process_end_of_file();
1264 /* NOTREACHED */
1265 }
1266
1267 if (
1268 ttype != comment &&
1269 ttype != newline &&
1270 ttype != preprocessing &&
1271 ttype != form_feed) {
1272 process_comment_in_code(ttype, &force_nl);
1273
1274 } else if (ttype != comment) /* preserve force_nl through a comment */
1275 force_nl = false; /* cancel forced newline after newline, form
1276 * feed, etc */
1277
1278
1279
1280 /*-----------------------------------------------------*\
1281 | do switch on type of token scanned |
1282 \*-----------------------------------------------------*/
1283 check_size_code(3); /* maximum number of increments of code.e
1284 * before the next check_size_code or
1285 * dump_line() is 2. After that there's the
1286 * final increment for the null character. */
1287 switch (ttype) {
1288
1289 case form_feed:
1290 process_form_feed();
1291 break;
1292
1293 case newline:
1294 process_newline();
1295 break;
1296
1297 case lparen: /* got a '(' or '[' */
1298 process_lparen_or_lbracket(dec_ind, tabs_to_var, sp_sw);
1299 break;
1300
1301 case rparen: /* got a ')' or ']' */
1302 process_rparen_or_rbracket(&sp_sw, &force_nl, hd_type);
1303 break;
1304
1305 case unary_op: /* this could be any unary operation */
1306 process_unary_op(dec_ind, tabs_to_var);
1307 break;
1308
1309 case binary_op: /* any binary operation */
1310 process_binary_op();
1311 break;
1312
1313 case postfix_op: /* got a trailing ++ or -- */
1314 process_postfix_op();
1315 break;
1316
1317 case question: /* got a ? */
1318 process_question(&squest);
1319 break;
1320
1321 case case_label: /* got word 'case' or 'default' */
1322 scase = true; /* so we can process the later colon properly */
1323 goto copy_id;
1324
1325 case colon: /* got a ':' */
1326 process_colon(&squest, &force_nl, &scase);
1327 break;
1328
1329 case semicolon: /* got a ';' */
1330 process_semicolon(&scase, &squest, dec_ind, tabs_to_var, &sp_sw,
1331 hd_type, &force_nl);
1332 break;
1333
1334 case lbrace: /* got a '{' */
1335 process_lbrace(&force_nl, &sp_sw, hd_type, di_stack,
1336 (int)nitems(di_stack), &dec_ind);
1337 break;
1338
1339 case rbrace: /* got a '}' */
1340 process_rbrace(&sp_sw, &dec_ind, di_stack);
1341 break;
1342
1343 case switch_expr: /* got keyword "switch" */
1344 sp_sw = true;
1345 hd_type = switch_expr; /* keep this for when we have seen the
1346 * expression */
1347 goto copy_id; /* go move the token into buffer */
1348
1349 case keyword_for_if_while:
1350 sp_sw = true; /* the interesting stuff is done after the
1351 * expression is scanned */
1352 hd_type = (*token.s == 'i' ? if_expr :
1353 (*token.s == 'w' ? while_expr : for_exprs));
1354
1355 /* remember the type of header for later use by parser */
1356 goto copy_id; /* copy the token into line */
1357
1358 case keyword_do_else:
1359 process_keyword_do_else(&force_nl, &last_else);
1360 goto copy_id; /* move the token into line */
1361
1362 case type_def:
1363 case storage_class:
1364 prefix_blankline_requested = false;
1365 goto copy_id;
1366
1367 case keyword_struct_union_enum:
1368 if (ps.p_l_follow > 0)
1369 goto copy_id;
1370 /* FALLTHROUGH */
1371 case decl: /* we have a declaration type (int, etc.) */
1372 process_decl(&dec_ind, &tabs_to_var);
1373 goto copy_id;
1374
1375 case funcname:
1376 case ident: /* got an identifier or constant */
1377 process_ident(ttype, dec_ind, tabs_to_var, &sp_sw, &force_nl,
1378 hd_type);
1379 copy_id:
1380 copy_id();
1381 if (ttype != funcname)
1382 ps.want_blank = true;
1383 break;
1384
1385 case string_prefix:
1386 process_string_prefix();
1387 break;
1388
1389 case period:
1390 process_period();
1391 break;
1392
1393 case comma:
1394 process_comma(dec_ind, tabs_to_var, &force_nl);
1395 break;
1396
1397 case preprocessing: /* '#' */
1398 process_preprocessing();
1399 break;
1400 case comment: /* the initial '/' '*' or '//' of a comment */
1401 process_comment();
1402 break;
1403
1404 default:
1405 break;
1406 }
1407
1408 *code.e = '\0';
1409 if (ttype != comment &&
1410 ttype != newline &&
1411 ttype != preprocessing)
1412 ps.last_token = ttype;
1413 }
1414 }
1415
1416 int
1417 main(int argc, char **argv)
1418 {
1419 main_init_globals();
1420 main_parse_command_line(argc, argv);
1421 #if HAVE_CAPSICUM
1422 init_capsicum();
1423 #endif
1424 main_prepare_parsing();
1425 main_loop();
1426 }
1427
1428 /*
1429 * copy input file to backup file if in_name is /blah/blah/blah/file, then
1430 * backup file will be ".Bfile" then make the backup file the input and
1431 * original input file the output
1432 */
1433 static void
1434 bakcopy(void)
1435 {
1436 ssize_t n;
1437 int bakchn;
1438 char buff[8 * 1024];
1439 const char *p;
1440
1441 /* construct file name .Bfile */
1442 for (p = in_name; *p != '\0'; p++); /* skip to end of string */
1443 while (p > in_name && *p != '/') /* find last '/' */
1444 p--;
1445 if (*p == '/')
1446 p++;
1447 sprintf(bakfile, "%s%s", p, backup_suffix);
1448
1449 /* copy in_name to backup file */
1450 bakchn = creat(bakfile, 0600);
1451 if (bakchn < 0)
1452 err(1, "%s", bakfile);
1453 while ((n = read(fileno(input), buff, sizeof(buff))) > 0)
1454 if (write(bakchn, buff, (size_t)n) != n)
1455 err(1, "%s", bakfile);
1456 if (n < 0)
1457 err(1, "%s", in_name);
1458 close(bakchn);
1459 fclose(input);
1460
1461 /* re-open backup file as the input file */
1462 input = fopen(bakfile, "r");
1463 if (input == NULL)
1464 err(1, "%s", bakfile);
1465 /* now the original input file will be the output */
1466 output = fopen(in_name, "w");
1467 if (output == NULL) {
1468 unlink(bakfile);
1469 err(1, "%s", in_name);
1470 }
1471 }
1472
1473 static void
1474 indent_declaration(int cur_dec_ind, bool tabs_to_var)
1475 {
1476 int pos = (int)(code.e - code.s);
1477 char *startpos = code.e;
1478
1479 /*
1480 * get the tab math right for indentations that are not multiples of
1481 * tabsize
1482 */
1483 if ((ps.ind_level * opt.indent_size) % opt.tabsize != 0) {
1484 pos += (ps.ind_level * opt.indent_size) % opt.tabsize;
1485 cur_dec_ind += (ps.ind_level * opt.indent_size) % opt.tabsize;
1486 }
1487 if (tabs_to_var) {
1488 int tpos;
1489
1490 check_size_code((size_t)(cur_dec_ind / opt.tabsize));
1491 while ((tpos = opt.tabsize * (1 + pos / opt.tabsize)) <= cur_dec_ind) {
1492 *code.e++ = '\t';
1493 pos = tpos;
1494 }
1495 }
1496 check_size_code((size_t)(cur_dec_ind - pos + 1));
1497 while (pos < cur_dec_ind) {
1498 *code.e++ = ' ';
1499 pos++;
1500 }
1501 if (code.e == startpos && ps.want_blank) {
1502 *code.e++ = ' ';
1503 ps.want_blank = false;
1504 }
1505 }
1506
1507 #ifdef debug
1508 void
1509 debug_printf(const char *fmt, ...)
1510 {
1511 FILE *f = output == stdout ? stderr : stdout;
1512 va_list ap;
1513
1514 va_start(ap, fmt);
1515 vfprintf(f, fmt, ap);
1516 va_end(ap);
1517 }
1518
1519 void
1520 debug_println(const char *fmt, ...)
1521 {
1522 FILE *f = output == stdout ? stderr : stdout;
1523 va_list ap;
1524
1525 va_start(ap, fmt);
1526 vfprintf(f, fmt, ap);
1527 va_end(ap);
1528 fprintf(f, "\n");
1529 }
1530
1531 void
1532 debug_vis_range(const char *prefix, const char *s, const char *e,
1533 const char *suffix)
1534 {
1535 debug_printf("%s", prefix);
1536 for (const char *p = s; p < e; p++) {
1537 if (isprint((unsigned char)*p) && *p != '\\' && *p != '"')
1538 debug_printf("%c", *p);
1539 else if (*p == '\n')
1540 debug_printf("\\n");
1541 else if (*p == '\t')
1542 debug_printf("\\t");
1543 else
1544 debug_printf("\\x%02x", *p);
1545 }
1546 debug_printf("%s", suffix);
1547 }
1548 #endif
1549
1550 static void *
1551 nonnull(void *p)
1552 {
1553 if (p == NULL)
1554 err(EXIT_FAILURE, NULL);
1555 return p;
1556 }
1557
1558 void *
1559 xmalloc(size_t size)
1560 {
1561 return nonnull(malloc(size));
1562 }
1563
1564 void *
1565 xrealloc(void *p, size_t new_size)
1566 {
1567 return nonnull(realloc(p, new_size));
1568 }
1569
1570 char *
1571 xstrdup(const char *s)
1572 {
1573 return nonnull(strdup(s));
1574 }
1575